• 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 daily reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Integrated Amplifier Comparison: $90 vs $9000

OP
G

GXAlan

Major Contributor
Forum Donor
Joined
Jan 15, 2020
Messages
3,753
Likes
5,799
It doesn’t matter what graph you show. As long as there is no proof of causation it’s just speculation.

1698513951798.png


Speculation is normal in your conclusion. This is from Dr Toole’s paper on room reflections talking about Seraphim’s measurements and speculation in 1961 and how his work in 1989 provides added insights, supporting this,

You generally tend to stop at speculation. And I think that is a problem, because it invites others to speculate and extrapolate, leading to conclusions that might not be correct.

Except speculation leads other scientists to evaluate and explore.
I also have no problem with speculation. I have a problem when it stops there.Yes, let’s go then!
The beauty of ASR and message boards is that it’s an ongoing. You need only look at my history to see that the investigations continue. Dr Toole was a full time audio research. I am a simple hobbyist sharing my notes along the way :)
Can one extract the impulse response from the recordings in Delta wave? If so, it should be trivial to create a filter that converts the Fosi impulse to the Marantz impulse.

That would be interesting. But that filter cannot easily be applied in real time. The filter MAY have input level sensitivity which would be super interesting.

The scenario would have to be
1) Impulse response at a fixed level to match two amplifiers
2) Apply the fixed VST filter, and compare different music at low medium and high volumes.

For something like tube non linearity, you might have a single transfer function to apply at all levels?
 

voodooless

Grand Contributor
Forum Donor
Joined
Jun 16, 2020
Messages
10,058
Likes
17,423
Location
Netherlands
The beauty of ASR and message boards is that it’s an ongoing.
Yes, I agree.

I think the biggest critique I have is that you just dropped it as is, and did not state that further efforts are needed to figure out what is going on here. Offerings some next steps would have been a good start.

That would be interesting. But that filter cannot easily be applied in real time.
Sure, it’s just a convolution filter. You’ll just need a second one with same length and a clean impulse at the same spot for the original track to have them times exactly the same (or a simple delay will also do).

But realtime is not really needed. I would just start with the two files you already have. Fix them and then ABX.
 

antcollinet

Master Contributor
Joined
Sep 4, 2021
Messages
7,171
Likes
11,860
Location
UK/Cheshire
Clarified above. That does not apply but is a great thought for the next round of testing.
Thanks for the clarification it's interesting testing. I'll be really interested to see any comparisons of electrical versus acoustic testing.

Edit: I may even get the kit to do it myself one day :)
 
OP
G

GXAlan

Major Contributor
Forum Donor
Joined
Jan 15, 2020
Messages
3,753
Likes
5,799
Yes, I agree.

I think the biggest critique I have is that you just dropped it as is, and did not state that further efforts are needed to figure out what is going on here. Offerings some next steps would have been a good start.
Ah. Then we are in agreement. For the record, this is all pilot data and a lot of next steps are considered.

Sure, it’s just a convolution filter. You’ll just need a second one with same length and a clean impulse at the same spot for the original track to have them times exactly the same (or a simple delay will also do).

But realtime is not really needed. I would just start with the two files you already have. Fix them and then ABX.

Wait. If you do this, your convolution filter will be perfect and you’re just ABX’ing the same signal twice.

Waveform A compared to Waveform B = convolution filter.

Waveform A + convolution filter = Waveform B.

So of course they will be similar (and I have done plenty of ABX tests where I couldn’t hear differences).

The real question is for what envelope does that convolution filter work? Only that portion of the music? For certain frequencies better than others?

What would be interesting is to do this with two SPEAKER recordings. If you took two different speakers, recorded them, and then convoluted them what is the envelope that you could take two different speaker recordings and then convolute them?

Of course, this is something anyone else can set up too. Happy to run the blind testing.

The other challenge is that until I win the lottery, I have found my end game speakers for the foreseeable future, which are active speakers, so it may be a while before I have a lot of gear again to play with and measure.
 

voodooless

Grand Contributor
Forum Donor
Joined
Jun 16, 2020
Messages
10,058
Likes
17,423
Location
Netherlands
Wait. If you do this, your convolution filter will be perfect and you’re just ABX’ing the same signal twice.
Well, that would prove that the audible difference between the amps is just frequency and phase response.

The real question is for what envelope does that convolution filter work? Only that portion of the music? For certain frequencies better than others?
Yeah, that will depend highly on the amount of information derived from the bit of music and how well the impulse response can be constructed from that.

What would be interesting is to do this with two SPEAKER recordings. If you took two different speakers, recorded them, and then convoluted them what is the envelope that you could take two different speaker recordings and then convolute them?
Yes, that should work. Although it will be highly dependent on the listener position. It must very close to the mic position. With quite a few caveats you can probably make one speaker sound like another in a totally different room, at least for that tiny sweet spot.
 
Last edited:

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,453
Likes
37,807
Location
Gold Coast, Queensland, Australia
I'm late to the party, but just to add to this.

