Importance of output phase orientation

[KarenColumbo]

Dear knowledgable fellows: How important is the phase orientation of a stompbox's output? I just stumbled upon a setup (consisting of a fet gain stage that reverses 180 degrees and a "tube emulation" with a BJT that doesn't) that sounds just about right. Do I need to put in another stage that makes a 360 out of the 180? From what I gathered so far there'd be problems with pedals that are phase-ey, like chorus and such ...

Edit: Seems I will have to put in at least a recovery stage anyway - have to Fender-Bassman-EQ the thing ...

But the question remains: Must I put out in phase with the input?

[mth5044]

As far as I know, it doesn't matter in your linear chain of pedals. The issue comes in when you try to do a stereo setup and your amp outputs different phase signals. For example, 'they' say you can't jumper the normal and vibrato channels in a blackface fender because the normal channel has two phase inversions, and the vibrato has three as there is a reverb recovery gain stage.

I suspect there are plenty of single transistor, and even more complicated pedals, that invert phase, without any issues when run in a chain.

[ElectricDruid]

As far as I know, it doesn't matter in your linear chain of pedals.

+1 agree. In a linear chain it doesn't matter. And that includes with chorus or phaser or whatever. Do you know which way up your guitar signal was when it went in anyway?!?
It only makes a difference if you split the signal chain and then mix it back together again. In that case, you might get a cancellation effect between the straight and inverted signals, or at least between the parts of the signals that are the same.

For example, say you have an inverting delay in parallel with an overdrive and then mix the two signals back together. (I'm not saying that's a good sound, it's just an example ). The overdrive tone won't be exactly the same as the dry signal coming through the delay, but it'll be similar enough to cancel to some degree and will probably make the overdrive sound weak and fizzy. The delays will come after and could be in phase or out of phase anyway, so they won't be so affected. If that sounds like something you might try then you maybe need to concern yourself with making sure your pedals are non-inverting. If that sounds like a highly artificial example to demonstrate a mostly-theoretical problem, then don't worry about it.

I'd say making a pedal non-inverting is "best practice" but not 100% essential. I'd always make a pedal non-inverting if I can, but I'm unlikely to add an extra single op-amp IC just to "fix" it if it isn't. Not worth it, imho.

Tom

[stallik]

Even if you do match input phase with output, all pedals in a stereo setup would have to be the same way to guarantee no issues. That's OK if all pedals are your builds but what happens if you add someone else's pedal?
The cost of an extra IC may not mean so much to DIY builders but it will to mr big commercial builder so I can't see a universal standard being adopted any time soo.

[KarenColumbo]

Thx guys - I think I "overthought" the matter Still in the picture-instead-of-words-phase.

[R.G.]

There have been many studies in the hifi luna-tweako circles trying to prove that absolute phase can be heard. Last I touched this issue, it had been pretty conclusively proved that the luna-tweakos are wrong about this one. And in amplifying and effecting rock guitar, even more so. Humans in general cannot distinguish absolute signal phase.

And that holds right up until you mix some inverted-phase stuff with noninverted. The cancellation you'd worry about does happen in the bass registers. It's much more elusive, approaching invisible as frequency rises. So if you mix any noninverted path with the same signal through an inverted path, what happens mostly is bass loss. That's at least partially because there are no analog filters that do not involve phase shift in the filter. In fact, you cannot make an analog filter that selects frequencies without having a phase shift. This is one of the central problems in speaker crossover design, as the phase shift from a high pass filter is generally the opposite direction from a low pass filter and when the two signals are added together in the air after passing through speakers, you CAN hear a hole-in-the-middle effect unless you've done a really good job on designing the filters NOT to phase cancel.

So every time a signal passes through a capacitor (or an inductor, but they're rarer) frequencies the signal phase around the filtering rolloff point of the filter get shifted. Phasing stages take some pains with this effect, setting themselves up so they shift all frequencies by some amount, but don't change the amplitude any. Phase stages are sometimes called all-pass filters for this reason. So - any frequency response change has a phase change, and it's not a simple invert/non-invert. Many filters can add up to cancellations or reinforcements - we use this effect in phasers and flangers.

