Benchmark DAC3 HGC D/A preamplifier-headphone amplifier Associated Equipment

Sidebar 2: Associated Equipment

Analog Sources: Thorens TD 124 turntable (reconditioned by Schopper AG); Ortofon RMG-212 (rebuilt), Thomas Schick 12" tonearms; Ortofon 90th Anniversary SPU cartridge.
Digital Sources: MacBook Pro computer (early 2015); Roon Optimized Core Kit running on Intel NUC computer with SSD drive; Benchmark DAC1, Meridian Explorer2 USB, Mytek Brooklyn, PS Audio DirectStream DACs.
Preamplification: Auditorium 23 Standard step-up transformer, EAR 834P phono stage.
Power Amplifiers: PS Audio BHK Signature 300 Mono monoblocks.
Integrated Amplifier: Leben Hi-Fi Stereo CS-600.
Loudspeakers: Alta Audio Titanium Hestia (review forthcoming), DeVore Fidelity The Nine.
Cables: USB: AudioQuest Carbon & Cinnamon & Coffee, Comprehensive Connectivity DXLRP-DXLRJ-6EXF. Interconnect: Auditorium 23, Chord (UK) Chorus & Chameleon Silver Plus, Mogami Gold, Sescom/Canare (XLR). Speaker: Auditorium 23. AC: Belden 14-gauge shielded, HOSA 14-gauge IEC, stock IEC.
Accessories: PS Audio Power Plant P10 power conditioner, Oswalds Mill slate plinth (under turntable), Chilton's Durham media console in cherrywood, Auralex ProPanel Fabric-Wrapped Acoustical Absorption Panel (4' by 2' by 2", beveled).—Jim Austin

COMPANY INFO
Benchmark Media Systems, Inc.
203 E. Hampton Place, Suite 2
Syracuse, NY 13206
(800) 262-4675
ARTICLE CONTENTS

COMMENTS
tonykaz's picture

NwAvGuy was describing the Benchmark and using it to evaluate his own DAC designs.

Mark Waldrip relies on Benchmark gear

I've heard of a good many Pro Audio folks tout Benchmark as being the Benchmark.

I've been trying DACs for some years now, I can't discover any advantage behind some of the super pricy DACs.

I've blamed my hearing, to the point of having my hearing evaluated by Audiologists at the University of Michigan. My hearing tapers off above 8k but I can still hear significant differences in 12AU7 preamp tubes. I still can't hear greatness in super expensive DACs.

Those that can hear significant improvements in Super Expensive DACs are living with the Audiophile Curse. ( the King's New Suit Curse )

As an Engineer, I've demonstrated Zip Lamp Cord vs. Monster Speaker Cable vs. Bruce Brisson's MH-750 to astonished Engineers.

Back in 2015 Tyll had a headphone gear Shootout where Stax, Sennheiser, and Audeze were the finalists. He also had the finest DACs including the highly touted Antelope. Nobody could detect any DAC performance advantage.

Then there's Chord and their Field Programable Gate Array devices which I suspect they've designed to sound better but probably not sound accurate ( like Tubes that make music nicer sounding ).

So, go for Accuracy with Benchmark or better sounding with Chord.

Nice reporting, sir, I think.

Tony in Michigan

barrows's picture

I would suggest that there is not much evidence to support your assertion that Chord DACs are not designed to sound accurate, take a look at their measurements here at stereophile.com...

On this review in general, despite the lack of difference heard vs the DS, it is interesting to note that Jim found an immediately apparent difference to the Benchmark DAC1, subjectively speaking...

Staxguy's picture

Accurate? Chord DAC's can't even handle (do) 24-bit. Even their best DAC, the Chord DAVE can't even do 24-bit. 21-bit, that's it!

Their 17-order noise shaper can do 350 dB DNR, but that's internal aah only.

It's taken forever for this review to come out. Back when we (I) was recommending the Benchmark DAC1 HGC in my magazine as a recommended component, The Absolute Sound was only recommending DACs with sub-par 16-bit performance and not even mentioning it, even though it had came out.

Well, it's great to see this review. I might buy the Rogue Audio RH-5 as a headphone amplifier, reviewed just before this one, which likely functions worse (no better) than the Aurorasound HEDA, because it looks good on the desk!

People like benchmark, and good for/on them. It's got a great summing DAC (multiple DAC) design.

I'd rather have the GTE Trinity DAC (the one with the actual triangles not the newer ahh one), but that's for myspace!

barrows's picture

Your reply is kind of in error. Here is a quote from JA's measurements on this Benchmark:

"...the increase in bit depth dropped the noise floor by more than 30dB (fig.5), indicating that the Benchmark's resolution is at least 21 bits. This is as good as a DAC can currently get!"

There is no DAC of which I am currently aware that achieves better than 21 bit resolution at its analog outputs. The resolution measured here for the Benchmark is at the same level as Chord DAVE (look it up on this site). When DAC makers say they are 24 bits, or 32 bits, that is what level the DAC operates at in the digital domain, but no DAC achieves that resolution at its analog outputs.

I am curious, do you have any references for measurements of the Trinity? I would expect them to be way worse than that of the Benchmark, not saying anything against the Trinity as far as sound quality is concerned though, just, it is likely not nearly as "accurate" being an R2R DAC...

Sal1950's picture

"There's a danger of being misled, of repeating the same mistakes again and again, of spending way too much money on things of little value. My point is that, as a hobby, industry, and avocation, we may have shifted too far toward the subjectivist side."

What a breath of fresh air in these pages! When I first read this review in my subscriber copy my jaw hit the floor. Thank you Jim, we can only hope as audiophiles dedicated to the accurate reporting, that blind listening sessions become much more widespread and prevalent in the pages of Stereophile. Sighted listening is much too fallible to be the sole basis for accurate evaluations.

mrkaic's picture

I hope this year, this month, and this review mark the beginning of the end of the tyranny of subjectivists.

ChrisS's picture

...listening/reviewing of this component was done blinded.

He did it the way we all do.

He listens.

He thinks about it.

He writes about what he hears.

That's what they all do at Stereophile.

ChrisS's picture

"Yes. Listening is what Stereophile reviewers do."

mrkaic's picture

There is only one question in audio and it is rhetorical -- why buy anything but Benchmark?

They are the best, nothing comes close.

Charles Hansen's picture

You claim to have a PhD and you make the most egregious beginner's mistakes imaginable. It is laughable that even Streophile would publish this garbage.

First you start of the article with a story about "hum and buzz" from "poor quality control" in the interconnects. If you knew anything at all about balanced circuitry and how it works, you would realize that a truly balanced source sending a signal to a truly balanced downstream device wouldn't make a bit of "hum and buzz" if the cables were shielded or not.

But since you clearly have no idea about anything, you start with the wrong assumption and leap to an incorrect conclusion. [Flame deleted by John Atkinson]

Then you try to "compare" the Benchmark against your PS Audio. And although you could "easily" hear differences between the Benchmarks 1 and 3, you could hear no difference whatsoever between the Benchmark 3 and the PS Audio. Why don't you look at your ridiculous methodology before jumping to more false conclusions?

First you said you connected "sent the output of the two DACs to different channels of my PS Audio BHK Signature preamp". WTF? So you sent the left channel of one DAC to one input and the right channel of the other DAC to the other input? Or do you not know the difference between a "channel" and an "input"? At this point you are looking beyond unprofessional, beyond amateurish, and all the way to "knowing just enough to be dangerous". Or does your preamp have only one input? Or does it have more than one but the inputs were full, and you were too lazy to disconnect them?

Anybody serious would know that the only FAIR way to do this type of test is to not only use the same input on the preamp (different inputs will exhibit different amounts of break-in, as the trace on the PCB needs a signal to dissipate the residual electrostatic charge after it is tested with many kilovolts for UL compliance), but to also use the exact same pair of interconnects as they need to be not only the same brand and model, but also have the same amount of break-in time.

And finally you lose all rationality. (We may have to send the men in white coats to save you from harming yourself.) You send the analog output of the PS Audio DAC into the analog input of the Benchmark DAC. At the very least this signal is now going through the analog circuitry and volume control of the Benchmark, which is clearly coloring the sound if it sounds "identical" to you - either that or your hearing ability sucks or your playback system sucks. And if (like the version 2) the version 3 also has a "hybrid digital/analog" volume control then the volume control wouldn't even be active unless the analog input were being digitized by whatever cheap A/D chip Benchmark is using with their cheap IC op-amp based analog circuitry. More cheap crap to color the sound of the PS Audio, and you can't figure ANY of this out...

Finally it is clear that you cannot even add 2 + 2 to equal the correct answer. First you claim that the Benchmark "moves interpolation off-chip" to eliminate problems with "intersample overs", but that the latest version sounds better because of "lower passband ripple, facilitated by the new chip's superior filter choices". So which is it, Mr. PhD? Does it sound better because the processing is done off-chip or does it sound better because the on-chip processing has lower passband ripple? You can't have it both ways.

[Flame deleted by John Atkinson]

You are the one that makes a joke out of the high end, not the people who make claims you can't understand - you can't even understand the simplest of claims, as this so-called "review" clearly demonstrates.

AJ's picture
Quote:

Anybody serious would know that the only FAIR way to do this type of test is to not only use the same input on the preamp (different inputs will exhibit different amounts of break-in, as the trace on the PCB needs a signal to dissipate the residual electrostatic charge after it is tested with many kilovolts for UL compliance), but to also use the exact same pair of interconnects as they need to be not only the same brand and model, but also have the same amount of break-in time.
And finally you lose all rationality.

