Benchmark AHB2 Power Amp: The Quietest and Cleanest Amp on Planet Earth?

Sonnie said:

Everyone is only trying to help... from years of very extensive experience and testing from various levels.

Click to expand...

So what is your verdict on the AHB2s?

I am thrilled with the AHB2s because:

• The run cool
• Have been completely reliable
• Have no mechanical (transformers bug me) nor audible noise at my speakers
• Are transparent
• Have excellent indicators to be confident that there is no clipping
• Have excellent protection to be confident that they will not harm speakers (a big deal IMO)

I am in the camp of more apparent detail but without an A/B comparison it's obviously an opinion.

Someday, there may be standardized tests into simulated reactive loads.
Amps that rely solely on negative feedback increase distortion when driving difficult loads and their distortion increases with frequency.
The AHB2 performance may be overkill or it may provide headroom to keep distortions at bay.

- Rich

I conducted an amp test using the MiniDSP SHD. It turns out that device is an excellent tool. It has 4 DACs, allows output channel duplication 1 to 3, and 2 to 4, in this case. The DSP permitted level matching that turned out to be 0.06 dBV when comparting the AHB2 is rated at 100 WPC into 8 Ohms and the AT525NC Hypex NCore (class D) is rated at 200 WPC.

The ATI is not as clean but measured very well on ASR.

Here is the configuration of the test:

• The MiniDSP SHD is used as the source for both amplifiers. Its DSP permits level matching between .1 dB. In actuality, matching was 0.06 dB.
• The MIniDSP outputs 1 and 3 are the left channel and 2 and 4 are the right channels.
• Both amps were level matched to approximately 2.83 Volts measure with a Fluke 87V sinewave voltmeter.
• Music selections are streamed by Roon to a Raspberry PI3 running DietPi and Roon endpoint to the MiniDSP USB DAC.
• This passive A/B switch was used to switch instantaneously between sources using the Nobsound Stereo switch (Amazon)

So the connections connection looks like this:

• Roon -> RPI3 (DietPi Endpoint) -> MiniDSP USB Input
• MiniDSP out 1 & 2 -> ATI525NC -> Nob switch Input A
• MiniDSP out 3 & 4 -> AHB2 -> Nob switch Input B
• Nob switch out -> Revel M123Be speakers

This configuration has nearly perfect level matching with 0.06 dB and instantaneous switching.
During the test, the MiniDSP channels were switched and no differences were found. This was done to make sure the MiniDSP was not affecting the test.

I have not yet done the single-blind-test (SBT) part so, for now, it is me listening and quick switching.

The AHB2 and AT522NC do sound different. These are most easily recognizable playing close-mic'ed female vocals, and percussive instruments, and recordings with good stereo imaging.

In all cases where they differed, I find the AHB2 to the cleaner sounding amp. The sound stage is more defined. The ATI sounds does not produce the sibilants as naturally and seemingly dulls the leading edge of drum hits and guitar strings. As a result the AHB2 sounds more lifelike and louder.
This was surprising, so more than once, the outputs were re-measured with 0 dBFS sinewaves. At 2.83 volts and the ATI is a just a hair louder, 0.02 volts which is insignificant. It was not a gain issue.

One album exemplified the differences more than the others, but all showed the same results.
Feist's "Metals" has excellent percussive sounds, up front female vocals, and wide imaging.
The AHB2 was exemplary and the ATI just did not excel.
To be fair, I am sure that if I walked in the room, I would not know which was playing but with the A/B switch it was not difficult.

I also changed the level match to -3.0 dB and -8.5 dB to make sure that both channels were using the DSP processing. This made no difference.

I am actually surprised by these results. The expectation was the AT525NC would be indistiquishable.

This is a bit long winded for a response but the AHB2 produced 2.83 volts (1 watt into 8 Ohms) at SHD Volume -26. It played all source tracks without clipping at 100 WPC, volume level -06. These tracks have peaks at or near 0 dBFS.
At -05.5, the AHB2 clip lights begin to flash.

The AT525NC peak indicators did not flash, even at Volume 0, 400 WPC, though the amp must be clipping.
The Revel M125Bes sounded much better with the AHB2 at -06, and with these sources, I'd rather listen to it than the AT525NC. I found the same to be true at all listening levels from below 1 watt up, though easier to discern at louder levels.