I'm really struggling with the comparison data being recorded via a microphone. It seems to me that doing this there are just too many environmental variations that can cause measured differences that are not due to the amplifiers under test.

A plane going overhead (or dog barking in the distance, or... any of a million other possible environmental noises) - a different recording.

Same for any changes in the room causing changes to reflections - such as the person carrying out the test not being in exactly the same position, or small movements of the measurement mic.

What about heating devices switching on/off - (speculation warning) does sound passing through a volume of warm/less dense air get measurably impacted? Even if only enough for a small phase shift to alter interference patterns?

Surely if we are looking for differences in electronics it would be far more accurate to measure electrically at the speaker terminals. Probably easier too once an appropriate filter has been built.

Actually, the audio output is where the rubber hits the road. It is a calibrated measurement microphone- no doubt a lot more accurate than your ears. An electrical (at the speaker terminals) recording tells you absolutely nothing how the interaction with the loudspeaker actually translates to the sound in the room.

The fact that each recording is a tiny bit different is good. It doesn't matter about a dog or two barking, other than as a 'tell' of which recording is which. After all, the differences in sonic reproduction is what he was listening for. You can listen 'through' any aberrations to hear changes such as that.

His experiment is similar to many I have done over the years and the outcome is much the same. There are two camps it seems: the ones who think all amplifiers sound the same and the ones who know they do not. Funny thing is, the former category is always the noisiest with ASR members when someone questions their long held belief, generally without ever presenting a shred of evidence to support their own position.

Basically, if you have an amplifier that modifies its transfer function dynamically into varying loads, at varying frequencies and power outputs, it will likely sound different to one that does not have those characteristics. An amplifier that exhibits the former characteristics, is not HiFi, is not accurate and should be excluded from any self respecting audiophiles wish list.
 
OP
G

GXAlan

Major Contributor
Forum Donor
Joined
Jan 15, 2020
Messages
3,753
Likes
5,799
Well, that would prove that the audible difference between the amps is just frequency and phase response.

That just proves that math is math? Once noise is taken out of the equation, there is not much left that can be different between two recorded waveforms if you can manipulate FR (amplitude) and phase?

if you have an amplifier that modifies its transfer function dynamically into varying loads, at varying frequencies and power outputs, it will likely sound different to one that does not have those characteristics.
+100

And along those same lines, a superb 1 kHz 5 W SINAD measurement and FR sweep into a resistor does not inform you if there is variability in the transfer function in real-world speaker loads with musical content…. But I’m not proposing any sort of voodoo or intangible. It’s all measurable/recordable with an ADC or even a microphone.
 

voodooless

Grand Contributor
Forum Donor
Joined
Jun 16, 2020
Messages
10,058
Likes
17,423
Location
Netherlands
Once noise is taken out of the equation, there is not much left that can be different between two recorded waveforms if you can manipulate FR (amplitude) and phase?
In this particular case, that would be the hypothesis.
 

antcollinet

Master Contributor
Joined
Sep 4, 2021
Messages
7,171
Likes
11,860
Location
UK/Cheshire
Actually, the audio output is where the rubber hits the road. It is a calibrated measurement microphone- no doubt a lot more accurate than your ears. An electrical (at the speaker terminals) recording tells you absolutely nothing how the interaction with the loudspeaker actually translates to the sound in the room.

The fact that each recording is a tiny bit different is good. It doesn't matter about a dog or two barking, other than as a 'tell' of which recording is which. After all, the differences in sonic reproduction is what he was listening for. You can listen 'through' any aberrations to hear changes such as that.

His experiment is similar to many I have done over the years and the outcome is much the same. There are two camps it seems: the ones who think all amplifiers sound the same and the ones who know they do not. Funny thing is, the former category is always the noisiest with ASR members when someone questions their long held belief, generally without ever presenting a shred of evidence to support their own position.

Basically, if you have an amplifier that modifies its transfer function dynamically into varying loads, at varying frequencies and power outputs, it will likely sound different to one that does not have those characteristics. An amplifier that exhibits the former characteristics, is not HiFi, is not accurate and should be excluded from any self respecting audiophiles wish list.

What I am struggling with:

I find it difficult to see how a tiny (inaudible) variation in the electrical level can result in a larger variation (audible) in the acoustic level (except perhaps if resonance can amplify the difference? - but then it is a combination of amp and speaker, not just amp variation).

Effects of modified transfer function with that load are visible at the speaker terminals.

Comparing waveforms graphically is surely invalid. You cant tell an amplifier created difference from an artefact of ambient noise.
 

Geert

Major Contributor
Joined
Mar 20, 2020
Messages
1,908
Likes
3,404
And along those same lines, a superb 1 kHz 5 W SINAD measurement and FR sweep into a resistor does not inform you if there is variability in the transfer function in real-world speaker loads with musical content…

How do you mean? If FR response measurements of an amp for 4 and 8 loads vary substantially, as is in the measurements of the DUT, then we do know that this can be audible in practise since the impedance of most speakers varies even more. The impedance of the JBL's you used varies between 3 and 22 ohm. The basic measurements can't exactly predict the magnitude of the problem as it depends on the character of the load, but you can't say they don't inform.
 

