Is there any good transistor substitution methodology -- the complete question

[Norm Dresner]

Let's say that I've found a really neat audio circuit on someone's website
that I just have to build -- toinight! -- but I don't have some -- or all --
of the exact transistors specified in the diagram. Let's assume for
argument's sake that the ones I'm missing are the transistors in the
long-tail differential input part of the circuit. Let's also say that I have
a stock of, perhaps, two dozen different transistor types, (almost) all
2N-types

What's a good substitution methodology for arriving at a working circuit?

Clearly polarity (NPN/PNP) matching is critical. Clearly breakdown voltages
have to be sufficient.

Where do I go from there? What's the next most important characteristic?
Frequency? Gain? What am I forgetting?

I suppose I could use a Spice-based circuit-analysis program -- but this
assumes that the program has models for all of the transistors involved.

How easily could I substitute, say, a switching transistor, perhaps one of
the 2-3 dozen 2N2222's or 2N2102's that I have, for an "analog" transistor,
say, just picking one I saw the other day, 2N5550? Is this foolhardy? Or
is it worth putting in a socket and trying it?

All suggestions, hints, and even well-meaning flames graciously accepted.

Norm

 

[Winfield Hill]

Norm Dresner wrote...

Let's say that I've found a really neat audio circuit on someone's website
that I just have to build -- tonight! -- but I don't have some -- or all
-- of the exact transistors specified in the diagram. Let's assume for
argument's sake that the ones I'm missing are the transistors in the
long-tail differential input part of the circuit. Let's also say that I
have a stock of, perhaps, two dozen different transistor types, (almost)
all 2N-types

What's a good substitution methodology for arriving at a working circuit?

Clearly polarity (NPN/PNP) matching is critical. Clearly breakdown voltages
have to be sufficient.

Where do I go from there? What's the next most important characteristic?
Frequency? Gain? What am I forgetting?

I suppose I could use a Spice-based circuit-analysis program -- but this
assumes that the program has models for all of the transistors involved.

How easily could I substitute, say, a switching transistor, perhaps one of
the 2-3 dozen 2N2222's or 2N2102's that I have, for an "analog" transistor,
say, just picking one I saw the other day, 2N5550? Is this foolhardy? Or
is it worth putting in a socket and trying it?

All suggestions, hints, and even well-meaning flames graciously accepted.

My answer is to read data sheets. But perhaps you're asking what to
look for on the data sheets? OK, depending on the application, after
polarity and case style, examine the maximum collector voltage, the
maximum current-carrying capability, thermal resistance and capacity,
collector capacitances, high-frequency transition frequency f_T, and
lastly, the current gain. For preamps, I'd also look at noise specs.

Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)

 

[Joerg]

Hi Norm,

The other critical parameter is hfe at the current they are run at. You
can usually see from their position in the circuit how much you need.
There shouldn't be a need to fire up SPICE just for that. It would work,
of course, but it's like swinging at a fly with a ball peen hammer.

Other than that, I am constantly using switching transistors in linear
circuits, mostly for cost reasons. Transistors are often designated to a
certain market by, you guessed it, marketing folks. I have to be careful
here because I am married to one .

Then there is noise. But if that matters in audio apps at all it
probably will only be of concern for the first stage.

Matching: A good design shouldn't rely on matched parts. But if it does,
most meters would be enough to pick nice pairs from a large enough bag.

Regards, Joerg