Green Ringer phase splitter

[PBE6]

Been looking at the Green Ringer schematic at General Guitar Gadgets, and I'm wondering is there a reason why the phase splitter Q2 is a PNP transistor as opposed to an NPN? Also, is it important that it be low gain or would another 2N5088 work just as well? It is just creating two unity gain copies of the input signal after all.

[antonis]

Perhaps the designer didn't want to change the already oriented diodes...

(just kidding...)

[PBE6]

I've always been a bit sketchy on transistors, BJTs in particular. According to this basic info:

http://www.learningaboutelectronics.com/Articles/Difference-between-a-NPN-and-a-PNP-transistor

it seems that while an NPN is designed to turn on progressively with increasing base current, the PNP turns off at higher base currents and turns on progressively with decreasing currents.

So how does that work? The 2N5088 has a fair amount of current gain, so much of the time the current going into the base of the 2N3906 will be high. According to the above, that means Q2 will be turned off much of the time, only generating signal copies over a small range of input voltages. Is that right? Why is that necessary? I thought the point was to get two similar or identical copies, rectify them and add them back together?

[R.G.]

Been looking at the Green Ringer schematic at General Guitar Gadgets, and I'm wondering is there a reason why the phase splitter Q2 is a PNP transistor as opposed to an NPN?

It's more elegant and saves parts. The first transistor, the NPN, has its collector sitting at about the right voltage to properly bias the base of the PNP. Changing Q2 to NPN would require more parts and perhaps some component-gymnastics to get an NPN to work well there.

Also, is it important that it be low gain or would another 2N5088 work just as well? It is just creating two unity gain copies of the input signal after all.

It is not important that it be low gain. Any other similar modern **PNP** would work about as well. A high gain PNP like the 2N5086 or 2N5087, 2N4250, or other similar device would work as well.

[PBE6]

I'm still a bit confused about all this. As far as DC biasing goes, I think this is what's going on:

For Q1 (NPN) the base is sitting at 9*160k/(560k+160k) = 2 V, which means the emitter is sitting at 1.3 V. The emitter current is then 1.3/6k2 = 210 uA, which means the collector is sitting at 9-(210uA*18k) = 5.2 V.

For Q2 (PNP) the base is sitting at 5.2 V, which means the emitter is sitting at 5.9 V (is that right?). The emitter current is then (9-5.9)/10k = 310 uA, which means the collector is sitting at 310uA*10k = 3.1 V.

Now, I think Ic(sat) for Q2 is 9/(10k+10k) = 450 uA, and Vce(cutoff) = 9 V, while the Q-point is at 5.2 V. This is close to mid-point biasing, and I'm not 100% sure (...honestly, really not even 10% sure) what the AC-load line math would be for this amplifier, but I assume this is what RG meant when he said Q2 is properly biased.

If any of the above is correct, then is it the case that since the output of Q1 is inverted with respect to the input signal that Q2 needs to be a PNP in order to get the proper behavior? Because the "positive" current into a PNP is actually flowing out of the PNP base, so the negative swing from the NPN output makes things proceed in the right direction?

I'm guessing I'm still wrong, since if that's the case wouldn't it make sense to start with an NPN configured the same way as Q2 to start with, rectify and mix as before, then amplify on the back end? Is there a huge functional flaw with this approach, other than maybe too much noise?

Still confused..

[Digital Larry]

I am probably not going to answer any of your questions directly, but...

I've always looked at a phase splitter as an emitter follower combined with a unity gain inverting amp (output at the collector).  With no input, E would like to be about 1/4 Vcc and C would like to be about 3/4 Vcc.

[R.G.]

First:

I've always looked at a phase splitter as an emitter follower combined with a unity gain inverting amp (output at the collector).  With no input, E would like to be about 1/4 Vcc and C would like to be about 3/4 Vcc.

That's the right way to look at it for this kind of split load phase inverter. The emitter is a follower, and this produces a nearly-equal current in the collector resistor, but opposite in phase.  This is the transistor adaptation of the vacuum tube split load phase inverter.

For Q1 (NPN) the base is sitting at 9*160k/(560k+160k) = 2 V, which means the emitter is sitting at 1.3 V. The emitter current is then 1.3/6k2 = 210 uA, which means the collector is sitting at 9-(210uA*18k) = 5.2 V.
For Q2 (PNP) the base is sitting at 5.2 V, which means the emitter is sitting at 5.9 V (is that right?). The emitter current is then (9-5.9)/10k = 310 uA, which means the collector is sitting at 310uA*10k = 3.1 V.
Now, I think Ic(sat) for Q2 is 9/(10k+10k) = 450 uA, and Vce(cutoff) = 9 V, while the Q-point is at 5.2 V. This is close to mid-point biasing, and I'm not 100% sure (...honestly, really not even 10% sure) what the AC-load line math would be for this amplifier, but I assume this is what RG meant when he said Q2 is properly biased.

Pretty much, yes. Q2 is correctly biased. Well, acceptably biased in that it's well within the linear region.

If any of the above is correct, then is it the case that since the output of Q1 is inverted with respect to the input signal that Q2 needs to be a PNP in order to get the proper behavior? Because the "positive" current into a PNP is actually flowing out of the PNP base, so the negative swing from the NPN output makes things proceed in the right direction?

No. The output of Q1 is inverted with respect to the input, but the output of Q2 is intended to be both inverted and non-inverted, so the polarity of the signal it gets from Q1 is irrelevant.

I'm guessing I'm still wrong, since if that's the case wouldn't it make sense to start with an NPN configured the same way as Q2 to start with, rectify and mix as before, then amplify on the back end? Is there a huge functional flaw with this approach, other than maybe too much noise?

I think you're overthinking it.

The central point of this circuit is the two diodes being fed equal and opposite-phase AC signals so they can full-wave-rectify them. Any system to get that signal to the diodes with an appropriately low impedance will work.  The "inputs" to the diodes are AC coupled through capacitors, so the DC levels of the signals fed to the capacitors and on into the diodes don't matter. The problem to be solved is how to get equal-and-opposite-phase signals for these internal inputs, and of course to make that signal be the right size, not too big or too small, for the output to be heard as an octave.

It is possible to use two NPN stages in series to do this. Some gain is needed, so the simple thing to do is to make Q1 do all the gain, and make the second NPN have an inverting gain of 1 so its output will be the "equal-and-opposite". It turns out that getting a gain of very close to -1 (it has to be inverted, remember) is tough to make reliable. The simple way is to make the stage be a follower with a gain of nearly unity, and put an equal resistor in the collector side to force the collector side to have an equal and opposite signal. And that immediately results in the split load phase inverter, reinvented.

Otherwise, you have to give the second NPN stage some gain, try to use resistors to stabilize the gain, and then pad down either the input or output or both to get the signal back down to unity. If you attenuate the signal on the base side of this NPN, you get worse noise and a tricky adjustment problem that has to be set differently for each circuit board. If you attenuate the output, you're throwing away some signal output size, and that's already in short supply in 9V powered pedals. And on top of that, you have to add the biasing and coupling capacitor components between stages.

So yes, it can be done. But using two NPNs (or two PNPs) in series and trying not to use the split-load phase inverter results in a more complicated circuit that also needs individual adjustments. Using an NPN and a PNP lets you get gain from the first transistor, directly couple the second transistor base to the collector of the first transistor, and then use the second transistor as a split-load phase inverter directly. And it avoids at least two biasing resistors, a DC-blocking cap, and a potentiometer to tweak the signal level down.

[PBE6]

Ahh!! Thanks, that makes much more sense.