Audio Research LS5 preamplifier & BL2 input controller Measurements

Sidebar 2: Measurements

BL2 Input Controller: The BL2 had a very high input impedance of 1.14M ohms, a little lower than the specified 1.5M. At these high impedances, however, an accurate measurement is difficult to obtain. Output impedance was a very low 55 ohms across the band.

The BL2's gain into a 100k load was 5.93dB, fractionally under the 6dB to be expected from converting a single-ended signal into a balanced signal. DC offset levels were quite low, measuring 4.3mV (left channel) and 1.5mV (right channel). The input overload point was greater than 13V RMS.

The BL2 had perfectly flat frequency response all the way past the Audio Precision's 200kHz upper frequency limit. At 200kHz, the response was down 0.27dB. The BL2's channel separation was excellent, the crosstalk measuring nearly –120dB at 1kHz, and lower than –90dB at any audio frequency. Similarly impressive was the BL2's low noise level; I measured an unweighted S/N ratio of 93.4dB (left channel) and 96dB (right). (All S/N measurements were made with 22Hz–22kHz bandwidth and referenced to 1V output at unity gain.) These figures increased to 96dB and 99.6dB, respectively, when A-weighted. The BL2's noise and distortion levels (fig.1) were also extremely low—less than 0.003% across most of the band, with a slight rise above 2kHz.


Fig.1 Audio Research BL2, THD+noise vs frequency (right channel dashed).

Overall, the BL2 had textbook measurements—very low noise, extremely wide bandwidth, high channel separation, and vanishingly low distortion.

LS5 Mk.I preamplifier: I measured an input impedance of 197k ohms for the LS5, this within measurement error of the specified 200k ohms. I wondered why Audio Research chose an extremely high input impedance for the BL2 (1.5M ohms), but a much lower (but still highish) input impedance of 200k ohms for the LS5. The LS5's output impedance was a moderate 415 ohms, suggesting it should easily drive any power amplifier input. The output impedance measured at the tape-out jacks was a rather high 2k ohms. DC levels were a very low 0.4mV from the left and right channels, respectively.

I measured a gain of 29dB with the gain switch in the "+30" position, 11.3dB in the "+12" position. Although 29dB is a lot of gain for a line stage, the LS5 sounded its best in this position. The "+12" setting should therefore be considered a convenience when using very-high-output source components.

The volume-control tracking was excellent, measuring just 0.2dB channel imbalance at the low end of the scale (four detents up from fully counterclockwise)—a figure that decreased to 0.06dB over most of the rest of the range. This measurement, which shows how well-matched the left and right elements are in the volume control, implies that the matching is similarly good between the + and – halves of the balanced signal controlled by the four-element potentiometer. This is critical in a balanced preamplifier; any gain mismatch between phases of the balanced signal will decrease common-mode rejection.

The LS5's frequency response (fig.2) was flat within the audio band, with a 0.25dB rolloff at 50kHz. Channel separation (fig.3) was not as high as I measured from the BL2, but was still excellent. The center pair of traces (one solid, one dotted) are the LS5's channel separation when the unit is selected for 30dB of gain. The second pair are the LS5's channel separation when providing 12dB of gain. As you can see, the right-channel separation decreased (by nearly 10dB) in the low-gain mode, while the left-channel separation increased.


Fig.2 Audio Research LS5, frequency response (right channel dashed, 0.5dB/vertical div.).


Fig.3 Audio Research LS5, crosstalk (from top to bottom): R–L, 12dB of gain; R–L, 30dB of gain; L–R, 12dB of gain; L–R, 30dB of gain (10dB/vertical div.).

Measuring the S/N ratio in the 12dB gain mode revealed only slightly poorer performance in the right channel (76.9dB right, 78.2dB left, unweighted). In the "+30" position, the S/N ratio measured 81dB in both channels—a figure that increased to 84dB when an "A"-weighting filter was applied. The bottom pair of traces in fig.4 reflect the LS5's THD+N with 30dB of gain, the upper traces with 12dB of gain. As you can see, the level of spuriae increased in the lower gain position, dramatically so in the right channel (dotted trace). This correlates with the apparently higher right-channel crosstalk in the "+12" mode, and somewhat with the decreased S/N ratio in the right channel with 12dB of gain selected.


