Phase flipping and transformers

[rp]

I’ve built two pairs of mic preamps that include a phase flip switch. In my pair of NYD 6sn7 One-bottles, the phase is switched before the mic input transformer. In my pair of Hairball Lolas, the phase is switched before the output transformer (they are fully balanced throughout). Each pre does the flip by swapping the + and - legs of the signal with a DPDT switch or relay.

I have a problem, which is that all four pres sound different when the phase flip is engaged versus when it is not. The phase flipped version sounds like it is lightly comb filtered, similar to the sound of two slightly out-of-phase signals summing. There is a shift in low and midrange energy where lows seem accentuated and mids dip. The effect is subtle, but noticeable, and is stronger in the Hairballs.

Has anyone else experienced this? My hunch is that this a phenomenon inherent to transformers. Perhaps the way the primaries are wound? Or maybe it's component mismatches in the surrounding circuitry?

I’ve tried measuring the differences using Room EQ wizard. When measuring frequency response with a sweep, I get some abnormalities, but they are different with every measurement and disappear when multiple measurements are averaged. I'm unable to see any appreciable difference between in phase or out, even though I can hear it clearly.

 

[zamproject]

Hello

What happen with two pass of the same audio (one flipped) and a null test ?

Best
Zam

 

[JohnRoberts]

I’ve built two pairs of mic preamps that include a phase flip switch. In my pair of NYD 6sn7 One-bottles, the phase is switched before the mic input transformer. In my pair of Hairball Lolas, the phase is switched before the output transformer (they are fully balanced throughout). Each pre does the flip by swapping the + and - legs of the signal with a DPDT switch or relay.

To be a little pedantic, you are talking about reversing "polarity" not phase, but we know what you are talking about.

I have a problem, which is that all four pres sound different when the phase flip is engaged versus when it is not. The phase flipped version sounds like it is lightly comb filtered, similar to the sound of two slightly out-of-phase signals summing.

Rich comb filtering is generally created by interference between a slightly delayed stem and the original (aka Flanging), simpler  notch/bump combing can be generated by interference with a phase shifted stem (aka Phasor).

There is a shift in low and midrange energy where lows seem accentuated and mids dip. The effect is subtle, but noticeable, and is stronger in the Hairballs.

The obvious suspicion is leakage BUT mixing original with a polarity reversed stem will not comb but just reduce level. Are both paths always feeding through the transformers?

Has anyone else experienced this? My hunch is that this a phenomenon inherent to transformers. Perhaps the way the primaries are wound? Or maybe it's component mismatches in the surrounding circuitry?

I’ve tried measuring the differences using Room EQ wizard. When measuring frequency response with a sweep, I get some abnormalities, but they are different with every measurement and disappear when multiple measurements are averaged. I'm unable to see any appreciable difference between in phase or out, even though I can hear it clearly.

I like the idea of null testing. If your frequency/phase response bench test is not repeatable is the device under test, or the tester different each time?

JR

PS: Note some music sources sound subtly different after polarity reversal, but the phenomenon you describe does not sound that subtle.

 

[dmp]

Flip the wiring on the transformers on one of them and see if the problem follows. That will tell you if the effect is coming from transformers or something else

 

[efinque]

Flip the wiring on the transformers on one of them and see if the problem follows. That will tell you if the effect is coming from transformers or something else

Yeah I was thinking there could be an all-pass filter somewhere in there.

 

[moamps]

I have a problem, which is that all four pres sound different when the phase flip is engaged versus when it is not. The phase flipped version sounds like it is lightly comb filtered, similar to the sound of two slightly out-of-phase signals summing. There is a shift in low and midrange energy where lows seem accentuated and mids dip. The effect is subtle, but noticeable, and is stronger in the Hairballs.

Did you maybe test it using your voice and headphones?

