negative alw super regulator problem

[piewie]

I have just made four regulators based on ALW's design
The positive regulators seem to work OK, but the negative
act strange.

They both produce a smaller output voltage then the positive regulators: 14.7V versus 15.2V for the positive.
Supply voltage is -28V.
At a closer inspection something seems to be wrong.
The opamp gets very hot (75°), there is a differential voltage remaining at the inputs (6.7V and 6.9V at the inputs).
The green led is working and the voltage across the zener is correct.
Transistors are d45h8 and ztx651.
When shorting the collector and base of the ztx connected to the opamp (like in jung's original) design the voltage drops to +/- 12.5V.

Would anybody know what the problem could be? Bad pinout for ztx serie in protel, need for different resistor value's?

[piewie]

Schematics used are these:

Schematics

[Eva]

If both regulators are both the same circuit and use independent floating transformer windings to achieve negative regulation, then it would be a good idea to try disconnecting the ground of the positive regulator from the output of the negative regulator. You may also swap regulator circuits and see what happens

That topology is prone to parasitistic oscillation on the negative side due to capacitance/resonance between windings and sometimes requres common mode filtering to isolate inter-winding capacitance to work properly

[piewie]

i updated the wiring like this:
starground: to neg pcbs
to pos pcbs
to load
sense ground to gnd connection on pcb

i don't see any ripple on the load connections now, but the
ripple remains at the feedback input of the opamp.
Opamp remains hot and output is still lower then the positive regs

[piewie]

There seems to be a huge ripple on the collector of the lower ztx (2V ripple) of the negative regulator. On the positive i get pure DC. Any ideas what could cause this?
I just removed some ground loops on the pcbs, but the problem remains the same.
Also when disconnecting the power of the positive regulators nothing changes.

No matter how i rewire everything the problems inside the negative regulators remain. I checked it dozens of times but I cant figure out what's wrong with them. Could anybody help me with them please?

[ALW]

I do hope you haven't got C9 fitted in your schematic - a recipe for oscillation if ever there was one.

PCB layout is critical too, you shouldn't have ANY ground loops or critical shared current paths - this is a high performance, wide bandwidth, precision instrumentation job, not a low-frequency DC circuit. At first glance I'd point the finger firmly at a layout issue.

Also check op-amp choice - try the AD825 for a recommended baseline.

Andy.

[ALW]

I've asked for this thread title to be changed, I don't mind discussing this at all, or contributing help, but I wish to make totally clear this IS NOT an ALW regulator, the design of which includes the critical PCB element.

Andy.

[piewie]

Yes i did use C9 in the design, but I already had removed it, it changes nothing, the same with C6.
Groundplane has been removed around C9, R4, Q2 and C7 to remove the groundloop i had made.
I just removed C9, C6, C3 and ran a wire from output to R2, doesnt change anything.

[Fred Dieckmann]

Your schematic indicates you are using an OPA623. This is a high speed current feedback and will not work in this topology! Your value for R6 is 1K ohms which is wrong. Make it 10 ohms or you can jumper it out. A high speed regulator like this is very layout and parts sensitive and I believe Andy spent several years playing with the design to get it to the level of performance it achieves. Changing parts, topology and even PCB layout, is an invitation to trouble. Try the AD825 (or similar speed voltage feedback op amps), jumper R6, and remove C9 and possibly C6. If you keep C6, I would put a 1 ohm resistor in between the pin 4 - C6 junction and Vout. Similar changes should be made to the positive regulator if the topology and op amp are the same as the negative regulator.

[piewie]

I am using an ad825 in the design. I had to use another opamp from my protel library because AD825 wasnt in there. Pinout should be the same.
R6 is a jumper in final design.
I think all the components I used are the same as the original design.
C9 has allready been removed.
I put a 4.7ohm resistor on the supply line to C6, doesnt alter the output.
Is this still a standard oscillation problem? No matter what i change to the circuit, the problem seems to remain the same on the scope trace.
Could someone verify the placement of the ztx transistors?
The distortion on the output of the opamp is only small and not in proportion of the distortion on the ztx transistors. ( some mV with +/-8V offset on opamp vs some volt sine peaks on ztx transistors.
Thanks for the support so far.