Can't make this stuff up ;-).
Charles, your rational scientific evidence of this audible "break-in" outside the imaginations inside your head?
TIA

mrkaic's picture

It boggles the mind that someone who believes in cable break-in has the chutzpah to write about electronics. It really does.

But we can learn here, for sure. In the words of the previous occupant of the White House— it is a teachable moment—sans beer, in this particular case.

It is tempting to get angry at the omni-present neglect of science education and fume about the ignorance of individuals who believe in cable break-in, directional fuses etc. But it is more fun is to realize that those esteemed gentlemen, who believe, for example, in cable break-in, provide some free comic relief in these trying times.

So, don’t get mad, get entertained.

Anton's picture

I'd hate to see how worked up you get over real world issues.

barrows's picture

Jim. I too was disturbed by some of your methodology in your comparisons, and dismissed your results because of my concerns. While I may not go as far as Charlie, please note that I felt disturbed enough to dismiss the results.
I would suggest that if you want to make quick A/B comparisons of source components in future reviews, you should have a balanced switch box made, who's only function is to take two balanced inputs, and switch them to a single balanced output. I would also suggest that this be wired the AES way for consistency (XLR pin 1 to ground) and that the signal path internally be identical for both inputs, same length wiring, etc. A high quality switch box like this should be a reasonable expense for any any reviewer who is interested in making quick A/B comparisons of source components.
Also of import would be keeping everything else the same, same cabling, same cabling lengths, etc.
In the case of comparing these two DACs (DS and DAC 3) such a methodology could then go direct into the amps, removing any additional components (preamp) and maximizing the apparent sonic differences. You should also volume match the DACs by measuring the output voltage with a test signal (like that available on stereophile test CDs) without very accurate level matching any such comparisons are mostly meaningless.

supamark's picture

won't work as you think. You're going to add two add'l sets of connections (in/out of the box) which will color the sound. Not every component has balanced outs, and a "dirty" little industry secret is that a majority of XLR outs are derived from the unbalanced RCA outs since few components are actually fully balanced internally. Okay, not really a secret since the few mfg's who make components that are fully balanced from input to output will make sure you know about it.

You will also still need a volume control since not everything being reviewed will have a variable output (and reducing volume in the digital domain adds its own.... issues). Jim's method of using two line inputs of his high quality preamp (that I'mma assume Jim knows the sound of quite well) is a much better methodology. The only way I could think of to improve his methodology would be for him to have someone else connect the two devices and make sure he doesn't know which input is which device.

Let me tell ya another little secret that every recording engineer knows about critical listening (this has happened to EVERY recording engineer at least once): You can imagine differences in sound that are not actually there. Every engineer has at some point reached for the EQ, started adjusting it and heard the sound change, and I mean literally hearing the changes being made to the sound.... but there's a problem - you never actually engaged the EQ (there's an on/off relay switch on every mixing console to switch the EQ into the signal path). It was all in your mind. This, to me, makes two very different points:

1. blind A/B/X testing really is the only fully valid method, but unfortunately nobody's come up with a truly transparent comparator (and likely never will).

2. those expensive cables from like Nordost et.al. are mostly snake oil. A much better system would be 100% copper all the way - internal component wiring (already happens), line level plugs/jacks, speaker connections all 100% copper (or silver/gold/whatever but must all be the same metal and the same purity/alloy). When the electrical signal is traveling along and the medium (metal) changes the signal will change. Were I a high end mfg, I would make all my connectors copper and provide high quality all copper cables at no add'l charge (because, seriously, the mark-up on cables is crazy - 30 years ago I picked up 40' of Kimber Cable at dealer cost, $1 per foot).

barrows's picture

won't work as you think. You're going to add two add'l sets of connections (in/out of the box) which will color the sound. Not every component has balanced outs, and a "dirty" little industry secret is that a majority of XLR outs are derived from the unbalanced RCA outs since few components are actually fully balanced internally. Okay, not really a secret since the few mfg's who make components that are fully balanced from input to output will make sure you know about it.
Read more at https://www.stereophile.com/content/benchmark-dac3-hgc-da-preamplifier-headphone-amplifier#JLg070doRRMojDZv.99"

Only one xtra pair of interconnects, and this is not a problem, as long as they are the same, the difference will still be the same. Same thing with the switch box itself, the entire point is to give the same signal path for each DUT, the result being that one will only hear the difference. This set up, given decent construction and good interconnects will be transparent enough to hear any actual difference.
Additionally, while your contention that components are generally pseudo balanced may have been the case long ago, it is not common now-for example, the DACs mentioned in Jim's review are true balanced, as are his pre amp and amp.

And if you think Nordost cabling is snake oil, you are either deaf, or have not listened to it.

supamark's picture

It's not the 1 extra pair of interconnects (the extra wire is essentially meaningless*), it's the extra set of connections - jacks and plugs (along with the circuitry inside the box, including gain matching circuits, even if they're just variable resistors). You going with a relay or mechanical switch? Both degrade the sound in different ways. Adding that crap in will make it harder to hear any differences because you just added noise and made the signal path more complex. That is much worse than using two inputs of the same high quality preamp.

Most home audio gear is not fully balanced inside (nor should it be). Its adoption rate is higher in high-end home audio, but honestly it's not really necessary in most home environments. Pro audio, where there's much longer cable runs and a lot more electrical noise, is where it's really useful.

*when I was recording Austin Symphony Orchestra in Bass Concert Hall at The University of Texas in the early 90's, the mic cable run - from the stereo pair of Neumann U87 mic's through underground conduits below a couple other buildings into our machine room patch bay and into our Harrison Series 10 console was at least a quarter mile, probably close to 2,000 feet. The sound was not dull or even particularly weak. The U87's large presence boost probably helped but as I said, a couple feet of line level cable ain't gonna mean squat.

btw, if you think Nordost or anyone's cable can magically transform the sound that is otherwise running through traces on a f'ing circuit board inside each piece of your analog gear as it moves from cap to resistor to etc, you have a few things to learn about science (esp. physics). I mean, holy sh!t, they'll sell you little teflon and wood teepees for your speaker cables (for a pretty penny, 'natch). It's really no different than the Tice Clock or coloring the edges of your CD's with a green marker. Maybe Nordost has proprietary connectors that do a truly excellent job of passing signal with minimal alteration - great, that's a real upgrade and can be audible... but that wire between those connectors for which they're charging you several grand more than JA's preamp? that stuff is snake oil.

Buy a high quality power conditioner instead of fancy cables, it'll make a real audible difference and cost a lot less money.

barrows's picture

Are you nuts? I am talking about a passive switch box here, no active circuitry. Just two sets of XLR input jacks, some wire, a high quality switch (lets say a shallco) and two output jacks. That is all. No resistors, no circuitry of any any kind. Will this change the sound, perhaps a tiny, tiny bit, but not enough to obscure differences between the two DUTs, as the change will be identical to both products. A passive box like this will have much less influence on sound than a fully active preamp (which we eliminate from the chain with this testing method) which includes the wiring and switch, but also adds a power supply (noise source), transistors, resistors, etc. We do not need any circuitry, both DACs are designed to run direct to power amps, and we match levels via the DAC VC.

On cables, I never suggested that a cable can "magically transform sound"! Please do not put words in my mouth, neither did I make any comment re audio cable pricing. I just said that if you think Nordost cabling is snake oil, you are either deaf or have not listened to it, and I stand by that. To be more detailed: compare directly a Nordost cable to Mogami, and you will hear a significant difference. But the cable thing is OT here, so lets leave it for another place.

supamark's picture

Those added connections will in fact degrade the sound. You are adding 2 additional connections to each channel (XLR's, which really aren't as firm as RCA connectors - they wiggle a bit), each will degrade the sound. Also, since not every DAC has balanced outs, you'll want another box with RCA's (or just go unbalanced in the first place - the main advantage with balanced is if you have RF issues in your home or really long cable runs).

If you believe so strongly that wire has a sound, just use the same brand/length/termination/model for both and they'll be the same... it's cable, not 9' concert grand pianos (which actually do all sound different because they're handmade and no two are identical but brands do have "house" sounds - Hamburg Steinways are my favorite btw, and Baldwins are popular in rock because they're a little more mid-heavy with less sparkle than NY or Hamburg Steinways and therefore easier to cut through a dense mix).

Not every DAC has a volume control, and even for those that do you'll need a master volume control if you want adjust volume while keeping the levels matched - unless you think no volume adjustments should be made? Besides, the fixed output *should* always be superior (digital volume controls necessarily change the bits, and not for the better).

Also, what if the output impedences of the DACs are not close? going direct into a tube amp could cause differences with freq. response.

So, yeah, going into a high quality preamp is both simpler and more likely to yield consistent results with a much wider variety of DACs.

Fleschler's picture

The Mogami Gold Balanced cables in an EAR 864 pre, 890 amp and 324 phono as well as a COS Engineering D1v DAC in various combination yielded a very clean, articulate sound, lacking dimension, body, bass and dynamics. It was beneath the equipment quality and totally unacceptable for home use. However, several other high end brands were tried which yielded extremely superior results, with various colorations for some of the lesser name brand cables (try GroverHuffman for more neutral sound). I've heard Nordost higher end balanced cables and Kubala Sosna balanced cables in very good systems and they did not detract from the sound even if they were imperfect transmitters (these were different systems-I could not tell what, if anything, they were doing to reduce the sound quality). I do know that the Mogami's suck on moderately high end gear. Maybe they are just what a recording studio needs for a clean, clear sound which the mastering engineer can take and adjust afterwards to his preference.

