Amplifier Bakeoff: Purifi Eval1, McIntosh MA252 & Benchmark AHB2

[ahofer]

poor decay measurements

Can you point me to these amplifier decay measurements?

 

[Goodman]

@SIY has measurements posted for this amp here:

Fresh From the Bench: A Tale of Two Class-D Amplifiers Orchard Audio Starkrimson and Purifi Audio Eigentakt EVAL1

Stuart Yaniger examines new amplifier solutions from two disrupting audio companies. Purifi Audio from Denmark, founded by Bruno Putzeys, Lars Risbo, and Peter Lyngdorf, together with a remarkable engineering team; and Orchard Audio, with a product developed from the creativity and...

audioxpress.com

They correlate with all the measurements that I have posted in this forum. Basically what I posted has been verified by a third party.

 

[Goodman]

No one is contesting the measurement you have posted. What I am saying is: The Absolute Sound magazine duplicated your measurements and added a couple of useless comments and they call that a review. I am not criticizing your product , I am criticizing the magazine.

 

[agiletiger]

I haven't had the patience to read through all of it... But has this thread grown over 100 posts, because the OP refuses to do a DBT (or simply a BT) correctly?

Dude - feel free to remain in the warm embrace of your presumption, but if you want to be sure of the truth there is no replacement for a proper blind test, that a many here have advised you on how to do it.

Anyways - best of luck to you, I'm unwatching.

It was a blind test, not a DBT. As someone who does design experiments for a living, I don't see how a single blind test was done incorrectly in this particular case.

 

[agiletiger]

Some notes from testing methodology standpoint:

• Someone asked if a statistical analysis can be done on the test. One would need the following:
  • Results of all three reviewers for all the trials
  • Probably need to double the number of trials to strengthen the statistical significance of the results - might be taxing on the ears and on the patience of the evaluators
• Randomizing the song order in addition to the amp order would have been better and that "wrinkle" may have gotten us the "can't hear a difference" result that we all expected.
• Double blind is always best but even in peer reviewed and published tests, it's not always practical so single blind tests like these are done. This would probably further the need for more trials like I mentioned above. IMO, the single blind aspect of the test seems to have been conducted properly.

Notes on possible elements of bias:

• The calibration technique(s) used
• Song order and song choices

You had a bias/expectation of there being no difference which is commendable that your data showed the opposite and you sought to understand why. This is the process.

I am not familiar with these amps and only have knowingly heard the McIntosh before. From what others have written here, it seems like there is a wider element of human error involved with the AHB2 in terms of setting the impedance and gain properly. While some might use this variable to invalidate the study, I find it to be incredibly valuable information provided that you use it in the context of the scope and objective of this test. I gather that this was going to be used to help one of the evaluators pick one of these three amps to purchase. Based on this, If I had been a third party facilitating this test for the three of you, my recommendations would have been the following:

• If no difference was heard, buy the cheapest one (duh)
• Potential buyer is either adept at setting the AHB2 up properly or is willing and able to have someone do this setup properly - rerun test with AHB2 setup properly
• Potential buyer is NOT nor willing to have AHB2 set up properly - The human error of possibly setting up this amplifier is a significant factor in this context and therefore the potential buyer should choose one of the other two amps.

 

[Goodman]

Can you point me to these amplifier decay measurements?

https://www.isotopx.com/wp-content/uploads/2018/09/ATONA-TIMS-Revolutionising-signal-amplification-tech

 

[Goodman]

We recently got a few folks together to do as close to a 'blind test' as we could of the Purifi Eval1, McIntosh MA252 and the Benchmark AHB2. We had 3 reviewers and a 4th person who switched the gear.

Test Setup: Topping D90SE (balanced outputs) directly connected to each amplifier.
Streamer: Tidal Bluesound Node connected via optical SPDIF to Topping D90SE
Speaker: Magico S5 MKII
Randomization: For each song selection: Each amplifier was randomly assigned a number 1,2,3, A random amplifier order was used for playback.
Level Equalization: For each song/amplifier, a 1K sine wave was used to normalize the level to about 0.1 Db using a calibrated microphone, Focusrite 2i2 and REW before playback of each selection. The level was managed using the Topping D90SE in pre-amp mode and managed to about 90-95 db.
Process: 6 songs. Each song was played and the reviewers wrote comments and then rank ordered amplifiers. Between switches loud music was played to drown out any switching related noises that could indicate which amplifier was being connected.

