Hi Salas
As I want to test the BC327-40 (550 hFe) I found in my drawer (From an old paradise batch)... but as my BC560C sounds so good now, I decided to build another one... so the questions are coming up...
I have some uPA68H with 10mA Idss... should I set a BF256 CCS with 10mA so to bias the jfets at near Idss ?
As I want to test the BC327-40 (550 hFe) I found in my drawer (From an old paradise batch)... but as my BC560C sounds so good now, I decided to build another one... so the questions are coming up...
I have some uPA68H with 10mA Idss... should I set a BF256 CCS with 10mA so to bias the jfets at near Idss ?
Prefer your non input shorted best trimming result, then install the servo chip. But achieve your best insulation first.
You don't want elevated J3 dissipation for reliability, also not its inferior CCS stability when with much less degeneration resistor value & higher temp, even not to add higher current noise through the mirror's TO-92s. I had found the standard bias point very nice for overall measurements during the development anyway.
Hello Salas,
All good with the DCG3, but this post just got me wondering how subjectively the sound would be different with 2SK170 instead of UPA68h?
Anybody else using 2SK170 please comment?
Thanks.
nash
They bring down the THD even more due to higher gm (thus more feedback is created) but they tilt the technical benefit towards the lows and mids due to they are slower because of their higher internal pF. With the uPA a THD sweep across frequency is more even. I have a friend who swears by them but I haven't listened to his build after he substituted with those. I had only measured their traits in DCG3 during development. Try them if you got some to morph your own subjective opinion.
Hi I checked the voltage on the potential divider R8/R9, the good channel was 4.3V the bad was 4.27V so I would think that was ok.
What I did find was that despite both BF256 CCS's both giving 5mA. There was a huge difference in the voltage measured on the drain. The good channel had 212mV the bad channel had 8.68V. Both voltages measure relative to 0V. I presume there is something very wrong there. Any suggestions on how to proceed?
What I did find was that despite both BF256 CCS's both giving 5mA. There was a huge difference in the voltage measured on the drain. The good channel had 212mV the bad channel had 8.68V. Both voltages measure relative to 0V. I presume there is something very wrong there. Any suggestions on how to proceed?
Thanks Salas. I very well might try them. What do you feel is ideal Idss to get? I am using the Diotec BC 327 in the mirrors.
nash
In your shoes I would now measure each semiconductor's leg DC voltage in respect to earth and write them down vs the good channel's readings. Then I would replace what semi I haven't yet replaced which is showing weird voltages around it. Along that, I would also scan for possible traps like some unlikely insulation leakage on the Mosfets assembly, an oddly wrong resistor value installed somewhere, anything.
Also important is to thoroughly clean possible flux residue everywhere but mainly around the Jfet & Mosfet legs (gate legs especially). Leakage through dirt and/or baked flux for high impedance nodes can bring about very strange effects.
Any. It will be brought down to 2.5mA on its yfs vs ids curve by the tail ccs bias system. Down low they all have almost same transcoductance result. You just need samples with Idss >3mA. That way most of them become useful, no waste. Its good to have relatively alike idss between channels so they become even more matched down low in their curves when applied but its not a make or break deal term.
Same case as in the reply above for uPA but within each channel's pair there should be Idss match. Even better is curves match for those having curve tracer test gear. The uPA is a super matched monolithic following that condition by default. The K170s are individuals and need pair matching in a differential pair as used in this circuit. When uPA is a curve matched pair side by side in each channel no matter the idss reading as a package per sample.