AJ's picture
Quote:

You going with a relay or mechanical switch? Both degrade the sound in different ways. Adding that crap in will make it harder to hear any differences because you just added noise and made the signal path more complex. That is much worse than using two inputs of the same high quality preamp.

You just need to wave your hands around as you say this stuff, then finish with "believe me" ;-)

supamark's picture

I'm livin' in your tiny little head, you can't quit me. it's really sad, really really sad.

AJ's picture
Quote:

but unfortunately nobody's come up with a truly transparent comparator (and likely never will).

Your factual evidence please.

supamark's picture

that's just stupid. Why don't you go out and find/build one and prove me wrong?

AJ's picture
Quote:

Why don't you go out and find/build one and prove me wrong?

Because folks with a modicum of intelligence know where Burden of Proof lies ;-).
Ok, so you admit to having zero evidence for your specious claim.
The question was rhetorical.

supamark's picture

when someone is stupid enough to ask someone to prove a negative, it simply proves that they're just stupid.

Again, find me a transparent A/B/X comparator system... maybe get Nelson Pass (or his pre-amp guy), Dan D'Augustino, and the guys at Schiit on it, they're all excellent engineers.

AJ's picture
Quote:

when someone is stupid enough to ask someone to prove a negative

Except I didn't. Your specious BS claim:

Quote:

nobody's come up with a truly transparent comparator (and likely never will)

I asked for evidence to support your BS claim of (all known) comparators (never mind I can only think of 2, the current AVA and a discontinued QSC. The really silly "all future ones" is too funny) are non-transparent.
You admit you have none. IOW, your claim is total unsupported evidence free BS. "Stupid"? Well, you decide ;-)

claud's picture

JA, what I, and I would think Barrows and others, don't understand is why hasn't Stereophile designed and issued such "comparator test boxes" for their editors to utilize when reviewing hardware long ago. Barrows's description of such a test device seems very straightforward and should yield accurate comparisons, certainly far better than Jim's method for comparing the sound of the Benchmark with the PS Audio DACs. Again, I see little excuse for why the use of such an official but simple and carefully designed and passive Stereophile test box was not instituted long ago-at least when testing DACs-especially since most DACs equal to and certainly above the price of any Benchmark DAC have true balanced differential outputs.

AJ's picture
Quote:

Again, find me a transparent A/B/X comparator system... maybe get Nelson Pass (or his pre-amp guy), Dan D'Augustino, and the guys at Schiit on it, they're all excellent engineers.

Though you could never comprehend it, that sir, is your burden, not mine ;-)

Corvaldt's picture

Actually the burden of proof is on you here. Because it is possible to create a non-transparent comparator in an actually infinite number of ways, your claim that creating a transparent comparator IS possible is the less likely to be true.

supamark's picture

but where is it? Also, as I said before, you can't prove a negative. I made the claim (and I'm far from the only one), you can choose to accept/believe my claim, you can try (possibly even succeed) in proving my claim wrong, or you can just ignore it because life is short and this shit ain't important.

AJ's picture
Quote:

I made the claim

Bingo!
..and then the hands waved frantically...because the evidence was zero ;-)

supamark's picture

I have my very own stalker, how pathetic of you.

rschryer's picture

But, AJ, here is where your argument doesn't hold water: Supamark is saying that a truly transparent comparator DOESN'T exist. With that in mind, if I were to tell you that your inability to prove that wood fairies DON'T exist means that they do, would you agree with me? If not, then the burden of proving that a truly transparent comparator does exist rests on your shoulders, because you are insinuating, by way of your challenge to supamark, that one does. Do you know of a truly transparent comparator? If not, then supamark wins by default.

AJ's picture
Quote:

Supamark is saying that a truly transparent comparator DOESN'T exist.

Exactly!
Now both of you are burdened with evidence for your fabricated imaginary assertions.
Let's see it....or more frantic hand waving.

rschryer's picture

I may be burdened with fabricated imaginary assertions, but yours isn't one of them.

AJ's picture

Of course you would have no comprehension of Argument from Ignorance,
but your false assumption that his non-transparent comparator claim is true, thus the burden shifts to me to provide evidence to disprove it...is exactly that.
No worries, we're in the age of hand waving "believe me", it's expected ;-)

SeanS's picture

Analog inputs stay analog.

From the manufacturers manual:
HGC™ is Benchmark's unique Hybrid Gain
Control™ system. The DAC3 combines active
analog gain control, passive low-impedance
attenuators, a 32-bit digital gain control, and
a servo-driven volume control.
All inputs are controlled by the rotary volume
control. This volume control moves in
response to commands from the remote
control. Analog inputs are never converted to
digital, and digital inputs never pass through
an analog potentiometer. Digital inputs are
precisely controlled in the 32-bit DSP system.
The DSP system preserves precise L/R
balance, and precise stereo imaging, while
avoiding any source of noise and distortion.

Sigh's picture

I too was shocked by this way of comparing the PS-Audio to the Benchmark. Wether you hear a difference or not doesn't even matter because you (and we) don't even know what it is that makes the sound of one chain or the other, (The DAC sections? The Cable? The differences between the analog or digital volume control?). Read what Mytek has to say about the effect of digital and analog volume controls in their Brooklyn. I was eagerly waiting for this review since I am a big fan of all Benchmark DACs.
The lack of rigor and the seven pages of blabla are almost offensive to the companies that put years of hard work into their products and put them in your hands. Some of the comments could have been phrased more diplomatically but the nonsense one reads in Stereophile, 6moons, DAR... the contradictions, the approximations, the lack of methodology are infuriating because your opinions greatly influence the success of these products. The spectacular measurements are useful. But what exactly is there to learn from what you wrote?

claud's picture

I can't be sure how sensible are my comments and questions but I'd appreciate any advice on my situation.

In this review In this review of the stripped down version of the Benchmark DAC3-HGC, https://www.stereophile.com/content/benchmark-dac3-hgc-da-preamplifier-headphone-amplifier-benchmark-dac3-b , I was expecting JA to be as impressed as Jim Austin was with these two otherwise identical DACs. Admittedly, given the price differences between the JCA's PS Audio and JA's MLB DAC, JA almost certainly would come away even more impressed with how the DAC3B competed with both. That's very good news.

But what did JA mean by the DAC3B lacking the "ease' of the MBL DAC? I can only imagine that it's analogous to what Benchmark calls a "euphoric" effect laid over the sound, and goes on to explain why listeners would prefer this (with flawed music recordings and movie soundtracks) and describes the ways designers can add this sonic quality to their DACs and amps. https://benchmarkmedia.com/blogs/application_notes/124929287-how-should-i-buy-an-audio-system

But even with the ten-fold price difference, would JA discern this "ease" from the MSB DAC to be the same thing as euphoria? What could be lacking in the DAC3B that makes it incapable of generating this "subjective"-and thus non-measurable-quality, but which JA and presumably many listeners would find enjoyable?

While apparently not so in the MLB DAC, http://www.mbl-northamerica.com/wp-content/uploads/2017/06/Feature-List-N31-EN.pdf ,could this "ease" be due to the Class A output stage which the MLB and even far lower priced DACs have? https://bryston.com/digital-audio/bda3/
Could that have also been why Larry Greenhill raved about the size and proportions of its sound stage? https://www.stereophile.com/content/benchmark-dac3-hgc-da-preamplifier-headphone-amplifier-benchmark-dac3-b And as with the output stage in the previous BDA-2?
https://www.theabsolutesound.com/articles/bryston-bda-2-digital-to-analog-converter/

Then JA said that the sound stage shrunk and the balance got "lighter" (meaning what?) when he swapped in the Benchmark speaker cables. Okay, with 12x price difference that's understandable.

BUT why did he ALSO have to swap out the Parasound mono blocks for the Benchmark amps at the same time!? At least if he had kept the mono blocks we all may have learned how much the sound quality was affected by the cable swap alone.

Instead, JA's said that "I replaced the Parasound monoblocks with a Benchmark AHB2 stereo amplifier. (To cope with the DAC3's high maximum output level, I switched the AHB2's input sensitivity to the lowest setting.)" But why should that have been at all necessary when those Parasound monos seem to have ample adjustable control of their input sensitivity on their rear panels? https://parasound.com/jc1.php

Therefore, what's to make of his actions?

As for the DAC3B's functionality, due to its total lack of external level controls, which JA had to contend with during measurement sessions
https://www.stereophile.com/content/benchmark-dac3-hgc-da-preamplifier-headphone-amplifier-measurements , unless one's using a passive) preamp with an analog attenuator how else to control?

There's this remote sold separately.
https://benchmarkmedia.com/products/benchmark-remote-control-metal?_pos=1&_sid=e978775a0&_ss=r#shopify-product-reviews But from p. 10 in the manual it's not even clear if the remote can control the volume. https://cdn.shopify.com/s/files/1/0321/7609/files/DAC3_B_Manual_Rev_B.pdf?v=1635945656

And even if it can provide true analog actuated attenuation for zero resolution loss?