 

[Bo Deadly]

Measure it. Get a spectrum. Audio interface software usually comes with an FFT feature. Although it's usually pretty low res because they want it to look like it's working in real time. But if you can make a good recording I can run it through a program that will generate a very high res FFT plot. Specifically, generate white noise in your DAW, run it out of the audio interface and through an attenuator like this:

You don't have to put the resistors in a shell like that. You could just breadboard something but if you want to also measure noise, it's nice to make a good shielded attenuator.

Then run that now 150 ohm -40dB source into the mic pre. Adjust the noise level output so that output is just below the clipping point. Then record the output of the mic pre for a full 10 seconds and put that .wav file somewhere that I can get to. Make one for both phase inverted and not inverted. I will run them through a program that will generate a high res frequency spectrum and post the image here. If there is any filtering going on we will see it.

However, my guess would be that it's actually no filtering at all. Reversing polarity of a signal will change the sound a little bit. If you also simultaneously reverse the phase in your audio interface to negate the reversal, does it still sound different? Also, there is asymmetry in all natrual sounds like voice and a tube mic pre might be clipping asymmetrically. All of these things can contribute to a small difference in the sound. There could also be some psychoacoustic phenomenon going on.

 

[Speedskater]

You might make a polarity reversing interconnect cable.  Then you can test whether the difference is acoustic or electronic.

 

[fazer]

Absolute phase also creates a subjective difference in preceived sound.  Are you talking  about listening to your voice on a mic with headphones flipping phase and noticing a difference in sound?  Midrange is more forward to your ears in an absolute aligned system.  When phase is flipped on monitors,(both speakers at the same time) the sound is in phase with each other but absolute phase is opposite and will be precieved less midrange. It’s an acoustic abnormality.

 

[Martin Griffith]

ISTR that the sound from your vocal cords also gets to your ears through your flesh and bones which will mix with your acoustic HP sound, causing some change in some cancellations when you flip the phase

 

[JohnRoberts]

Absolute phase also creates a subjective difference in preceived sound.  Are you talking  about listening to your voice on a mic with headphones flipping phase and noticing a difference in sound?  Midrange is more forward to your ears in an absolute aligned system.  When phase is flipped on monitors,(both speakers at the same time) the sound is in phase with each other but absolute phase is opposite and will be precieved less midrange. It’s an acoustic abnormality.

Again its called absolute polarity.... I put a polarity switch on my (kit) phono preamps back in the 1980s, because recording chains were not very disciplined about absolute polarity back then...

Hopefully they are better now, but DIY can always repeat old mistakes.

JR 

PS: Indeed listening to your own voice in cans can be conflicted by internal parallel conduction paths, and very sensitive to polarity and path delay. Never attempt critical listening that way.

 

[efinque]

PS: Indeed listening to your own voice in cans can be conflicted by internal parallel conduction paths, and very sensitive to polarity and path delay. Never attempt critical listening that way.

I think stuff like this, ie. tuning by ear affected and eventually defined how regional scenes sound(ed).. I remember reading people used to make tunes for playback on specific systems because they were so popular, albeit by todays standards not "properly" engineered.

 

[Bo Deadly]

ISTR that the sound from your vocal cords also gets to your ears through your flesh and bones which will mix with your acoustic HP sound, causing some change in some cancellations when you flip the phase

Great point. If the OP is vocalizing into the mic and simultaneously listening to the monitor with headphones, there will be significant bleed both through tissue / bones as well as through / around even the best "over-ear" phones resulting in some cancellation. Quality in-ear phones (like ones with those clear rubber tips from Etymotic Research) will reduce bleed slightly.

 

[ruairioflaherty]

Absolute phase also creates a subjective difference in preceived sound.  Are you talking  about listening to your voice on a mic with headphones flipping phase and noticing a difference in sound?  Midrange is more forward to your ears in an absolute aligned system.  When phase is flipped on monitors,(both speakers at the same time) the sound is in phase with each other but absolute phase is opposite and will be precieved less midrange. It’s an acoustic abnormality.

Never in 20+ years of listening have I heard a polarity flip on both speakers result in less perceived mids. I just sat and flipped polarity on 5 or 6 beautiful acoustic and pop recordings, on large ATCs in a mastering room and there is no difference, zero.

