Listening Comparison Test: DSP Phono vs SUT

[Blumlein 88]

Here is the result with Deltawave FR EQ and phase EQ engaged. Makes me think most of the difference is in FR differences and phase.

BTW I had it set to ignore FR and phase differences below -70 db which is why the ultrasonics weren't corrected.

 

[Blumlein 88]

Here is the result if I let the software correct for FR and phase down to -120 db levels. This again showing difference in response between SUT and non-SUT. Obviously differences are mostly FR and phase. When those are corrected for there is not all that much difference.

 

[watchnerd]

Here is where I applied a 24 khz low pass filter. Remember the SUT is the reference. This is the difference in the non-SUT in response vs the SUT.

View attachment 30511

Does the tool allow you to create the inverse, i.e. make the straight connection the inverse reference?

Right now, if I mentally invert it, it looks like the SUT (relative to straight) has a hump in the very low bass and a roll-off in the top octave, right?

 

[watchnerd]

Hmmm, looking at what I think is a difference in the low bass...

I had speakers on during the rip. Can a SUT vibrate in sympathy to, say, subwoofers, in a way that would induce audible differences in the bass?

Or is that relatively unlikely / far-fetched?

 

[watchnerd]

Looks better...

Notes:

Had to use Audacity to change the duration of the files - 87.4 second maximum in the REW import.
Added 1 dB to the amplitude of the SUT file.

Observation:
Slight roll-off on the sweep with the transformer at low frequencies.

Might expect a straighter line - a trace above, post #129, appears to have a constant 3dB/octave droop in each channel (until the end), here, not consistent.

• RIAA defect in the Devialet?
• Is it in the recording?
• Response of the cartridge/loading?
• Phase cancellation when you summed to mono?

Other than that, the frequency response looks audibly identical.

View attachment 30442

Hmmm, maybe I'm reading @Blumlein 88 's graphs wrong, but the sweep above seems to show the SUT having less bass output than direct, but his null test seems to show the opposite (direct has less bass).

I'm also curious why the big difference >15k that shows up in the null test doesn't seem to appear hear....

Maybe I should do the long sweep again using the same mixed-mode method we used on the mono music tracks he analyzed?

 

[Blumlein 88]

Does the tool allow you to create the inverse, i.e. make the straight connection the inverse reference?

Right now, if I mentally invert it, it looks like the SUT (relative to straight) has a hump in the very low bass and a roll-off in the top octave, right?

Yes it lets you swap Reference and compare. In this case I was comparing the same file Left to Right channel. Swapping the channels here is the difference in SUT vs straight in. I didn't use any EQ or phase correction for this one. I did get much lower RMS difference levels. I suppose it is the difference in having to apply gain in one direction vs the other. The correlated nulling wasn't much different.

 

[Blumlein 88]

Hmmm, looking at what I think is a difference in the low bass...

I had speakers on during the rip. Can a SUT vibrate in sympathy to, say, subwoofers, in a way that would induce audible differences in the bass?

Or is that relatively unlikely / far-fetched?

I don't know if it is or not. I've done rips for people who used tube phono preamps. Having the sound on vs not made an audible difference as the sound did something like a reverb effect in the low frequencies of the tubes.

If you want to test this, put your cartridge on a junk record, and play something with ample low end off of CD while recording the phono. Just have it sitting there with the LP motionless. Do it via both devices or as here with one of each in each channel. You'll see if one picks up or is effected by the xfmr sub coupling and the other not.

 

[Thomas_A]

I did an experiment long time ago using a frequency sweep to measure impact of sound waves in the room on the vinyl playback. You can measure it, but unless you play very high volume, I doubt it will affect much. Peaks of disturbance coincide with standing waves between roof and record surface. Lid on reduces these somewhat. Whether it will affect tubes and what volume, no idea.

https://www.dropbox.com/s/6dfa59myt7canf5/Turntable lid on or off.pdf?dl=0

 

[Blumlein 88]

Here is the other file with channels reversed. Somewhat similar yet somewhat different. A bigger difference in the low end and the very high end seems to trend in the opposite direction. I'll do the dual SUT track in a minute comparing channels. This is one has the straight in as reference. Also we have to remember the straight in connection is wrong loading for the cartridge and may cause sweeps to show a little different trends if it were properly loaded.

 

[Blumlein 88]

Okay, this is with the SUT in both channels comparing one channel to the other. Left is reference in this case. It appears there is some difference in the channels. Especially in the upper frequencies. In fact it appears the differences in straight in vs SUT shown previously are mostly from the channel difference. So from those other views the difference in SUT or straight in appears to mainly be the low frequency difference. This squares with Ray's finding the treble was elevated in the right channel in his graphs earlier in the thread back on page 6 or 7. You could swap channels and do this again to figure out if it is something in the cartridge or SUT or the MM input. I doubt the latter, but following the signal path step by step is the only way to be sure.

That difference doesn't seem to become highly significant until above 14 khz as seen in this level matched spectra for each channel.

 

[RayDunzl]

I'm also curious why the big difference >15k that shows up in the null test doesn't seem to appear hear....

Well, don't know to which image you refer, but one file was 22.05kHz sample rate... so there was a big difference.

The most recent sweep files gave this:

Direct vs SUT, with 1dB added to the SUT data.

 

[Blumlein 88]

Those most recent sweeps both with SUT and direct give nulls between the channels of more than -120 db. Do you remember if you actually just did one channel and it was recording into both channels?

