Transistor Triple (like Clapton Mid Boost)

[sbirkenstock]

Hi Everybody,

got a question about this "transistor triple". (Ignoring Q1)
I attached the schematics of the Clapton mid boost.
Q2 and Q3 are NPN - Transistors and Q4 is a PNP Transistor.
Is that required? I do not unterstand why.
Could I use an NPN for Q4 as well or what changes would it require?
Any hints?



Also in this schematic I do not get the Vol1, Vol3, Vol2 thing in the middle.
Any hint appreciated!
Best regards,

Stephan

[ElectricDruid]

If the circuit uses a PNP transistor, it's because it needs to. You could probably re-arrange the circuit so that it used a NPN, but then it'd be a different thing.

The VOL1,2,3 and MID1,MID3 are connections to the control pots.

This circuit is incomplete, since we don't know what the pot values are, and we haven't got the off-board wiring. That wiring presumably includes MID2, the Mid control wiper, and also the rather important OUT!

I'd hunt for another schematic to compare that with, or at least an off-board wiring diagram.

HTH,
Tom

[PRR]

> Q2 and Q3 are NPN - Transistors and Q4 is a PNP Transistor. Is that required?

Darn good question.

The short answer is: if you don't know why, just do it like the designer did it. He had reasons.

The too long answer:

Amplifying devices can be made in two polarities: N-type and P-type.

It is often convenient to have both types.

However, for much of history, we had no choice. Vacuum tubes are all N-type. Early Germanium transistors were almost all P-type (PNP). Early Silicon was mostly N-type (NPN). Chips, for a long time, had only good N-type and poor P-type devices.

The first reason is gain advantage. Especially with BJT transistors. With all the same type, one stage can't directly drive the next, except by diverting current from a coupling resistor. With both N and P, the N's output current can flow directly as the P's input current. The gain advantage may be large or negligible, but is often-enough useful.

The other reason is DC levels. The output of a device is usually higher on the supply voltage than its input, or the input of a same-type device. Tube input at zero volts, tube output at 100V. Your plan, Q1, input near +1.5V, output near +6V. It is possible to divide-down or shift-down between stages; or tap a stack of batteries so each stage stands a little higher than the one before it (Loftin-White). Using both types allows really elegant DC-coupled amplifiers.

*Here*, Q2-Q4 is really an Op-Amp. You could put a TL071 there. (Might not break-up the same.) The DC coupling is not essential (ever! in audio). But a generation of us now-old guys were raised on $1 chip op-amps, and like to design that way even if we choose to build the opamp out of parts.

There is *surely* a way to build the same audio function with only NPN parts. Without knowing what hangs on the external MID pins, that path is not clear. Since 1972, small PNPs have cost the same as small NPNs, so there's little design reason to avoid them. (I understand you may have no PNPs in the house....)

[sbirkenstock]

Thanks a lot PRR,

I think I understood a bit.
Is it correct that both PNP and NPN transistors have the reversed polarity signal at the collector?
Just the collector is connected to ground with the PNP,
and to V+ with the NPN?

Then in the EC mid boost "example" R16, R19 and Q4 are (also) biasing the input of Q3. So we can omit an extra coupling cap.
Ok, and if the PNP works in that sense alike a NPN, that a large collector resistor and a smaller emitter collector give a higher amplification factor,
then it would add up for my understanding.
Are my assumptions correct?

thanks a lot,
Stephan
 

[Rob Strand]

If you replace the PNP (Q4) with an NPN, then in order to get full o/p swing the NPN's emitter would have to connect to the 0V rail (or trough a small resistors like the original ckt).  When you do that the base is stuck at 0.6V above 0V.   The problem now is the collectors of the differential stage transistor Q2 is pinned at 0.6V above 0V.  The differential stage cannot work correctly with collector voltages too close to 0V.   So DC-wise it all falls in a heap.

As PRR mentioned PNPs are slow.   When you want fast stuff like video and RF you try to use NPN's.
Here's a good example of an NPN version,
www.onsemi.com/pub/Collateral/NE592-D.PDF

The output stage is a buffer, so you lose the gain of Q4 your original circuit.  In order to compensate for the gain loss an extra differential stage is added.     These things have very wide bandwidth and usually operate with the first stage in open loop.  Not really designed to be stable within a feedback loop.

IIRC, some old oscilloscopes used these.

[PRR]

> PNPs are slow.

Not as a general thing; or not by much.

Certainly nothing we audio-people need to fret.

Older _IC_s were optimized for good NPN, which left the PNP lame. There was much experience with all-NPN wafer processing for logic circuits, and for a long time the fabs were reluctant to do the extra steps needed to put good PNP on the same wafer as NPN. (That '592, under another number 733, is VERY old.) They have to diffuse a little "island" under each PNP. But that became readily available over 20 years back.

[Rob Strand]

That '592, under another number 733, is VERY old.) 

Very similar if not the same.  Back in the day 120MHz bandwidth was amazing compared to the 1MHz / 2MHz stuff. IIRC they weren't expensive either.   Philips had some others, slightly more modern, which were 350MHz bandwidth - can't remember the numbers.

They have to diffuse a little "island" under each PNP.

Early ones were lateral PNPs but those might be vertical PNPs.  Tell you the truth, I don't read much of that stuff anymore.

[Edit maybe this one:  (another all NPN)
http://pdf.datasheetcatalog.com/datasheet/philips/NE5539D.pdf
]