Fig.4 Audio Research LS5, THD+noise vs frequency with (from top to bottom): 12dB of gain; 30dB of gain (right channel dashed).

To further investigate the difference in measured performance and sound quality between the two gain modes, I looked at the LS5's distortion spectrum in both gain settings. Fig.5 is an FFT-derived spectrum of the LS5's output when driving a 50Hz sinewave into a 100k ohm load at 1V RMS with 30dB of gain selected on the LS5. For comparison, fig.6 represents the same test conditions, but with 12dB of gain in the LS5. Although they're quite similar, in that the predominant harmonics present are the third and fifth, the power-supply harmonics at 180Hz, 300Hz, and 540Hz are higher in level in the 12dB gain mode, confirming that the difference in the fig.4 traces was due to increased noise rather than to distortion.


Fig.5 Audio Research LS5, 30dB of gain, spectrum of 50Hz sinewave, DC–1kHz, at 1V into 100k ohms (linear frequency scale). Note that the third harmonic at 150Hz is the highest in level, at –97dB (0.0014%).


Fig.6 Audio Research LS5, 12dB of gain, spectrum of 50Hz sinewave, DC–1kHz, at 1V into 100k ohms (linear frequency scale). Note that the power-supply components at 180Hz, 300Hz, and 540Hz are higher in level than any distortion harmonics.

To see how the LS5 behaved when driving a lower impedance, I performed the same test into 600 ohms with 30dB of gain (fig.7). The distortion spectrum changes, with the third-harmonic dramatically increasing in level and the second harmonic making an appearance. Loads this low are best avoided.


Fig.7 Audio Research LS5, 30dB of gain, spectrum of 50Hz sinewave, DC–1kHz, at 1V into 600 ohms (linear frequency scale). Note that the third harmonic at 150Hz is the highest in level, at –76dB (0.015%).

LS5 Mk.II preamplifier: I performed a complete set of measurements on the Mk.II, but have reported on its measured performance only when it differed from that of the LS5.

The Mk.II's volume-control tracking was amazingly good, eclipsing even the superb performance of the potentiometer used in the LS5. The best case was with the volume control barely open (1V in, 100mV out), when the left-right tracking was within 0.005dB. At any other volume setting, the left-right volume tracking was better than 0.008dB. This virtually perfect left-to-right tracking suggests that the Mk.II's four-element potentiometer is equally precise between the two halves of the balanced signal—a vital factor in a fully balanced preamplifier.

The LS5 Mk.II measured significantly better than the original in some key respects, the biggest difference being between the two units' signal/noise ratios. The LS5's unweighted S/N ratio was 76.9dB (right) and 78.2dB (left) (referenced to 1V output), the Mk.II's 94dB (left) and 96.7dB (right)—a whopping 19.8dB improvement (ie, a tenfold reduction in noise). (All measurements for both the LS5 and the LS5 Mk.II were made over a 22Hz–22kHz bandwidth.)

This vastly better noise performance was reflected in the Mk.II's THD+noise plots (fig.8). In the 30dB gain mode, the THD+N was 0.004%; with 12dB of gain, the distortion and noise was 0.016%—nearly an order of magnitude lower distortion and noise compared with the LS5. A comparison of the harmonic distortion spectra revealed the Mk.II's to be nearly identical to that of the LS5.


Fig.8 Audio Research LS5 Mk.II, THD+noise (%) vs frequency set to 30dB gain (bottom) and 12dB gain (top) (right channel dashed).

Interchannel crosstalk decreased slightly overall in the Mk.II (fig.9), and didn't exhibit the odd disparity between gain modes measured in the LS5. In the LS5 sample, the right-channel crosstalk decreased in the low-gain mode compared to the high-gain mode.


Fig.9 Audio Research LS5 Mk.II, channel separation set to 30dB gain (bottom) and 12dB gain (top) (10dB/vertical div.).

