run REW on different amps that measure similiar

[ta240]

I know it is an ongoing argument that 'testing doesn't reveal everything about an amp' but I'm curious if anyone has ever run REW on several different amps that measure very similar. I've done it with quite different amps and got results that overlapped in some areas and in other areas were way different and in some small areas were just a bit different. But I'm curious how close the results would be with different amps that measured very close.

 

[pozz]

Just keep in mind that your room isn't anechoic and that you'll encounter run-to-run variations when doing acoustics measurements.

@RayDunzl has posted plenty of REW measures of his Krell amps, but not multiple amp measurements in the same sitting IIRC.

 

[pma]

REW is IMO not good for detailed amplifier measurements, especially not for low distortion amplifiers. It is good for fast view at the amp, but that's all. JAVA support usually does not work at 24bits, with ASIO it is better but not all ASIO drivers are supported perfectly. There is no WASAPI option. Sweep measurement is fast, but not precise.

As an example, the distortion with frequency as you can see below does not exist in the DUT. It was created by REW.
Arta/Steps is much better for amplifier measurements.

 

[mitchco]

@ta240 Archimago uses REW for amplifier measurements. Here is one example: http://archimago.blogspot.com/2020/01/measurements-hypex-ncore-nc252mp.html
He is testing a handful of amps using his MOAR approach to be consistent. You could reach out to him on his blog (email at top) or PM as he is a member here to get his thoughts.

 

[Blumlein 88]

REW is IMO not good for detailed amplifier measurements, especially not for low distortion amplifiers. It is good for fast view at the amp, but that's all. JAVA support usually does not work at 24bits, with ASIO it is better but not all ASIO drivers are supported perfectly. There is no WASAPI option. Sweep measurement is fast, but not precise.

As an example, the distortion with frequency as you can see below does not exist in the DUT. It was created by REW.
Arta/Steps is much better for amplifier measurements.

View attachment 48879

I don't know how you specifically obtained this result. I've measured real DACs with real ADCs and obtained better results than you are showing.

I see Mitchco has also pointed to Archimago getting results that aren't like those shown in your post.

I don't know what the OP had in mind. If you are measuring with a microphone room noise and loudspeaker distortion is the limiting factor. If you are measuring amplifier outputs then you can use REW quite well for measuring.

 

[ta240]

I don't know how you specifically obtained this result. I've measured real DACs with real ADCs and obtained better results than you are showing.

I see Mitchco has also pointed to Archimago getting results that aren't like those shown in your post.

I don't know what the OP had in mind. If you are measuring with a microphone room noise and loudspeaker distortion is the limiting factor. If you are measuring amplifier outputs then you can use REW quite well for measuring.

Measuring with a microphone. It wouldn't be a test to see what was the best amp so with the room noise and loudspeaker distortion staying the same it would just be interesting to see the differences. There is nothing really scientific about it; just a curiosity of how close differently designed amps that test similarly perform throughout the frequency sweep through the same speakers in the same room.

I no longer have a microphone but the quick comparison I did ended up with curves that were quite different in the low frequencies and then matched up close through some and drifted apart in other sections. The most interesting part to me were the little changes in frequency where one amp's line dropped quickly and another was more of a curved drop. I would have needed to run multiple sweeps on each amp to make sure those weren't aberrations. But if the differences existed with multiple sweeps, to me, it would be interesting.

 

[pma]

If you measure same speaker, with same microphone position at the same place and change the amps and they are competently designed, then the distortion is same. I have tried it many times. If not, then you need to measure amp and speaker interaction. Measure the distortion directly at speaker box terminals. Use balanced input to prevent signal ground loop.

 

[ta240]

If you measure same speaker, with same microphone position at the same place and change the amps and they are competently designed, then the distortion is same. I have tried it many times. If not, then you need to measure amp and speaker interaction. Measure the distortion directly at speaker box terminals. Use balanced input to prevent signal ground loop.

Not being super familiar with REW and the terminology; is the distortion what shows on the frequency plot like this:

I'm just looking at how the frequency generated reaches the microphone (and the listener)

I guess what I'm looking at measuring then is the amp and speaker interaction and how much it can vary with comparable, quality amps. And if it does vary then that would mean those amps still have an audio signature.

 

[restorer-john]

I know it is an ongoing argument that 'testing doesn't reveal everything about an amp'

I'd take it a step further and it is something I've been meaning to do for a while. Use a digital recorder (mounted on a tripod in a fixed position) to capture the actual sound in your room with multiple level matched (electrically) amplifiers. The room and its vagaries are a constant.

I know I can hear differences between amplifiers that measure perfectly well and if simple recordings can demonstrate that, within the confines of the limitations of a digital recorder (small Tascam), then we may be able to investigate further.

