Review and Measurements of Benchmark AHB2 Amp

[JohnYang1997]

I remember in Douglas's book, the crossover notch is always there. And even when the distortion is down to 0.000x% with averaging, the crossover notch is still there. The graphs shown in the last post are not even in the same distortion category. The harmonic distortion and noise is overpowering the crossover notch. Unless we all have the same scale, it's hard to say the corssover distortion is eliminated.

 

[John_Siau]

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!

It is important to run these low power tests cold and hot, but you have me curious. I measured two samples of the same power amplifier (4 channels total). All four channels were similar. After running those tests, I acquired a number of used amplifiers which I have been hoping to test. I just need to find the time to do it!

 

[SIY]

I'm running some tests this weekend for an amp under review. Because these will be published, I can't yet name the specific amp beyond saying that it's a conventional AB amp designed according to the "Blameless" principles, but I'll post the cold and hot distortion residuals for comment.

 

[SIY]

I remember in Douglas's book, the crossover notch is always there.

In order to explore this, Doug had to deliberately underbias the amps because the residual distortion was so small (<0.001%). I suspect that's what you're recalling.

 

[JohnYang1997]

In order to explore this, Doug had to deliberately underbias the amps because the residual distortion was so small (<0.001%). I suspect that's what you're recalling.

This is the part from Douglas's book.

 

[JohnYang1997]

It is important to run these low power tests cold and hot, but you have me curious. I measured two samples of the same power amplifier (4 channels total). All four channels were similar. After running those tests, I acquired a number of used amplifiers which I have been hoping to test. I just need to find the time to do it!
View attachment 27018

How many times was the distortion/noise residue multiplied in your graph?

 

[SIY]

How many times was the distortion/noise residue multiplied in your graph?

1024x.

And the next page of the Self book continues with him saying that he had to underbias the amp to determine if crossover distortion increased with decreasing level. Note the machinations needed to visualize it at all!

 

[JohnYang1997]

1024x.

And the next page of the Self book continues with him saying that he had to underbias the amp to determine if crossover distortion increased with decreasing level. Note the machinations needed to visualize it at all!

Yeah. But it's still there. And I did say it's possible to reduce the artifacts by using feedback techniques to achieve higher ol at higher frequency. But it's inherently in the class b/ab. Only way to eliminate is to by pass/ takeover the region or to use cancellations, which are not necessarily conventional pure class b/ab amplifiers anymore.

 

[JohnYang1997]

If it's 1024x, the graph is probably a bit exaggerating. That's very severe crossover distortion.

 

[John_Siau]

For what it is worth, I ran my ABX listening tests at 0.01 W, producing an SPL of 67 dB at the listening position. I used a 1 kHz tone. Running at 1 W would have been intolerable. At 0.01 W the residual wave shape was similar and the distortion was audible. It was also audible at 1 W, but running a 25-trial ABX test at the resulting SPL (about 87 dB) would have driven the participants insane.

 

[SIY]

Did you try the same test with more complex program material like music?

 

[JohnYang1997]

For what it is worth, I ran my ABX listening tests at 0.01 W, producing an SPL of 67 dB at the listening position. I used a 1 kHz tone. Running at 1 W would have been intolerable. At 0.01 W the residual wave shape was similar and the distortion was audible. It was also audible at 1 W, but running a 25-trial ABX test at the resulting SPL (about 87 dB) would have driven the participants insane.

There should be better class ab amplifier that are really that noticeable. Probably 1/10th of the crossover distortion and harmonic distortion. I thought the graph was 10000x. Since it's 1024x that's not considered a good amplifier.

 

[John_Siau]

Did you try the same test with more complex program material like music?

No, I was trying to keep the test as simple as possible.

 

[John_Siau]

There should be better class ab amplifier that are really that noticeable. Probably 1/10th of the crossover distortion and harmonic distortion. I thought the graph was 10000x. Since it's 1024x that's not considered a good amplifier.

The crossover distortion was roughly -73 dB relative to 1 W. This would be -93 dB relative to full output (0.0022% THD which is not necessarily terrible). In our 0.01 W listening test, the distortion was reproduced at a calculated level of 14 dB SPL.

I just corrected the numbers above. I didn't realize that I had published the exact numbers in the paper:

"... the error signal has been amplified by 1024 (about 60 dB). At a 1 watt output, this class-AB amplifier produces a distortion waveform that measures 70 dB below the output level of the 1 watt test tone (see graph at the bottom of this application note). This means that the power produced by the THD+N is 70 dB below 1 watt. If we drop the level of the test tone by 20 dB, the output power is 0.01 W. At this output level the amplifier was still producing distortion at a level of 73 dB below 1 watt. At 0.01 watt the distortion waveforms look virtually identical to the 1 watt waveforms. This distortion was clearly audible when we used this amplifier to drive a speaker at 0.01 watt. For our tests we used a stereo pair of Benchmark SMS1 speakers with a sensitivity of 87 dB 1 watt, 1 meter. The tone was reproduced at a sound pressure level of about 67 dB (measured 67 to 68 dB SPL) at the listening position while the amplifier distortion was reproduced at a calculated sound pressure level of about 14 dB (87 dB - 73 dB = 14 dB). With a 0.01 watt 1 kHz test tone, the amplifier distortion was clearly audible through the loudspeaker. At this low power level, the speaker distortion is lower and much more musical than the ugly crossover distortion produced by the amplifier.

