Review and Measurements of Benchmark AHB2 Amp

[orangejello]

With Eigentakt on the horizon, this still seems like the best option.

In any case, I gotta hear it for myself and I know I won't get a half assed product. I can certainly built the NC400 modules myself, but I have no respect for the shoddy engineering of some of the NC500 amplifiers. This doesn't only include boutique brands but also NAD with their M22 which shows much worse measurements than the DIY implementation.

I am going to order the ABH2 as well. I have the M22 and it sounds great. It has moved my tube amps and class a ss amp to the side.

Where did you find measurements that show that the M22 is “much worse” than the ncore 400 diy? The stereophile measurements are quite good for the NAD v1? I can’t really compare the ASR measurements because they are done at different load impedance and output wattage and I don’t know how to extrapolate from one to the other. Further, l have the v2 and have not seen any measurements on it. I would like to think that NAD took the opportunity to clean up the v2 a little, but I have no way of knowing. Also the v2 has some very nice features, e.g. 3 choices for output gain. It also has a very nice design compared to diy.

In any case, the only way to know which I prefer is to A/B them in my system. Based on what I have hear from the M22 whose measurement are very good compared to the vast majority of amps, I am hoping that the ABH2 is is just another degree of clarity beyond the M22. I am really curious to know at what point improvements in measurements are no longer relevant. I suspect the the ABH2 may be that point. But the M22 v2 May already be there, at least for me with the other constraints of my system and budget.

 

[JohnYang1997]

Just found out that the high frequency distortion is much more impressive on NC400. Almost a magnitude lower at 10khz.

 

[JohnYang1997]

And damping fact or is also much much better. 30 times better at 10k, 100 times better at 10k.

 

[Soniclife]

I'd like to see the Devialet tested here. Their specs are are pretty interesting.

As a Devialet owner I would too, just out of academic interest as I'm very happy with the performance. The tests I've seen seem to show it better than the hypex, but not as good as the AHB2.

 

[JohnYang1997]

These tests are just "extreme conditions", that's how they should be done. 10khz is the extreme condition of possibly audible. And 6khz is another more realistic high frequency test frequency with first two harmonics within 18khz. It's not that it's audible or not but be able to say if perfect under these conditions it must be inaudible.

 

[scott wurcer]

And not to continually batter a deceased equine, here are more randomly grabbed residual measurements of well-known engineered AB amps (top to bottom: Parasound, Classe, Bryston, Pass Labs) that fail to show crossover distortion. It's almost as if the engineering of this sort of amplifier is actually understood!

View attachment 27014
View attachment 27015
View attachment 27016
View attachment 27017

This is an interesting discussion, I wish more folks would measure THD not just THD+N at low levels. In some cases this requires some extra work like synchronous averaging but I have been told some amplifiers would surprise you.

I once asked Nelson Pass about "optimum" AB bias and he said folks seemed to prefer just turning the bias up. I run 50W per channel quiescent on a fairly ordinary MOSFET PA. If you read the intro section of the THX patent the whole idea is getting the quiescent power down for multi-channel AV systems. Mr. Siau himself said, IIRC, that they suggested all the way class B for the main amp. You can't ignore this issue in comparing the two solutions.

 

[JohnYang1997]

This is an interesting discussion, I wish more folks would measure THD not just THD+N at low levels. In some cases this requires some extra work like synchronous averaging but I have been told some amplifiers would surprise you.

Averaged spectrum is better instead of just a number

 

[SIY]

This is an interesting discussion, I wish more folks would measure THD not just THD+N at low levels. In some cases this requires some extra work like synchronous averaging but I have been told some amplifiers would surprise you.

Hey, Scott! Hope you'll hang around and comment, it's a pretty good environment.

 

[amirm]

This is an interesting discussion, I wish more folks would measure THD not just THD+N at low levels. In some cases this requires some extra work like synchronous averaging but I have been told some amplifiers would surprise you.

THD components are readily available in the FFT in the dashboard. It is all the harmonics of 1 khz to the right of it.

 

[scott wurcer]

THD components are readily available in the FFT in the dashboard. It is all the harmonics of 1 khz to the right of it.

I meant the THD at say 10mW or even less. I have not confirmed the measurements but I have seen posted elsewhere that a certain Halcro amp has as much as 8% THD at levels that are still relevant to users of very high sensitivity speakers while specifying ppm distortions at high levels. Earl Geddes also explored the behavior of class A/B amplifiers as the output goes to 0. THD+N plots bury this behavior at low output levels.

 

[SIY]

I meant the THD at say 10mW or even less.

So 280 mV into 8 ohm? If so, I'll add that to the stuff I'm measuring on the high end amp (not Halcro) I've been testing and post it.

 

[restorer-john]

The protection circuits detect full amplitude sine waves and limit the amount of time these signals can be produced when driving a low impedance. The time limit is a function of output voltage and output current. @amirm may be able to tell us the total sweep time before shutdown. He is correct: If the sweep was from low frequency to high frequency, the protection would have activated on the high-frequency side of the graph. A faster sweep time would allow a complete sweep before triggering the protection.

