OT Tranny Polarity/PI Polarity

[fiftynine]

Gents,


If the input signal inverts for every gain stage, one side of the PI must be the exception to that or they'd cancel out. In an amp with 2 channels with the same pre-amp stages (one going in either side of the PI), how do you know which side of the OT to connect to which power tube (or pair of) so you're in phase with the input signal? I know the dots (attached) denote no change of phase, but if you're not aware of the phase at the PI, the dots mean nothing.


I guess I'm asking, which side of the PI gets inverted and which doesn't and how can you tell from the amp schematic to keep the speaker in phase with the input signal?


Thanks.

[jjasilli]

The guitar speaker does not have to be in phase with the input signal. 


For a guitar amp, signal phase is not relevant.  This is due to the narrow frequency range of the guitar.  By contrast, a hi-fi amp may process the entire audio frequency band from 20 - 20K Hz.  Elaborate home theater systems may go down to 5 Hz.  This is inaudible but will physically shake the house & widows.  Due to time delay/advance factors inherent in reactive circuits, there may be need for affirmative steps to keep all these wide ranging frequencies in phase-time with one another, and lip-sync'd to video.  But this is not needed in guitar amps.


Even with 2 or more channels in a guitar amp, phase remains irrelevant, UNLESS the channels are mixed.  (This also applies to the wet & dry mix when there's a reverb circuit.)  Usually we want the 2 signals to be in phase.  But some people state a preference an out-of-phase mix.   


In a 2 channel amp, each channel does NOT go to it's own side of the PI.  Rather, one channel is selected, or both are mixed, before going to the PI. 


It's easy to determine phase by simply counting the number of gain stages.


Live on stage, various speaker cabs may be out-of-phase with one another.  This is generally ignored, at least in a rock band.  It can be corrected with a phase SW; an adjustable delay feature built into pro PA equipment; or by shifting the physical location of the speaker boxes

[eleventeen]

" the same pre-amp stages (one going in either side of the PI)......


That's not what happens. You have a single phase signal  all the way thru your preamp & tone controls) which goes to a phase splitter which generates two opposite phase signals. Before that "box" in the functional diagram, you have only one phase of signal. Sounds a little complex maybe; but it's probably a result of our collective habit of speaking about it sloppily. We abbreviate it "PI" phase inverter instead of "PS" which would generally be taken as "power supply".


Now you have created the two separate phases and feed those signals to your push-pull output tube grids.


Now; your question: " how do you know which side of the OT to connect to which power tube (or pair of) so you're in phase with the input signal?"


You don't, and you can't. Unless you are a manufacturer and you have 1000 output trannies lined up and you know the brown goes to this tube and the blue goes to the other one, on each and every amp you build.


If you are a one-off builder, you simply have to hook it up one way, try it, then reverse the leads if you're wrong. 50/50 chance.


That is why we recommend that you do not trim the OT plate leads to length nor dress them neatly when you are done w/your build and are ready to fire it up. Leave them long, tack solder them to the output tube plates. You are either right or wrong. "Wrong" usually is immediately apparent as the amp will feed back or squeal or "motorboat"....there is usually nothing subtle about it. If you are wrong, you just flip the leads. IMO it does not pay to overthink it in advance. There is no fixing it other than flipping the leads if you are wrong, thus it's much easier to just allow for the possibility that you might have to flip the leads when you are ready to fire up the amp.

[fiftynine]

Wow. I didn't realise there was a 50/50 chance of getting your OT wires wrong if you don't have a negative feedback loop. I thought it would be the same either way, just a different phase at the speakers and that doesn't matter anyway as suggested by jjasilli.


There were some long discussions on The Gear Page where some players suggested the speaker phase in relation to the guitar signal is like night and day. I assumed this is why Matchless include a speaker phase switch. I'm with jj because you stick a pedal board in there and don't have a polarity switch last in the chain because 'it ruined your tone'.

[eleventeen]

"...a 50/50 chance of getting your OT wires wrong if you don't have a negative feedback loop...."


No it's if you *DO* have NFB. Simply put, if the NFB signal is incorrectly phased then you have POSITIVE feedback which is a req'ment for an oscillator. That should make sense.


The speaker phase aspect jjasilli is brought up is rather a separate issue. INSIDE the box of any one amplifier, the OT phasing has to be right or you do not have an amplifier: You have an oscillator.


Now, if you play through multiple amps, those amps cranking out the signal from ONE guitar might produce one phase going to one speaker and a different one going to another, for any of several reasons. This produces various effects that may or may not matter to you; but it is a separate issue than what you originally asked about.

[sluckey]

I assume you are talking about a typical LTP PI? The only time that phasing becomes an issue is when there is a NFB loop involved. If there is no NFB loop, then phasing does not matter. If there is a NFB loop, the transformer must have phase markings (usually a dot) in order to look at the schematic and know for certain how to connect the OT properly. Without knowing the transformer phasing you cannot know how to connect the plate leads. It becomes a coin toss guess.

As for the phase inversions going on in the LTP PI, you must use a point of reference to understand. Let's call one triode A and call the other triode B. Now put a reference signal into the grid of triode A. A's plate will be inverted 180° but B's plate will be in phase. OK, now put a reference signal into the grid of triode B. B's plate will be inverted 180° but A's plate will be in phase. See how that works?

[tubeswell]

...which side of the PI gets inverted and which doesn't... ?

For a Long Tail Pair inverter, the side you put the signal into is the inverting stage, and the other side is the non-inverting stage. (If you have two different signals - one going into each side, it still works the same way, its just that the LTP will amplify the difference between both sets of input signals - because of the type of gain stage it is.)

... and how can you tell from the amp schematic to keep the speaker in phase with the input signal?

