Review and Measurements of Benchmark AHB2 Amp

[restorer-john]

It's more complicated than that, an ordinary class A/B amplifier biased heavily in class B would be unlikely to have stellar THD performance without a healthy margin over the theoretical limit on slew rate. The internal slew rate in this case is larger than the output slew rate at the zero crossing. I suspect (not really, I know) the error correction schemes in the Benchmark are necessary to compensate for this. If one simply does not have quiescent power as a design parameter there are other paths to sub-ppm distortion amplifiers. No criticism implied or intended, just that if you allow for 50W quiescent power in a 100W amp it is easier to achieve linearity before any feedback or error correction.

Scott, the post above simply asked whether the AHB-2 rated amplifier slew rate of 16V/uS was "sluggish".

Clearly, the 16V/uS quoted is surely an outer loop (overall) worst case number for the AHB-2 amplifier as a whole, wouldn't you say? Wouldn't you consider 16V/uS overall (line in to speaker out) sufficient at large outputs to reproduce content in the audible bandwidth without slew-induced distortion rearing its head?

Other feedforward/'non-nfb'/direct distortion cancelling power amplifiers from the past specified huge numbers for slew rate and not just internal. My Denon POA-1500 is +/-400V/uSec from main in to speaker out, a S/N of 123dB and a FR of 1Hz-300KHz (+0/-3dB @ 1W) THD of <0.002% and IMD <0.0015% (60Hz/7KHz 4:1 at full power). That needed 150MHz fT transistors in the front end/driver and 35MHz output stages and it was made in 1983!

It uses a most interesting design and I wonder how it differs or is similar to other systems like the Benchmark? Perhaps you can comment on the topology and any similarities, this amplifier has always intrigued me (I have few of them) and apparently caused more than a few techs and designers to scratch their heads back in the day.

Schematic is attached, VAS is virtually a complete amplifier on its own. There are a few small schematic drawing errors in the power stage- ignore them, they don't translate to the actual amplifier.

 

[scott wurcer]

Scott, the post above simply asked whether the AHB-2 rated amplifier slew rate of 16V/uS was "sluggish".

Clearly, the 16V/uS quoted is surely an outer loop (overall) worst case number for the AHB-2 amplifier as a whole, wouldn't you say? Wouldn't you consider 16V/uS overall (line in to speaker out) sufficient at large outputs to reproduce content in the audible bandwidth without slew-induced distortion rearing its head?

It will take time to read the attachment to make any informed comment. As to the first point no, the input error signal to the closed loop amplifier is what matters. If there is heavy class B bias the internal node (the input to the output buffer) can (must) slew faster than the actual closed loop output voltage to cover the discontinuity at 0. This is of course for an ordinary amplifier architecture, the Benchmark has correction for this.

 

[DonH56]

It's more complicated than that, an ordinary class A/B amplifier biased heavily in class B would be unlikely to have stellar THD performance without a healthy margin over the theoretical limit on slew rate. The internal slew rate in this case is larger than the output slew rate at the zero crossing. I suspect (not really, I know) the error correction schemes in the Benchmark are necessary to compensate for this. If one simply does not have quiescent power as a design parameter there are other paths to sub-ppm distortion amplifiers. No criticism implied or intended, just that if you allow for 50W quiescent power in a 100W amp it is easier to achieve linearity before any feedback or error correction.

Yes, of course, agreed. I was just answering the question and adding a bit. I don't think my addition was wrong but I've been wrong before. I also agree that Benchmark's feedforward circuit probably compensates for the (hypothesized) heavy class-B bias (not disagreeing, don't know) and allows a lower net (external) slew rate.

I tried a conceptually similar FF circuit in an RF design some years ago but it was for a pulse application, thus needed very high slew rate and bandwidth, and aligning the FF circuit's output to the main signal path over the required bandwidth and PVT (process, voltage, temperature) variations was difficult. It worked but did not aide as much as hoped.

Which just goes to say I haven't your expertise with audio circuits.

 

[restorer-john]

It will take time to read the attachment to make any informed comment.

VR3 and VR5 are for trimming for lowest THD at 1KHz and 20KHz. Look forward to any comments, Scott.

 

[Blumlein 88]

So Spectral's 500 v/microsecond and 600 v/microsecond slew rates with a megahertz or two of bandwidth weren't good? They had me fooled.

 

[scott wurcer]

VR3 and VR5 are for trimming for lowest THD at 1KHz and 20KHz. Look forward to any comments, Scott.