 

[watchnerd]

Those most recent sweeps both with SUT and direct give nulls between the channels of more than -120 db. Do you remember if you actually just did one channel and it was recording into both channels?

If my posting notes are accurate, I:

"-Set the Devialet to mono, so channel imbalances should be moot
-Locked Audacity in at 24/96khz (although I might have left the dither on)
-Picked a better test track on a different record with a 1 khz warning signal before the sweep begins (not recorded), followed by 120 second sweep track"

Would it be worth doing the long sweeps again in the leftSUT/rightdirect vs rightSUT/leftdirect mode?

 

[watchnerd]

Well, don't know to which image you refer, but one file was 22.05kHz sample rate... so there was a big difference.

The most recent sweep files gave this:

Direct vs SUT, with 1dB added to the SUT data.

View attachment 30547

I'm having a hard time reconciling this sweep with what I hear and @Blumlein 88 's data.

I'm not blaming the methodology or graph, I'm just trying to figure out:

--If there are big FR variances that show up in music, why isn't it showing up in the FR graph for the sweep?

--If the FR graph is as uniform as the sweep would seem to imply, why don't they sound identical?

 

[watchnerd]

Okay, this is with the SUT in both channels comparing one channel to the other. Left is reference in this case. It appears there is some difference in the channels. Especially in the upper frequencies. In fact it appears the differences in straight in vs SUT shown previously are mostly from the channel difference. So from those other views the difference in SUT or straight in appears to mainly be the low frequency difference. This squares with Ray's finding the treble was elevated in the right channel in his graphs earlier in the thread back on page 6 or 7. You could swap channels and do this again to figure out if it is something in the cartridge or SUT or the MM input. I doubt the latter, but following the signal path step by step is the only way to be sure.

View attachment 30545

That difference doesn't seem to become highly significant until above 14 khz as seen in this level matched spectra for each channel.
View attachment 30546

I think the simpler answer is that I switch to a mono cartridge for the next round of experiments.

The caveat is that is that it's a spherical stylus, so the comparisons should be expected to be very different at the high end. But SUT vs direct should still be relevant, at least in the low end.

 

[Blumlein 88]

I think the simpler answer is that I switch to a mono cartridge for the next round of experiments.

The caveat is that is that it's a spherical stylus, so the comparisons should be expected to be very different at the high end. But SUT vs direct should still be relevant, at least in the low end.

I don't see that this will help, and you can get the same result without swapping cartridges. For instance just pick one channel of the cartridge and run it with sweep and/or music. Let us say you chose the left channel. Now also using only the left channel output, run everything else through the right channel chain. That would tell us if the SUT or the direct or some of both had different response in one channel vs the other. You could repeat using only the right channel output, but more than likely that won't be needed.

 

[Blumlein 88]

If my posting notes are accurate, I:

"-Set the Devialet to mono, so channel imbalances should be moot
-Locked Audacity in at 24/96khz (although I might have left the dither on)
-Picked a better test track on a different record with a 1 khz warning signal before the sweep begins (not recorded), followed by 120 second sweep track"

Would it be worth doing the long sweeps again in the leftSUT/rightdirect vs rightSUT/leftdirect mode?

It might be worth left SUT/right direct using sweeps and also the reverse.

However, you would have been better not to select mono in the Devialet. What I'd hoped to find is if there are channel FR differences. When you switched to mono you made that impossible. So repeating the SUT on both channel sweeps, and Direct on both channel sweeps, but keeping them in stereo would tell us the answer I think. It will let us know if there are differences in the channels anyway. That also likely will let us see if there are FR differences in the two signal paths SUT vs Direct. Finally with music, at upper frequencies the noise may be intruding on the measurements enough to be a false reading if the sweeps look just fine. That could apply to the below 50 hz frequencies if there isn't much music there and we are simply seeing noise mostly there as well. Most mastering mono'd frequencies below 50 hz on LP.

 

[watchnerd]

It might be worth left SUT/right direct using sweeps and also the reverse.

However, you would have been better not to select mono in the Devialet. What I'd hoped to find is if there are channel FR differences. When you switched to mono you made that impossible. So repeating the SUT on both channel sweeps, and Direct on both channel sweeps, but keeping them in stereo would tell us the answer I think. It will let us know if there are differences in the channels anyway. That also likely will let us see if there are FR differences in the two signal paths SUT vs Direct. Finally with music, at upper frequencies the noise may be intruding on the measurements enough to be a false reading if the sweeps look just fine. That could apply to the below 50 hz frequencies if there isn't much music there and we are simply seeing noise mostly there as well. Most mastering mono'd frequencies below 50 hz on LP.

I may have misunderstood him, but I did the mono thing per @RayDunzl 's request.

 

[RayDunzl]

I'm having a hard time reconciling this sweep with what I hear and @Blumlein 88 's data.

It's a comparison of the sweep files you produced for analysis.

I may have misunderstood him, but I did the mono thing per @RayDunzl 's request.

I requested playback of a mono disk.

Initial files were music - figured the test sweeps later were mono, so didn't mention it again.

Did not request pushing the mono button on the devialet, if that's what happened.

 

[watchnerd]

It's a comparison of the sweep files you produced for analysis.




I requested playback of a mono disk.

Initial files were music - figured the test sweeps later were mono, so didn't mention it again.

Did not request pushing the mono button on the devialet, if that's what happened.

Yes, the mono button was pushed on long test sweeps.