The LS5's and LS5 Mk.II's input impedances, output impedances, gains, frequency responses, and DC levels were all similar.

I was greatly impressed by the Mk.II's much better technical performance—particularly the superb volume-control tracking and the 20dB increase in the S/N ratio.—Robert Harley

Audio Research Corporation
6655 Wedgwood Road N., Suite 115
Maple Grove, MN 55311
(763) 577-9700

JRT's picture

(edited to clarify)

More properly noise sums as RMS sums of the contributions of the various sources of noise.

Random noise separately generated on different signal paths does not correlate in a downstream difference sum, so does not cancel.

Within an amplifier you get linear distortion and nonlinear distortion and inherent noise (also a nonlinearity), and in the many ways that these are modulated and shaped and combined.

This inherent noise is the noise that I am referring to. I am not referring to external interference.

Jim Austin's picture

Is your argument semantic? Are you questioning the use of the word "noise?" Whatever you want to call it, noise or something else, if an unwanted signal is induced--eg, due to changing e/m fields in the vicinity of a cable--is the same or very similar in the two wires, then the noise--or the portion of it that is the same--will cancel.

This is a well-established, widely accepted advantage of balanced connections. It is why balanced connections and circuits are ubiquitous in the pro-audio world, where longer interconnections are often needed.

Jim Austin, Editor

JRT's picture

To contend with EMI... Using balanced interconnection, a differential pair of conductors with impedance balanced (very similar) with respect to ground is designed such that the varying electromagnetic field through which the interconnection cable passes will induce similar currents on the pair, and those currents passing through those two similar impedances will cause similar change in signal voltage on the pair. In this case, being similar is a combination of one fraction of that change in signal being exactly the same on both conductors, and the remaining fraction of the change in signal being different on both conductors. The fraction of the similar change that is exactly the same is in the common mode and the remaining fraction is in the differential mode. The balanced interconnection terminates downstream into a differential input, which largely cancels the common mode, rejecting it, and passes the differential mode though the input of the amplifier.

Key in that balanced interconnection is in managing the unwanted change by getting most of it into the common mode, equal, same, so that the differential input can reject most of that common mode. The balanced interconnection most usually utilizes STP, shielded twisted pair. That is a pair of insulated wires twisted and wrapped with a shield covered by an outer jacket. The grounded shield reduces the EMI entering the pair of wires. The The EMF carrying the EMI induces current on the pair of wires, and the twist and the balanced impedance causes the induced currents to be more nearly equal. That EMI current through the balanced impedance causes the related change in signal voltages to be nearly equal on both wires, places most of the unwanted change in the common mode. The differential input passes the difference and blocks most of the common mode.

That described above is not the largely random inherent noise that I was referring to.

The inherent noise generated within the amplifier is largely a random nonlinearity. A differential pair of paths through a bridged dual mono pair of amplifiers will generate different random noise on each of the pair. Take the difference at the output and the random noise does not cancel, because it is random, without coherent phase, uncorrelated in the difference sum.

It is very much more complicated than this gross simplification. There are a lot of different types of noise, and any of it can be further modulated and shaped and combined. It is a big subject suitable for graduate level coursework, and I am not suitably qualified to author those textbooks.

Somebody more like Bruno Putzeys could teach all of us much.

Bogolu Haranath's picture

Speaking about Bruno Putzeys ....... The new NAD M33 integrated amp being shown in CES 2020, uses the new Purifi's 'Ultraquiet Amplification Technology' ....... M33 is scheduled to be available this spring 2020 ......... Jim Austin could review it :-) ......

John Atkinson's picture
JRT wrote:
A differential pair of paths through a bridged dual mono pair of amplifiers will generate different random noise on each of the pair. Take the difference at the output and the random noise does not cancel, because it is random, without coherent phase, uncorrelated in the difference sum.

If the noise in the two signal paths is uncorrelated but truly random, you still get some noise reduction in the differential output signal. This is because the sum of the uncorrelated random noise is 3dB greater than the individual noise signals compared with the 6dB increase in the audio signal level. In effect, therefore, you get a 3dB reduction in the noise level.

