• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Review and Measurements of Benchmark AHB2 Amp

RichB

Major Contributor
Forum Donor
Joined
May 24, 2019
Messages
1,946
Likes
2,611
Location
Massachusetts
We could do more with respect to measurements. Question is whether we confuse the readers and overwhelm them with information. For example, some reviewers use square wave and show those waveforms. The ones that don't know what they are doing use fast rise-time square wave which cannot happen in our audio-band music.

With respect to difficult loads, we could add reactive ones. I tried to build one but faced resistance on them being too difficult than typical speaker loads. At the end of the day, they just lower the impedance of the load at higher frequencies.

Note that sine waves are more difficult load for the amp than music. They have every low crest factor, pushing the amp to deliver more power than music would. So our simplistic test here is actually more extreme than the real application. The capacitors in the power supply for example work a lot more effectively with ups and downs of music than sine wave we use for measurements.

SINAD and the clear focus on measurements is a good/best way to find the state-of-the-art components.
Whether they be square-waves or sine waves these are tests of the design robustness.
All of that makes sense.

Benchmark's has clearly focused on reducing distortion and noise, increasing phase accuracy. Other design aspects include a tightly regulated power supply that may have had an impact that is not directly reflected by anyone's amplifier measurements.

These amplifiers certainly measure differently and IMO sound differently.

Ultimately, the pursuit of amplification perfection should lead to audible improvements.
Many believe amplifiers reached that threshold, but this amplifier has changed my mind on that.

- Rich
 

SEKLEM

Active Member
Joined
Apr 26, 2019
Messages
275
Likes
326
Location
Indiana
Sure, amps with similar architecture and good measured performance can be indistinguishable.
I am also searching for the reason for observed differences, as others have stated, in so many words.

The systems are described here:
https://www.audioholics.com/amplifier-reviews/ati-at4002

Game Room: Revel M20's, Oppo UDP-205, Oppo HA-1 (Preamp) sourced my Roon (network)
Living Room: Revel Salon2's, Oppo UPD-205 Emotiva XMC-1 (Stereo Reference mode) sourced by Roon (USB-DAC)

My systems have changed as of late:
  • Sold the HA-1 for the game room. Going Oppo UHD Direct (thanks to Roon)
  • Added an LA-4 for a two channel mode: Oppo UDP-205 -> LA4 -> Salon2's.
In the game room, the switch between the AHB2 and AT522NC in less-than 20 seconds.
Directly comparing these amps in my systems, I find it unremarkable that completely different architectures (which also measure differently) sound differently.

Are you in the Boston area? If so, perhaps I could demonstrate.

- Rich

In terms of audio memory, 20 seconds is likely too long a period of time. Never mind that appears from your description that this is a sighted comparison. In order to make the best determination the switch should be immediate and the comparison should be blind to remove bias.
 

RichB

Major Contributor
Forum Donor
Joined
May 24, 2019
Messages
1,946
Likes
2,611
Location
Massachusetts
In terms of audio memory, 20 seconds is likely too long a period of time. Never mind that appears from your description that this is a sighted comparison. In order to make the best determination the switch should be immediate and the comparison should be blind to remove bias.

I suppose implicit in this skepticism is the following assumptions:
  • Amplifier measurements are completely characterize their performance into all loads.
  • The complete list of measurements has been established.
  • For each measurement, there is an published and accepted criteria at which further improvement has no audible benefit.
  • Probably more, just don't have much time now
The testing was single blind (SBT) and I don't think my friend and I had any tells :p
If changes took 2 seconds, I suspect you would remain unconvinced. That's fine by the way :p
I am not publishing a paper here. These are my subjective views.
That said, I have probably done more than most...

- Rich
 
Last edited:
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,376
Likes
234,550
Location
Seattle Area
In terms of audio memory, 20 seconds is likely too long a period of time.
Correct. In Harman speaker testing, the shuffling time of speakers is about 4 to 5 seconds and I found that excruciatingly long even though speakers have a strong sonic signatures.

In detecting small impairments, codec artifacts and such, even one second switching time is hugely disruptive. My best detection ability is to have the switching be in milliseconds, ideally instantaneously.

