Digital Processor Reviews

Weiss Engineering Helios D/A processor Measurements

Sidebar 3: Measurements

I measured the Weiss Helios with my Audio Precision SYS2722, repeating some of the tests with the higher-resolution APx555. All the measurements were performed with the Helios's DSP bypassed. Apple's USB Prober utility identified the Helios as "Helios" from "Weiss_Engineering_Ltd." with the serial number string "0.0.1." The USB port operated in the optimal isochronous asynchronous mode, and Apple's AudioMIDI utility revealed that the Helios accepted 32-bit integer data sampled at all rates from 32kHz to 384kHz. The AES3 and coaxial S/PDIF inputs accepted data sampled at rates up to 192kHz; the TosLink input was restricted to sample rates of 96kHz and lower.

With the balanced output mode set to Loudspeaker, the Helios's maximum output level at 1kHz feeding 100k ohms was 16.4V with the level set to "0dB"; 10.3V set to "–4dB"; 6.5V set to "–8dB"; 4.1V set to "–12dB"; 2.6V set to "–16dB"; 1.63V set to "–20dB"; 1.03V set to "–24dB"; and 648mV set to "–28dB." Each nominal reduction of 4dB in the output level reduced the actual level by 4dB. As expected, the maximum levels from the unbalanced outputs were half those from the balanced outputs. Switching the output to Headphone reduced all the maximum levels by 12dB. With its polarity button set to Normal, the Helios preserved absolute polarity (ie, was noninverting) from all of its outputs. The balanced output impedance was 93.5 ohms at all audio frequencies in both Loudspeaker and Headphone modes; the unbalanced output impedance was 47 ohms in both modes.


Fig.1 Weiss Helios, impulse response (one sample at 0dBFS, 44.1kHz sampling, 4ms time window).


Fig.2 Weiss Helios, wideband spectrum of white noise at –4dBFS (left channel red, right magenta) and 19.1kHz tone at 0dBFS (left blue, right cyan) into 100k ohms with data sampled at 44.1kHz (20dB/vertical div.).

Fig.1 shows the Helios's impulse response with 44.1kHz data. It is typical of a conventional linear-phase filter with a symmetrical ringing before and after the single full-scale sample. This filter's ultrasonic rolloff (fig.2, magenta and red traces) reaches full stop-band attenuation at 24kHz with complete suppression of the aliased image at 25kHz of a full-scale tone at 19.1kHz (cyan, blue). The harmonics associated with the 19.1kHz tone all lie below –107dB.


Fig.3 Weiss Helios, frequency response at –12dBFS into 100k ohms with data sampled at: 44.1kHz (left channel green, right gray), 96kHz (left cyan, right magenta), and 192kHz (left blue, right red) (1dB/vertical div.).

Fig.3 shows the Helios's frequency response with data sampled at 44.1, 96, and 192kHz. The response with all three sample rates is down by just 0.1dB at the top of the audioband with then a sharp rolloff just below half of the two lower sample rates. The response with 192kHz data continues the relatively gentle ultrasonic rolloff, reaching –3dB at 71kHz.


Fig.4 Weiss Helios, spectrum of 1kHz sinewave, DC–1kHz, at 0dBFS with volume control set to the maximum (left channel blue, right red; linear frequency scale).

Channel separation was superb, at >122dB in both directions below 3kHz, decreasing to a still-superb 113dB at 20kHz. Fig.4 shows the spectrum of the Helios's low-frequency noisefloor as it drove a full-scale 1kHz tone with the volume control set to its maximum and the level set to "0dB." The level of the random noise is extremely low, and there are no AC supply–related spuriae present. Reducing the maximum level didn't increase the level of the noisefloor.


Fig.5 Weiss Helios, left channel, 1kHz output level vs 24-bit data level in dBFS (blue, 20dB/vertical div.); linearity error (red, 2dB/small vertical div.).


Fig.6 Weiss Helios, spectrum with noise and spuriae of dithered 1kHz tone at –90dBFS with: 16-bit data (left channel cyan, right magenta), 24-bit data (left blue, right red) (20dB/vertical div.).

