Curves aren't going to tell you much. There isn't much variation between induvidual transistors, in the parameters that matter.
Well, hopefully in circuits that are made insensitive to those parameters.
Well, "high end audio", who knows. You'll have to find (or trace) the schematic and analyze it for stability and such.
The headline parameters are the more important ones. Vceo (actually Vces under current, Vceo is a minimax parameter) is measured like a zener. Ic(max) is more or less where hFE falls over. hFE and Vce(sat) are measured at typical conditions (not at the same time, mind: hFE drops sharply at low Vce, for obvious reasons!). Those will more or less fix the size of the junction: hFE falls over at high current density, and junction thickness (not necessarily die thickness) sets Vceo.
Junction size and thickness also set capacitance, more or less. (I don't think anyone is using super fancy doping profiles in amplifier transistors, and switchmode transistors have shit SOA so hopefully wouldn't be found here.)
That leaves the biggest degree of freedom: fT or t_r and such. A ring base connection, like the ancient mesa 2N3055, has shit all fT and Vce(sat), due to large Rbb'. Modern (80s+) epitaxial are planar, made with interdigitated emitter and base, much faster. A lot of power supplies, amplifiers and such fell victim to that, where they sang like a bird after repair...
To test that, set up an amplifier and measure the current gain as a function of frequency. It'd be probably a resistor from func gen to base, and collector load of a cascode (with a known faster, equally ampy) transistor. Or just a low resistance (noninductive) so Miller effect is still negligible.
Switching behavior isn't quite the same because of the highly nonlinear conditions, but that's easy to test, too, just set up the right turn-on and turn-off base currents. The sharpness of turn-on/off is distinctive of different types.
Also noise, but power transistors probably don't need to worry about that. If you're replacing input transistors, you may need to consider it. Testing noise figure is kind of a pain, you'll want to look up some references.
Tim
--
Seven Transistor Labs, LLC
Electrical Engineering Consultation and Design
Website: https://www.seventransistorlabs.com/
wrote in message
news:7fe0e947-7965-45ff-bb2b-49172f9de534@googlegroups.com...
I find myself in a situation where I am repairing some high end audio amps
and cassette decks that have house numbered semiconductors, mainly
transistors as well as JDEC p/n that are NLA.
I have found a few transistors that are dead and need to be replaced.
Fortunately, they are in parts of the amp where the other channel has an
identical transistor.
Having one transistor as a base will allow me to match something close to
the original.
I have made (in the past) the typical transistor curve tracer that shows
the curves on a scope - cute but not precise enough for my intentions. I am
more interested in a lab type instrument.
Basically what I am looking for is a curve tracer that has a micro attached
to it that have some A/D's on them that can record the data of interest
(e.g. Ib (vb), Vce, etc.
I've considered an arduino based approach but they have the limitations of
5V max on the input for the A/Ds and don't want to scale it.
I've looked for old Heathkit and Tek curve tracers that I could augment with
a micro but they seem to be very scarce, in addition to having transistors
in them that are NLA.
I could design one myself but in all honesty, I'd prefer to find something
where someone has 'plowed the ground' and at the very most, I'd have to
augment it with a micro. Yea, I am a bit lazy - the intent was to fix the
amp, not develop a curve tracer....
Thanks in advance
J