Edit : And yes, if OP is listening to mic pres live on headphones it absolutely 100% sound different.  When I tracked I would do that test with every singer and sometimes the difference was small, sometimes much larger.

 

[fazer]

Well  ruairio I would trust your impression on speakers but headphones, seems we all agree there can be issues with ear canal and such. 
The speaker issue came from I believe An REP magaizine article from years ago with some engineer talking about the kick sounding different than the night before on a mix.  It was traced to maintenance hooking speakers with reverse absolute polarity.  When the speakers were corrected the midrange point on the drum came back.  God I’d hate to pass on fake engineering news.   

You set up lots of very high end monitors.  Do you not care about absolute polarity?  Meaning  a kick drum with microphone in front should create a positive going initial wave front so speaker starts by pushing and then pulling.  Do you feel that doesn’t matter as long as there in phase?

As I said I would trust your ears. 

 

[rp]

It seems like auditioning in headphones is most of the problem. My findings from poking around further:

- Running two passes of the same material with the polarity flipped and summing them nulls completely using the Hairball pres. I tested with noise and a pulsed square wave.

- Doing the same with the tube pres doesn’t completely null, but I think that could be explained by asymmetrical triode curves.

- Recording a mic pointed at a speaker and nulling two passes of the same material also gets about as close to null as one could hope in a city apartment with a lot of background noise.

- Vocalizing into a microphone with headphones on, the effect is quite pronounced, the reason for which seem obvious now that you've all mentioned it!

- Listening back to a recording of my voice with alternating polarity flips, I can still hear a difference though, both in headphones and on monitors. Maybe just a quality of my voice. Or voices generally? I’m unable to distinguish between recordings of alternating flipped polarity versions of other sources like noise played out of my monitors, so it seems source dependent.

I think most of the effect I described comes down to auditioning things with headphones when there’s bleed. I do a lot of recording in less than ideal spaces where I’m using headphones while tracking others. Thinking back to other times I’ve noticed the effect, they’ve always been in monitoring-compromised places - when I’ve been in the same room as the source, or in the next room but with a lot of bleed. It makes sense that the sum of the external sound and the headphones would create the effect I hear, and that the effect would be particularly strong when talking into a mic with headphones.

Thanks all for the comments and clarifications! Trying out your suggestions made me eliminate some variables, and the conclusion is that there may be some perceived effect due to whatever the source and listening conditions are, but that the preamps aren't really a factor.

 

[ruairioflaherty]

You set up lots of very high end monitors.  Do you not care about absolute polarity?  Meaning  a kick drum with microphone in front should create a positive going initial wave front so speaker starts by pushing and then pulling.  Do you feel that doesn’t matter as long as there in phase?

As I said I would trust your ears.

Of course I care about absolute polarity, it's part of a competent system setup.  But, that does not mean that you can hear the difference in 99% of cases.

The general fascination/obsession with the kick drum causing a positive going wave is a little absurd.  Anyone who has ever actually played a kick drum can tell you that it still sounds like a kick drum from the other side..

 

[scott2000]

Interesting stuff...  I always thought I heard differences in the low mids/low end when polarity was reversed......thought  Maybe it was the drivers reacting differently??? It's been a while since I've listened....

So you recorded  with a mic pointed at the speaker with one pass of each polarity and then nulled them,  and it didn't show anything outside of the expected???

Neat

 

[abbey road d enfer]

If the OP is vocalizing into the mic and simultaneously listening to the monitor with headphones, there will be significant bleed both through tissue / bones as well as through / around even the best "over-ear" phones resulting in some cancellation.

The dominantleakage  return path is bone conduction. Acoustic return is a significant problem only with open cans.

 

[abbey road d enfer]

Indeed listening to your own voice in cans can be conflicted by internal parallel conduction paths, and very sensitive to polarity and path delay. Never attempt critical listening that way.

I believe the problems related with this "method" are well-known and still people continue to use it! I think we should make a warning poster in red font size 28.