I have no doubt THD suffers with mismatched emitter resistors but I've seen first hand many times when an output device runs considerably warmer than the rest. I don't bother to try to measure actual resistance value so maybe my examples were more extreme. I put a bunch of resistors in series and measure the voltage drop across them to match them.
I do not understand the excitement about resistor matching.
I think it is far more important to match transistors.
Furthermore, resistors from the same batch come matched, much better than their nominal tolerance. Of course this is not guaranteed, they usually deviate from their nominal value, but all on the same side by a close amount.
I think it is far more important to match transistors.
Furthermore, resistors from the same batch come matched, much better than their nominal tolerance. Of course this is not guaranteed, they usually deviate from their nominal value, but all on the same side by a close amount.
It depends how hot you are running the amp. Output resistor matching becomes very apparent wit +/-75V rails on smaller heat sinks for example. Thermal runaway of a single device isn't good.
Thanks everyone contribute your thought. Of course, we wish an evenly distributed output among the output devices and between the upper and lower rail for low distortion, no runaway, identical phase between upper rail and lower rail for good sound quality and at least satisfy a DIYer's soul. Further consideration: Will be it possible to take one more step to adjustment for distribute the Re among the parallel transistors in a single rail? To equalize the individual current output (equal - heat) ? (If we can get the summed current equal btw the + rail and the - rail is great also.)
Say, appoint a slightly lower Re to a transistor has slightly lower Hfe (lower Ic/Ib)? Because we can not obtain identical hfe values in the power transistor match selection, 1% error is excellent. Will this final adjustment helps? Assuming errors in hfe and Re do exact, no matter how small. This is against my first assumption that all power transistors are identical in Hfe.
Say, appoint a slightly lower Re to a transistor has slightly lower Hfe (lower Ic/Ib)? Because we can not obtain identical hfe values in the power transistor match selection, 1% error is excellent. Will this final adjustment helps? Assuming errors in hfe and Re do exact, no matter how small. This is against my first assumption that all power transistors are identical in Hfe.
Last edited:
Hi Fjc,
It seems to me you've got a number of misunderstood points here:
1) Idle current of each individual transistor does not depend on hFE. BJT transistor is a voltage-controlled device. Bias is a voltage, leading to certain idle current. For even current distribution among the paralleled transistors, you need to match Vbe (not hFE). Anyway, emitter resistors make the current distribution more even, so normally - nothing to worry about, however, it always makes sense to check, measuring the voltage drop over each individual emitter resistor.
2) On a working amplifier, having zero DC offset at the output (controlled by the negative feedback and/or a servo or a trimmer), the sum of the idle currents for the positive and the negative half of the output stage will be always equal automaticclly. Even if you use the double-value for the resistors in one of the halves - the sum of the currents for the positive and the negative half will be equal (as soon as they are not equal you lose your zero offset, but it's controlled outside the OPS, as I mentioned above).
So - use the equal value emitter resistors (say, 0.22R), measure the voltage drop over each of them on the working build, see if some measurement is noticeably lower or noticeably higher and replace the corresponding transistor with the one with more suitable Vbe. Altering the resistor value will not give you the result you expect. +/-5% idle current deviation is perfectly normal. Even +/-10% deviation is still ok.
Cheers,
Valery
It seems to me you've got a number of misunderstood points here:
1) Idle current of each individual transistor does not depend on hFE. BJT transistor is a voltage-controlled device. Bias is a voltage, leading to certain idle current. For even current distribution among the paralleled transistors, you need to match Vbe (not hFE). Anyway, emitter resistors make the current distribution more even, so normally - nothing to worry about, however, it always makes sense to check, measuring the voltage drop over each individual emitter resistor.
2) On a working amplifier, having zero DC offset at the output (controlled by the negative feedback and/or a servo or a trimmer), the sum of the idle currents for the positive and the negative half of the output stage will be always equal automaticclly. Even if you use the double-value for the resistors in one of the halves - the sum of the currents for the positive and the negative half will be equal (as soon as they are not equal you lose your zero offset, but it's controlled outside the OPS, as I mentioned above).
So - use the equal value emitter resistors (say, 0.22R), measure the voltage drop over each of them on the working build, see if some measurement is noticeably lower or noticeably higher and replace the corresponding transistor with the one with more suitable Vbe. Altering the resistor value will not give you the result you expect. +/-5% idle current deviation is perfectly normal. Even +/-10% deviation is still ok.
Cheers,
Valery
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.