The red trace in fig.5 plots the error in the analog output level as a 24-bit, 1kHz digital tone stepped down from 0dBFS to –140dBFS. Even at the lowest level, the amplitude error is <1.1dB, which implies very high resolution. An increase in bit depth from 16 to 24, with dithered data representing a 1kHz tone at –90dBFS, dropped the Helios's noisefloor by 33dB (fig.6). This implies a resolution between 21 and 22 bits, which is the highest I have encountered, greater even than that of the Weiss DAC502 and of the Merging Hapi MKII that KR reviewed in January 2024.


Fig.7 Weiss Helios, waveform of undithered 1kHz sinewave at –90.31dBFS, 16-bit data (left channel blue, right red).


Fig.8 Weiss Helios, waveform of undithered 1kHz sinewave at –90.31dBFS, 24-bit data (left channel blue, right red).

When I played undithered data representing a tone at exactly –90.31dBFS, the waveform was symmetrical, with negligible DC offset, and the three DC voltage levels described by the data were free from noise (fig.7). With undithered 24-bit data (fig.8), the Helios's very low analog noisefloor means it can output a clean sinewave, even at this very low signal level.


Fig.9 Weiss Helios, 24-bit data, spectrum of 50Hz sinewave, DC–1kHz, at 0dBFS into 600 ohms (left channel blue, right red; linear frequency scale).


Fig.10 Weiss Helios, 24-bit data, HF intermodulation spectrum, DC–30kHz, 19+20kHz at 0dBFS into 600 ohms, 44.1kHz data (left channel blue, right red; linear frequency scale).

Even set to its highest output level, the Helios produced very low levels of harmonic distortion with full-scale data even into the punishing 600 ohm load (fig.9). The subjectively benign second and third harmonics were the highest in level, but each lay close to a negligible –120dB (0.0001%)! Intermodulation distortion with an equal mix of 19 and 20kHz tones at –6dBFS was similarly extremely low (fig.10), with the difference tone at 1kHz into 600 ohms lying at –130dB (0.00003%)!


Fig.11 Weiss Helios, high-resolution jitter spectrum of analog output signal, 11.025kHz at –6dBFS, sampled at 44.1kHz with LSB toggled at 229Hz: 16-bit AES3 data (left channel blue, right red). Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz.


Fig.12 Weiss Helios, high-resolution jitter spectrum of analog output signal, 11.025kHz at –6dBFS, sampled at 44.1kHz with LSB toggled at 229Hz: 24-bit network data (left channel blue, right red). Center frequency of trace, 11.025kHz; frequency range, ±3.5kHz.

The Helios offered excellent rejection of word-clock jitter. Fig.11 shows the spectrum of the Helios's output when it was fed high-level 16-bit J-Test data via AES3. All the odd-order harmonics of the undithered low-frequency, LSB-level squarewave lie at the correct levels, and no other sideband pairs are visible. The central spike that represents the high-level tone at one-quarter the sample rate (Fs/4) is broadened at its base, which was not the case with USB and network data (fig.12).

Weiss's DAC502 performed supremely well on the test bench, but its measured performance was exceeded by that of the Weiss Helios!—John Atkinson

COMPANY INFO
Weiss Engineering Ltd.
Florastrasse 42
8610 Uster
Switzerland
weiss@weiss.ch
(416) 638-8207
weiss.ch
ARTICLE CONTENTS

COMMENTS
georgehifi's picture
Submitted by georgehifi on February 23, 2024 - 1:56pm

Wow!!! stunning set of measurements JA, must have a listen to this one.
Direct in to the poweramp/s with this baby, (hope you got to do this with the Halo's JA??), nothing says to me more that active preamps are obsolete and a backward sonic step when used with this dac, like these two sentences of your measurements.
JA: "
"Helios's maximum output level at 1kHz feeding 100k ohms was 16.4V with the level set to "0dB"
"The balanced output impedance was 93.5 ohms at all audio frequencies in both Loudspeaker and Headphone modes; the unbalanced output impedance was 47 ohms in both modes."

Cheers George

cognoscente's picture
Submitted by cognoscente on February 23, 2024 - 11:53pm

But then again, that price? How does that compare to everything else? Are such prices still in proportion to everything else in life? Someone who buys this, how much does he pay for a bottle of good wine? For a good coat? For an armchair? For a car? 800? 4K? 20K? 400K? Then it would be proportionate. Of course everyone has their own thing, choice (freedom), hobby and pleasure. Naturally! I just wonder.

georgehifi's picture
Submitted by georgehifi on February 24, 2024 - 1:17am

At least with this, you'd save thousands by not having to have a preamp and be better off sonically for it.