MaxwellsEq

Major Contributor
Joined
Aug 18, 2020
Messages
1,503
Likes
2,243
I understand why people think in-room measurement is a valid direction for measurement, since we have been exposed to decades of magazine reviews based on well-known reviewers swapping a Yamaha amp with a Marantz amp and reporting a substantial difference (which is often not evident in the lab report). Hence: "there must be a way to reproduce this effect repeatably", and eliminate all the human biases at work in sighted listening tests.

But on the other hand, I struggle with how to do this in a controlled way. Given what we know about the gross effect of millimetre difference in positioning of speakers, listening position and furniture, I would struggle to trust a limited run. Perhaps the answer is to make hundreds of runs of one DUT over a long period interspersed with hundreds of runs of the other DUT and then to use the averaged results for the hundreds of runs. We also need to ensure absolute voltage level matching at the speaker terminals and ensure that neither DUT is being operated outside its performance envelope.

But even then I would want to understand what causes the difference.
 

Geert

Major Contributor
Joined
Mar 20, 2020
Messages
1,908
Likes
3,404
An electrical (at the speaker terminals) recording tells you absolutely nothing how the interaction with the loudspeaker actually translates to the sound in the room.

What interaction differences would be hidden from speaker terminal measurements when comparing 2 different amps with the same speakers?
 

DSJR

Major Contributor
Joined
Jan 27, 2020
Messages
3,154
Likes
4,266
Location
Suffolk Coastal, UK
Not really on topic and I can't afford to purchase one merely for the purpose, but has anyone measured the humble Yamaha AS 201 amp, which offers some power and facilities for not much money? I gather the chassis isn't the same at all as the better known models higher up the scale and it could be tested and judged as a low price 'traditional amp design which is widely available from dealers' reference.
 

Galliardist

Major Contributor
Joined
Jun 26, 2021
Messages
2,547
Likes
3,244
Location
Sydney. NSW, Australia
What interaction differences would be hidden from speaker terminal measurements when comparing 2 different amps with the same speakers?
Well, you may be left with a detailed electrical measurement of the difference between two unknowns. Can you tell just from those measurements which is the better representation of the music (for whatever meaning you ascribe to "better")?
 

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,453
Likes
37,807
Location
Gold Coast, Queensland, Australia
Not really on topic and I can't afford to purchase one merely for the purpose, but has anyone measured the humble Yamaha AS 201 amp, which offers some power and facilities for not much money? I gather the chassis isn't the same at all as the better known models higher up the scale and it could be tested and judged as a low price 'traditional amp design which is widely available from dealers' reference.

I've got a Yamaha A-S300 here. Same as the A-S301 apat from the inbuilt D/A.

What parameters do you want to know specifically?
 

DSJR

Major Contributor
Joined
Jan 27, 2020
Messages
3,154
Likes
4,266
Location
Suffolk Coastal, UK
I've got a Yamaha A-S300 here. Same as the A-S301 apat from the inbuilt D/A.

What parameters do you want to know specifically?
The 300/310 is a similar base chassis to the larger ones isn't it? The 201 looks to be totally different and actually more powerful possibly. It may have been mentioned here a good while back with internal pics and I'll have to search for it.

Thanks for offering :)
 

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,453
Likes
37,807
Location
Gold Coast, Queensland, Australia
The 201 looks to be totally different and actually more powerful possibly.

Not sure. Haven't had one to play with. I assumed Yamaha model numbers would have put it well below a 300/301.

Thanks for pointing that out.
 

DSJR

Major Contributor
Joined
Jan 27, 2020
Messages
3,154
Likes
4,266
Location
Suffolk Coastal, UK
Here it is - and I don't think it's bad at all on the standard tests this review did -


Price is anything from £200 to £300 here, but the bigger models (301 and even the 501) in the range don't seem much more expensive, so no idea on value these days. The test above indicates around 100WPC or more at 1kHz on the slightly raised 'distortion knee at -70dB or so, so Yamaha being Yamaha, I'm assuming a clean minimum 70WPC across the frequency range.
 

Geert

Major Contributor
Joined
Mar 20, 2020
Messages
1,908
Likes
3,404
Well, you may be left with a detailed electrical measurement of the difference between two unknowns.

But the difference would still be indicative of what might happen to the sound, like one amp might sound brighter than the other. The magnitude of the audible difference might not be obvious, but that's still a different story than "tells you absolutely nothing".

Can you tell just from those measurements which is the better representation of the music (for whatever meaning you ascribe to "better")?

Don't know where 'better' entered the discussion. 'Better' is subjective. Does a microphone measurement shows what sounds better, and also above the Schroeder frequency (since you found differences in what you called the 'vocal region')?
 
Last edited:
OP
G

GXAlan

Major Contributor
Forum Donor
Joined
Jan 15, 2020
Messages
3,753
Likes
5,799
The basic measurements can't exactly predict the magnitude of the problem as it depends on the character of the load, but you can't say they don't inform.

Agreed. My original statement had too much hyperbole.
 
Top Bottom