 

[Blumlein 88]

I'd take it a step further and it is something I've been meaning to do for a while. Use a digital recorder (mounted on a tripod in a fixed position) to capture the actual sound in your room with multiple level matched (electrically) amplifiers. The room and its vagaries are a constant.

I know I can hear differences between amplifiers that measure perfectly well and if simple recordings can demonstrate that, within the confines of the limitations of a digital recorder (small Tascam), then we may be able to investigate further.

So is this going to be a stereo recording at the normal LP, going to be a recording 1 meter from the speaker or what do you have in mind?

While I've not always investigated fully, good amps that sound different have normally come down to FR differences due to output impedance or volt/amp limitations when actually attached to complex speakers.

 

[restorer-john]

So is this going to be a stereo recording at the normal LP, going to be a recording 1 meter from the speaker or what do you have in mind?

While I've not always investigated fully, good amps that sound different have normally come down to FR differences due to output impedance or volt/amp limitations when actually attached to complex speakers.

I figured I'll put the recorder on a tripod in a position where the recording sounds pretty decent with not too much room interaction to muddy the waters. I'll experiment and find a single spot and not move the recorder once located. What amazed me last time I recorded my system was the room acoustics were really obvious although when listening "live" my brain simply locked it all out.

I'll pick a level that will give me a decent lift above ambient noise but not try for silly SPLs. That way, various amplifiers of wildly differing power outputs can be compared all at a similar nominal (normal) level. Preset all the amplifiers to exactly the same output level (V across a common resistor, just in case output impedance is an issue) and use the comparator or simply plug and unplug each amp. Obviously, all the amplifiers will have their tone bypassed, line direct on and no filters or loudness engaged. Any amplifier with a significant variation in frequency response will not be included in the experiment.

Assuming it doesn't turn into an all weekend affair, I'll put the FR plots up for each amplifier (at the preset volume position to ensure variations from pot position are not an issue in differentiating the "sound") and do an output impedance (damping factor) vs Freq if I get time.

The listening room is a mess anyway, so I might as well have more fun.

What do you suggest in terms of period for each sample? 30 seconds? Less or more? I'll pick something where I think the recording is good and use a reasonable pair of 6.5" two ways and identify the speakers once I've sorted the room/recorder/speaker interaction for the "best" sound at the LP.

Now I know people will say "we can hear stuff the recorder can't so the experiment may not prove anything" To that I say, it doesn't matter. The recorder should capture any audible differences between the various amplifiers if they are as clear as I think some can be. I'm prepared to be wrong and admit it. Maybe, depending on how good and identical the file lengths and levels are, perhaps PKane's difference software may come in handy to look at the results?

 

[solderdude]

What amazed me last time I recorded my system was the room acoustics were really obvious although when listening "live" my brain simply locked it all out.

Have the same experience. Even when you record with a dummy head it doesn't quite sound the same as when you listen 'live'.

 

[Blumlein 88]

I figured I'll put the recorder on a tripod in a position where the recording sounds pretty decent with not too much room interaction to muddy the waters. I'll experiment and find a single spot and not move the recorder once located. What amazed me last time I recorded my system was the room acoustics were really obvious although when listening "live" my brain simply locked it all out.

I'll pick a level that will give me a decent lift above ambient noise but not try for silly SPLs. That way, various amplifiers of wildly differing power outputs can be compared all at a similar nominal (normal) level. Preset all the amplifiers to exactly the same output level (V across a common resistor, just in case output impedance is an issue) and use the comparator or simply plug and unplug each amp. Obviously, all the amplifiers will have their tone bypassed, line direct on and no filters or loudness engaged. Any amplifier with a significant variation in frequency response will not be included in the experiment.

Assuming it doesn't turn into an all weekend affair, I'll put the FR plots up for each amplifier (at the preset volume position to ensure variations from pot position are not an issue in differentiating the "sound") and do an output impedance (damping factor) vs Freq if I get time.

The listening room is a mess anyway, so I might as well have more fun.

What do you suggest in terms of period for each sample? 30 seconds? Less or more? I'll pick something where I think the recording is good and use a reasonable pair of 6.5" two ways and identify the speakers once I've sorted the room/recorder/speaker interaction for the "best" sound at the LP.

Now I know people will say "we can hear stuff the recorder can't so the experiment may not prove anything" To that I say, it doesn't matter. The recorder should capture any audible differences between the various amplifiers if they are as clear as I think some can be. I'm prepared to be wrong and admit it. Maybe, depending on how good and identical the file lengths and levels are, perhaps PKane's difference software may come in handy to look at the results?