John Atkinson
Technical Editor, Stereophile

Jim Austin's picture
that you were disagreeing with something Harley wrote in the review. As far as I can tell, what you and Harley wrote are consistent. Jim Austin, Editor Stereophile
JRT's picture
Jim_Austin wrote:

I assumed that you were disagreeing with something Harley wrote in the review. As far as I can tell, what you and Harley wrote are consistent.

I took issue with something Harley had included in that article, which I thought was unintentionally misleading, and was feeding what I perceive to be a common misperception.

A dual mono system bridged to differential will cancel what is shared in common, but that will be mostly some of the 2nd order harmonic distortion, and almost none of the inherent noise generated within those separate gain stages. In reaction to his comment quoted below, I was only trying to point out that the inherent noise is not shared similarly in common, so does not really cancel.

Robert_Harley wrote:

The advantages of a preamplifier being truly balanced are the elimination of two active stages (the differential amplifier at the input and the phase splitter at the output) and the fact that any noise or distortion common to both phases of the balanced signal will cancel when combined.

It is true what he said that any noise shared in common will cancel, but it is also misleading because the vast majority of the inherent noise is not shared in common, not enough to matter much, because those two sets of inherent noise are randomly generated inside two separated amplifier functions prior to the differential sum at the end.

Bogolu Haranath's picture

According to Benchmark blog (see, balance connections), common mode noise is 3rd harmonic (odd-order) :-) .......

JRT's picture
Bogolu_Haranath wrote:

According to Benchmark blog (see, balance connections), common mode noise is 3rd harmonic (odd-order)

Do you have a link to that reference? I would like to read it.

I suspect that Benchmark, in referring to "3rd order" or "odd order", likely used the word "distortion" and did not use the word "noise", as nonlinear distortion and noise are two different categories of nonlinearities, and they very well understand that. Nonlinear distortion can also ride the common mode, not just noise.

Bogolu Haranath's picture

Jim Austin posted a link to Benchmark's blog in one of his replies on Benchmark LA4 pre-amp review web page :-) .......

Bogolu Haranath's picture

According to many audiophiles 2nd order harmonic distortion is quite pleasing to the ear :-) .......

Bogolu Haranath's picture

If you are talking about 'active noise cancellation' (ANC) technology used in some headphones ....... they use small microphones to pick up noise and send a 180 degree opposite phase signal to cancel the noise :-) .......

JRT's picture

in this.

Bogolu Haranath's picture

Your initial comment did not clearly specify ...... So, I posted my comments about ANC in headphones :-) .......

Bogolu Haranath's picture

BTW, as an additional note ..... Some automobiles are also using somewhat similar ANC technology to cancel the noise :-) .......

Bogolu Haranath's picture

Most of the headphone audiophiles don't like potential ANC interference for the FR of music anyway .......... They prefer 'passive noise isolation' instead, if needed :-) .......

tonykaz's picture

Hmm, I wonder.

If this thing is as good as the reviewer's adjectives it'll still be in his music system today. Is it?


does ARC gear go off as easily and quickly as it's tubes performance.

Reading stuff like this has me pondering who wrote it, probably an aspiring Marketing Exec.

I've taken in a wide range of ARC gear in Trade, none of it was anywhere near as good as the Product Reviewer's claims.

ARC is nice gear, I could easily live with an SP6 B thru E. but it's not as lovely as an Audible Illusion Modulus.

Tony heading for Home and a nice 83F swim in the pool.

ps. even so, I wish that William Z was still around.

spivechild's picture


JRT's picture

I find that the print editions of magazines are very compatible with the system that I use for reading those, which is me. I prefer to read and turn real pages rather than virtual pages. And if I hang onto some print magazines for a few decades, I don't have to be concerned about future incompatibility with reader software and future operating systems, only the space claim of the print magazines (which for me is admittedly considerable). If I cannot read the print edition, then I probably cannot read the computer screen.

