If you need matched pairs to reduce distortion, especially OpAmp should be used for their selection: once to assemble the system for precise adjustment, and to benefit all life. My simple equipment for exaction һ21е and check Vce of the transistors I have collected 20 years ago, since then I have no problem with that. Well, if you have no problems with it.
I have not found that.
I can select for hFE (in a thermally coupled LTP pair) and when I test for equality of collector current they can be far out from balance. This collector current error is due to Vbe differences.
Hi Andrew,
Vbe differences will cause very little change in the collector current of a pair as long as they match in beta and are held at the same temperature. My experience has been about opposite to yours. Mind you, I'm looking at parts from the same batches so that Vbe is close anyway. Once you start playing with differing batches or even manufacturers, getting matches becomes a lot more difficult.
The jig I proposed compensates for Vbe and beta since the transistors are held in a diff pair arrangement with nothing external to affect their balance. I used 0.1%, 10 K resistors from each base to common as well. You can even calculate the beta with it. Also incorporated was a current source for each diff pair, this test current is adjustable from 100 uA to 20 mA in steps. That allows the balance to be repeatable. The current selection occurs in binary steps, more or less. They are basically a red LED-BJT type where the LED and transistor are also in tight thermal contact and some foam.
It works very well, the most expensive parts being the 8 x 0.1 % resistors it requires. You could probably use 1% values and still end up with very tight matches if you match the resistors by hand. They have to match the other in a pair, not the listed value.
At least the concept has gone from using the xistor test feature on cheap multi-meters to jig where the temperature term is cancelled. The actual beta value isn't nearly as important as the fact that the transistors match very closely. With all the other questionable tweaks out on the net, at least this one delivers. Anyone with a half decent THD meter can verify this work. Additional support comes from D. Self's book. It was nice to get confirmation I was on the right track - beyond my own testing of course. I've been matching transistors for 30 + years. On Semi also verified my output matching activities. They claim up to 10 x less THD generated in the output stage if you match those parts together. That is also true.
-Chris
Vbe differences will cause very little change in the collector current of a pair as long as they match in beta and are held at the same temperature. My experience has been about opposite to yours. Mind you, I'm looking at parts from the same batches so that Vbe is close anyway. Once you start playing with differing batches or even manufacturers, getting matches becomes a lot more difficult.
The jig I proposed compensates for Vbe and beta since the transistors are held in a diff pair arrangement with nothing external to affect their balance. I used 0.1%, 10 K resistors from each base to common as well. You can even calculate the beta with it. Also incorporated was a current source for each diff pair, this test current is adjustable from 100 uA to 20 mA in steps. That allows the balance to be repeatable. The current selection occurs in binary steps, more or less. They are basically a red LED-BJT type where the LED and transistor are also in tight thermal contact and some foam.
It works very well, the most expensive parts being the 8 x 0.1 % resistors it requires. You could probably use 1% values and still end up with very tight matches if you match the resistors by hand. They have to match the other in a pair, not the listed value.At least the concept has gone from using the xistor test feature on cheap multi-meters to jig where the temperature term is cancelled. The actual beta value isn't nearly as important as the fact that the transistors match very closely. With all the other questionable tweaks out on the net, at least this one delivers. Anyone with a half decent THD meter can verify this work. Additional support comes from D. Self's book. It was nice to get confirmation I was on the right track - beyond my own testing of course. I've been matching transistors for 30 + years. On Semi also verified my output matching activities. They claim up to 10 x less THD generated in the output stage if you match those parts together. That is also true.
-Chris
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.