Cheers George

Scintilla's picture
Submitted by Scintilla on February 24, 2024 - 9:59am

I ran direct for many years from an NAD M51 using its volume control, then a D90 for another year and finally a Holo May for about 6 months before adding back a Holo Serene preamp to provide better gain-matching for my system. Not only did that addition drastically improve the sound quality it also allowed me to have proper gain matching. There are plenty of utterly transparent preamps on the market and they serve a useful purpose beyond source switching. I believed the dogma for too long that no preamp was the best preamp and I have seen the light. So have a lot of us that went all digital a long time ago and have ben using digital gain structures. The May and the Serene are designed for each other and work and sound magnificent, and much better than the May direct using a digital volume control in HQPlayer. This hobby isn't about "saving thousands" it is about getting the best sound quality -for the dollars- and that doesn't mean eliminating a useful component because it costs money. Listening isn't about theories, its about listening, with your ears, not your ideas.

georgehifi's picture
Submitted by georgehifi on February 24, 2024 - 10:33am

If you go direct and have enough volume level for listening up to your loudest level with the quietest recording you have, then you have enough gain. (and are "gain matched")
Any more gain by adding an active preamp is not going help it sound any better "what so ever" if anything it's adding more noise and distortion.

And as far as impedance matching, the Weiss Helios D/A processors output stage is perfectly impedance matched for any poweramps input impedance.

And I repeat: "At least with this, you'd save thousands by not having to have a preamp and be better off sonically for it."

Cheers George

JRT's picture
Submitted by JRT on February 24, 2024 - 11:10am

I want to preface my comment with sincere thanks for another well considered and well written review in combination with extensive measurements and associated commentary (a lot of work, and very well done), before continuing with some negative criticism of the component that was the subject of the review.

"...balanced and single-ended analog outputs. But on closer inspection, the Helios lacks the earlier processor's headphone jack. It can still be used with headphones, however: Using optional adapter cables ($495), the Helios can drive headphones from its balanced and unbalanced outputs." — J.Atkinson

That would seem to imply that changing output from loudspeakers to headphones requires some cable swapping, detaching the cables interconnecting the rear ports to the loudspeaker power amplifiers, and connecting an adapter cable to those rear ports to power the headphones. To me that seems like a truly shameful asnine engineering solution on a product marketed at a $22k pricepoint, a workaround that might be acceptably accomodated on some kid's entry level cheap-fi system. This should have proper headphone ports on the front panel, classic conventional 1/4 inch (6.35mm) TRS for single ended stereo headphone connection, and the newer 4.4mm Pentaconn for "balanced" (differential) stereo headphone connection, while using internal relays for the switching, and a proper control system.

Scintilla's picture
Submitted by Scintilla on February 24, 2024 - 11:11am

I know what gain matching is. I also know how to use a network analyzer. In my system, I apply room correction filter in a convolution engine and I have insertion loss so having 6db of clean, class-A gain available without resort to digital is very helpful so I can keep well under the threshold for digital overs/clipping. You do it your way and Ill do it mine. But in my system, I know which way sounds better because I have listened.

georgehifi's picture
Submitted by georgehifi on February 24, 2024 - 11:31am

Scintilla: "I can keep well under the threshold for digital overs/clipping" "George, I'm an EE"
http://tinyurl.com/225y6skz

JA: Even set to its highest output level, the Helios produced very low levels of harmonic distortion with full-scale data even into the punishing 600 ohm load"

Cheers George

Glotz's picture
Submitted by Glotz on February 24, 2024 - 10:08pm

the Swiss Army knife of digital.

After reading this review I thought that this product is a landmark in the field of modern DACs.

I can see a lot of owners feeling like pros when it comes to tailoring their sound like no other. 20 grand is a steal for what it does after considering how expensive DACs can get.

georgehifi's picture
Submitted by georgehifi on February 24, 2024 - 10:58pm

"20 grand is a steal for what it does after considering how expensive DACs can get."

Yeah, I've seen the preamp "aficionados" spend that much and more on active preamps they don't need, that are just adding more noise, colorations and distortions to the sound, that wouldn't happen if they just went direct instead.