The ATI NCore is class D, has a linear power (non regulated supply), and relies heavily of feedback to curb distortion. I have no idea what is going on but this amp cannot resolved music as well as the AHB2, though this is not expected given the measurements. I would not recommend this amp for Revel Be speakers. It may not hold for other NCore implementations and class-D amps.

- Rich

Sonnie said:

Thanks for the info Rich... very interesting comparison of the two.

I'm probably like you, if I walked into the room I doubt I would be able to tell you which amp is playing. However, when you are able to switch quickly while listening, I can see where it would make it easier to notice the detailed differences like you did. I may still not notice because my ears are not that good, but guys with good ears like you would be able to pick out the differences.

I rely on Wayne and Dennis in my home to help me make sure my system sounds good. They've been here plenty of times over the last 7-8 years. It may sound a bit silly to rely on others ears if I can't hear it, but it is at least psychologically pleasing.

Click to expand...

I bi-amp the Salon2s with AHB2s and for movies, they don't clip, even concert videos like Yesterday.
That may be partly because there are 5 speakers firing.

For music, I can clip when I try but they are still very loud.
There is another point, you can use a volt-meter and 0 dBFS 1 kHz tone to determine the level that is 2.83 volts.
That can be matched with a sound meter at your listening position (I used mono-times for both speakers).

With those numbers in hand, you know can compute the voltage gain using this equation (obtained from Benchmark) =20*LOG(X/2.83) where X is the voltage. For example, bridged into 4 Ohms, the 45.4 volts. A 0 Ohms 87 dB efficient speaker can produce 111 dB before clipping, aside from speaker induced compression. This is essentially, the Salon2. I have found that room gain and two speakers produces about the 86 dB at my listening position. So with bridging, 106 dB and bridged, 111 dB.

What are your speaker specs?

- Rich

I just saw the gear drop-down. That's a very nice feature.
Those ESLs present a tough load at higher frequencies.

It would be interesting to use the volume measurement to determine the level at 2.83 volts and the level where your clip lights illuminate to determine if they are triggering earlier than expected due to the load.

- Rich

Well, this is obviously not a blind test, much less randomized, so I'm taking with grain of salt. Certainly appears level matching ok. I'm also suspect of the switch, has it been tested for decoupling/isolation? Never heard of it (Nobsound).
Salons? AB2s? Hmmm, seems you should be able the afford a AVA ABX, around $1k iirc, which accounts for these issues.
I would be slightly less skeptical then.

cheers,

AJ

Not the droids we're looking for.
This, OTOH, looks closer https://www.amazon.com/Nobsound-Amplifier-Splitter-Switcher-Selector/dp/B073RCRR82?ref_=ast_sto_dp

• Left and Right channels to ground, suitable for different audio atmosphere

Sonnie said:

Ahhh... yes, not sure why I was thinking the XLR version when he was clearly showing the toggle switch model.

Click to expand...

Low vs high level signal, difference well beyond switch.

Sonnie said:

Seems like it was serving the purpose he was trying to accomplish for his own benefit at least.

Click to expand...

When you start seeing N-rays and cold fusion, best to check methods/surprises.

cheers,

AJ

Sonnie said:

I have the Van Alstine ABX box Rich... if you would like to borrow it for some testing, although it seems like you did a pretty good job of being honest about it.

Here is testing on the Nobsound... I use one in my system to switch back and forth between my music and movie processors.

Nobsound 3-IN-1-OUT XLR Audio Switch Review

This is a review and detailed measurements of the Nobsound MC-103 Pro 3-IN-1-OUT XLR Audio Switcher. The unit I have came with the brand Douk on it so that is what I used in my notations below. I suspect multiple companies put their name on it. I purchased it a few days ago from Amazon. The...

www.audiosciencereview.com

Click to expand...

The ATI AT525NC will be installing it a a vacation home in a week or so.
I did this test to make sure all channels worked so it was not much more effort to try the A/B amp compare.

