The capacitors should be more important than the sense lines. They
provide part of each regulator's frequency and phase compensation.
I would be suspicious if not hooking up the sense lines has an effect.
They are all shunted to the main lines via low value resistors from 47
to 470 ohms. Could one or more of those resistors be partially open?
Or maybe the pole-zero compensation in one of the regulators (resistor
and capacitor in series like R3278/C3278 and R3289/C3289 in the -15
volt regulator) is bad.
The -15 and -50 volt regulators are more liable to oscillate because
their NPN output power stages have gain. They have extra frequency
compensation of their output stages to help control that.
On Fri, 26 Oct 2012 20:48:35 -0000, "Albert" <aodiversen@...>
wrote:
Hi Jerry,
Nothing helped to get rid of the oscillations when using my testload. True oscillation, about 12 MHz. So as a final attempt I also made sense lines connections, something I never did before. Against my expectation this cured the problem! There still is some switching noise in the rhythm of the inverter, but the DC voltages are fine now.
This is the final configuration (|| means in parallel):
+/-50V both 470R || 1k5, both 10uF to GND
+/-15V both 22R || 22R, both 100uF to GND
+5V 6R8 || 6R8, 100uF to GND
+5VL 5R6 (to GND Lights, I did not connect pin 3 and pin 7).
No other ground wires, except earth from chassis PS to chassis main frame to be on the safe side. All loads via the 10-pin LV connector and the 6-pin Sense connector. So nothing at the 54V lines for the HV supply.
I didn't try it yet with less load, and I added the caps (values taken from the Main Interface Board) before I added the sense lines. So maybe these are not (all) needed. Anyway the result could be used as a guide I think. I used forced cooling; some resistors had to dissipate a little more power more than rated for.
Albert
---
I would expect the sense lines to be needed? They are directly connected on the interface.
Jerry Massengale
David Hess