Phase and NFB confusion

[Randall]

I read somewhere that the OT plate wires have a correct and an incorrect way to be wired to the output plates, which determines whether you get a NFB or a PFB in say, an AB763 type amp. So I put a tone through my DR AB763 build and followed it from stage to stage with an o-scope. And I'm left unsure of exactly what is going on.

I see where the normal and vibrato channels are out of phase with each other because of the extra inverting reverb stage, I get that. The tone hits both the PI tube grids either in phase or out of phase with respect to which channel input, ie in phase for the normal channel and out of phase for the vibrato channel.

I understand the Phase Inverter tube sends (with respect to it's input) an in phase signal to one output and an out of phase signal to the other output tube. It is then I get murky. I think the output tubes flip the signal yet again with each feeding one side of the OT. On my amp, the OT secondary green wire is wired to the speaker jack and the 820R feedback resistor. The signal there is in phase with respect to the input of the PI tube. This seems wrong to me. Do I in fact have a positive feedback loop? The amp makes no evil sounds. Would swapping plate wires correct this, and if so how?

[g1]

You seem to be thinking of a single input to the PI, in fact there are 2 and they are out of phase (the 2 grids).
So it depends which you want to call "the" input.
And what about the OT itself? Is the secondary in or out of phase with the primary?
Much easier to test than conceptualize. Lift the NFB, if gain increases then it is connected correctly.

[Randall]

Now I'm REALLY confused. I have the same, in phase wave on both PI grids. When I lift the 820R, gain increases.

[Leo_Gnardo]

At the OT, bottom line is, if you have the phase incorrect, your amp will likely howl, squeal, honk, chatter, groan, putt-putt and any combination of the above. All artifacts of positive feedback - you create an oscillator of unknown/dubious frequency(ies). With multi tap OT's, easiest to swap primary leads. You'll notice on some Fenders and many old Ampegs, black secondary goes to speaker output and green to ground. For OT's with one secondary you can invert the polarity by swapping secondary leads - pick whichever is easiest.

One more reason to like amps without feedback loops.

[Enzo]

Sorta.

There is not really a correct or not way to wire the OT primary by itself, it only matters when you add the feedback line. In a push pull amp, the two sides are opposite polarity. So the speaker output will be in phase with one and opposite phase with the other. Now we have to pay atention to whether that feedback opposes the output or agrees with it. Agree and you have an oscillator, disagree and you have NFB.

When you scope, the scope generally will trigger on whatever you are looking at. So if your scope is set for positive trigger, then the display will start on the left with the waveform rising. You can look at two signals of opposite polarity, but the scope will still trigger each on the rising side of teh waveform, so the picture will be the same. If you use two probes and scope the opposite points at the same time, THEN your scope will show that one is opposite polarity of the other.

Your scope is going to display the waveform at any point, it will have no idea what the original waveform might have been at the input.


Most triggered scopes have an EXTERAL TRIGGER input. You can conect that to your input test signal, and now your scope will show relative phase. That because when the scope sweep is always triggered by the same thing, then the waveform will be or whatever polarity it is at that stage.

Looking at your DR circuit. The signal enters the PI through the 0.001uf cap. Each tube stage inverts unless it is a cathode follower. So the upper PI stage inverts, then the upper power tube inverts. The transformer should be positive on the top of both primary and secondary. So with a positive at the .001, we get negative to the power tube and aagain positive at the power tube plate, so the output agrees in phase with the input to the PI. ANything in the preamp has no bearing on this relationship. So don't mix up the relative phases in the preamp and the phase relations in the power amp. ABsolute phase means the actual positive or negative on the real world signal. Relative phase means the relationship of two signals in the circuit, regardless of the absolute phase. In other words, the NFB is still NFB, even if we invert the preamp.

The lower side of the PI and power tubes are both opposite polarity from the top two.

It may not be orthodox, but here is how I look at the bottom half of the PI. The top half is just a gain stage more or less. The lower triode has a huge 0.1uf to essentially ground. That grid then is kept at ground (except for the NFB). The cathode is moving up and down to the music via the current from the upper triode. So just like any other triode, the lower one has a varying signal voltage between grid and cathode, it is just that the cathode is doing the moving rather than teh grid. Since the signal at the cathode of the upper triode is same polarity as its cathode, signal wise, that means a positive going waveform will make the other triode cathode more positive too. That is equivalent to making the lower triode grid more negative. So in that manner, we now have esentially a negatove going input to the lower triode. which inverts, then its power tube invertsw, and the lower end of teh OT primary is thus negative - opposite of the top end.