For my hopelessly overdue 5.3 system, my plan was to avoid MCH processors under ~ $7K, which may use lesser DAC chips plus connectivity and other things I don't need, and find a 8-channel DAC-but one with ample output voltage to offset the nearly 20db gain losses KR said he encountered with DIRAC Live 3.
https://www.stereophile.com/content/topping-pre90-line-preamplifier

KR apparently found a satisfactory solution using this DAC and three Pre90 preamps. https://www.stereophile.com/content/exasound-s88-multichannel-da-processor

Though using three Bryston BDA-3 or even the BDA-2 DACs would have likely delivered the most satisfying sound, their balanced outputs max out at 4 volts. But I thought that I might still get better sound than the Exasound DAC (the latter with the same chips https://benchmarkmedia.com/blogs/application_notes/dac3-introducing-the-new-es9028pro-converter ) with using three DAC3B which could easily offset DIRAC gain losses with need of an active preamp. But it looks like that can't happen either unless that optional remote can sufficiently attenuate the DAC's output at least down to very low listening levels with no resolution loss. How likely is that??

And unless I got four DAC3B I'd also have to Y-connect my three subs to one output among the three DACs, though given the DAC3B's high output the combined load impedance might not impact sound quality.

But probably the biggest question is how noise and clock error-free will a 5.1 PCM signal decoded by JRiver from the DTS MA of a BD disc remain it's fed via three different USB ports to three DAC3B DACs? How would it sound compared to the Exasound s88 , or even this 8-channel DAC? https://www.oktoresearch.com/dac8pro.htm
https://www.stereophile.com/content/okto-research-dac8-pro-da-processor-measurements

Any advice much appreciated.

pma's picture

Nice review Jim and also thanks to John Atkinson for a valuable set of measurements, as always. We can see that the review has initiated quite strong reactions of a prominent high-end designer as it obviously targets very good objective parameters that tend to be overlooked and replaced by rather mystic beliefs.

I would like to encourage Stereophile team in doing more controlled tests. Though there may be objections to AB switch box transparencies, the same applies to 'high-end' components and usually at much higher degree. Please continue the good job and do not let you make disappointed from the offensive comments.

Last but not least, the switch box transparency might be well evaluated by SYS-2722 redaction system.

arve's picture

@John Atkinson: Have you considered adding a test to see how gracefully DACs handle intersample overs? While it's possible to create waveforms with a True Peak value with an arbitrary value above 0 dBFS by approaching Nyquist, here is one that generates a peak that's +3dBFS - the below example is done in Audacity, but any sample editor with similar functionality should suffice.

  1. Generate a 0 dBFS sine at 1/2 Nyquist, so each cycle is represented by the four (floating-point) values 0,1,0,-1
  2. Use "Change speed" in the effects menu, and set it to 0.5.
  3. Zoom way in on the start of the waveform, so individual samples become visible. Now, chop off the first sample
  4. Again, use "Change speed" and set the speed to 2.0
  5. Add a short fade-in and fade-out to the beginning and end of the generated tone
  6. (Peak) Normalize the track to 0 dBFS

This has the overall effect of introducing a 45 degree phase shift in the period of the sample, and the peak of each sinusoid will now occur halfway between the two equal-valued samples, causing a true peak that's ~3dB above the sample values in the file. A DAC with less than graceful handling of intersample peaks should thus exhibit clipping of the sinusoid, while a DAC like the Benchmark should still show a proper sine wave.

barrows's picture

arve, love you're idea here. I think testing for this problem would be great and I would love to see it in Stereophile measurements.
When we at Sonore were developing the OSF used in our USB interface we encountered this problem, and adjusted our filter parameters appropriately. With a lot of current music releases featuring full scale 0 dB signals, or even clipping (what are these recording engineers doing) it is important for DAC designers to take this problem into account.
Do note, as I recall, Ted Smith (main designer of the PS Audio DAC referenced here) has noted that the DS DAC uses 6 dB of headroom in its DSP stage. So it is not true that Benchmark is the only company addressing this problem.

arve's picture

@barrows : As a general comment (I haven't looked into the PS Audio DAC), but merely a general observation: DSP headroom isn't necessarily the same as DAC headroom - when processing a signal, you use headroom to prevent downright clipping of the processed signal, but if that DSP outputs digitally to a DAC, but if the DSP itself is _capable_ of outputing a sample value with an amplitude of "max value" for the DAC, there still needs to be headroom, as the intersample peaks occur _between_ the samples.

As an example: In my second setup, I use a little homegrown box running on shairport-sync, with full room correction using BruteFIR. Prior to DSP, that system uses (for the specific filters I use) -9.2 dB of attenuation to guarantee against digital clipping, it provides no guarantee against inter-sample overs, so that job has to be left to the DAC, as the DSP can still output 0dBFS.

But as said: I haven't tried to delve into precisely what Ted Smith and PS Audio is doing with a DSP in their DAC - so they might very well have tackled the problem

arisholm's picture

I think also Lyngdorf has attempted to address this issue with their "Inter-sample clipping correction" ICC. http://lyngdorf.com/tdai-2170/

claud's picture

Couldn't find any indication of this in the manual or elsewhere online.
https://lyngdorf.steinwaylyngdorf.com/downloads/lyngdorf-tdai-1120-owners-manual/ I don't think the more expensive Lyngdorf models do either. I asked owners of same about this at the AVS forum threads; no replies.

JimAustin's picture

See footnote 4 in the review.

Best,
Jim

JimAustin's picture

I've been having trouble posting, or I would have replied sooner.

I think what makes sense is to do some preliminary tests, see how common this problem is. I'll fool around with this a little, test some DACs I have on hand.

I did eventually manage to get good data contrasting the DAC1 and DAC3 response to a test signal similar to the one you describe. It's quite dramatic. Red is the DAC1; orange is the DAC3.

InersampleOvers

Jim

arve's picture

@JimAustin: That's quite dramatic, and something I'd readily take to be audible. Now on to convince JA to include that 11025 measurement as one of his standard measurements.

JimAustin's picture

this is a test signal. Absolutely audible. It's less clear how often intersample overs affect the sound of music--although if Benchmark's John Siau is to be believed, it's very often. (In his Manufacturer's Response, he agreed that the DAC1 sounds brighter than the DAC3--and attributed the difference entirely to intersample overs.) If you haven't already, read the essays on Benchmark's site:

https://benchmarkmedia.com/blogs/application_notes/intersample-overs-in-cd-recordings

https://benchmarkmedia.com/blogs/application_notes/13545433-audio-that-goes-to-11

https://benchmarkmedia.com/blogs/application_notes/13740017-why-audio-goes-to-11

TBD, IMO, is how common this problem is in recent/new DACs. JA and no one else will decide what to add to regular testing, but before that's even considered, I think it makes sense to do a little preliminary investigating.

Best,
Jim

supamark's picture

probably are not audible - when I was recording rock bands in the early 90's, I used an Aphex Dominator to prevent overs (it's an analog brick wall limiter). I'd mix a track, check the Sony DAT and verify no overs (I set the limiter to kick in at 0.5dB before 0 using a test tone) and would still occasionally see overs on playback but they were inaudible (Genelec 1031A monitoring). My aim was usually to light it up a few times during the mix to make sure I maxed out the s/n but never used it as a mix buss compressor like some did.

JimAustin's picture

But take a look at John Siau's Manufacturer's Comment. John's not one to claim audibility without evidence. He attributes the DAC1's relative brightness to overs. I intend to do some more listening myself. I did some for this review--but not after I confirmed I was actually getting them--i.e., that the server was sending bit-perfect output. As the review documents, I had a little problem there.

Best,
Jim

supamark's picture

and the D/A converters were nowhere near as good as they are now. I'm also talking about a few overs per song in rock music and generally on a drum strike. I can see it being a much bigger problem with today's mastering techniques, but then again with the aggressive use of hard limiting today how would you hear it separate from all the clipping?

You probably listen to a lot more classical/acoustic music than I do, and I bet it's a lot more audible in that context. I spent a couple years recording classical (always with just a stereo pair of Neumann mics), but 2 channel stereo just can't capture that surrounded by luscious reverb sound of being in the hall and always dissapoints me.

waynel's picture

Hi Jim, did you prefer the sound of the DAC3 direct to the amps or through a pre amplifier?

Thanks
Wayne

JimAustin's picture

Wayne, with apologies, I'm going to provide a somewhat complicated, and probably unsatisfying, answer. Because the most obvious technical virtue of the DAC3 is its extremely low noise (hence, high dynamic range), I'd be inclined to try to use it in the way that would take full advantage of that virtue and not ruin it by adding in extraneous noise. That would, of course, mean leaving out the preamp. Another problem, though, is that the noise level of most amplifiers would tend to swamp the noise advantages of the DAC3, even if you leave the preamp out. That would then suggest not only leaving out the preamp, but also using a very low-noise amplifier.

Of course this is all focusing only one one aspect of the DAC3's performance--noise--when there's surely much more to it than that. Still, it's the path I headed down, by asking Benchmark to send me a pair of their extraordinarily low-noise AHB-2 amplifiers. The AHB-2 amp is unusual in that when you bridge the channels (making a monoblock) you further improve the SNR. (Bridging in most amps increases the power but also to noise.)

So that was, and is, the plan--to create the ultimate low-noise system and see what it sounds like. Unfortunately, while I have all the parts here, I have not done that listening yet. But I'll do it soon, and then I'll write about it.

By rights, every system should sound better without unnecessary gain stages. Preamps always add some noise, and they can only lose--not add--musical information. So getting rid of noise should always be a good thing, assuming you're not giving up something else that matters more--and yet. I want to probe that--to try and figure out what, sonically, is gained with lower noise and what, sonically, is sometimes gained by adding a preamp.