Cheers George

Charles E Flynn's picture
Submitted by Charles E Flynn on February 25, 2024 - 2:20pm

It appears that a URL was not finished at this point in the text:

"After the Q Acoustics speakers had been returned, I used my content/kef-ls50-anniversary-model-loudspeaker">KEF LS50 minimonitors,"

John Atkinson's picture
Submitted by John Atkinson on February 25, 2024 - 3:20pm
Charles E Flynn wrote:
It appears that a URL was not finished at this point in the text: "After the Q Acoustics speakers had been returned, I used my content/kef-ls50-anniversary-model-loudspeaker">KEF LS50 minimonitors,"

Fixed. Thanks for the catch.

John Atkinson
Technical Editor/Part-time Web Monkey, Stereophile

Charles E Flynn's picture
Submitted by Charles E Flynn on February 25, 2024 - 3:29pm

You are welcome, and it is interesting to see that you are still using those KEFs.

georgehifi's picture
Submitted by georgehifi on February 25, 2024 - 7:35pm

Hi JA: Interested to know what preamp you used with the JC1 Halo's, as you didn't state if you did or went direct.?

Cheers George

John Atkinson's picture
Submitted by John Atkinson on February 26, 2024 - 4:22am
georgehifi wrote:
Interested to know what preamp you used with the JC1 Halo's, as you didn't state if you did or went direct.

I don't use a preamp, instead going direct and using the processor's digital-domain volume control. I have the Parasounds set to low sensitivity so the volume control is operating near the top of its range.

John Atkinson
Technical Editor, Stereophile

georgehifi's picture
Submitted by georgehifi on February 28, 2024 - 12:37am

http://tinyurl.com/27j9vdq7

ok's picture
Submitted by ok on February 26, 2024 - 1:07am

but digital attenuators subtract information.

Glotz's picture
Submitted by Glotz on February 26, 2024 - 9:25am

Many of them now.

ok's picture
Submitted by ok on February 26, 2024 - 1:45pm

..are less lossy according to their creators - but lossy nevertheless especially at lower volumes. I have extensively experimented with direct-to-poweramp modern dacs and I first thought they sounded clearer.. but they just sound simpler; your mileage may vary.

Glotz's picture
Submitted by Glotz on February 26, 2024 - 2:14pm

They sell a preamp module purposely for that need. To assume that modern active preamp all have undesirable levels if noise, etc. is naive. Technology has come a looong way.

Benchmark is a great example. They too, like most DAC makers, feel that their digital volume control sounds worse and measures worse than their stand alone active preamps. My experience and reviews find that totally true. DCS and others would be the exception. They clearly have spent enough and designed well enough to ignore a pre.

ok's picture
Submitted by ok on February 26, 2024 - 2:36pm

..but I generally agree that most dacs tend to sound better when connected to a decent active preamp. Added distortion is negligible, noise less so, but dynamics and nuance compensate for both.

georgehifi's picture
Submitted by georgehifi on February 28, 2024 - 4:37pm

Going through an active preamp adds less detail, more noise, more distortions and more colorations.

Cheers George

ok's picture
Submitted by ok on February 28, 2024 - 11:00pm

instant swapping between direct mode and preamplification, so nothing to worry about in case of change of heart. Anyway there are also practical reasons for going preamp. First of all I use multiple sources and outputs (cd player, turntable, dac, tape, headphone amp) which render a control center sine qua non. Moreover with no preamp buffering an accidental push of a little button in the remote can turn on the dac's full output mode and blow the whole system apart.

georgehifi's picture
Submitted by georgehifi on February 29, 2024 - 10:40pm

Even at the lowest level, the amplitude error is <1.1dB, which implies very high resolution. An increase in bit depth from 16 to 24, with dithered data representing a 1kHz tone at –90dBFS, dropped the Helios's noisefloor by 33dB (fig.6). This implies a resolution between 21 and 22 bits, which is the highest I have encountered

ok's picture
Submitted by ok on February 29, 2024 - 10:57pm

..I read and hear much better than you do. Now get off my back. I have no time for talking to a tape loop.

georgehifi's picture
Submitted by georgehifi on March 1, 2024 - 5:52pm

Yeah right!! you hear real good.

Glotz's picture
Submitted by Glotz on February 26, 2024 - 9:31am

I have read that the latest ESS ES9039MSPRO chip specifically addresses the 'IMD hump' in the past chip iteration.

