TNY256
I designed a 10W flyback regulator using the TNY256 controller. It works very well except at temps below -30 deg.C it will not start up. I discovered that if I installed the resistor string for UVLO it will start consistently at -40 deg.C, which is the requirement. Unfortunately I cannot use the UVLO string because the unit must work from 96VAC to 288VAC, which it does without the UVLO string. How can I resolve these two incompatible issues with the TNY256?
Comments
Thanks for your reply. I have discovered in the interim that the biasing of the optoisolator diode is critical to this problem. I am measuring the differential voltage across the biasing resistor and equating that to the diode current change with temperature. It so happens that if the biasing current goes below 50u amp the IC will not start up. What the current transfer ratio is at that low temp is not at all clear from the opto data sheets(ANY of them), but I know that the IC needs more than 50u amp to start. By experimenting with the bias resistor value I've been able to achieve a start at -40 deg.C, but not after a cold soak at that temperature. The best I've been able to achieve after cold soak is at -30 deg.C.I will send the schematic or the relevant part of it as soon as I can. Thanks.
Ultimately there was no way to start the power supply at -40deg.C no matter what was done to the control loop, no matter what innovation was applied, nothing. The IC simply WILL NOT START AT -40deg.C because of the dual function of the Enable circuitry, i.e. UVLO vs. "normal" function. So it was necessary to think "outside the box" and disregard "normal" function. What works- amazingly well- is to add a high voltage capacitor of 5.5nf from positive rail to Enable, to FORCE current into the Enable pin as the capacitor charges but then to get out of the way after that. It does, however, affect the output ripple, thus the value of the capacitor selected. That keeps the output ripple down to + or - 20mv.
The internal architecture of the IC creates some major problems across a large temperature span.

I am not sure why this would be the case. Can I see your current schematic?
Also keep in mind that all other components must have the appropriate tolerance at this low temperature.
I can help more once I see the schematic.
-Wesley