That's a long-winded way of not answering your question. But hopefully you can see where I'm going with it, and hopefully I'll be able to share more insight soon. I'll leave you just with this: In a system with such a phenomenally quiet DAC, I'd be inclined to try and exploit its greatest virtues; it just seems wrong to add a noisy pre--unless it sounds better, and it just might. Stay tuned.

claud's picture

Jim, How were your DAC3 experiences yea/nay when paired with the preamp-and which one make/model? Did you get to try this test with more than one preamp? How did any particular recordings change in perceived sound quality with the preamp (s) in/out between the DAC3 and your power amps?

AJ's picture

And you believe yourself to be the arbiter of all what is and isn't audible?
Comedy gold ;-)

supamark's picture

No? Then STFU troll.

arve's picture

… let me show you a pathological case

https://imgur.com/ACqrucC

That's the _additional_ inter-sample clipping introduced into a track (Muse - Map of the Problematique). Intersample peaks below 0.1 dB are ignored. Every vertical red line is an inter-sample over.

While AB(/X) testing this takes some effort, because you need a controlled and calibrated testing setup with precisely matched levels, I would be _very_ surprised if that track didn't reveal differences between a DAC with inter-sample headroom and one without.

supamark's picture

that is some of the worst brick wall limiting I've ever seen - the mastering engineer (Howie Weinberg, who really should know better) should be taken out behind the woodshed and beaten.

arve's picture

Not being a reviewer with access to expensive gear: I tried this on various DACs I have lying around. All but of them will clip the signal,, showing a distortion spectrum similar to yours.

Edited: In re-testing - all of them exhibit clipping when digital volume is set to beyond -3 dBFS

Bubbamike's picture
Quote:

This is the kind of scientistic nonsense that's so common this world--a just-so story (ad-hoc fallacy) that attempts to explain subjective impressions via nice stories or casually observed phenomena while never subjecting those claims to serious tests.

Waiting for these kind of tests for MQA that you and the others at Stereophile seem to have swooned over without ever conducting any scientifically serious tests.

John Atkinson's picture
Bubbamike wrote:
Waiting for these kind of tests for MQA that you and the others at Stereophile seem to have swooned over without ever conducting any scientifically serious tests.

Starting with the January 2018 issue, we are publishing a series of articles examining MQA's claims one-by-one.

But "swoon" - I respectfully suggest your comprehension has been a wee bit colored by what you might have read from ill-informed hot heads on the InterWebs. :-).

John Atkinson
Editor, Stereophile

Bubbamike's picture

Sorry, but I think swoon is a fair description. Not the first time, even you admitted, just an issue or two ago that you and the magazine over empathized the effect of HDCD. I think this MQA mania is of the same enthusiasm. I have never seen a negative word applied to MQA from the staff of Stereophile. Perhaps I'll be surprised by your January issue. I hope so.

tonykaz's picture

There aren't any, are there ?

Sure, there are negatives. Everything has a negative of some sort even if it's quite minor.

MQA's prominent negative is that so many other things have been heavily promoted but not achieved universal acceptance that dubious consumers are remaining skeptical.

After all, isn't Vinyl 'still' the finest music Format for the group of Audiophiles that began during vinyl's Big Era? A good many of these folk remain Digital Deniers although even 'that' position is getting difficult since 2009 or so ( when HP of TAS was started blessing digital stuff ).

Another negative is Bob Stuart and Meridian. I've loved Meridian and Boothroyd Suart since the 1980s but I feel very much alone in this.

MQA being British is not a good thing for us Yanks, why couldn't one of our sharpies figure this out, someone like Edison or Einstein. Dam it, why another Brit thing?. That Linn guy and that LP12 was probably all the Brit we could take. We're the RedWhiteBlue Team and we deserve to win one, don't we?

Maybe the worst part of all this MQA business is that 'our' Warner was the first Record Label to get on-board. phew. Now it seems everyone is in a hurry to do MQA stuff. ( except some Neanderthal outfits that steadfastly refuse to advance into this 21st Century, I won't mention any names except Shit. There are a few more. )

The MQA Positive for those who remain negatives: Noboby has to buy it and nobody will hold it against you, MQA is just better RedBook. As far as I can tell.

Besides, if a person can't hear any MQA difference it only means that they have lesser gear or hearing ( like an old geezer ).

Ancient Tony in Michigan

Bubbamike's picture

You realize that MQA starts out with a high resolution file? It isn't meant to replace Red Book but to allow the streaming of Hi Res files over the internet with reduced bandwidth. Well among other issues, such as DRM and loss of data. But that you didn't know that is an indictment of Stereophile's coverage of MQA. If you look around you'll find explanations and critiques of the method, as well as Bruno Putzey's recent criticism of the lack of reliable tests of MQA.

tonykaz's picture

I accept, MQA is high resolution transmitted via RedBook.

Loss of Data, hmm, is that true?, maybe folks like Bruno Putzey will explain these things or someone else.

So far, MQA seems to be blessed but not universally adopted.

If I were making gear, I'd accept MQA.

As I stand today, I'm happy with RedBook and happy that I don't have to have a vast vinyl collection to enjoy music.

I feel like I'm winning and I'm not letting anyone take that away from me.

Tony in Michigan

arve's picture

I accept, MQA is high resolution transmitted via RedBook.

It's not. "Redbook" is entirely specific to data stored on an Audio CD, and governs all topics related to storing audio on that CD, including data structures and data encoding. The sample rate is only one such aspect. A file stored on a computer or audio streamed over a network can never be redbook.

But, the point you were trying to make was probably related to 16/44.1 audio being stored or streamed. Which is also entirely wrong for MQA. MQA uses a 24-bit container, rather than a 16-bit container.

Loss of Data, hmm, is that true?, maybe folks like Bruno Putzey will explain these things or someone else.

Yes, the data loss is true. John Siau has already done this dissemination of the format for you: https://benchmarkmedia.com/blogs/application_notes/163302855-is-mqa-doa

tonykaz's picture

Ok, I just had a brief look at that Benchmark Media report on "is mqa doa".

Firstly, thank you for pointing out that its a 24Bit container, didn't know ( or probably care ). Egads, can I accept or embrace 144db of dynamic range?, seems way too much ( even on a Battlefield Re-inactment )

Lossless? Sure, it looses the old file as it folds it up. I don't know what to think beyond that. Is it like flour stops being flour as the Pizza is made?

MQA is a just a Streaming System, isn't it? MQA is for the Record Company and the Streaming Company, it's not something built into the CD that we buy, as far as I can tell.

So, the Record Company & MQA devise a way to distribute their owned music to us 'Renters' of their convenient listening system.

We can create our own tiny SD memory cards and own our own Astel & Kern players and not bother with Streaming.

We can also collect vinyl and own vinyl playback gear.

We can have Tape Machines and buy Tape from Acoustic Sounds.

We seem to have a wide range of options.

MQA is just another option.

I don't see the reason for all the fuss.

Tony in Michigan

ps. as far as Streaming listeners are concerned, that Newspaper study showed how people had a hard time hearing 320 being different than hi-res.

Camilo's picture

Arve has proposed a very pertinent and concrete suggestion with regards to your test routine of D/A converters, and its merit and rationale very much deserves your consideration.

The suggested test would allow to recognize the innovation made by Benchmark with regards to intersample overs, and would allow to document and evidence the performance of other D/A converters in this respect.

It would shed more light into the differences between D/A converters, and thus provide a better resource of information for consumers, which is - I believe - the objective of carrying out measurements as well as the mission of a dedicated consumer product magazine.

Best,

Camilo

John Atkinson's picture
Camilo wrote:
Arve has proposed a very pertinent and concrete suggestion with regards to your test routine of D/A converters, and its merit and rationale very much deserves your consideration.

We used routinely to examine digital filter headroom, as you can see from reviews on this site that were originally published in the early 1990s. But I admit that was before the Loudness Wars, when CDs were mastered so that there were never consecutive samples at 0dBFS.

Modern digital audio workstations do calculate the waveform on the assumption that it would be processed by a typical digital filter and I have looked at some modern CDs. Inter-sample overs are rarer than many people believe, though I have found a few pathological examples.

But yes, I think I will start looking again at a DAC's digital filter overload characteristics.

John Atkinson
Editor, Stereophile

arve's picture

Inter-sample overs are rarer than many people believe, though I have found a few pathological examples.

This is a bit preliminary - as there are still non-music (and much quieter) data in the corpus (voice memos, audio books, podcasts and a couple of test tracks), but I have examined a corpus of 14411 tracks for intersample overs. Of those tracks, 6082 tracks reported true peaks below 0 dB, 1237 tracks has true peaks exactly at 0.0, and 7092 tracks have intersample overs. In other words, very roughly half of the tracks contain such overs.

I have not tried to do a more qualitative analysis of the overs, such as how many overs there are in a track, or counting whether some of them have overs that last more than one sample.

claud's picture

If we are to credit Benchmark's John Siau's Ultra-lock jitter reduction system as being sufficiently effective (or not?), then why did you find it necessary to insert that AudioQuest Jitterbug device between that Roon server and the DAC's USB input? I could see doing an ABX listening and/or measuring test with the Jitterbug inserted/removed/inserted two or more times, but I don't think you made any such tests. Please explain.

John Atkinson's picture
claud wrote:
If we are to credit Benchmark's John Siau's Ultra-lock jitter reduction system as being sufficiently effective (or not?), then why did you find it necessary to insert that AudioQuest Jitterbug device between that Roon server and the DAC's USB input?

