The 7.5V dip is measured at the power supply itself. This causes the switching to stop for a brief moment, which trips the power good circuit on the secondary side. This resets the motherboard. I could fudge the power good circuit, but ideally it would be able to power through the dip.Originally posted by Probedude_2000:
7.5V for microseconds? I understand that there are instantaneous high peak current loads but I can't see that your power supply would need to work down to 7.5V when it only lasts for microseconds.
Have you monitored the output during this brief dip and seen that it is affected? Seems that if it was you could stiffen the caps on the output to carry you through the dip.
Lastly at 125Khz and a 60W load, what is your peak primary current? Seems like it would be way more than 30A. Have you monitored the current sense resistor and verified that you are not saturating the core under load? (esp at 7.5V)
How hot is your core getting? (even when idling without air flow for say 10mins) The biggest problem I had with a similar design was finding a core that could handle the peak current ratings without saturating or being so inefficient that it would get scorching hot without much of a load connected.</STRONG>
I'm not sure of the peak current, I don't have any current mode control on it, hence no sense resistor. I haven't blown any transistors yet. Oh and by the way, there are two primary windings switched simultaneously, sharing the load between the two, as well as two transistors. They need a fan on them, but with a fan at ~80W they are about 120deg F. (Finger test, FWIW)
The core does not get perceptibly warm to the touch, even under high load. If the power supply is 'on' but with no load I think the idle current is around 150mA, not bad at all, although I haven't actually measured it, just read it off my test supply with an analog meter.
I've noticed some weird oscillation in the MHz at some duty cycles, I think it is my catch diode on the primary side, I'm going to try another and see if I can get rid of it.
Was the design you tried a Forward Converter or a Flyback? I could see core saturation with a Flyback especially if the core is not gapped.
I attempted a Flyback Converter a few months ago, and procedeed to blow most of the components at one point or another. Hence the Forward Converter topology I have now!