Modulation distortion between inverted preamp samples

It's tough to fit into a subject heading.

So I was doing some sample tests on various preamps around here. I was feeding a line signal out through a -25dB impedance converting pad. I recorded the same thing twice and phase inverted one of the samples.

Within the remaining noise floor, remnants of the original audio were audible but sounded like they were being run through a ring modulator. the music was there but was being reproduced with additional significant amounts of non harmonic information which was correlated to the original signal.

I tried this with three different preamps, all with the same result (though with slightly different characters for each). The preamps in my 828 MKII did not exhibit this characteristic (though the gain was constantly drifting within a range of a few 100ths of a dB).

Is this normal? Is this an indication of inability to accurately resolve very low level information? In each case, the remnant signal was below the noise floor of the preamp, but definitely audible.

The only thing I can think of is that by nulling the signal, you left the remnants of the signal that were modulated by 60 cycle PS ripple and noise, which would sound like what you describe.

That was one thought I had... If this is the case, is this normal for PSUs? Two of the examples tested were using the JLM Powerstation, the other was an ART Pro MPA.

I would be suspicious of errors in the recording media.

You can perform similar null tests in real time in the analog domain. That might help isolate the offending path.

JR

Sounds like latency to me. We the recordings sample accurate?

RonL

I recorded two takes of the same playback file out of and back into the same program. The playback/recording was exactly synchronized.

And what I was hearing was modulation distortion, not simply comb filtering.

I'm interested to know if high end commercial units exhibit the same behavior, but I don't have any of that. Maybe I'll ask my friend who has a 3124 and Prism Orpheus to do the same test.

[quote author="JohnRoberts"]I would be suspicious of errors in the recording media.

You can perform similar null tests in real time in the analog domain. That might help isolate the offending path.

JR[/quote]

I'll try this, though I should hope it's not the media. I'm outputting from an 828 MKII clocked to a Rosetta 800 and recorded back in through the R800.

After thinking about it some, there's no way I can do the exact test I'm thinking of in real time. If I phase inverted a single it cancels out perfectly of course, but the thing about this test is that it requires two separate passes through a preamp module, and somewhere within lies the issue.

If anyone cares to try this test (particularly with a commercial API unit or other high end preamp) I'd be very interested to hear about the results.

Here's an idea. Try the same experiment -- pass the signal twice through a wire rather than a preamp, then synchronize the recordings and see if you get the same odd sounds.

My guess is that the synchronization isn't perfect. Even if it is, though, tiny thermal drifts in the preamp (or the A/D or D/A converter) could change the gain a fraction of a decibel, possibly from moment to moment as the recording proceeds. That would account for the sound you're hearing.

Peace,
Paul

I did try it straight from output to input of converters with none of this sort of characteristic.

And actually, an interesting followup (for me at least). I'm doing some more testing this morning and performed the same test with much more signal cancellation.

Could be the weather...