For consistency in my auditioning of USB-input D/A processors. This included not just the Benchmark DAC3 B but also the Mojo Mystique that I was auditioning during the same period and comparing with it.

John Atkinson
Technical Editor, Stereophile

SeanS's picture

Hi John,

Eagerly anticipating the MQA article(s).

As an audiophile, I have to say I am not interested in the benefits of smaller file sizes for hi-res files--the origami, etc.

What I am really interested in is what, if anything, MQA can offer to better sound quality vs. a well implemented hi-res AD-DA conversion chain that doesn't use MQA. Like, for example, if you simultaneously created two 24/96 digital recordings of a jazz band using the same recording and playback equipment, with one recording having MQA enabled throughout the chain, and the other without, could you tell the difference? I would really appreciate your expert opinion coming from this perspective.

Thanks,
Sean

NeilS's picture

With all the expert technical scrutiny, even reverse-engineering that MQA has been subjected to on sites like computeraudiophile and Archimago this year, it's hard for me not to get the sense that MQA and its related claims are already technically very well understood by anyone who wants to know.

So when I read about Stereophile announcing at the end of October 2017 that it plans to run a series of articles beginning in 2018 on MQA to test its claims, it sure seems like a lot of effort, but it also seems to me like a lot of effort a lot too late.

Raving about/"swooning" over MQA before testing its claims may be a reminder that sometimes, like putting on socks and shoes, the order in which things are done matters.

tonykaz's picture

Paul McGowan of PS Audio is now doing a Question & Answer Series on YouTube.

Paul just answered the Question about different sounding DACs.

Egads, he's taking the time to answer all kinds of Audiophile questions.

This is well worth checking out.

He even tried to reason thru StillPoints.

Tony in Michigan

Camilo's picture

Paraphrasing the sentence that immediately links to your 5th footnote, I would say: "If you're REVIEWING 24-bit DACs and hi-rez downloads, you'd best get your noise level down to where you can hear at least some of that extra resolution, and that's harder than you might think".

What surprised - and outraged - some who read your review, is that you failed to hear a difference between the Benchmark DAC3 and the PS Audio Perfect Wave DAC, which leads to conclude that if there is a difference, it is not audible and doesn’t matter.

Although your 5th footnote hints a clue to why you - no matter how hard you tried - couldn't hear a difference between the Benchmark DAC3's 21 bit state of the art resolution, and the rather poor 17bit performance of the PS Audio Perfect Wave DAC, your review fails to take into account precisely that advice offered by your own footnote.

That advice obviously draws attention to the even poorer performance delivered by the PS monoblocks, acting as bottleneck when it comes to rendering the 21 bits of resolution of the Benchmark DAC and even the 17 bit performance of the PS Audio DAC, let alone their difference.

As Benchmark's John Siau wrote in a application note back in March 2014: "Anyone who thinks they can hear the difference between 16-bit and 24-bit digital audio through a "17-bit" power amplifier is fooling themselves."("What is high resolution audio? - part1" https://benchmarkmedia.com/blogs/application_notes/13174001-what-is-high-resolution-audio-part-1)

Nevertheless, the lenghty passage you dedicated to the efforts taken to hearing the difference between the Benchmark and PS Audio DACs, can suggest to some that this your flawed and misleading conclusion are not not merely innocuous omission or mistake. You appear to ignore your own advice, and completely disavow the fact that your monoblocks aren't capable of delivering even 16 bits to your speakers.

I think it is fair to criticize your review and to dismiss your conclusion regarding the audible difference between the Benchmark DAC and the PS Audio DAC. I also believe it is fair to demand that in the future, stereophile reviews take into account basic specifications - which I want to believe reviewers understand but apparently and consistently fail to apply.

I also believe it is impossible and unacceptable to excuse the obviously flawed attempt to establish a difference between the mentioned components and the consequently misleading conclusion, with a statement like: "Yes, listening is what Stereophile reviewers do." If people read reviews here and elsewhere, it is based on the expectation that they will be offered more than casual listening impressions that completely ignore basic science, let alone the specifications of the components reviewed.

It is in this particular case, quite obvious that you need an amplifier that matches and even exceeds the performance of a DAC to deliver the resolution the DAC has to offer to the speakers, yet that was blatantly ignored by the review. Worse, to reinforce the apparent ignorance - I will nevertheless refrain from concluding foul play or second intentions here - of the reviewer with regards to importance the specifications of the components used have for his review, he proceeds to write a lengthy passage showing what lengths he went to in order to hear a difference that the equipment he used is clearly unable to render. This is also not the only review to be found on Stereophile or Audiostream, with this exact same flaw.

This is poor work, and ultimately undermines the credibility of Stereophile, as well as the effort made by John Atkinson to measure equipment as a way to offer transparency and accountability with regard to manufacturer specifications and in many cases supplement the lack thereof.

I am in my mid fourties and, after thousands of hours behind a drumkit, exposure to loud environments and extended listening periods, cannot argue to hear extremely subtle differences. But even having a well trained ear as a musician, I would not be above relying on the due diligence of reading the specs and setting up valid review, audition or test conditions first, and before I make conclusions regarding the audible differences between two audio components or recordings.

I own a Benchmark DAC2 DX and have owned a DAC1. I had the chance to audition them both side by side with recordings that are known to have intersample overs, and I could very much hear the difference using my Sennheiser HD 800s.

I would hope for a clarifying response from you, with regards to taking into account the components used for a valid review of the performance of specific components. I would also like to see a response from John Atkinson with regards to the very pertinent suggestion made by arve - further up in the thread - with regards to introducing a test that takes into account intersample overs. Atkinson made no remarks to this excellent and concrete proposal, and instead threw in the infamous MQA topic, which distracted from arve’s suggestion.

Updating measurement routines as components introduce new improvements and features that need to be account ted for in order to do fair comparisons between components, would only benefit the objectivity of the measurements and reliability of what Stereophile publishes as such.

Benchmark has clearly separated itself from all other DAC manufacturers I have knowledge of, by effectively dealing with the audible downside of intersample overs, and has clearly introduced a substantial improvement to sound quality that sets new standards. This can only be acknowledged by introducing the corresponding measurement to account for this innovation and the – recording industry – problem of intersample overs, whose existence has been acknowledged and documented, as well as as put forward by this review.

Best,

C. Rodriguez

griff2's picture

When talking about bit depth, "resolution" is not the word to use; bit depth merely determines the noise floor. To say a 21 bit DAC (actually I'd say 20 bit DAC, 21 bits give an S/NR of -126.4 dB, whereas 20 bits gives a S/NR of -120.4 dB) should be paired with a speaker amp of similar or better is meaningless. It would be totally impossible to achieve a S/NR of -126 dB in any acoustic environment apart from an anechoic chamber. Even through closed headphones that sort of S/NR would be pushing it. Having said that, how many recording can boast a S/NR of anywhere near -120 dB anyway? In short, even with a recording that genuinely had a S/NR of >20 bits, It would be totally impossible to differentiate between 17 bits (-102.3 dB) and 21 bits (-126.4 dB) through a pair of loudspeakers, in a conventional listening environment.

Robocop's picture

I have owned both the DAC1 and DAC2L over 15 years. This review actually tells me little I havn't already read from other reviewers and the Benchmark web site.

What I find most disconcerting is the DAC2 was the current reference DAC for Benchmark which I own.

Why was the DAC2 not compared directly to the new reference DAC3 in listening sessions?

Comparison with the DAC1 is obsolete, well surpassed by the DAC2.

"How much audible improvement do these changes add up to? "I'm quite certain that there should be no audible difference between a DAC2 and a DAC3 given a single pass through the converters," Siau wrote to me in an e-mail."

What does this mean from John Siau Benchmark? Is he saying no audible improvement over the DAC2?

If it sounds the same, well why bother!!!

I really want to know how much better is the DAC3 over the 2 to justify its increased purchase price.

There must be a sound improvement, the Sabre 9028 chip is alone a sonic upgrade over the 9018. This must be audible and not just measured.

It is at the end of the day all about the "SOUND" compared to live instruments in an acoustic space.

Robert

griff2's picture

I wouldn't be so hasty to dismiss the DAC1. I own both the DAC1 and DAC3 L, and whilst the DAC3 L is the superior DAC in terms of being less forward and a little more refined than the DAC1, they are both very good highly transparent DACs.

Also, it is not about "sound" compared to live instruments in an acoustic space. Unless the recording deliberately sets out to emulate a live acoustic event (such as some of the Chesky Binaural recordings) the recording is produced as an artistic interpretation of what the producer/engineer/artist envisage; similar to what a photographer, painter or movie director will do to create a work of art.

I purchased the Benchmark DACs because I wanted to hear the music as transparently as possible and wanted to hear what the artists/producers etc., intended (as opposed to a what some hi-fi manufacturer interpreted as "music").

Sigh's picture

It seems that the 9028 is the new 9018 that might be discountinued soon, I can see why Benchmark made the change even if it offers only a minute improvement that isn't audible. Maybe they should have called it DAC2.2 (Although they did have a little update to the DAC2 called 2.2 already) or DAC2+ like Mytek with their Brooklyn+.

To compare the PS-Audio to the DAC3 in a set up that would have made some sort of sense one could run the analog-out of one DAC3 into the analog-in of another DAC3 :) Is the reviewer suggesting that this too would have sounded exactly like a single DAC3 on its digital volume control?

