Transistor Matching Setup

[CJ]

this is just a Public Service message on a quick and dirty way to screen transistors if you are replacing, say, 20 of them onto a carvin b1500 or some other heat box,

you just take some readings on what base current sets up what collector current at a fixed voltage across the emitter and collector,

this can make the transistors last longer by even load sharing, and it is fun! 

use a heat sink otherwise the collector current will creep up as the transistor warms up,

you will need a dual pwr supply, or a single supply with a pot if you are clever,

this bench looks like this (also does mosfets at top of test jig)>

 

[CJ]

you will need to observe polarity when switching from NPN to PNP,

here is the schematic i use, base resistor is 470 ohms to help control base current, this resistor will change depending on which type of transistors you want to screen, it will go up in value for small signal stuff,

note that if you are replacing old parts, you might find a device that is the same only in a lead free package  for cheaper,
for example, a MJW21193 sells for about $4.50, it's new lead free p/n NJW21193 sells for about $2.50 in quanities of 10 ea.

that is all i can cough up for now, unless you want to add something or point out a glaring omission as to look good at my expense,  ;D ;D ;D

 

[joaquins]

When I read the subject I said, no again... When I saw the poster I got in. 

I'd use a 1K resistor for the base for easy reading, or any resistor that works for matching an analog meter scale I have on hand, so I read directly in the correct unit  8)

I don't know what you are reading, if it's base current vs colector current that should be enough. Would be more stable adding a resistor on the emitter but I guess would made measurement less sensitives, unless you do it for the final topology.

JS

 

[Gus]

How many points do you match?
It looks like a manual curve tracer
As joaguins posted 1k makes easy calculations, 1V across 1K is 1ma etc

 

[mjrippe]

Reminds me of this ancient doc from Moog...

 

[CJ]

yes, that 470 was chosen rather arbitrarily, which means

    forthwith  promptly    swiftly    at short notice    expeditiously    immediately    on the spot  peremptorily    readily  speedily without waste, it was determined by chance, whim, or impulse, and not by necessity, reason, or principle,  ;D

 

[PRR]

> Reminds me of this ancient doc from Moog...

Yes, but that is Vbe matching.

CJ seems to be doing hFE matching.

In a good power amp, neither is super-critical. In a bad power amp, either or BOTH may be critical.

Gus is half-way to a super-simple hFE match. Add about 20 ohms in the collector, take the 1K from collector to base. Measure the collector voltage. You can (small caveats) know the hFE with a little math (uck). More important, you can sort a bucket of devices into hFE bins just with the one voltage reading. If three devices show 5.23V, 5.20V, and 5.74V,  you have one pretty-good pair and one outlier.

A second meter to read Vbe can help you sort for that. If the collector voltages are very different (implying very different collector currents), the match is crummy; but if the collector volts fall near each other, the Vbe readings are valid for most non-precision work.

 

[CJ]

yes, we use to screen 2N4124 transistors for low collector voltage when hammered on by saturating the base,

customers had PLC (programmable logic controllers)  that would screw up if there was more than 0.2 volts across the  transistor when switched on,

about 20 percent of the 4124 devices will do .18-.20 across the collector,

we were replacing relay contacts with solid state contacts,

 

[JohnRoberts]

In a good power amp, neither is super-critical. In a bad power amp, either or BOTH may be critical.

+1  there's nothing like large scale manufacturing to refine a design for repeatability.

Emitter degeneration resistors pretty effectively force sharing between bipolar power transistors and can minimize Vbe effects.  It helps to be able to buy components graded for beta and fairly tightly controlled Vbe. At Peavey the Motorola power devices were close enough that they didn't even use trims for output power stages (they did use a special dual-diode in the bias string that was specified for forward voltage.)

I actually  matched Vgs in pairs of JFETs used in a phono preamp front-end back in the '80s but JFET Vgs parameters vary more than bipolar Vbe.

JR

 

[ricardo]

This is not very different from how most digital multimeters measure hfe.

It's also exactly how my 1970's analogue multimeter does it.  8)

 

[CJ]

that hfe spec seems to change depending on how much base current you bias at,

gain seems to go up with base/collector current,

 

[PRR]

> hfe ... gain seems to go up with base/collector current,

And then it goes down.

At very low collector current, recombination (you might say leakage) eats your base current, leaving less for the intrinsic hFE to multiply-up. At high current things get crowded in the emitter and multiplication is less efficient.

In the end, hFE is zero at either zero current or infinite current. There's no sudden change; hFE is always different at different currents.

See Shea 1952. In those days the two effects ran into each other, there was a fairly narrow range of good operation. Today we can buy power devices with fairly flat hFE from 10mA to 1A. We usually peak "power" amps well into the crowded zone where hFE is falling; a fatter driver is cheaper than adding fatter/more output devices.

There's some hints on this in Cordell's book, although his main approach is to go for fairly hi-performance devices instead of working-around lesser parts.

 

[mjrippe]

> Reminds me of this ancient doc from Moog...

Yes, but that is Vbe matching.

CJ seems to be doing hFE matching.

In a good power amp, neither is super-critical. In a bad power amp, either or BOTH may be critical.

Correct as usual PRR.  I was really just adding that doc for any future searchers looking for transistor matching info in general.  But as you say, both methods may be useful/critical.