So John, with all that SOA in silicon and heatsinking, a beautiful regulated SMPS, do we also have a protection circuit that won't let the amplifier deliver >~133W@4ohms in a continuous manner into a purely resistive load? Seriously? If the amplifier is shutting down prematurely as the sweep time on the AP is a bit long, perhaps the protector is a bit pessimistic/overactive wouldn't you say?

I theorized (earlier in the thread) that the sweep from high to low was possibly being interpreted by the 'protection' as excessive HF or an indicator of possible oscillation, or perhaps it looks at the 'history' of the high level signal and decides enough is enough, but it seems otherwise.

I guess I'm old school. One third power preconditioning for one hour. Then full power testing. Headline specifications like this example below, where the rated power is sustained for no less than 5 minutes:

Your Benchmark is, after all, a proper 2ch amplifier, so doesn't have to duck and weave around the ratings like multichannel AVRs and subwoofers. It shouldn't need to have one rating for music signals and then be de-rated for actual testing should it? (sounds like the VW emissions scandal in reverse doesn't it?)

What is the real, continuous rating of the Benchmark into 8/4/2 ohms within the FTC mandated Amplifier Rule (1974)? So we can compare today's apples, with apples produced and tested for the last 40+ years.

 

[restorer-john]

I have not confirmed the measurements but I have seen posted elsewhere that a certain Halcro amp has as much as 8% THD at levels that are still relevant to users of very high sensitivity speakers while specifying ppm distortions at high levels.

I always thought Candy was clever by using the parts per billion designation. It was different to the string of zeroes in a THD number. (DM-68 'report' excerpt below). I think stereophile tested them at one point?

(4:1) 8 volts @ 7KHz. Funny figure to stop at. Maybe things did go seriously off the rails below that?

 

[scott wurcer]

So 280 mV into 8 ohm? If so, I'll add that to the stuff I'm measuring on the high end amp (not Halcro) I've been testing and post it.

I am suggesting something like second, third, etc. order intercept behavior of a weakly non-linear system, that is the distortion vanishes by a simple power law to 0 at 0 output. Some folks have made an argument that this can be a differentiator in amplifiers that is overlooked. Any zero crossing discontinuity will show up dramatically.

 

[scott wurcer]

I always thought Candy was clever by using the parts per billion designation. It was different to the string of zeroes in a THD number. (DM-68 'report' excerpt below). I think stereophile tested them at one point?

View attachment 27292

(4:1) 8 volts @ 7KHz. Funny figure to stop at. Maybe things did go seriously off the rails below that?

Look familiar, Walt and I did this in 1983? BTW Candy's stuff on metal detectors is absolutely SOTA.

 

[restorer-john]

I am suggesting something like second, third, etc. order intercept behavior of a weakly non-linear system, that is the distortion vanishes by a simple power law to 0 at 0 output. Some folks have made an argument that this can be a differentiator in amplifiers that is overlooked. Any zero crossing discontinuity will show up dramatically.

Where residual noise doesn't dominate (and obliterate the plot- Class D I'm looking at you), how is this different from the 'not to scale' CRO/DSO traces showing the output from the analyzer with the fundamental notched out? The levels are just way lower?

Look familiar, Walt and I did this in 1983?

AN-245, your application note. Just the ticket for my old Leader LDM-171 and Panasonic VP to extend their useful lives. Have you got a practical circuit with values, basic layout rules etc? I'd love to knock up that front end. Saves me buying an AP.

It seems they are available on the secondary market (eBay) but not in DIP16 anymore except NOS.

 

[Panelhead]

I meant the THD at say 10mW or even less. I have not confirmed the measurements but I have seen posted elsewhere that a certain Halcro amp has as much as 8% THD at levels that are still relevant to users of very high sensitivity speakers while specifying ppm distortions at high levels. Earl Geddes also explored the behavior of class A/B amplifiers as the output goes to 0. THD+N plots bury this behavior at low output levels.

This is the power level I am interested in also. My speakers are 105 dB and are supported with an active sub. The volume at 10 mW is normal serious listening volume. 100 mW is loud.
I look at the 1 watt output noise floor, and the power supply harmonics. But a test at 10 mW would be much closer to normal operating conditions.

 

[neutralguy]

With my 86db speakers, I use 5mW at 60db at listening position, 40mW at 70db. Computed based on measured voltage and nominal impedance.

 

[restorer-john]

...With my 86db speakers, I use 5mW at 60db at listening position, 40mW at 70db...

Wow. That's hardcore flea-power. Do you run your speakers directly off your iPhone?

 

[neutralguy]

The power used doesn't depend on amp, it depends on the speakers and volume levels, and 60db-70db rms is imo reasonable for music. This power number is consistent with calculations based on specs. The speaker spec says 86db at 1.3W, so at 66 db it's 1% of that which is 13mW. That's at 1 meter anechoic whereas I listen farther away, but then there's a few db of room gain to compensate. This calculation is also in line with Benchmark's application notes on this topic.