I've owned every Benchmark, DAC1, USB, PRE, HDR, DAC2 and changed more for the new features (USB, remote, second line in) than the negligible sound differences. The one thing I didn't like with the DAC1s was that they ran hot. With the DAC2 I feel like they reached a perfect product, great interface, hybrid gain control, runs cool, asynchronous USB.

It probably makes no sense to trade a DAC2 for a DAC3, yet at the same time it would make no sense for Benchmark to continue using a older chip.

I find that whatever my questions are, the best way to get them answered is to ask Benchmark themselves. They answer promptly and won't push a new product onto you.

Charles Hansen's picture

Once again, I am baffled by Benchmark, designer John Siau, and reviewer Jim Austin. The existence of intersample overs has been well known for over 20 years. I'm surprised that JA did not catch this much earlier.

For evidence of this fact, we only need to look at the datasheet for the once-popular Pacific Micronicss PMD-100 digital filter. In the datasheet is a an unambiguous statement:

The PMD-100 has a design attenuation of 1 dB to allow for filter overshoot on transients.

During the '90s when Robert Harley was technical editor of Stereophile, he would routinely show CD players with no internal headroom, and how they would clip the "ringing" (Gibbs phenomenon) on the tops of a 1kHz, 0dBFS square wave. In contrast were other players that used (for example) the PMD-100 digital filter (and many other designs) that provided headroom to prevent internal overload from intersample overs.

While this phenomenon is understood more clearly now, with a better understanding of the degenerate (worst possible) case, this is hardly some sort of "breakthrough" as Siau and Benchmark would have us believe. In fact it's more surprising to me that he was previously unaware of a well-known issue regarding digital audio playback.

Camilo's picture

Knowing the problem does not automatically guarantee there to be an immediate solution, and blaming those who finally solve a pervasive and known problem for not having solved it before and for taking credit for it, doesn't seem like the right place to aim your criticism at.

The solution for the audible artifacts of intersample overs that Benchmark came up with, is not something trivial, as it is an inherent flaw of D/A chips including the ES9018 and ES9028PRO used in the DAC2 and DAC3, respectively:

"Every D/A chip and SRC chip that we have tested here at Benchmark has an intersample clipping problem! To the best of our knowledge, no chip manufacturer has adequately addressed this problem. For this reason, virtually every audio device on the market has an intersample overload problem. This problem is most noticeable when playing 44.1 kHz sample rates." ("Intersample Overs in CD Recordings"-John Siau https://benchmarkmedia.com/blogs/application_notes/intersample-overs-in-cd-recordings)

Benchmark had to develop and implemenmt a solution that is not an option provided by the D/A chip, but which addresses a common problem that originates in D/A chips:

"It is possible to build interpolators that will not clip or overload, but this is not being done by the D/A and SRC chip manufacturers. For this reason, Benchmark has moved some of the digital processing outside of the D/A chip. In the Benchmark DAC2 and DAC3 converters we have an external interpolator that has 3.5 dB of headroom above 0 dBFS. This means that the worst-case +3.01 dBFS intersample peaks can be processed without clipping. We also drive the ESS D/A converter chips at -3.5 dB so that no clipping will occur inside the ES9018 and ES9028PRO converter chips." ("Intersample Overs in CD Recordings"-John Siau https://benchmarkmedia.com/blogs/application_notes/intersample-overs-in-cd-recordings)

It is not out of mere negligence that Benchmark, or other manufacturers who rely more on the D/A chips - and who in some cases even list the specs of D/A chips rather than the measured performance of their components, and thus not the result of their implementation of D/A chips - haven't offered an effective solution to the intersample overs problem before. Being Benchmark the ones who came up with it,
hardly makes them the target to blame for said negligence, let alone undeserving of the merit for effectively addressing the issue of intersample overs.

If anything, it is other manufacturers who have not yet implemented a similar solution who could deserve some criticism in this respect.

As to the developing and final implementation of a solution to intersample overs, it is debatable if - according to your appreciation - it took the industry too long. It is however not debatable that Benchmark's solution is very much a welcome breakthrough for the performance and sound quality of D/A converters, and one that clearly sets them apart from all other manufacturers.

SeanS's picture

Wouldn't Siau's solution have to sacrifice dynamic range in the digital domain? He says they then bump up the 3.5db in the analog domain to bring the level back up. Definitely not perfect. Sounds like your trading dynamic range on all input data in order to correct the consequences of "bad" input data. On the other hand, maybe the amount of dynamic range lost is less than the noise floor?
Sean

arve's picture

Due to thermal noise, no DAC can achieve the full 24 bits of dynamic range inherent to such program material - at _best_ the analog outputs of a DAC will offer a bit above 21 bits of signal, with the three lowest bit drowning in the thermal noise. By adding 3.5 dB of attenuation to the signal, Benchmark is throwing away approximately the lowest half bit of the signal - this is signal that was already lost because analog electronics won't let us do better.

The loss in level is made up for through analog means adding the 3.5 dB of gain back after conversion. The risk here is that adding this gain causes the noise floor to rise, and thus lose dynamic range. However, as seen from JA's measurements, this is not the case with the Benchmark: The analog outputs have class-leading state-of-the art performance with regards to the noise floor.

SeanS's picture

Arve, the content of your message seems to elaborate on the thoughts I was having when I posted.

Throwing away the last half bit of the signal = throwing away dynamic range in the digital domain.

However, the low level thrown away was already lost because of the physical limitations on analog electronics.... = it is below the noise floor.

I was hoping to get opinions on the solution.

Sean

nikolaos's picture

First of all I'm a huge fan of Ayre and had several components from you and always enjoy whenever I hear Ayre gear. I'm also a huge fan of Stereophile and consider the mag.. as very serious. I will also say I enjoyed reading the review of DAC 3.

Regarding your comments (sorry I'm not english so forgive me for errors in the writing) I had to reply regarding things that where said.

I had lot's of equipment and tested a lot and my conclusion is that whatever the inputs or cables has to say to what I'm hearing does not even come close to other factors like room, components and the ultimate headroom in the system itself.
I had some quite expensive cables and I can for sure not say some where better than others even if the price difference was huge. My concusion is that it is so minor differences in cables that you will gain much more elsewhere in the system. I have heard amazing systems with cables that where the cheapest and I heard the worst systems with the most expensive cables.
I had simelar experience as the reviewer regarding hum with unshield Nordost cables but as Charlie say as they are balanced it can not happen. I guess the system was not truly balanced.
I agree that the reviewer or the editor maybe should have picked that up.
Regarding Benchmark it is well know that they have made great products for some time and as J.A show in the meassurements of this DAC it is a well designed product. I don't understand why a DAC should cost as much as some charge. AYRE used to have their top of the line DAC in around the same price range as Benchmark. You say Mr. Hansen that the DAC 3 is made of cheap components and it seams like you consider Benchmark bellow average quality.

I don't buy to much into all that voodoo anymore about lot's of the claims in the high-end industry. Not to say that from time to time some come up with new and better solutions.
Audio is to me much more about measurements and real world facts. Even if the reviewer did something not considered right by connecting one DAC into the other I'm not so sure it would be so easy to hear coloring of the sound. It would for sure act as a filter but so does a preamp.

As said I'm not English and maybe I'm stupid but I did not get the problem you had with how the reviewer explained how he connected the output of the DACs to the preamp.

It is a little industry. Let's enjoy the music and be friends. Everytime I meet someone with the same hobby it is always a pleasure. In a forum things are very unpersonal and I know people would not act like this when met for real.

Doctor Fine's picture

As so many of us (myself included) own a Benchmark DAC and these have a fundamental "house" sound I propose we use them for all future comparison between DACS du jour.
My other DACS for the last ten years or so have been two different models of Lucid Audio/Symmetrix Pro DACs bought for studio use.
And then I thought it wise to buy an "upgrade" of the Bryston BDA-1 DAC which offered two significant technical advantages---upsampling AND class A analog output circuitry to sweeten up the sound and make it less plastic y.
The room I used for playback held on to energy quite well, was a little bit too small for my monitors AND was a bit hard sounding.
As a result the warm Lucid DACS gave the best results.
The Benchmark IN THAT ROOM sounded too crisp and lacked balls.
The Bryston sounded much clearer than the Lucids also. But it sounded over processed especially when upsampling was turned on.
So the increase in clarity both brought to the table was not appreciated.
My conclusion was the Benchmark sounded thin and the Bryston sounded fake. I scratched my head at the rave reviews these two products were getting from all quarters. You couldn't prove it by ME.
Then I moved to a much larger home with a wonderful listening room four times larger with unequal walls, sloped ceiling and great listening position.
The listening room resembled a small concert hall and the sound was open and relaxed and with a few pieces of Sonex and some properly placed furniture the reverberation times were felicitous.
So I decided to see if my opinion had changed about my DAC collection.
In a nutshell---YES!
The Lucids NOW sounded outclassed. Warm? Yes. But I could really use more upper octave energy without grain and here is where the Benchmark and the Bryston really came to show what they could do.
The Benchmark added at least an octave to the playback.
It still sounded a bit "cheap" which is to be expected as the company does not invest a lot of money in their analog circuitry instead relying on rather basic function over sophistication.
The strength of the Benchmark unit was its astonishing clarity, not its warmth. It was clearly twice the DAC at twice the price of the Lucids.
Next up the Bryston offered similar high frequency articulation but with even more precise soundstage AND with upsampling turned on a certain harshness and edginess which I assumed was inherent in digital suddenly evaporated.
The Benchmark could still be said to be a bit thin compared to the Bryston but then there was that amazing Benchmark clarity. Still, the Bryston with upsampling was also making its super clear digital sound relaxed and like really good analog---which quickly becomes addictive as you get spoiled by the lack of glare and edge!
OK so the Bryston at three times the price of the Lucids was in deed three times as good IN A BETTER ROOM.
It has me wondering if in fact as hobbyists we are approaching the time where our gear is getting sophisticated enought that simple adjustments to listening rooms can alter sound in a more meaningful way than "upgrading" endlessly.
Back when I started in this hobby all you had to do to get first class results was purchase McIntosh gear some good speakers and a good sounding cart to play your records.
We all knew rich friends that owned Marantz 10Bs, MC275, AR3as, Quads and THAT was the simple upgrade path for home audio.
Spend more dough on your "components", get more jollies.
We are at the point where finding recordings is as near as the internet.
Formats are improving to the point some people can no longer even hear improvements because things are that good already.
And here I am finding out my old DACS that I didn't like are actually now my favorites.
I hand it all to Benchmark for resetting the standard for what one may get out of digital. Everybody should own at least ONE Benchmark something.
But after that happens and you learn the limits of your room perhaps we simply all need to purchase new HOUSES.
It worked for ME.
I imagine my next purchase will be a nice tube amp, a Mytek Brooklyn with MQA and a good bottle of scotch.
Give me another ten years and I'll decide which of the three gave the best improvements in my listening enjoyment.
I already know the scotch will make everything sound more "liquid" ha ha.