And with all that misdirection, we get back to the advice you've already been given. As stated, if you do your own effects, and never split an effect path and later recombine them (even acoustically, in the air outside two amplifiers) you don't need to sweat inverting or non-inverting. On the other hand, if you design effects for other people, and want to cover as many possibilities as you can, I'd advise going to as much as half a dual opamp worth of trouble to make it overall non-inverting, at least at bass frequencies where this will matter most. I do in fact do exactly this. In any new designs for my day job, I wll always make it be noninverting. Same background facts, but my personal preferences fall on the side of non-inverting overall.

[Elektrojänis]

Then there is one case in the other direction... High gain and true bypass.

On a high gain pedal you might get squeeling feedback if tho output and input are in phase. This usually doesn't happen when you are just testing the circuit whithout any bypass swiching, but when you add the bypass switch in the mix, it will start as you turn up the gain. The positive feedback path from the output to input is capacitive coupling inside the bypass switch... It's the spot where the output and input signal paths are closest to each other and you can't easily move them away from each other. If you invert the phase of the output, the problem can dissapear... Not always but for simple drive/distortion/fuzz it's worth a try.

This only happens if you have really a stupid amount of gain, but I've had it happen (when I wanted a stupid amount of gain).

[R.G.]

Good point, Petri. It can happen.

Many things come to mind related to this issue.
- Wiring an input wire to a stomp switch can easily require a shielded wire, for exactly this reason.
- In really bad cases of high, high gain, shielding the output wires can be needed.
Both of these reduce the cross coupling by limiting the un-shielded lengths of the two wires.
- This situation can also be looked at as an argument for cold-switching. Cold switching is the technique of putting the actual signal switching stuff over on the PCB with the signal, not run out to the operator panel on signal carrying wires. Something like a CMOS switch, JFET switch, or on-PCB relay can do this job nicely, leaving only a non-signal-carrying wire running off the board to the stomp switch. Relays are available that are a bit smaller than two dual opamps, and a suitable CMOS switch can be had in an 8-pin or 16-pin IC.
- These techniques become necessary at some gains no matter what you do. With enough gain, everything oscillates. This is because phase shift accumulates as frequency increases, and at very high frequency every conductor has a capacitance to every other conductor that makes oscillation possible. As an example, common NPN transistors will have usable frequency responses in the hundreds of MHz. JFETs easily go higher. If you are unlucky enough to have just the wrong conditions, your circuit can oscillate at frequencies that a normal low end 20MHz oscilloscope can't even see. And in these cases. the overall phase situation doesn't matter, because the phase at ultrasonic and RF frequencies is a product more of the stage to stage filter-type phase shift much more than whether your low audio phase was inverting or non-inverting.

[marcelomd]

Every filter will change the phase of the signal. A simple single pole will shift the phase 90 degrees (plus or minus, if high- or lowpass) for frequencies greater than the filter's center frequency. More complex filters have weird phase shifts.

So it may be the case that even if you keep the same relative phase orientation between two signals, you will have both constructive and destructive interference wherever you mix them.

[karbomusic]

There have been many studies in the hifi luna-tweako circles trying to prove that absolute phase can be heard. Last I touched this issue, it had been pretty conclusively proved that the luna-tweakos are wrong about this one.

I felt the same until someone showed me how easy it is to hear in headphones last year - I can dig up that old thread from another forum (with examples) if needed but.... I still agree abs phase doesn't really matter in most of our lives and hearing it headphones was more of a card trick than real-world issue (albeit was easy to hear if memory serves).

As far as the thread title, I see no reason not to keep the outgoing phase = incoming for no other reason than being a good citizen as you hand the signal off. There is no reason to be the reason someone else needs a polarity/phase switch - let that be someone else's fault.

[samhay]

>I felt the same until someone showed me how easy it is to hear in headphones last year...

Was that inverting phase of one channel relative to the other? I can believe you could hear that, but it isn't absolute phase. If you switch the phase of both channels at the same time, can you hear the difference?

Back to the OP. I like to design effects that don't invert the signal, but won't add an extra stage to correct it if does. However, I don't do this for a living and would somewhat expect a modern commercial effect (e.g. those designed by RG) with multiple amplification stages to preserve phase. I think you'll find this is generally the case.

[karbomusic]

>I felt the same until someone showed me how easy it is to hear in headphones last year...

Was that inverting phase of one channel relative to the other? I can believe you could hear that, but it isn't absolute phase. If you switch the phase of both channels at the same time, can you hear the difference?

No, it was straight up absolute phase testing - that's what the thread was about IIRC - but what I don't remember is the signal used - hence the point about a card trick, meaning I think it was a manufactured test signal, not music content - I'll see if I can find it to confirm.