Some of the selections used:
1. Africa -Toto
2. The Street only knew your name (extended version) - Van Morrison
3. What a wonderful world - Eva Cassidy
4. Beethoven: Triple Concerto in C Major, Allegro - Mutter/Ma/Barenboim
and a few others.

Results: 2 out of 3 reviewers had very similar results with the Purifi in 1st place and the Benchmark in last place in all or a majority of the time, while one reviewer had mixed results from which we couldn't determine a specific pattern.

I was one of the reviewers and I picked the exact same order 100% of the time of the Purifi in the top spot followed by the McIntosh and the Benchmark as a distant 3rd place. What I noticed that distinguished the best (Purifi) from the worst (Benchmark) in my case was the following: The Purifi had better clarity, detail and high end compared to the other two, especially the Benchmark and it was noticeable. It was especially easy to hear differences for me with complex stringed instruments like a strummed acoustic guitar or cello and cymbals and triangles.

This wasn't the result that we were anticipating going into the comparisons. Based on the ASR reviews all of us went into thinking that Benchmark would be far ahead of the other 2 amplifiers based on its measurements. One of the reasons for the bakeoff was that one of the reviewers was seriously considering buying the Benchmark but wanted to do a comparison before. He decided to pan the Benchmark and buy the Purifi instead.

I hate add fuel to this debate about whether amplifiers with great specs can sound different, as I went into this believing that it wasn't possible, but some of the differences we heard weren't subtle. I am really curious as to whether there is some psycho-acoustic reason for this or limitations in what and how we measure these systems to begin with. I actually bought the Purifi Eval 1 and an NAD 2200 which I upgraded based on the measurements from ASR with the assumption that I wouldn't be able to hear a difference and I was quite surprised by the fact that there were differences. One thing I did notice was that my ability to hear a difference was dependent on the speakers used. On a few run of the mill speakers, I really couldn't tell much of a difference as they themselves just don't have the detail resolution of the S5 or the Magnepan LRS.

I must add it was loads of fun and I would highly recommend comparing the Benchmark and Purifi since in theory there should be no real difference that you should be able to detect to see if you can.

Welcome to the club, with this sentence, "One thing I did notice was that my ability to hear a difference was dependent on the speakers used", you have just become a subjectionist.

 

[radix]

I guess I'll wade in here.

It seems like you would want 3 preamps and level adjust them once. Then simply switch the spdif cable. Or is that what you did? From the OP, it seemed like they re-adjusted the D90se for each switch, i assume based on the D90se display? I'm not sure how accurate that is.

Does playing loud music during the switch affect the outcome? (i.e. does a person's perception of the previous amp depend on the interlude music/volume?)

In post #41, the OP said they used high gain on the ABH2. I am not sure why this was necessary. Mid-gain is for 4V balanced input, which is what the Pre90 should be putting out.

 

[pogo]

Yeah it would be nice to have had a 'switch' as that would have made the process faster allowing for more samples also. We will think about that for next time.

But exactly with a 'switch' you build in a systematic error, which would have the greatest effect especially with the purifi with the very high DF.
Has this influence ever been measured by a tester, who have performed similar tests with a switch?

 

[pogo]

Based on my post #64, the amps will become more equal when using a speaker 'switch' with an assumed contact resistance of 0.05mOhm (2x0.025) and that is less desired in a comparison! Idealized calculation approach:

AHB2 (8ohms, 6.5kHz): DF > 65 (without 'switch'), > 47 (with 'switch')
Purifi (8ohms, 6.5kHz): DF > 1500 (without 'switch'), > 145 (with 'switch')

 

[SIY]

Based on my post #64, the amps will become more equal when using a speaker 'switch' with an assumed contact resistance of 0.05mOhm (2x0.025) and that is less desired in a comparison! Idealized calculation approach:

AHB2 (8ohms, 6.5kHz): DF > 65 (without 'switch'), > 47 (with 'switch')
Purifi (8ohms, 6.5kHz): DF > 1500 (without 'switch'), > 145 (with 'switch')

Still haven't bothered doing the back of the envelope calculation yet, apparently. Anyone with the integrity to do so will quickly find out how ludicrous your (as yet totally unsupported) sonic claims are.