The SHD has 4 DAC outputs and DSP so it can level match to .1 dB in the digital domain.
Wire was on-hand but decided to spend $80 on a passive A/B switch:
Amazon.com: Nobsound Stereo 2(1)-IN-1(2)-OUT Amplifier/Speaker Banana Female Audio Splitter/Switcher Passive Selector: Electronics

Today, I redid the level match (after SHD software update) using the Fluke 8V using stacking banana plugs connected to the speaker outputs.
There was not alteration to the connection or leads when adjusting the gain.
Playing a 1kHz 0 dBFS sine-wave on the ATI varied between 2.83 volts and 2.86 volts and settled in at 2.856 volts.
The AHB2 remained stable at 2.859/2.856. The are very tightly matched.

An ABX box would be interesting but I prefer a simple make-and-break switch with no analog components.
After a few listening session today, the two amps were closer but the AHB2 is still the cleaner sounding amp on close-mic'ed female voices and percussion.

I am satisfied with the result and it is certainly better than most that compare amps.
I have read posts where some have sold their AHB2 and want to Hypex because they felt they were more dynamic.
The AT525NC requires -5.3 gain (after SHD software update) to level match so suspect this is source of that impression.

I can say with certainty that even at clipping, the NCore amp is not more dynamic driving the M126Bes and the AHB2 remains clean and uncompressed even after the initial illumination of the clip indicators.
The AHB2 clip-indicators are precise, lighting up just at 100 WPC, where -26 is 1 watt (2.83 volts) and 100 wats is exceeded at -.6.
The first clip-light occurred at -5.5.

FWIW,

Rich

RichB said:

An ABX box would be interesting

Click to expand...

It would make the test blind and randomized, the only valid method. Unlike sighted A/B.
It would also isolate the DUTs, rather critical for amplifiers.
Results may not be surprising though, a lot less exciting.

Sonnie said:

Doesn't seem to be any doubt that I'm apparently pushing my ABH2's very close to the limit with my lights flashing constantly. At first I didn't think much about it. However, there are a few songs that I like to crank on up from time to time, and I'm at 0dB on the SHD... volume full throttle, and ABH2 warning lights flashing... and I'd like a little more volume, but I don't have it.

I spoke with a couple of different guys that are very knowledgeable about amps, etc... all are involved in manufacturing amps, and all from different companies, but I was getting pretty much the same answer from all of them. One manufacturers NCore amps, and he ultimately felt it best for me not to purchase an NCore amp for my 15A's because I would possibly have less power than I do now, and would likely run out of power and/or continue to experience issues. Example... the NCore 1200 is rated at 700 watts at 4 ohms, but that's max and at 1% THD. While the AHB2 in it's bridged state is NOT rated into 4 ohms (for a reason)... if it were it would be somewhere in that 700 watt range... considering it loses some wattage when halving the impedance load. However, the difference is it would be rated at .0003% THD, not 1%.

• 200 Watts into 16 Ohms, bridged mono
• 380 Watts into 8 Ohms, bridged mono
• 480 Watts into 6 Ohms, bridged mono

The AHB2 is conservatively rated at an output level where THD+N is < 0.0003 % instead of the more typical 1% THD+N. Power at 1% THD will be higher.

So these ratings are at .0003% and the NCore amps are at 1%. Surely the AHB2 is considerably more powerful than the NCore when compared at 1% THD.

However... ultimately the AHB2 is simply not designed to be bridged into 4 ohms, hence the lack of that bridged 4 ohm rating.

Another issue... even though the 15A is rated a nominal 4 ohms, it does drop to .52 ohms at 20kHz, and even though 20kHz is way on up there and there's no audible sound, it apparently matters to an amp, as it is still receiving noise, even if inaudible to the human ear. The AHB2 minimal impedance rating is 1.4 ohms, so this may very well be part of the issue with it nearing its limits more often than I prefer.

Click to expand...

I think really the question is do you prefer an amp with an unregulated supply and higher noise and distortion. These will be able to handle momentary peaks better.

The AHB2 will certainly drive 4 Ohm loads and ASR measured the AHB2 bridged at 500 WPC but as you can see that is not much more than 480 WPC into 6 ohms. Current become the limiting issue. I think the lack of a 4 Ohms rating is to be conservative since long term tests may exceed the cooling capabilities and cause the amp to shutdown.