Bogolu Haranath's picture

Is your computer screen 8k ultra, ultra hi-res, Dolby vision, IMAX enhanced? ....... Just kidding :-) ........

spivechild's picture

Says the guy responding in the comment section on a webpage.

New decade, new Stereophile. Let’s concentrate on.....print media! The only way to make sure Stereophile is flat line once all the baby boomers are dead or have cataracts which shouldn’t be long.

Jim Austin's picture
Your message has been received, loud and clear, by the editor and the publisher of the magazine--mainly because of the email you sent. The rest is spam, so please refrain from multiple postings in future. Jim Austin, Editor Stereophile
Ortofan's picture

... truly necessary, in that same era alternatively one could have purchased the $1200 Onkyo P-388F.
The P-388F also had unbalanced inputs (and outputs) - without the need for a separate box - as well as separate phono sections for MM and MC cartridges. For the price of the ARC unit, one could have afforded to include the M-388F power amp plus the DX-788F CD player.

OTOH, if you needed to spend the entire amount on a line stage with balanced inputs and outputs, then there was the Accuphase C-250.

Bogolu Haranath's picture

LS5 has just a touch of 3rd harmonic distortion ........ See, Fig.7, measurements ....... Do the Onkyo and Accuphase have that? :-) ........

Jason Victor Serinus's picture

Is everyone aware that this review is 26 years old?

Bogolu Haranath's picture

"The more things change, the more they remain the same" ........ Alphonse Karr :-) ........

JHL's picture

Surely everybody knows that musical pleasure derives directly from beating digital Stereophile about the head, neck, shoulders, head, face, and neck; that Bogu never, ever logs out; and that pedantry itself was invented in these comments threads.

Whatever joy an Audio Research component gave 26 years ago is completely irrelevant to the world in general, my good man.

Why, just ask us.

Ortofan's picture

.. joy?
Having owned the ARC preamp for the past quarter century or having opted for the far less expensive Onkyo unit and having invested the $5K price difference in an S&P 500 index fund, where that $5K would have grown to about $25K today?

Bogolu Haranath's picture

Money alone can't always bring happiness ....... OTOH, that $25K could buy the new ARC Ref. 160S ..... Not a bad idea :-) ........

JHL's picture

speculating on hindsight beats hifi hands down. Vaguely reminds me of the guy hectoring others with the wholly superior sound of the loudspeaker he's never built, the rubes.

Never change, digital S'phile commentariat, never change.

Ortofan's picture

... none of the above.

The stereotypical audiophile would never be satisfied with the less expensive piece of equipment (and letting the excess funds accrue interest in some account), nor would they have kept the ARC unit(s) for several decades.

Instead, they would have continually failed to summon up the courage to step off the perpetual trade-in/trade-up merry-go-round for fear of possibly missing out on something better.

Bogolu Haranath's picture

If we humans stopped looking (and making) for 'something better', we would still be living in caves (not wine caves) :-) ......

JHL's picture

As if it's still somehow not clear, digital S'phile commentariat are the only thing standing between order in the universe *and people doing what they want with their money*.

Oh the humanity.

Jim Austin's picture

>>As if it's still somehow not clear, digital S'phile commentariat are the only thing standing between order in the universe *and people doing what they want with their money.

So you think that if it were not for the commentariat, people would not do what they want with their money?

Jim Austin, Editor

JHL's picture

I doubt the chronic second-guessing contingent sees far enough past its black and white preoccupation with perceived sins and ills to realize normal people aren't listening. To it, I mean.

Glotz's picture

They still rock!

I've heard these ARC power amps several times, and they are completely musical and throw a pretty convincing illusion. They may have possessed a slightly whitish character with some noticeable noise vs. more modern equipment, but they are testament to the ARC lineage and it seems like splitting hairs for arguments' sake.

They were fantastic for it's time, even more so with the Ref600's! I heard this alongside the MP-1 prototype around the same time (the no display/switch-only version in '95/'96). I can't remember the exact name of it, but I think JGH reviewed it? (Searching for that is a 'pita', btw.)