 

[pogo]

how ludicrous your (as yet totally unsupported) sonic claims are.

Here is an interesting video in german from a sound engineer, who explains which technical data are actually relevant and audible nowadays:
Link

You can also add a subtitle in your native language.

 

[SIY]

Here is an interesting video in german from a sound engineer, who explains which technical data are actually relevant and audible nowadays:
Link

You can also add a subtitle in your native language.

Ich kann Deutsch sprechen und lesen.

Better than you can do a simple back of the envelope plausibility calculation, apparently.

 

[mocenigo]

https://www.isotopx.com/wp-content/uploads/2018/09/ATONA-TIMS-Revolutionising-signal-amplification-tech

"Sorry, that page can’t be found"

 

[mocenigo]

Ich kann Deutsch sprechen und lesen.

Hiermit bestätige ich die Korrektheit dieses Satzes!

 

[xaxxon]

We recently got a few folks together to do as close to a 'blind test' as we could of the Purifi Eval1, McIntosh MA252 and the Benchmark AHB2. We had 3 reviewers and a 4th person who switched the gear.

Test Setup: Topping D90SE (balanced outputs) directly connected to each amplifier.
Streamer: Tidal Bluesound Node connected via optical SPDIF to Topping D90SE
Speaker: Magico S5 MKII
Randomization: For each song selection: Each amplifier was randomly assigned a number 1,2,3, A random amplifier order was used for playback.
Level Equalization: For each song/amplifier, a 1K sine wave was used to normalize the level to about 0.1 Db using a calibrated microphone, Focusrite 2i2 and REW before playback of each selection. The level was managed using the Topping D90SE in pre-amp mode and managed to about 90-95 db.
Process: 6 songs. Each song was played and the reviewers wrote comments and then rank ordered amplifiers. Between switches loud music was played to drown out any switching related noises that could indicate which amplifier was being connected.

Some of the selections used:
1. Africa -Toto
2. The Street only knew your name (extended version) - Van Morrison
3. What a wonderful world - Eva Cassidy
4. Beethoven: Triple Concerto in C Major, Allegro - Mutter/Ma/Barenboim
and a few others.

Results: 2 out of 3 reviewers had very similar results with the Purifi in 1st place and the Benchmark in last place in all or a majority of the time, while one reviewer had mixed results from which we couldn't determine a specific pattern.

I was one of the reviewers and I picked the exact same order 100% of the time of the Purifi in the top spot followed by the McIntosh and the Benchmark as a distant 3rd place. What I noticed that distinguished the best (Purifi) from the worst (Benchmark) in my case was the following: The Purifi had better clarity, detail and high end compared to the other two, especially the Benchmark and it was noticeable. It was especially easy to hear differences for me with complex stringed instruments like a strummed acoustic guitar or cello and cymbals and triangles.

This wasn't the result that we were anticipating going into the comparisons. Based on the ASR reviews all of us went into thinking that Benchmark would be far ahead of the other 2 amplifiers based on its measurements. One of the reasons for the bakeoff was that one of the reviewers was seriously considering buying the Benchmark but wanted to do a comparison before. He decided to pan the Benchmark and buy the Purifi instead.

I hate add fuel to this debate about whether amplifiers with great specs can sound different, as I went into this believing that it wasn't possible, but some of the differences we heard weren't subtle. I am really curious as to whether there is some psycho-acoustic reason for this or limitations in what and how we measure these systems to begin with. I actually bought the Purifi Eval 1 and an NAD 2200 which I upgraded based on the measurements from ASR with the assumption that I wouldn't be able to hear a difference and I was quite surprised by the fact that there were differences. One thing I did notice was that my ability to hear a difference was dependent on the speakers used. On a few run of the mill speakers, I really couldn't tell much of a difference as they themselves just don't have the detail resolution of the S5 or the Magnepan LRS.