allhifi's picture

"We can make him stronger, faster ... " (oh, wrong Austin -thinking Steve)

Mr. Jim Austin's review of BM's DAC-3 touched on (and off it appears) a number of issues/concerns in the hifi business. Siau

he first order of business is this "subjective/Objective" argument. Any and all great audio designer's must (and do) sek key circuit design elements -followed by rigorous testing/measurements.

And one can rest assured that a designer will implement design choice(s)/considerations they feel represents the better sounding circuit -as opposed to the better 'measuring' circuit.

No such logic or sensibility exists with Siau-inspired products. And it shows.

Jim Austin's review started off with this (what some see as a) dilemma; create/listen/tweak/release or the Benchmark mantra of "spec's say so let's build it" ! Too funny.

Jim does, not surprisingly, note considerable sound variation between the 14-year old DAC-1 design, and the current Gen.3. But ask Siau and he'll tell you that nothing really meaningful exists (between the two) so the two should be indistinguishable from the other! Even funnier.

Jim (Austin), you mention how you admire this in the "objectivist" crowd -noting their/science's: "... traditions of precision, self-depracating ... understatement" type disposition.

YES, indeed, so much so 'they' go on record to tell the world how much they know and how foolish it is to entertain anything that can't be readily measured by test equipment ! For sure Jim, these "objectivists" are the "unassuming" types.

I would also say that the near identically priced Schiit "Yggy" annihilates the DAC-3 beyond recognition. The 23-lb. "Yggy" (using "old" DAC architecture) vs. the 2.3 Kg. BM-3!

Forget the build quality distinctions (but only for a moment), and consider that the "Yggy" is 23-lbs. worth of brilliant Moffat design using top quality parts within a full-size chassis, built like a tank.

Mr. Moffat understands (very well) the utility/necessity of lab work/ results, but his inner (music-loving) sensibilities are the real driving force behind great products -whether the spec's" say so, or not. This talent is something all of the best designs/designers share.

Back to Siau-products; may I suggest a nice pair of world-renowned Bose loudspeakers with your BM DAC along with your choice of lamp-cord speaker wire color -but do make sure it is #18/#16 gauge.

peter jasz

griff2's picture

Having owned (and got rid of) several "well regarded" Schiit headphone amps and DACs I can quite happily say that the "c" and "s" should be transposed and one "i" removed from the company name.

allhifi's picture

Lol. You may be right. Over-estimating Schit's/Moffat's capability is something I regret: competent (design) yes, but nothing more.

Sooner or later, a company's 'mantra' exposes itself. With respect to the Norse-inspired chaps that run this operation, it became quite clear quickly the arrogance on offer -customer service is deplorable.

In any case, my point was to say the best gear goes through rigorous listening evaluations before deciding final production models. Such listening evaluations become complex (the sheer variables involved)but absolutely necessary. But not according to Benchmark/Siau types.

I chuckle thinking about some hi-fi brands where the principals and engineers have no concept/understanding of hi-fi 'cables', 'power cords' etc.
But slowly, surely, some come around to acknowledge the errors of their ways -and implement the necessary product changes for much improved sound -whether or not superior specifications are achieved.

It's easy to pick out such brands; they almost always brag about their "technical" direction, not "swayed" by concepts they don't understand -or can 'measure'.

The problem with that is many fold, but the biggest slap to their faces arrives when future (on-the-horizon) discoveries will reveal why cables/power cords/speaker wire etc. has such a profound impact upon our subjective experience/interpretation -and even give the White-Coats the numbers and spec's they absolutely must have to build into their product. Sadly

Such folk are simply very insecure: unable to wrap their head around that they may feel/hear considerable sound variation but then look to 'reasonable' explanations for their subjective interpretations. But then again, some brands exist based on nothing to do with the science/art of fine sound quality but simply a business opportunity that "fell-into-their-laps" (so to speak).
This type of product/company will never have product (unless by complete shit luck) a product to savor over simply because of the designers indifference -or misunderstanding of cutting-edge thinking, research and design.

Technical excellence (circuit design/loudspeaker performance criteria) must be a design objective. But also taking into account that which is not presently understood very well, if at all. And that, right there is the 'objectivist's' failing. And it's a detrimental one that's often reflected in their products.

Bottom line: Listen to verified/authentic audiophile voices along with company's that speak of listening sessions as part of the design process. The chance of enjoying a superior product (hi-fi) is much greater.

Man, I talk too much. lol

pj

Marcel-073's picture

Hi all,

New to this forum.

I'm in the process of buying a DAC and have read some reviews on different sites. Looked at the Benchmark DAC review after having read Mark Waldrep's recently released book "Music and Audio: A User Guide To Better Sound".

When reading the chapter dealing with DACs I had the feeling that I was missing PS Audio's DirectStream. Unfortunately the review on the Benchmark turns out not to be an A/B comparison to the PS Audio DS, though the latter has been used "in the chain".

Anyone know of an A/B comparison done between these two?

Moreover, the Chinese company Auralic seems to have received several good reviews on their Vega G2. Anyone experience with this unit?

@Stereophile team: is a review of the Vega G2 - perhaps couple to the Leo GX reference clock - foreseen in the near future?

Thanks,

Marcel from the Netherlands

lesmarshall's picture

Is the Charles Hansen who commented above the late Charles Hansen of Ayre audio ?

John Atkinson's picture
lesmarshall wrote:
Is the Charles Hansen who commented above the late Charles Hansen of Ayre audio ?

Yes, I believe so.

John Atkinson
Technical Editor, Stereophile

hemingway's picture

..and to learn it is still well regarded from a subjective point of view by John A.

"With the Benchmark amps and cables, the balance was slightly lighter, the soundstage somewhat less enveloping, but the overall presentation was still involving."

I am curious if the slightly lighter sound is attributed to the amp, or if the swap in cables made any difference? This analysis was consistent with what I am remembering from your comments about the Schiit monoblocks, compared to the parasounds, compared to the Benchmark stereo power amp, which suggests it is the amp.

Seems like there is a lot of shade directed to the measurements crowd, right or wrong. I for one appreciate Benchmark's ethos and transparency. They discuss performance of their products in engineering terms and offer explanation of how or why the products performs as it does. Their claims about cables relate to rejection of noise and distortion, e.g., and they charge reasonable amounts of money for the products compared to competitors. Its a breath of fresh air in this industry/hobby.

Could other products sound "better"? Sure, but when considering the measurements of the Benchmark products, you have to question if the better sounding component is "accurate" in the sense that input=output, or if that better product is adding a distortion that sounds nice. If the other one sounds better then fine. But these products require the audiophile to consider their biases and priorities in listening. So props to Benchmark, from my POV.

msmucr's picture

Just one remark to John Atkinson's follow-up with DAC-3 B.
All (AFAIK) their converters incl. DAC-1, 2 and 3 has internal pads with 0, -10 and -20 dB of output level attenuation. You can set that via internal jumpers at PCB right next to XLR connectors. It's a normal voltage divider, so it naturally affects output impedance, but IME it works very well to accommodate converter output level to sensitivity of particular downstream device. You can combine it with digital volume control and optimize for good dynamic range.

Michal

BudhaNL's picture

You are correct MSMUCR, Benchmark recommends to lower the output impedance as much as possible in order to lower the amount of distortion and/or to adapt to the optimal range of amplification from your pre or integrated. I've the DAC-3B at -20 dB feeding a Luxman 590X to very satisfying results.

SeanS's picture

I believe Sterephile's review practice is not to make any adjustments that require opening the case. Whether it is for better or worse is debatable, but it has been their practice from the beginning, AFAIK.

michelesurdi's picture

do read the manual.it explains most clearly how to set the internal jumpers which
modify the xlr output to suit.

claud's picture

Can JRiver player do this? While I think it stinks to have to rely on a pay forever to use solution do Roon, HQ Player or other players have filters that will satisfactorily reduce intersample overs distortion in any of my commercially issued music CD recordings?

X