Matching transistors - Measuring hfe

[jacco vermeulen]

Originally posted by taj
Do these folks have any reputation I should be concerned about?

Nope.
Don't expect 2 DMMs to give the same reading, not even same model/brand ones.
GR hFe is >200. Getting a reading as low as 100 tells you something about your DMM, not about the Tosh.
The hFe function on a DMM is good for matching/checking, not for accurate hFe measuring.

[AndrewT]

Hi,
I am not sure, but I think a DMM hFE test uses a constant current feeding the b-e junction and measures the resulting collector current.

This is very much different from setting constant collector current (Ic) and measuring Vbe and base current (Ib).

The constant Ic test set up is very easy to do on the bench IF you use a constant voltage supply.
This will give far more consistent results than the presumed DMM hFE method assumed above.

Tom,
copy the PASS method shown on his web site.

From the few tests I have carried out I consistently find a much bigger spread with PNP & Pchannel than the tight groupings I find for NPN and Nchannel. I think this points to the production difficulties in achieving good and reproduceable PNP and Pchannel characteristics.

[taj]

Originally posted by jacco vermeulen


Nope.

...Getting a reading as low as 100 tells you something about your DMM, not about the Tosh.
The hFe function on a DMM is good for matching/checking, not for accurate hFe measuring.

Cool. Thanks Jacco.

[taj]

Is it important (or worthwhile) to match diff-amp collector resistors as well? Or is a 1% mf resistor close enough?

..Todd

[anatech]

Hi Todd,

Is it important (or worthwhile) to match diff-amp collector resistors as well?

It can be, depending on your application. Remember that this is one small part of the circuit.

Or is a 1% mf resistor close enough?

Generally speaking, I would say yes. As long as you are taking a signal from both collectors, otherwise it will not matter.

-Chris

[AndrewT]

Hi,
I am not able to test/measure the tiny errors (distortions) that occur when the diff amp (LTP) runs with unbalanced collector currents.

The current injected from the VAS can be very significant and if using the collector loads for measuring and balancing the LTP then the VAS current introduces an error.
According to D.Self, the required balance accuracy is quite critical to minimising this first stage distortion and when the currents are as little as +-1% there is a very significant increase in distortion (he shows his graphs with the current difference and +-1%=2% difference).
If you want to better that and I think Self implies you should aim better than +-1% then the currents should be measured at the emitter resistors and I suggest matching to 0.5% if you think LTP balance is important to your design.

[anatech]

Hi Andrew,
As the LTP is the error detector/corrector in an amplifier, I would say it's important.

Get things as closely balanced as you reasonably can. That is the only advice I can give. I realise we may not see eye to eye on this point Andrew.

-Chris

[AndrewT]

Hi Anatech,

I realise we may not see eye to eye on this point Andrew

I suspect we do see eye to eye on many topics, matching of emitter resistors for future balancing measurements is important to me.
I was trying to offer a reasonable viewpoint that would be acceptable to most. Personally, since it costs nothing but time, I try to match (not absolute acurracy) to +-0.1% if I can see a benefit.
There is nothing worse than seeing/finding a voltage error and not knowing if the error is in the resistor used for measuring or in the semiconductor upstream.

Have I interpreted the VAS base current contribution correctly and do you agree that Re is better than Rc for LTP balance measurements?

[anatech]

Hi Andrew,
I agree that the effect of Vas base current must be compensated for. If emitter resistors are used, they will be very effective if matched. You may also unbalance these to compensate for the tendancies of the rest of the amp.

I try to make everything happy sitting near ground with minimal diff amp action first. Then your front end is more concerned with correcting errors than keeping the entire project from going DC.

I think we are in full agreement then. I just wanted to point out that there are other factors which may force the LTP from a balanced condition even though everything is matched in that location.

-Chris