Resistor in Sziklai pair

[Fancy Lime]

Hi there,

I've been messing around with Sziklai pair based fuzz circuits a bit and have a question: There seems to be an "optional resistor", that is there in some Sziklai designs, but not in others. Here it is shown dashed in gray:

https://1.bp.blogspot.com/-5t0y2KuHQt4/Vnb7Hk2RedI/AAAAAAAABww/GxAonBlvl0c/s1600/sziklai_pairs.png

The Jordan Bosstone has it, too (R5):

https://www.diystompboxes.com/pedals/bosstone.gif

In addition, the Bosstone has a cap in parallel (C3). I'm guessing wildly that the cap is supposed to suppress the MHz oscillation that seems to plague Sziklai pairs (don't remember where I read that). Is that so? And what does the resistor do?

Thanks,
Andy

[PRR]

Can you write-down the currents in each transistor by inspection? With and without the added resistor?

That would be a first step in figuring out what the resistor does.

[amptramp]

The resistor helps shut off TR₂.  Without it, the only way charge carriers in the base of TR₂ get swept away is by recombination.  It helps to have something to take the base below the Vbe cutoff to speed up the process.

[iainpunk]

i heve used a sziklai based drive/fuzz with a combination of a NPN Si and a PNP Ge (2n2222 and OC171) and i experimented with the resistor value and leaving it out. i found that the sound was more `tweed like` with no resistor, and more `round´ with the resistor, but your mileage may vary depending on the circuit.

[Fancy Lime]

The resistor helps shut off TR₂.  Without it, the only way charge carriers in the base of TR₂ get swept away is by recombination.  It helps to have something to take the base below the Vbe cutoff to speed up the process.

Ah, I see... that makes sense.

Thanks,
Andy

[Fancy Lime]

Can you write-down the currents in each transistor by inspection? With and without the added resistor?

That would be a first step in figuring out what the resistor does.

Paul, you can't help but answer a question with a counter question, can you? Very educational question, though. So I'll try, Sensei:
I cannot figure out the currents quantitatively because i don't know how and don't have the time to look that u right now. I'll do so as soon as I have some time. But qualitatively, it goes like this (NPN version):
If B of TR1 swings high, TR1 opens and wants to draw lots of current. How much it can draw, however is limited by how much will flow out of the base of TR2, unless there is the resistor, in which case the resistor is in parallel with the current through B of TR2. If B of TR1 swings low, TR1 shuts off and B of TR2 should float high to shut off TR2 but cannot do so easily unless the resistor is there because the C-B current of TR2 limits how fast B of TR2 can move up. This latter point seems the important one, as Ron has also pointed out. Correct? Slightly off? Complete bollocks?

Thanks,
Andy

[PRR]

> don't have the time to look that u right now.

"Look it up"?? Where?

Before a question like this you should have *some* understanding of transistors. Most practical transistor stages can be estimated "by inspection".

Clear out anything not related to the DC condition.


At first, all we know is that Q1 Vbe is around 0.6V. 0.5V-0.7V depending on current, not a big spread. If the whole thing is going to be happy, Q1 must work to force the top of R4 so that the R4 R3 R2 divider has 0.6V at Q1 base. Neglecting base current, top of Q4 should tend to be 5V.

Now knowing the voltage across R6 is 9V-5V= 4V we know the current in R6. Probably most of this flows to Q2, but that's not important yet.

Assuming Q2 base current is small, the current in Q1 must be Q2's Vbe of 0.6V over R5 18k. 0.030mA.

Now take out R5. The only current Q1 can get is Q2's base current. We need an estimate of Q2 hFE. I was taught "assume >50" but with newer parts it will be over 100. And seeing that R6 and Q2 are flowing 0.220mA, we take a cheap/lazy trick, say "hFE=220?", and the only current in Q1 must be 0.001mA. If Q2 hFE is only 110, 0.002mA. Or 1uA-2uA.

hFE and speed fall off with current. What is the current gain of a transistor at 1uA? For most transistors this is right off the graph. Our best friend 2N5089's graph only goes to 10uA. Before 2N5089 price fell to a penny we would be shopping bulk generic transistors to build pedals for a profit, and some of these would crap right out (not work like a good transistor) near 1uA. Our odds of building a "good" pedal are very much better working Q1 at 30uA.

If you don't know any better, a ballpark value for R5 will be to use Shockley's Law and hFE to estimate the input impedance of Q2, and pick a similar resistor. At 0.22mA, Shockley predicts hie of Q2 is 120 Ohms, times hFE of 220, is 26k. If using cheap transistors, hFE=110, then 13k. 18k is right in there. 2:1 difference is not too important.

Doug Self touches on the "optimum" in his Amplifier Institute (bottom, "DC-bootstrapping"). More extensive experiments in his Small Signal Audio Design book which I highly recommend. (I realize you can buy store-bought pedals for less than the price of his book; but note that used copies fetch full price so a lot of people must think it is worth it. )

[amptramp]

This does bring up another point - at high temperatures, the leakage current out of TR₁ could get high enough to turn on TR₂, meaning it would never turn off.  I have seen people use a Sziklai pair as an SCR where the device does not turn off until the voltage and current go to zero.

[Fancy Lime]

Hi there,

very helpful explanations, thanks!

@Paul

"Look it up"?? Where?

Well, exactly. I did not know where, which is why it would have taken me a while. I meant "Look up the basics that allow me to figure out the specifics in my own". Thanks to your book recommendation, I shall henceforth know where to look. I am clearly lacking some of the necessary background on transistors. But your explanation is easy enough to follows.

Thanks,
Andy

[PRR]

> at high temperatures, the leakage current out of TR1 could get high enough to turn on TR2, meaning it would never turn off.

Yes. You very-very-rarely see this done in Germanium. If you follow my analysis, just a few uA leakage in TR1 will saturate TR2. GE Transistor Manual 1964 shows it done twice, both oscillators (violent overdrive of Q1), both with low-Z loads (so much higher nominal currents). Well-designed Ge amps "always" have a base drain resistor.

Small Silicon after 1975, you could be very very sure of <1uA leakage and you could try micropower without that resistor. However it is easy to get so low that transistor gain falls-off and speed won't cover the audio band (and not predictably enough to use as a filter).

Define transistor operating point with Resistors. (Or Vbe differences, but that's an advanced topic.)