Review and Measurements of Benchmark AHB2 Amp

[Music1969]

No this is not a problem at all. Capacitive or inductive speaker loads will not cause stability issues.

From the amplifiers perspective, -45 degrees at 4 Ohms is equivalent to driving a 2.8-Ohm resistive load. This can be driven to full output voltage in stereo mode or in mono mode.

-50 degrees at 4.1 Ohms is equivalent to driving a 2.6-Ohm resistive load. This can be driven to full output voltage in stereo mode and can be driven to within 0.1 dB of full output in mono mode without flashing the TEMP lights. Even in mono mode, it would be virtually impossible to trip the protection with music although it could be done with a 100 Hz sinusoidal test tone if it was played at, or slightly above, full power for a few seconds.

Wow, thanks for all this info John. Amazing product and thanks for taking the time to share so much info here.

 

[SIY]

My question to @John_Siau re the first watt white paper- the distortion residual seems to show that the bias was incorrectly set, since there's a spike at zero crossing. With the bias at an optimal level, the distortion residual maxima ought to be happening at the gm doubling points. What am I missing?

@John_Siau Not sure if my question got lost in the morass. Am I looking at this incorrectly?

 

[restorer-john]

Frequency response is exceptional too as expected:

-0.3dB @20KHz is hardly 'exceptional' in a power amplifier idling along at just 5W@4ohms...

 

[amirm]

-0.3dB @20KHz is hardly 'exceptional' in a power amplifier idling along at just 5W@4ohms...

It is 3 dB down at 200 khz John.

 

[restorer-john]

My question to @John_Siau re the first watt white paper- the distortion residual seems to show that the bias was incorrectly set, since there's a spike at zero crossing. With the bias at an optimal level, the distortion residual maxima ought to be happening at the gm doubling points. What am I missing?

The lovely depiction below that conveniently omits the most important text stating "distortion products depicted are not to scale"?

Does Joe Average know that the distortion analyzer outputs a typical 1V for FSD on whatever scale it is on? Does Joe Average know the CRO's top trace is 2V per division and the bottom trace range could be anything to get a nice visual of those distortion products for marketing purposes?

To be fair, it, and the text below it was the only thing in that 'white paper' that annoyed me.

Edit: it has been pointed out to me that somewhere in the article, there is clarification of the 60dB amplification of the residual.
Edit2: It seems there are two articles, one mentions the 60dB gain, the other does not.

 

[restorer-john]

It is 3 dB down at 200 khz John.

I have plenty of power amplifiers that have -3dB points well above that, but they don't start drooping off at midrange frequencies to be 0.3dB down when still in the audio band. 20-20Khz should be ruler flat at such low powers.

I'm happy to run some 5W@4ohm 20-20K plots over the weekend for comparison if you like. That I can do quickly and easily with my old AudioLab hardware.

Don't get me wrong, I think the Benchmark amplifier is a lovely product and I think the involvement of John here is absolutely wonderful. That said, I am merely bringing some historical perspective into this, these numbers are not earth shatteringly better than everything that came before.

 

[amirm]

I'm happy to run some 5W@4ohm 20-20K plots over the weekend for comparison if you like. That I can do quickly and easily with my old AudioLab hardware.

How accurate is the Audiolab with respect to flatness of frequency response?

 

[Daverz]

So do we need to abandon the Audio Critic maxim of (I'm paraphrasing from memory), "If the input impedance is high, the output impedance is low and the distortion is low, you can't tell the difference between amps."?

 

[GrimSurfer]

So do we need to abandon the Audio Critic maxim of (I'm paraphrasing from memory), "If the input impedance is high, the output impedance is low and the distortion is low, you can't tell the difference between amps."?

Ummm, yup.

Albert Einstein once said that it is important to make things simple, but not too simple.

 

[Daverz]

Found the full quote:

As I have said, and written, innumerable times, any two amplifiers with high input impedance, low output impedance, flat frequency response, and sufficiently low distortion and noise will sound exactly the same at matched levels if not clipped.

http://www.biline.ca/audio_critic/audio_critic_web2.htm#ba

If "sufficiently low distortion" means as low as the ABH2, then in practical terms all amps will sound different.