[karbomusic]

Meh, I can't find the ^old thread but I may run across it later, it's not really that important because I walked away not thinking it was anymore important than before, it was just that the headphone demo was interesting at the time.

[samhay]

^I stand corrected.
Internet says you can hear the difference if the sample is sufficiently asymmetrical.
Just wrote some code to confirm and it turns out the internet was correct - had to happen once.

I guess this makes some sense if your ear has a non-linear sensitivity. Which of course it very much does.

Edit - but I still don't think this makes enough difference to warrant one to religiously design phase-correct effects.

[karbomusic]

Edit - but I still don't think this makes enough difference to warrant one to religiously design phase-correct effects.

Yes, I agree on that and I also agree it's not bad to send the signal out the same as it entered, just because I think it is being a good citizen.

[anotherjim]

I do think it's important to keep signal polarity thru the box. Of course, someone out there has an inverting effect in a parallel loop and is totally happy with the sound, maybe even doesn't like an alternative design that does keep TP (And that could be because the other side of the loop contains an inversion!). It's chaos out there so you can't always win.

With non-inverting high gain, you do risk feedback. Screened wiring is not hard to fit though is it? But even then, there can be enough capacitive coupling between contacts in the stomp switch to make it unstable. The 3PDT wiring scheme with the LED switching on the middle set of contacts should be used.

Those who don't like op-amps can use a phase splitter/concertina stage to get an extra inversion without extra gain.

[pinkjimiphoton]

i used to notice issues with combining boss and dod pedals, always seemd the dod's were out of phase with the boss pedals. tone would get thin, sustain would be weird...

the other place phasing matters, maybe the most, is if stuff is in your effects loop. if its out of phase things can get seriously weird and unstable in my experience.

usually wrong, but... fwiw 

[Danich_ivanov]

For dirt pedals it doesn't matter quite as much, unless you're a bass player for example, trying to mix clean signal with an od of some sort, it will be an issue. I personally like experimenting with dry wet, had obvious issues because some of the pedals were out of phase, so i would say it is better to keep things in phase, which is what i try to always do, and think through every circuit i'm working on. And of course tone controls alter the phase, but they also change the overall sound when they do so, so that the difference will be more obvious rather than distracting, if you know what i mean. And delays, modulation effects, some sort of a clean preamp, boost pedals definitely should be in phase i think. Although to be fair in reality nothing is perfectly in phase, it's just that 180° is what is it.

[Transmogrifox]

Something not mentioned yet is a good utility stompbox is a phase inverter.  "ON" puts the signal through an inverting op amp, "BYPASS" is either TBP, or switches the signal over to the non-inverting side with correct unity gain correction for both paths.

Add a volume knob and some gain to the whole thing and you have a nice little utility for stereo or parallel mix chains.

Make a few of them if you do a lot of parallel and/or stereo processing.

The ONE instance I can imagine in which this utility could be of value in a linear chain is for the case where you rest your headstock on the amp.  Inverting or non-inverting is not the key here, but the ability to flop phase and hear where the feedback sounds the best when you rest your headstock on the amp.

On the points about ear detecting absolute phase absolutely requires nonlinear ear response and asymmetric signal.  In a typical song asymmetry changes depending on the instruments, the interlude in the song and what is present/absent in the mix.  In other words, for a given song you might be able to hear a subtle difference in the A/B when specifically listening for this, but for playing rock 'n' roll or hearing music on the radio there is no practical difference.  Compared to all of the non-subtle things (like radio broadcast compression and EQ settings) this is just too far down in the weeds to care about (at least for me).

I think the effect of atmospheric pressure is of the same order of magnitude of influence on how you perceive music.  Listen to the same recording on the same stereo at a different altitude and some audiophile junky will tell you the "sweet spot" is 22,000 feet above sea level.  The real reason it sounds epic is because you have just reached terminal velocity and your cheeks are stretched behind your ears, obviously optimizing the frequency response characteristic of your headphones   But still, you can DEFINITELY tell atmospheric pressure is what makes the MOST difference.

[PRR]

If you took me to 22,000 feet I'd pass out.

I've been well over 10,000 feet. Baby-pictures in Colorado Springs and Pikes Peak. One visit my grandmother did pass-out. Later riding a DC-3. As a young adult, walking around the van at the Wyoming Continental Divide made me dizzy. And I have not been over 200 feet in years.