As you, whenever someone says they listen to one thing, than go listen to something else minutes, hours or even days later, I completely dismiss their results. Your brain would blow up if it had to capture all fidelity information it hears in long term memory. :)
 

RichB

Major Contributor
Forum Donor
Joined
May 24, 2019
Messages
1,946
Likes
2,611
Location
Massachusetts
Since the exact measurements and criteria for inability are not established, it makes sense to include SINAD measurements in purchasing decisions. I assume that is the rational by many here.

There is a lot a DAC reviews here and I would definitely buy high on the curve.
However, as of late, I have noticed the impact that filters make to the sound...

I am not a fan of MQA and suspect that what folks are hearing is a different filter. The rest complete B..., um marketing.

- Rich
 

LTig

Master Contributor
Forum Donor
Joined
Feb 27, 2019
Messages
5,760
Likes
9,442
Location
Europe
Speakers are not a passive load. They generate its own voltage which then feeds back the amplifier. Amplifier then tries to damp these parasite signals the best way it can - but it's never perfect and various amplifiers do this with various success. Measurements being done with passive resistance don't cover what will happen with real speakers. Does it generate additional distortion, this feedback from speakers? Why, of course it does - there's a parasite signal that should not be there at all in an ideal case - which never happens in reality. [..]

All in all, one who looks in whole complexity of amplifiers, understands that set of measurements performed simply cannot seriously cover all what it takes to make a correct, objective evaluation of the amplifier. The only proper way to assess the amplifier is to actually connect it to the speakers you want to pair it with...and listen to it. [..]
Knowing how unreliable the human hearing sense is listening is not a good bet. Connect the amplifier to the speaker and repeat all measurements done previously with a resistive dummy load makes much more sense. Just make sure that the speaker is not driven with too much power.

It would be interesting to connect the AP both to the output of the amp and the input of the speaker and compare the data depending on the cable.
 

DDF

Addicted to Fun and Learning
Joined
Dec 31, 2018
Messages
617
Likes
1,355
With respect to difficult loads, we could add reactive ones. I tried to build one but faced resistance on them being too difficult than typical speaker loads. At the end of the day, they just lower the impedance of the load at higher frequencies.

Ken Kantor's load is fair and accurately represents tweeter inductance though of an older variety before shorting rings:
https://www.stereophile.com/reference/60/index.html
 

RichB

Major Contributor
Forum Donor
Joined
May 24, 2019
Messages
1,946
Likes
2,611
Location
Massachusetts
Knowing how unreliable the human hearing sense is listening is not a good bet. Connect the amplifier to the speaker and repeat all measurements done previously with a resistive dummy load makes much more sense. Just make sure that the speaker is not driven with too much power.

It would be interesting to connect the AP both to the output of the amp and the input of the speaker and compare the data depending on the cable.

I know it is not a fair comparison but consider if we did this for automobiles.
We could measure there performance with a wide battery of tests but insist that they not be driven.
If you must drive it, then you have to be a blind-folded passenger. :p

It may not be clear, but I am advocating for more testing mainly because I believe that there are audible differences and I have always liked the notion of comparing the input of the amp to the output of the amp at the speakers.

Here is an Audioholics article that may be of interest to some:

https://www.audioholics.com/audio-amplifier/audio-amplifiers-sound

- Rich
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,376
Likes
234,550
Location
Seattle Area

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,588
Likes
38,290
Location
Gold Coast, Queensland, Australia
Here is an Audioholics article that may be of interest to some:

https://www.audioholics.com/audio-amplifier/audio-amplifiers-sound

The square wave test is well known and has been used by just about every manufacturer in their glossy brochures since Adam was a boy.

Amplifiers in the 1970s got to the point where any of them could demonstrate a pretty much perfect square wave, even up into silly frequencies.

Comparing input and output is as simple as using a scope X/Y in add and invert. Large differences show up pretty quickly...
 
Last edited:

RichB

Major Contributor
Forum Donor
Joined
May 24, 2019
Messages
1,946
Likes
2,611
Location
Massachusetts
Class-D can have different looking 1kHz square waves:
https://www.stereophile.com/content/theta-digital-prometheus-monoblock-power-amplifier-measurements

1560386867732.png

ASR also exposes this high-frequency switching noise.
I don't suppose that most speaker have LPF on the tweeters, does this matter?