If you are incorporating a global NFB loop from the OT secondary back to the LTP and it squeals when you fire it up, then switch the OT primary winding ends around to the opposite output tube plate pins. That's really about the only 'fool-proof' way of dealing with it.

Otherwise - what everyone else already said.

[dmp]

If the speaker is out of phase with the guitar, will it reduce acoustic feedback?

[jjasilli]

Wow. I didn't realise there was a 50/50 chance of getting your OT wires wrong if you don't have a negative feedback loop. I thought it would be the same either way, just a different phase at the speakers and that doesn't matter anyway as suggested by jjasilli.


There were some long discussions on The Gear Page where some players suggested the speaker phase in relation to the guitar signal is like night and day. I assumed this is why Matchless include a speaker phase switch. I'm with jj because you stick a pedal board in there and don't have a polarity switch last in the chain because 'it ruined your tone'.

[HotBluePlates]

If the speaker is out of phase with the guitar, will it reduce acoustic feedback?

Only if you don't move. Even then it's unclear...

Your speaker will pump back & forth to output a sound. That sound will propagate through the air, at a speed determined by air density (sea level & average humidity, it will be about 1,000 ft/sec). If you have a single pitch (say 1kHz), you can figure out a spot at a single distance from the speaker where that sound wave is pushing/positive-polarity or pulling/negative-polarity. Get the right orientation positive/negative, and have a loud enough sound at the speaker, maybe it will shake the string and result in acoustic feedback.

But the magic distance depends on how fast sound moves through the air, and how frequently the sound cycles through a complete positive/negative cycle. Put this together, and a 1kHz wave goes through a complete cycle in about 1 foot (+/- for air pressure, humidity and rounding error). So you might get feedback at 1kHz at 1 ft, 2 ft, 3 ft, 4 ft, 5 ft, etc.

But the distances will be different for 500Hz: it goes through a complete cycle half-as-fast, so distances are twice as long. Feedback at 2 ft, 4 ft, 6 ft, 8 ft, etc.

Every other pitch has its own different wavelength for a single cycle. That means different magic distances to result in acoustic feedback. And we haven't even considered how sound bounces off of walls, floor, ceiling... Each reflected path between speaker and string/mic amounts to a different-length path, which might/might not be the correct phase to encourage acoustic feedback. And longer paths may be lower intensity wavefronts, not loud enough to push the string appreciably.

So orienting the amp's electrical phase one way or another doesn't make much sense if you're attempting to get an amp that does/doesn't feed back... unless you play only one pitch and stand in one spot. 

[jjasilli]

I specifically left NFB out of my prior post, because its phase in relation to that of input/preamp signal doesn't matter in guitar amps.  This is because NFB from the OT goes only to the PI, and not further back into the preamp.  If the OT is out of phase with the PI, then there's positive feedback which causes howling.  The simple cure is to  swap power tube B+ leads.  The PI does not care what phase signal it is receiving from the preamp.  Whatever that phase is, the PI splits it into 2 equal & opposite phases (in an ideal world). 


None of the following apples to guitar amps:  Some amps, especially hi-fi amps,  may source NFB from late in the amp (OT secondary or power tube plates), and feed it back into earlier preamp stages, before the PI.  This cuts the preamp stage gain, but there is usually excess gain anyway.  It also broadens frequency response & lowers various types of distortion.  The problem in this scenario is that due to phase-time differences in frequency bands within the full audio band, plus gain stage phase inversion, some or all bandwidths in the "Negative FB" loop may actually go Positive, wreaking havoc.  Not an issue in guitar amps.

[jjasilli]

Re hotblue's post:  FWIW, during soundcheck, Slash has his guitar tech roam around the stage seeking the spot of maximum resonance where the guitar will sustain indefinitely, and mark that spot with tape. 

[dmp]

Re: hotbluestuff
Good stuff - thanks.

[fiftynine]

Edited. Probably not a fair comment. Suffice to say I'm with you HBP.

[HotBluePlates]

Please go and find Steve Kimock on TGP, give him a slap and tell him what you just told us.

I don't know Steve. It might be easier to point him to a Physics of Sound textbook, as what I described is (probably more simply) one of the first things covered in such a course.

Also, audio engineers know/use this when considering how to move/place microphones where phase cancellations happen when 2 mics pick up the same signal but are placed at two different points in a room. Additionally, PZM microphones have their unusual designs to minimize strange phase cancellations due to reflected sound.

[PRR]

I don't believe absolute phase is very important in most musical situations.

If you think it matters, put a polarity switch in your speaker line (easiest place).

As HBP says, any time you go through AIR, and compare the wave against the original, it goes in-and-out of phase every few feet/inches. On a broadband signal, the mix will "comb-filter".

Marking the stage for "maximum feedback" probably also needs to specify the note (even string/fret) for which feedback is wanted. On a crowded stage there may be a generally-loud area, but the hot-spots and nulls will still be different for different frequencies. (However for soprano guitar, leaning several inches one way or the other may get off a null and near a peak.)

Harmonica (mouth harp) players do often report significant differences when polarity is flipped. They typically work very close to their amps, and right on the edge of howling. Statistically there will be about the same peaks/nulls but at alternate frequencies depending on polarity. One set may work better than the other, depending on the music, the player, and the harmonica's reed-balance. 

[Willabe]

Marking the stage for "maximum feedback" probably also needs to specify the note (even string/fret) for which feedback is wanted.

I saw a picture years ago of Ten Nugents stage set up, I think it was in Guitar Player?

There was a large, 20'(?) simi circle/crescent, white gaffers tape(?) across the front of the stage floor. It had cross hatch marks of tape with a note next to the mark. The notes did not go in order.