What magic has Weiss wrought from a the '38 and would they see lower IMD from the new chip? Would it matter? Kudos to Weiss for creating a very special DAC.

Glotz's picture
Submitted by Glotz on February 26, 2024 - 9:44am

Hilarious name even beyond 'Schiit' and released in December of 2023. Their new flagship. It even includes MQA-CD from coax and boasts a striking new design with tempered-glass fascia. The price point $1000 and measurements I've seen could be worth a listen and test. There does seem to be one note issue with DOA power after a week or two.

That being said, I would even rather the reviewers to test and listen to the new Holoaudio Cyan 2 at $1199 (or $1350 with shipping). "Equipped with R2R network and vector step resistor network at the same time. The R2R network is used for primary digital-to-analog conversion of the PCM stream, and the vector step resistor network is used for primary digital-to-analog conversion of the DSD stream."

There is support for DSD 1024x and up to PCM 1.536M (32bit). Being more at my price point for digital, I am impressed with either DAC's details.

RustyGates's picture
Submitted by RustyGates on February 26, 2024 - 7:27pm

Hi JA, what was the RMS voltage level at the output when the impulse response was captured? I would like to know if the unit was able to rise to full 0dBFS when given an impulse sample, like the Mola Mola Tambaqui.

John Atkinson's picture
Submitted by John Atkinson on February 27, 2024 - 7:27am
RustyGates wrote:
Hi JA, what was the RMS voltage level at the output when the impulse response was captured?

The Weiss's maximum output voltage was set to "-12dB" (4.1V RMS) for the impulse response measurement.

RustyGates wrote:
I would like to know if the unit was able to rise to full 0dBFS when given an impulse sample, like the Mola Mola Tambaqui.

It didn't have a problem in this regard.

John Atkinson
Technical Editor, Stereophile

bracondale's picture
Submitted by bracondale on February 29, 2024 - 4:46am

at about $20k I would want to hear the forthcoming Grimm.

weiss2496's picture
Submitted by weiss2496 on March 1, 2024 - 10:04am

Just a short comment on how good a digital level control can be these days. The quality depends on the quality of the D/A Converter. It should be as linear as possible and as low noise as possible. With today's high quality DACs the digital level control can yield extremely good results.
Of course dithering noise has to be applied. Dithering means adding a low level noise signal to the audio signal and then do the quantisation. Quantisation is required because the input word-length of a DAC is not high enough to accommodate for the output after the digital multiplication (i.e. the level control). As an example - if we have a 16 Bit audio signal and multiply that with a 16 Bit gain factor the result is 32 Bit long. For a 16 Bit D/A converter those 32 Bits have to be reduced to 16 Bits. This can be done by cutting them off (i.e. truncation which is bad) or by adding that dithering noise before cutting the Bits off. Dithering causes the quantisation error to become de-correlated from the audio signal and becoming pure noise.
Here are three listening examples with an 8 Bit (!) quantisation.

This example cuts off the bits after the level control to a mere 8 bits without adding dither noise:
http://www.weiss.ch/linked/digital-level-control/nodither.mp3
The quantisation distortion can be easily heard.

This example adds dither noise before the 8 bit quantisation:
www.weiss.ch/linked/digital-level-control/flatdither.mp3
Already much better. The music is undistorted, just noisy because of the 8 bit resolution. The noise is not modulated by the music.

This example adds dither using a shaped noise, i.e. a noise which has a lower level at frequencies which can be heard and higher levels and very high frequencies:
www.weiss.ch/linked/digital-level-control/shapeddither.mp3
An astonishing quality considering that the resolution is 8 bits (256 level steps).

Daniel Weiss

Glotz's picture
Submitted by Glotz on March 3, 2024 - 10:46am

SQ depends on the quality of the D/A converter. Price of the product factors in heavily, I would assume.

georgehifi's picture
Submitted by georgehifi on March 4, 2024 - 1:03am

And it doesn't depend on cost, I've seen the same ESS ES9038PRO d/a converter used in the $899usd Topping D90SE dac with stunning objective measurements.
https://tinyurl.com/22rpo5bw
And then also used in the $40,000usd Gryphon Ethos D/A processor, 44 x the price, would be interesting to do a subjective listening A/B between the two.

Cheers George

X