- Rich
 

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,588
Likes
38,290
Location
Gold Coast, Queensland, Australia
I don't suppose that most speaker have LPF on the tweeters, does this matter?

Class D proponents of course say no, it doesn't matter.

Those of us who favour completely noiseless High Fidelity equipment might say otherwise. Me, I like my waveforms clean, not fuzzy traces polluted with switching garbage.

That said, not many people listen to square waves or pure sine waves (I do on my bench, but that's me..).
 

DDF

Addicted to Fun and Learning
Joined
Dec 31, 2018
Messages
617
Likes
1,355
I am well aware of that. It is a low power load though and is only used for frequency response tests. Is that that useful?

No, agreed, not useful for that. It will only reflect frequency response issues if amp source impedance is high, and then its actually not so useful because that can be easily calculated just measuring complex source impedance.

Its useful to draw more current than an 8 ohm load and stress components (thermal, power) with a more realistic load. Granted sines are not reflective of music dynamic range, so perhaps its fair to say sine source use provides a similar stress.

It can be made high power through proper component choice, but not so high in priority.
 

DDF

Addicted to Fun and Learning
Joined
Dec 31, 2018
Messages
617
Likes
1,355
Class D proponents of course say no, it doesn't matter.

I've yet to investigate these, but I wonder, can the constant presence of the switching frequency:
- lead to heating of the tweeter coil? At what voltage does this become a non issue?
- cause any sort of demagnetization, especially neo (probably not but I'm anal with my engineering).
 

SIY

Grand Contributor
Technical Expert
Joined
Apr 6, 2018
Messages
10,386
Likes
24,752
Location
Alfred, NY
I've yet to investigate these, but I wonder, can the constant presence of the switching frequency:
- lead to heating of the tweeter coil?

Voice coil inductance is your friend here. And it takes a LOT of current to demagnetize things.
 

DDF

Addicted to Fun and Learning
Joined
Dec 31, 2018
Messages
617
Likes
1,355
Voice coil inductance is your friend here. And it takes a LOT of current to demagnetize things.

I've inadvertently significantly reduced the B of a neo magnet in a tweeter used in an initial proto speaker design, driving it with excess power (loud but not ridiculous levels in a large room) from a high wattage class d amp.
 

SIY

Grand Contributor
Technical Expert
Joined
Apr 6, 2018
Messages
10,386
Likes
24,752
Location
Alfred, NY
I've inadvertently significantly reduced the B of a neo magnet in a tweeter used in an initial proto speaker design, driving it with excess power (loud but not ridiculous levels in a large room) from a high wattage class d amp.

Wow, how much power did it take to do that?
 

DDF

Addicted to Fun and Learning
Joined
Dec 31, 2018
Messages
617
Likes
1,355
Wow, how much power did it take to do that?

I was surprised as it was only 300W or so but it was a first proto and I crossed it too low. I would never have thought to see this side effect but speaking with a friend that designs drivers on contract commercially, he's seen this as well
 

zalive

Active Member
Joined
Sep 4, 2018
Messages
263
Likes
38
Knowing how unreliable the human hearing sense is listening is not a good bet. Connect the amplifier to the speaker and repeat all measurements done previously with a resistive dummy load makes much more sense. Just make sure that the speaker is not driven with too much power.

It would be interesting to connect the AP both to the output of the amp and the input of the speaker and compare the data depending on the cable.

Hearing music and enjoying the sound is about personal liking, not necessarily even about fidelity. One thing is guaranteed: if you pick equipment by hearing, you'll be applying the very same mechanism you use when you listen to and enjoy the music. It's apples to apples. Numbers and graphs vs listening to music, that's not apples to apples. No guarantee you'll like it accordingly when you listen to it.

But at least measuring it with the speakers connected should be more meaningful than with resistive dummy load. Yet better, some measurements which include signals which are closer to actual musical content than continuous sine wave. However, ok, we can measure it like that...but where's reference? Where's comparison to other gear? So unless we get all the gear we want to compare to measure them at home using the same metodology so we can actually compare apples to apples...

Seems much simpler to just listen to it and pick the sound you like the most :)
 
Top Bottom