My+Your Opinion And Experience On Matched Transistor and MOSFETS

Status
Not open for further replies.

MR Overclock

Banned
2016-06-12 11:17 pm
Heres my exprience with mathcing trasntiors, I have matched a tip41C and tip42c today, precisly matched.

After putting it into the amp I can hear a improvement in sound, sounds a little bit more crisp and clear , and also more clean compared to before.
Although Its only two transitors matched and the outputs and dirvers aren't matched it makes a diffrence to sound.

Cause I am a sensitve guy I reckon its a massive diffrence, slightly better is big to me.

Whats your exprience and or oppnion on matched trasnsotrs and mosfets? :D

I have tired using diffrence operating voltage 45V vs 52V and theres a diffrence in the two gains when matching (BC voltages on both varry more voltage accross my load changes, not close to 0 mV any more)
, So if its matched at 45Volts its not matched any more at 52V, or not precisely matched
So its improtant to use the voltage your going to operating the trasntior with and use the biasing your going to have to match the trasntistors.

A matched trasnsitor using 12V may likely not be a matched pair once you use 50Volts with them

Heres exctally what I did to match my stuff, Find two matched resistors for bias, connecting a resitor from C to B.
Resitors connected in common collector mode, 1X pnp and npn using duall supply, connect Emiter legs together through a 4 OHMM 5w wire round resitor, and connect other end to ground. Hook everything up and measure voltage accross 4ohm to be closest to zero volts and have the BC voltage simullar on both trasnsistors. Wen't through about 8 pnp transistors and only two matched, 4mv and 1mv at .Worst I got is a varry of about 22mV at load, which is equivlent to 22Volts diffrence at BC voltages.
I get about 45Volts with my protection bulb on and 52Volts without the bulb, and gain changes once the voltage or bias varries

If I was to match a common base or emitter, I would match bias resitor or zener/diodes, depneding on which your gonna use.

Also you should use a PCB, and put wires comming out of the commponents that you need to connect to the trasnsitors, as you solder resitors and they have a temp change, they can degreade and could cause them to change values and the resitors don't match any more

Then measure BE to be same,
(not entierly sure my self which is equivlent to BC voltage in common emitter and base set up)

The load resitors also need to be the same, matching some 10K reistors. As common base and common emiter use the collector as out put, we cannot short them toghet to get zero volts(If you don't belive me you can try, I think the two trasnitors will blow up, cause I tried it before), other wise it will blow up, so they way i match them is to find same BC voltage
 
Last edited:
I reckon the best place to measure if your devices match is in the final amplifier. But that is generally inconvenient to taking measurements and requires repeated swapping into and out of a complex environment.

Instead I suggest you build a test jig that mimics the main operating conditions that the devices will meet in the final amplifier. Vce & Ic and Tc are the three main parameters that need close control to make matching worthwhile.
 
My 2 cents is that it's not perfectly clear at all.

What do we mean when the term "matching" is used?

Most people consider matching when the bias points are the same.

I consider devices matched when the curves match when seen on a curve tracer.

Finding N and Pch devices that have complimentary curves is often not so easy.
There are a few devices I know of that actually do have complimentary curves, and then finding "matched" pairs requires access to at minimum a 100pcs of each (unless you get lucky in a smaller batch)

Notwithstanding whatever feedback will do to cure ills, there's no doubt in my mind that having the best open loop performance before applying feedback is always a good idea. Truly complimentary parts that are matched P and N are a plus. Some circuits undoubtedly will benefit more than others. Which will and which won't in theory is a rather complex subject requiring a lot of EE type expertise.
 
Don't try to measure current.
Instead use a resistor and measure the voltage drop across that resistor.

An LTP always has at least one collector load. Use that collector load as your current monitoring device.
Most Builders have access to a DMM that can resolve down to 0.1mVdc (on the 199.9mVdc scale)
A 1k0 collector load with 0.1mVdc is a resolution of 0.1uA. That allows reasonable accuracy down to 1uA.

Still with that LTP, use two collector loads in your measuring jig. Match those resistors to better than 0.1%
Now you can measure the voltage DIFFERENCE between the two collectors. You are now measuring DIFFERENCES between the two Ic values of 100nA And you can swap the probes over to check it really is 0.1mVdc that you are reading.
Then you can swap the two transistors across and remeasure the DIFF @ collectors.
When you have these 4 meausrements, a little bit of averaging gives you a resolution of better than 100nA for Ic diff.
 
Last edited:
Status
Not open for further replies.