There is way too much going on in that schematic to just look at it and make any definitive comment. There's some kind of local feedback one at low frequency and one at a higher frequency. Picking it apart and doing a simulation would probably give one a feel for what is basically going on.

 

[Sal1950]

 

[restorer-john]

View attachment 29959

That is just great- I love that cartoon!

 

[DonH56]

So Spectral's 500 v/microsecond and 600 v/microsecond slew rates with a megahertz or two of bandwidth weren't good? They had me fooled.

Define "good". I had problems with Spectral amps and stability, as well as RFI (though not necessarily the Spectral's fault), but maybe I was the only one. In any event I stated general precepts I have found to be true in my world which apparently is irrelevant to audio anyway.

 

[Kal Rubinson]

That is just great- I love that cartoon!

Me, too, and it is relevant to this day.

 

[anmpr1]

AC Cartoon

Very funny. Very true. Problem is, where the hell can you go to test tubes, anymore? I remember the old man pulling glass from the television and driving to a local hardware store. We'd test the tubes, and underneath the tester machine was a storage bin of replacements.

Phono was a hoot. VTA, SRA, tracking force, azimuth, which alignment geometry to use, should I go with the Shibata/MR/Gyger? What about this new and improved mat, that 'blows away' whatever it was that came before. Record clamps. Dynagroove or DMM mastering? Ahhh...those WERE the days!

 

[JohnYang1997]

Well I am certainly the same. I used to test 20 different opamps in o2. And measure them. When I see the low value distortion, I'm happy.
Recently I had measure my other amplifier every day when I see those damn low distortion I'm in heaven. But a few days ago I found my measurements had issue. Then I fuss around the grounding point, the arrangement of supply rail. And finally got very low distortion again. Ah I don't listen to music, i watch distortion spectrum.

 

[Sal1950]

Well I am certainly the same. I used to test 20 different opamps in o2. And measure them. When I see the low value distortion, I'm happy.
Recently I had measure my other amplifier every day when I see those damn low distortion I'm in heaven. But a few days ago I found my measurements had issue. Then I fuss around the grounding point, the arrangement of supply rail. And finally got very low distortion again. Ah I don't listen to music, i watch distortion spectrum.

Maybe you should consider seeing a professional on this obsessive compulsive disorder?
https://www.webmd.com/anxiety-panic/understanding-obsessive-compulsive-disorder-treatment#1

 

[John_Siau]

Not really.

For operation without slew-induced distortion, the following formula can be used:

View attachment 29905
where f= frequency and V= the peak voltage

It's a nominal 100W amplifier @ 8ohms. That's 40V peak.

Plugged into the formula at 20KHz (max rate of change of a 20KHz sine is 0.125V/uS), we need only around 5V/uS.

Or it is happy out to above 60,000Hz at full power.

If we look at the maximum rated voltage swing of 56.57Vrms (for the bridged mono into 16ohm) , that is 80V peak.

Plugged into the formula, at 20Khz we need around 10V/uS.

Or it is happy out to above 30,000KHz.

Obviously, at low powers and low voltage swings, 16V/uS will push the slew induced distortion free response out to several hundred KHz.

One correction: The slew rate is twice as high when running bridged mono. This means that there is no change in the full-power bandwidth when running in bridged mono.

 

[g29]

Due to high demand we are currently out of stock.

Looks like ASR may have had a hand in modifying Benchmark's supply and demand chain.


Benchmark AHB2 OUT OF STOCK

 

[nemesisrobot]

Due to high demand we are currently out of stock.

Looks like ASR may have had a hand in modifying Benchmark's supply and demand chain.


Benchmark AHB2 OUT OF STOCK

No kidding, I'm still waiting for my order to ship :'(

 

[anmpr1]

No kidding, I'm still waiting for my order to ship :'(

When I bought my AHB two months ago their was a waiting list. They sell them faster than they can make them. The typical high end company would raise the price in an opportunistic way if they had this 'problem.' Good things come to those who wait. And it is worth the wait, I can tell you for sure.

 

[Kal Rubinson]

Looks like ASR may have had a hand in modifying Benchmark's supply and demand chain.

One hand of many.

 

[DonH56]

One hand of many.

Stick up your hand, Kal. It was you who got me interested in them a while back. Fortunately, or unfortunately, house work and retirement savings has preempted other spending for just a bit.

 

[nemesisrobot]

When I bought my AHB two months ago their was a waiting list. They sell them faster than they can make them. The typical high end company would raise the price in an opportunistic way if they had this 'problem.' Good things come to those who wait. And it is worth the wait, I can tell you for sure.

I have no doubt it it will be worth it. My HPA4 is just getting lonely .