Hi, Please provide a copy of your schematic and the waveform you are observing (either as a JPG, BMP, TIF or a PDF file). This will help us to understand your problem. Are you using a Ultra Fast diode on your 48V output? Regards PI-Sarek

Yes, the 48 VDC output is using an ultra-fast diode. The 12VDC output is schottky. I am seeing the low voltage problem only on a minority of assemblies. The non-failing units all have very similar output voltages at the same loading. The failing devices also all have similar output voltage 42VDC. I need to keep the output above 47VDC.

Cross regulation errors are due to a mismatch in secondary leakage inductance, secondary resistance and load differences. When the 48 Volt output is loaded, the 12 Volt output peak detects on the pedestal. Since this is the regulated output, the 48 Volts will decrease. I am concerned that the "good" units may have the same problem. The reason that it regulates better is that the drain clamp diodes may be clipping the reflected output voltage on the primary. This will reduce efficiency and may cause D6 and D7 to overheat. To test this theory, replace D6 and D7 with 75 Volt diodes on a good unit. If the cross regulation begins to behave as a bad unit, this will tell us why the good units regulate better. To improve the cross regulation, you may want to wind the transformer to minimize leakage inductance. Also regulating off of the 48 Volt output should help the cross regulation.

I have been down this path; that is to alter the primary snubber. In fact I have removed D6, D7, D9, C82 and C111 on 'good' units and 'failing' units prior to me making my initial post on this forum with no affect. I.E., no snubber at all! The low voltage level on the failing units remains low a the same level 42VDC and the non-failing units remain at 48VDC or slightly above. I have also swapped multiple trnsformers from a working unit to a failing unit [and vice versa] without success! Any other ideas?

BTW...I have also swapped out the TOP247, all other diodes, and filter capacitors. I don't get why the problem doesn't follow the component changes. It is like the PCB itself is part of the problem. It is the same board blank for all units as well.

Just a thought - I noticed that the LM431 doesn't have a supply resistor fitted. Would you try a 1 k resistor across the opto just to rule that out.

Cheers

PI-Chekov

Thanks for the advice. I did try a resistor across the opto; no affect. However, after thinking about this and realizing that some things just did not make sense; one being why the poor regulation problem did not follow swapping components...especially the transformer, I made a discovery. The transformer was not actually swapped. My tech having difficulty removing the transformer, cut the pins, discarded it, and replaced it with a transformer from stock. So I was not really swapping the original transformer. But, this discovery did lead me to the source of the problem. We have an issue with a certain 'date code' of transformer. Prototype samples and the first lot operate as expected. The third lot, now having tested many do not operator properly in the circuit. Just as pointed out in another reply, the coupling between secondary windings is different from lot to lot! We are having the supplier investigate!

Okay...now I am really confused. I had the transformer manufacturer disassemble and test transformers from different lots and I had 20 products tested with two lots of transformers. The transformer company provided me with analysis information that indicated that the transformers of all lots were constructed the same and the measurements of devices in all lots were nearly identical. My testing revealed that the cross-regulation problem followed a particulat lot of components. So now I have a situation where I am convinced a given lot of devices is causing the problem but the device analysis doesn't prove this out. I have no reason to distrust the transformer supplier but I also have no solution to the problem. Since the transformer is the only device exchanged [20 devices tested] and no other components were changed and the problem followed the transformer the problem must be the transformer...right? But how? The leakage of the windings and the DCR of the windings are all within 10% and to top it off one lot is not consistently higher or lower than the other, i.e., a lower leakage transformer of the 'bad' lot does not work where a higher leakage transformer of the 'good' lot does work! Any suggestions?