I suggest being no more than 2 meters from the speaker. For this experiment moving speakers so you can record 2 meters from each in stereo might work better than in their normal position. With your Tascam that will basically mean a triangle 2 meters to a side with the Tascam at the apex.

If you have a pair of external mics to feed the Tascam, I'd suggest using speakers where they are and miking each one at 2 meters (1.5 would be better depending upon how close the drivers are).

My opinion is 30 second snippets are just about optimum for this sort of thing.

And yes, you experienced what people mean by our ears mostly ignore the room. Our ears ignore quite a few early reflections/echoes live. The microphones don't, and when you play the recording those recorded reflections are coming from the speaker as direct sound and our ears don't ignore those. This unmasks the reflections you normally are incapable of hearing. Which is why you need to record closer than you listen.

I don't know what will happen. I've made such recordings though not while switching out amps. I'd probably throw in one amp that is really way out in left field to see how obvious it is or isn't.

 

[restorer-john]

If you have a pair of external mics to feed the Tascam,

Plenty of mics, but only a few non-vocals mics and no matching full range pairs (2). Plenty of electret elements, but I don't feel like building a low noise front end as well. If I do that, I might as well go direct into the PC M-Audio card. I guess I could use the Scarlett 2i2 and use the phantom power...

The little Tascam is easy, like a point and shoot- just works, think I'll stick with it.

 

[restorer-john]

I suggest being no more than 2 meters from the speaker.

Speakers are about 8ft apart, so 2400mm, I'll try an equilateral triangle and see what it sounds like. If it's crap sounding, I'll try some other positions. I'll go through some speakers to find the ones that sound the "best" on the recording and with the content, but they will likely be smallish speakers as moving big ones and all the amplifiers in this room is not a good idea...

If it's not conslusive or sounds dreadful, I'll pull the pin as that won't really prove anything. Hopefully I'll end up with a bunch of decent quality 30 second files we can play with to see if anyone can hear amplifier differences.

 

[Blumlein 88]

Speakers are about 8ft apart, so 2400mm, I'll try an equilateral triangle and see what it sounds like. If it's crap sounding, I'll try some other positions. I'll go through some speakers to find the ones that sound the "best" on the recording and with the content, but they will likely be smallish speakers as moving big ones and all the amplifiers in this room is not a good idea...

If it's not conslusive or sounds dreadful, I'll pull the pin as that won't really prove anything. Hopefully I'll end up with a bunch of decent quality 30 second files we can play with to see if anyone can hear amplifier differences.

I don't want to complicate your experiment too much. It probably is worth your time to give a quick read of Williams Stereo Zoom article.
https://microphone-data.com/media/filestore/articles/Stereo zoom-10.pdf

It helps you pick an angle between the microphones which will capture the speakers the way you want if you think of them as musicians. I don't know if your Tascam lets you use a variable angle or not. If it has an M/S setup, you'll benefit form perhaps changing the ratio of mid and side after the recording to see what captured the speakers best.

Which Tascam are you going to use?

 

[pma]

Measuring with a microphone. It wouldn't be a test to see what was the best amp so with the room noise and loudspeaker distortion staying the same it would just be interesting to see the differences. There is nothing really scientific about it; just a curiosity of how close differently designed amps that test similarly perform throughout the frequency sweep through the same speakers in the same room.

OK, as you probably know, I like hard data. I have just made a measurement of the speaker in my room with 2 different amplifiers.
Conditions:
- microphone on a mic stand, position is fixed for both measurement
- speaker position is fixed
- SPL level made same for both amplifiers

As a result, measured frequency responses overlap. There is no difference. It looks like one plot, but they are 2 plots, red and blue. There is a lot of room modes and reflections, as might be expected. Amplifiers used
http://pmacura.cz/pa2.htm
http://pmacura.cz/powersym_en.htm

The result

Edit: SPL axis not calibrated, voltage level is 8V

 

[Blumlein 88]

OK, as you probably know, I like hard data. I have just made a measurement of the speaker in my room with 2 different amplifiers.
Conditions:
- microphone on a mic stand, position is fixed for both measurement
- speaker position is fixed
- SPL level made same for both amplifiers

As a result, measured frequency responses overlap. There is no difference. It looks like one plot, but they are 2 plots, red and blue. There is a lot of room modes and reflections, as might be expected. Amplifiers used
http://pmacura.cz/pa2.htm
http://pmacura.cz/powersym_en.htm

The result

View attachment 48963

How about showing it with distortion.

 

[pma]

How about showing it with distortion.

Attached (I expect it is clear it is measured on acoustical side, as speaker distortion). Measured at 8V at speaker terminals.

Attached.

 

[restorer-john]

I like it Pavel.

Golden ears will hear that sweetness and change in the treble distortion you know...