 

[restorer-john]

Thanks for pointing this out. I will add that note to the feed forward application note. Not trying to be deceptive. Our goal is to present accurate and useful information.

For the layperson, the classic annotation below the image: "distortion products depicted are not to scale", would be sufficient.

 

[DDF]

Our adapter cables wire the XLR+ back to the RCA pin. The XLR- is wired to the RCA shield. A separate shield conductor connect the AHB2 chassis tot the RCA shield. This wiring allows the use of the balanced differential amplifier at the XLR input. This wiring provides significant rejection of ground-loop noise. An RCA to RCA connection provides no rejection of ground-loop noise

Hello, if I read this correctly, same as Rane's recommendation attached?

 

[LTig]

The K&H O300D studio monitors gave you exactly what should have been delivered. Sound emanating from a virtual speaker exactly centered between the two physical speakers and on the same plane.

This is what I thought as well. It must be that way. BTW these are my own.

The K&H monitors use trimmers to match the high-frequency drivers to precise tolerances. This gives them a wonderful ability to keep high frequencies tightly imaged into the phantom center. We have a pair of K&H O300 monitors and I have spent lots of time listening to them and comparing them to other studio monitors (see photo of Benchmark's listening room below). If you listen closely you will discover that the phantom image is several inches wide, it is not quite a pinpoint. I attribute this to the excessive distance between each of the three drivers in the O300D. The three-way design and the size of the drivers dictated a larger driver-to-driver spacing than you would find in a two-way nearfield monitor. In our listening tests, this diver spacing spread the width of the phantom image by a few inches. Each of the two-way monitors did better in this regard than the K&H O300.

My listening distance is 3.6 m, so I think it's not relevant here.

Nevertheless, at the very high end of the audio spectrum, the K&H were able to keep high frequency sounds well focused in the center. I attribute this to the very precise driver trimming in the K&H monitors. On some of the other speakers shown in the photo, cymbals and other high frequencies sounds tended to move to the physical speakers. The degree to which the high frequencies move is an indication of the accuracy of the tweeter matching.

Yep. Recordings with ambience sound wonderful. There are some recordings done by our local opera house and their ambience is fantastic. Later I found out that their control room also uses K&H O300D. One could say that here the circle of confusion of @Floyd Toole has been broken. And it's one more reason for home users to buy pro speakers.

And coming back to the topic: I really would jump and buy the AHB2 on the spot if I still had passive speakers. It's an example of perfect engineering. Unfortunately I have disposed of using passive speakers since 15 years, after I got the O300D. I just hope that the same kind of engineering trickles down into the amplification of active speakers. Even Neumann now again offers a monitor with DSP cross over (the KH80DSP) after being on the forefront with the K&H O500C around 2001. I actually expect that DSP usage in active speakers is inevitable.

 

[JJB70]

As lovely as this amplifier is, and as impeccably engineered and well measuring as it is, I am also of the view that active speakers are the future. If I was to buy a new system (and my existing gear won't last forever) then I don't think I would go with passive speakers. However this clearly is a fabulous amplifier and demonstrates again that there is substance and engineering depth in Benchmark products. Given that I actually think that they are nicely priced compared to most audiophile gear. I would love to see Benchmark do some system speakers.

 

[AnalogSteph]

The crossover distortion was roughly -73 dB relative to 1 W. This would be -93 dB relative to full output (0.0022% THD which is not necessarily terrible). In our 0.01 W listening test, the distortion was reproduced at a calculated level of 14 dB SPL.

Which is just 53 dB below 67 dB SPL, so given the generally ugly spectrum of crossover distortion, it's no surprise that it was audible. Amplifiers like this one are precisely why people started harping on about the importance of the first watt.

But is this really the norm for competently built traditional amps? Like some others here, I'd have my doubts. I just pulled up some measurements for an NAD C356BEE stereo integrated amp (100 Wpc, 750€ in 2011). At 1 W, we are seeing dominant H3 at about -100 dBV (-109 dBW). Below 1 W, all distortion components from H2 to H5 fade away to around or below the -110 dBV (-119 dBW) mark, limited by the (AP) equipment used. Mind you, NAD gear tends to be about as much above average in measurements as their parts quality is below average, but still.
Alternatively, digging up a review of the Yamaha A-S700 (another well-reputed mass-market integrated amp, about 130 Wpc), we see dominant 2nd-order distortion at the 1 W / 4 ohm level, with H2 at about -83 dB. That would, at the very least, still beat the crummy test amplifier by 10+ dB.

I mean, I wouldn't be surprised to find the odd amplifier like this in circles where testing often is less than rigorous. But using one as a typical example of a traditional AB design, eh, I don't know. At least I'd hope this is not that typical, since that would be bad. At 10 mW / 8 ohms, we are talking like 50 mA peak. Typical EF output stages should still be in Class A at this point (even if potentially just barely), so I'd guess it was some CFP job with less than generous bias current, hinting at a PA amp. I wonder how a trusty ESP P3A would fare, being a common (and known decent but less-than-perfect) DIY topology... or an LM3886 "gainclone" even.

Anyway, kudos for bringing the first "sane" RCA-out to XLR-in cable to market (that I know of). It really was about time. No idea why cable manufacturers have been showing precious little interest.

 

[Music1969]

I would love to see Benchmark do some system speakers.

I would imagine THX are planning/hoping to partner with speaker makers.

Their technology has caught the attention of the amplification world and probably those making or interested in making active speakers.