I must add it was loads of fun and I would highly recommend comparing the Benchmark and Purifi since in theory there should be no real difference that you should be able to detect to see if you can.

That's great, but the correct following step is to measure the difference. If you can hear it you can easily measure it. Figure out what's different in the measurements and then see if it can be adjusted for to see if it's just a defaults preference or an intrinsic part of the amp's design. For example, does the purifi have a house sound? Maybe it's a sound that was good in that setting but if it's an inaccuracy and it is something you could EQ in for another amp if you wanted that change, then it's not necessarily ideal to have it baked in.

Of course you could always find that you just didn't control well enough in the experiment, too.

Then you can have amir do the test and when he finds it (since apparently it's pretty obvious) he can add the measurement to his tests.

 

[Galliardist]

That's great, but the correct following step is to measure the difference. If you can hear it you can easily measure it. Figure out what's different in the measurements and then see if it can be adjusted for to see if it's just a defaults preference or an intrinsic part of the amp's design. For example, does the purifi have a house sound? Maybe it's a sound that was good in that setting but if it's an inaccuracy and it is something you could EQ in for another amp if you wanted that change, then it's not necessarily ideal to have it baked in.

Of course you could always find that you just didn't control well enough in the experiment, too.

Then you can have amir do the test and when he finds it (since apparently it's pretty obvious) he can add the measurement to his tests.

Why on earth are you reopening this three year old thread when the test was garbage in the first place?

 

[xaxxon]

Why on earth are you reopening this three year old thread when the test was garbage in the first place?

Let's say I gave you an answer. What would you choose to do with that answer?

Before you actually type the answer think about what you're going to type and ask yourself if it is productive. Whatever you're thinking about typing you could have typed in your first response but instead chose to go with something that wasn't productive.

 

[Galliardist]

Let's say I gave you an answer. What would you choose to do with that answer?

Before you actually type the answer think about what you're going to type and ask yourself if it is productive. Whatever you're thinking about typing you could have typed in your first response but instead chose to go with something that wasn't productive.

Let's just say that if you want "productive" in any way, here and now is not the place it is going to happen. This thread is three years old. The person who posted their test was last here over two years ago. No way is the person who isn't here going to send amps they possibly have no access to three years later, to be tested by Amir.

Also, I stand by my original complaint about this test on page 1 - since they could only hear a difference between the amplifiers when they used a difficult to drive speaker, the most likely cause was that one or more of the amps was being used outside of their performance envelope. Other criticisms of the test have also been raised here.

I really, really do not see the slightest reason for your post whatsoever. So, let's say you gave me an answer. You might even enlighten me. I can't however imagine what you might say that is, as you put it, "productive".

If you want "productive" then maybe direct your attention to currently active conversations with people who are still here rather than waking up old threads that were never really productive in the first place.

 

[MAB]

Let's just say that if you want "productive" in any way, here and now is not the place it is going to happen. This thread is three years old. The person who posted their test was last here over two years ago. No way is the person who isn't here going to send amps they possibly have no access to three years later, to be tested by Amir.

Also, I stand by my original complaint about this test on page 1 - since they could only hear a difference between the amplifiers when they used a difficult to drive speaker, the most likely cause was that one or more of the amps was being used outside of their performance envelope. Other criticisms of the test have also been raised here.

I really, really do not see the slightest reason for your post whatsoever. So, let's say you gave me an answer. You might even enlighten me. I can't however imagine what you might say that is, as you put it, "productive".

If you want "productive" then maybe direct your attention to currently active conversations with people who are still here rather than waking up old threads that were never really productive in the first place.

Yeah, none of it is going to be revisited.
Agree, test had multiple layers of fail. Also, the level matching was done at 110 dB, this is high power for the PuriFi, and well beyond what the Benchmark and McIntosh can do. No chance the trials were actually level-matched.