Unlike other amps, the AHB2 is giving accurate clipping indicators.
I pushed the AT525NC to 400 WPC and the "Peak" indicators are not lighting.
Driving the 126Bes this is not of much value because after about 100 watts, in the 106 dB range, they are compressing so your not getting much more out of them, that sounds good anyway.

Perhaps, I am off base here. but at 20kHz the magnitude (voltage) of the signal is very low.
What might require a watt with the average speaker may 6 watts for the 15As, I don't see how this could be a problem.

If you measured your speakers determine the volume level for 2.83 volts, the AHN2 should provide 20 dB not bridges and something like 26 dB bridged. If your not getting that gain, then the 15As current requirements are limiting the AHB2.
Which clip indictors light?

- Rich

AJ Soundfield said:

It would make the test blind and randomized, the only valid method. Unlike sighted A/B.
It would also isolate the DUTs, rather critical for amplifiers.
Results may not be surprising though, a lot less exciting.

Click to expand...

I am not exited and I dare not ask if you are

It is not as easy to get folks together for SBT as it was pre-COVID.
I may yet do so, but in the end, people have opinions that are well forged and even, if I passed an ABX test, that will not change.
Jon Iverson put it well, such tests are those people, that gear, that day that room.
A negative finding proves nothing and so does a positive finding

- Rich

RichB said:

people have opinions that are well forged and even, if I passed an ABX test, that will not change.

Click to expand...

That wouldn't be opinion. It would be a strong data point and certainly worthy of further investigation as to why.
If ifs and buts were candy and nuts, it would be Christmas every day.

Sonnie said:

Doesn't seem to be any doubt that I'm apparently pushing my ABH2's very close to the limit with my lights flashing constantly. At first I didn't think much about it. However, there are a few songs that I like to crank on up from time to time, and I'm at 0dB on the SHD... volume full throttle, and ABH2 warning lights flashing... and I'd like a little more volume, but I don't have it.

I spoke with a couple of different guys that are very knowledgeable about amps, etc... all are involved in manufacturing amps, and all from different companies, but I was getting pretty much the same answer from all of them. One manufacturers NCore amps, and he ultimately felt it best for me not to purchase an NCore amp for my 15A's because I would possibly have less power than I do now, and would likely run out of power and/or continue to experience issues. Example... the NCore 1200 is rated at 700 watts at 4 ohms, but that's max and at 1% THD. While the AHB2 in it's bridged state is NOT rated into 4 ohms (for a reason)... if it were it would be somewhere in that 700 watt range... considering it loses some wattage when halving the impedance load. However, the difference is it would be rated at .0003% THD, not 1%.

• 200 Watts into 16 Ohms, bridged mono
• 380 Watts into 8 Ohms, bridged mono
• 480 Watts into 6 Ohms, bridged mono

The AHB2 is conservatively rated at an output level where THD+N is < 0.0003 % instead of the more typical 1% THD+N. Power at 1% THD will be higher.

So these ratings are at .0003% and the NCore amps are at 1%. Surely the AHB2 is considerably more powerful than the NCore when compared at 1% THD.

However... ultimately the AHB2 is simply not designed to be bridged into 4 ohms, hence the lack of that bridged 4 ohm rating.

Another issue... even though the 15A is rated a nominal 4 ohms, it does drop to .52 ohms at 20kHz, and even though 20kHz is way on up there and there's no audible sound, it apparently matters to an amp, as it is still receiving noise, even if inaudible to the human ear. The AHB2 minimal impedance rating is 1.4 ohms, so this may very well be part of the issue with it nearing its limits more often than I prefer.

Click to expand...

You analysis is way off. Too lengthy to correct this evening.

Sonnie said:

I can vouch for blind testing of amps being about as boring as mud. We did it ... results here: https://www.hometheatershack.com/th...event-reporting-and-discussion-thread.115674/

Where are you located Rich? Is Alabama too far?

Click to expand...

I am in Mass. too far.
Your in one of the those free states

- Rich

AJ Soundfield said:

That wouldn't be opinion. It would be a strong data point and certainly worthy of further investigation as to why.
If ifs and buts were candy and nuts, it would be Christmas every day.

Click to expand...

I was asked to do this by others and ASR to try this experiment.
I posted it here to support Sonnie is a good guy, also my opinion.
It was interesting and your opinion was anticipated but not the tactless delivery

- Rich