Now your NFB wire. It comes from the speaker positive, same phase as the input to teh PI at the .001. The NFB goes to the small resistor on the cathode circuit. We just mentioned how a positive signal on the cathode is like a negative signal on the grid of the lower tube. So the positive phase NFB line is really feeding the lower triode, which is negative with respect to the input, Hence negative feedback.

Now ther may be some glaring technical errors in that rundown, but for purposes of making it intuitive, that is how I look at it in my head.

[g1]

If your gain increased when you disconnected the NFB loop, then you are phased correctly.
How are you checking the phase at PI grids?
2 channel scope, one channel to each grid?
If the 2 grids are in phase, then the 2 plates would be in phase and your phase inverter would not be inverting .

Edit: sorry, was simul-posting with Enzo, just with a lot less relevant info.

[Randall]

" So the positive phase NFB line is really feeding the lower triode, which is negative with respect to the input, Hence negative feedback."

Woot! There it is! Thank you yet again Mr. Enzo.

Now what about the part where the primary plate wires get reversed, what happens then?

You guys are fantastic, I have learned much from you.

[Randall]

If your gain increased when you disconnected the NFB loop, then you are phased correctly.
How are you checking the phase at PI grids?
2 channel scope, one channel to each grid?
If the 2 grids are in phase, then the 2 plates would be in phase and your phase inverter would not be inverting .

I am measuring with a dual trace scope. I get in phase looking at both grids together, and amplified and inverted looking at both plates of the PI. What gives?

[Alan0354]

I read somewhere that the OT plate wires have a correct and an incorrect way to be wired to the output plates, which determines whether you get a NFB or a PFB in say, an AB763 type amp. So I put a tone through my DR AB763 build and followed it from stage to stage with an o-scope. And I'm left unsure of exactly what is going on.

I see where the normal and vibrato channels are out of phase with each other because of the extra inverting reverb stage, I get that. The tone hits both the PI tube grids either in phase or out of phase with respect to which channel input, ie in phase for the normal channel and out of phase for the vibrato channel.

I understand the Phase Inverter tube sends (with respect to it's input) an in phase signal to one output and an out of phase signal to the other output tube. It is then I get murky. I think the output tubes flip the signal yet again with each feeding one side of the OT. On my amp, the OT secondary green wire is wired to the speaker jack and the 820R feedback resistor. The signal there is in phase with respect to the input of the PI tube. This seems wrong to me. Do I in fact have a positive feedback loop? The amp makes no evil sounds. Would swapping plate wires correct this, and if so how?

If the phase of the input signal to the PI is the same phase as the output signal feeding back to the 820 ohm resistor, you have the right phase. If you reverse the wire, your amp will sound like a horn with or without input.

It is not easy to understand just by tracing the signal with the scope because it's NFB. The phase of the signal interact and can be very confusing. If you want to understand then read on.

To look at it more technically, the power amp is sort of like an opamp that you have a +ve input ( where the signal from preamp goes in) and a -ve input ( where the 820 ohm resistor goes to the PI stage). The 820 ohm is the NFB resistor. The gain is set by the 47 ohm resistor at the tail of the PI stage. You'll see the 820 ohm connect to a 100ohm to ground, that's the one. This is the typical opamp circuit connected in the non-inverting configuration. The theoretical gain is 820/47+1=18.4. BUT due to the low forward gain from the PI stage to the power tube and OT, your observed gain will be lucky if you to have about 10. But it is an opamp. Look up the non inverting opamp and you'll find tons of explanation how that work.

Here is one article:http://www.electronics-tutorials.ws/opamp/opamp_3.html. You can find tons of informations by just google "non-inverting opamp".

Spend some time studying this, then you'll understand how "presence" and "resonance" circuit work also.

[J M Fahey]

Now I'm REALLY confused. I have the same, in phase wave on both PI grids. When I lift the 820R, gain increases.

I am measuring with a dual trace scope. I get in phase looking at both grids together, and amplified and inverted looking at both plates of the PI. What gives?

Why, you are seeing right, of course

They are in phase ... and that´s right !!!

Don´t let the tree hide the forest

I don´t care about the OT, we might even not have one, so I'll use a transformerless amp to explain it.

First let´s see what we have here:
in a Tube amp we start with an LTP , commonly called PI or Phase Inverter ... which to begin with it is not, very poor choice of name.

A single common cathode (emitter) gain stage is a Phase Inverter .... that´s exactly what it does after all.

What we call PI in fact is a Phase Splitter: it gets a signal input and puts out 2 signal outputs, one IN Phase and another OUT of phase .