 

[Blumlein 88]

Found the full quote:

As I have said, and written, innumerable times, any two amplifiers with high input impedance, low output impedance, flat frequency response, and sufficiently low distortion and noise will sound exactly the same at matched levels if not clipped.

http://www.biline.ca/audio_critic/audio_critic_web2.htm#ba

If "sufficiently low distortion" means as low as the ABH2, then in practical terms all amps will sound different.

Well when your speaker dips to .75 ohm in the treble that output impedance will need to be small. Yes my speakers do that.

 

[Daverz]

Well when your speaker dips to .75 ohm in the treble that output impedance will need to be small. Yes my speakers do that.

Right, that's where the "if not clipped" caveat comes in.

 

[GrimSurfer]

It all depends on what you think is the most plausible:

- Two amps, whose basic measurements are the same, sounding identical; or

- Two amps, whose basic measurements are the same, having different components, signal paths, topologies, operating temperatures, parts and sub assemblies of differing quality etc. sounding, to somebody with a trained ear, different.

Now setting aside those issues for a moment, what is the timeline for comparison? Listened to for five minutes? Tested after a few hours of use? Retested after a few years of use? Does durability and stability not count anywhere in the equation?

And even if you get past that, do we look for tests and conditions that favour one amp over the other (because it can't compete at a certain output or operating profile)? Do we hook up the better equipped amp with RCA connections instead of balanced ones because the cheaper one doesn't have them? Select 8 ohm speakers because one cannot drive 4 ohms reliably?

And what do you mean by flat frequency response... over what bandwidth? And what do you mean by sufficiently low distortion? What constitutes "sufficiently close" and how does this render the comparison meaningful or meaningless?

The quote, it would appear, is an over simplification of a very complex issue. Like any over simplification, it's shortfalls and discontinuities become apparent when examined in practical terms.

And if you believed any of this, why did you buy a Bryston SST when any number of amps with the same basic output, gain, distortion specs etc. can be found for FAR less?

 

[SIY]

Well when your speaker dips to .75 ohm in the treble that output impedance will need to be small. Yes my speakers do that.

The guy I was visiting last night is a fan of Wilson speakers, which are infamous for that. In my mind, that's a pathologically bad design, but if one is committed to them, yes, a more elaborate power amp is needed than for driving what I would consider better engineered speakers. The amps I use wouldn't hack it.

With speakers having a non-pathological impedance curve, Aczel's maxim remains true.

 

[SIY]

The lovely depiction below that conveniently omits the most important text stating "distortion products depicted are not to scale"?

However, the text then states:

"With an audio analyzer we can extract the error waveform (lower trace) and amplify it so that we can see what our ears are hearing. In this case, the error signal has been amplified by 1024 (about 60 dB). "

My question (still not answered) is why the waveform looks as it does, when in an AB amp with properly adjusted bias, as far as I'm aware, that spike at zero crossing shouldn't be there and the limitation should be gm doubling.

 

[restorer-john]

How accurate is the Audiolab with respect to flatness of frequency response?

For fixed level FR plots, it's really accurate. 9 (manual software controlled hardware switched) ranges from 100mV to 100V RMS with 8 bit sampling (256 levels per range) and a true RMS converter, 492 sample points in a 20-20K log sweep. Digitally generated swept sine, variable level. Linear and log sweeps with step and start/stop freqs. It doesn't care about the actual frequency of the sweep signal- it measures the amplitude of it. Because the only signal 'should' be the one it generates and it knows what that is and is synced to it.

To give you an idea, these are a few plots of one of my Denon preamplifiers (PRA-1000) and a Sony preamplifer (TAE-77esd) channels offset for clarity at around 1V output. FR deviation of those preamps is virtually one LSB's worth (1 pixel=10mV on the 2.5V plot). These plots are 8 years apart and the calibration routine had just been run IIRC.

The loop-back is hardly flatter than that Denon!

There's a 10-20KHz loopback on the 1V scale, DC coupled. The cross hair shows a few pixels above 0.5V, so a 4mV approx deviation which is essentially a run to run variation.