In fact, the input we normally use
>>>to be continued later, very busy now<<<

[loudthud]

If the 2 grids are in phase, then the 2 plates would be in phase and your phase inverter would not be inverting .

This is not true. The reason is that in the LTPI, the dual triode acts like a differential amplifier. WTF you ask? An ideal differential amplifier only responds to the difference between the two inputs. In most guitar amps if there is a 1 Volt peak signal at the input from the preamp, the feedback signal will be in phase with the input but only (about) 0.9 Volts peak. The net difference is a 0.1V signal (of the same phase as the input). The plates of the dual triode will be of opposite phase.

Another way to look at is this. If you touch a DVM lead to the grid on the preamp input side, the speaker will jump one way. If you touch the meter lead to the grid on the feedback side, the speaker will jump the other way. (The big honkin .1uF cap gets in the way but you get the idea.) What is happening is that touching the meter to the grid pulls it towards ground because of the DC on the grid and the (usually) 1M grid leak resistor. The feedback is in phase with the input, but the differential amplifier makes it look like negative feedback because it reduces the effect of the input.

[g1]

Thanks LT, I realize the error of my statement, but am still having difficulty conceptualizing why the 2 plate signals are out of phase.
The amplification of the net difference still does not explain to me the phase inversion between the plates.
Enzo's explanation steers me toward thinking of the second grid being basically grounded and the second triode getting it's input via the cathode. Am I way off base again?

[Randall]

This is the most difficult part of an AB763 to understand for me. Hurts my poor head, it does.

[Alan0354]

This is the most difficult part of an AB763 to understand for me. Hurts my poor head, it does.

Believe me, this is the easy part of designing amps. Wait until you get to tweaking the sound, now you talk about mojo!!!

There is no short cut for this. As I said, you get through this, quite a few things can get a lot easier to understand. You cannot just trace the signal and use easy explanation for NFB. If you really want to understand, start with the link I posted. Might take you two or three days. But if you understand the non- inverted opamp configuration, you look back to this, it's a piece of cake.

Ha ha, the difficult part to understand is......why Fender sound so good!!!!

This is the text book I studied opamp 36 years ago. I used the much older version of this book. This book really explain NFB and non inverted opamp in a very easy way. I never went to school, this was my gold book for a while.Amazon.com: Buying Choices: Electronic Principles

I always buy used books on Amazon and I am very happy. The whole thing including shipment is less than $10.

[loudthud]

... I ... am still having difficulty conceptualizing why the 2 plate signals are out of phase.
The amplification of the net difference still does not explain to me the phase inversion between the plates.

The LTPI is not a perfect differential amplifier, but it's about 90% of the way there.

Ignore for a moment what it takes to properly bias the tubes. That isn't important to how the LTPI performs it's magic. Just assume that both tubes have a grid leak resistor to a constant 44V, the cathodes are tied together with a 22K Tail resistor to ground. The tail current is 2mA. The current splits evenly so 1mA goes through each tube and causes 100V drop across a 100K plate resistor on each tube.

In the normal common cathode preamp gain stage circuit a positive signal on the grid causes the tube to pull more current from ground through the cathode circuit. This increase in current causes more drop across the plate resistor and the plate voltage goes down.

In the differential amplifier, the Tail resistor is much larger than a normal cathode resistor so current can't increase very much through that resistor. But because the input side's cathode is also connected to the feedback side's cathode, the input side cathode can steal current from the feedback side tube. This theft of current causes the plate voltage on the feedback side to go up (Current goes down, plate voltage goes up). So the net effect is that in current on the input side is increasing and the current on the feedback side is decreasing and the plate voltages respond accordingly.

The above assumed that the voltage on the feedback side's grid was constant. In the LTPI it is the difference between the input and the feedback grids that determines how much signal gets fed to the power tubes. There is a small increase in current through the Tail resistor. This is a deviation from ideal operation and some tricks are used to minimize it.

In the LTPI there is some feedback connected to the bottom of the Tail resistor. This keeps the voltage across the Tail resistor almost constant. A constant voltage across the Tail resistor means the current through it is constant. This helps the PI balance. If the current through the Tail can be made to remain constant, the balance of the PI will be improved. The two tubes act like a current steering mechanism. An increase of current in one side causes an equal decrease in current of the other side. If the plate resistors are the same resistance, the voltage increase on one side will equal the voltage decrease on the other side.

The 82K and 100K plate resistors are a compromise supposedly to improve balance. Probably not ideal for all versions where they are used. Dumble amps have a pot to adjust the balance.