Certainly no AP, but I've had the old girl since 1996 and what it does actually do, it does well.

AudioLab is tethered (via serial RS-232) to an old W95 box that sits on my network in the cupboard below my bench, so I can pull the data files off it. I've recently (thanks to Blumlein88) been able to finally get the raw data values into excel to create damping factor vs frequency plots externally (doing an unloaded vs loaded sweep). It can also measure capacitance, inductance and do impedance sweeps.

For distortion I use a Leader LDM-171 (quick tests to -80dB max THD/S/N), my Panasonic VP-7723a or PC sound card front ends for low level FFT. Other gear on my bench are a few CROs, A DSO, a Kenwood FL-140 W&F meter for TTs and Tape Decks, Freq counters, PSUs, audio oscillators, a Panasonic VP-8177A Sig Gen, DMMs and some hot, pointy ended things for melting holes in plastic.

 

[restorer-john]

"With an audio analyzer we can extract the error waveform (lower trace) and amplify it so that we can see what our ears are hearing. In this case, the error signal has been amplified by 1024 (about 60 dB). "

OK, there are two articles, the first watt mentions the 60db gain, the feedforward article does not.

https://benchmarkmedia.com/blogs/application_notes/feed-forward-error-correction

 

[RichB]

Here is the Stereophile measurements for the Parasound JC-1 measured into 8, 4, and 2 Ohms.
https://www.stereophile.com/content/parasound-halo-jc-1-monoblock-power-amplifier-measurements

It looks to be about 0.2dB down at 20kHz.
Not that I wouldn't like to see lower output impedance from the AHB2, but they sure sound good, and yes different, in a good way.

For $58,000 you can get the Verity Audio Monsalvat Amp-60 power amplifier (that's a mouthful):

https://www.stereophile.com/content/verity-audio-monsalvat-amp-60-power-amplifier

This is 1 watt and it's down 0.25dB.
Here is the notched out crossover distortion:

Did I mention, this is a 60 WPC amp?

- Rich

 

[Blumlein 88]

The guy I was visiting last night is a fan of Wilson speakers, which are infamous for that. In my mind, that's a pathologically bad design, but if one is committed to them, yes, a more elaborate power amp is needed than for driving what I would consider better engineered speakers. The amps I use wouldn't hack it.

With speakers having a non-pathological impedance curve, Aczel's maxim remains true.

Well that is one of the reasons I think class D amps work well with my Soundlabs. They have such a small output impedance. But there is the output filtering too. And I agree wholeheartedly about it being a pathological design.

 

[Blumlein 88]

It all depends on what you think is the most plausible:

- Two amps, whose basic measurements are the same, sounding identical; or

- Two amps, whose basic measurements are the same, having different components, signal paths, topologies, operating temperatures, parts and sub assemblies of differing quality etc. sounding, to somebody with a trained ear, different.

Now setting aside those issues for a moment, what is the timeline for comparison? Listened to for five minutes? Tested after a few hours of use? Retested after a few years of use? Does durability and stability not count anywhere in the equation?

And even if you get past that, do we look for tests and conditions that favour one amp over the other (because it can't compete at a certain output or operating profile)? Do we hook up the better equipped amp with RCA connections instead of balanced ones because the cheaper one doesn't have them? Select 8 ohm speakers because one cannot drive 4 ohms reliably?

And what do you mean by flat frequency response... over what bandwidth? And what do you mean by sufficiently low distortion? What constitutes "sufficiently close" and how does this render the comparison meaningful or meaningless?

The quote, it would appear, is an over simplification of a very complex issue. Like any over simplification, it's shortfalls and discontinuities become apparent when examined in practical terms.

And if you believed any of this, why did you buy a Bryston SST when any number of amps with the same basic output, gain, distortion specs etc. can be found for FAR less?

Maybe it is because in the Swedish LTS series amplifier testing only one amplifier has ever been straight wire transparent in both sighted and blind listening and that amplifier is a Bryston SST.

http://mjaudiolab.pl/images/stories/Bryston/swedish14bsstreview.pdf

http://www.sonicdesign.se/amptest.htm