> >Could you have a saturating core? > > are you saying that the humming is a symptom of a > saturating core? Could be, I'd want to slap a current probe on the primary winding, and see that the current ramp is linear. If not, you're probably saturating. > yes. actually, i calculated #primary turns based on maximum > flux. according to mfg. data sheet, for my desired power >dissipation and > operating frequency, i can be running at over 300mT in a bridge. > since i am using a forward converter, my flux is in only one >direction, > and i should be safely able to double that flux density. but i >didn't. > so i should be safe. 300mT is a lot of flux! I think you may be in trouble here. Especially about doubling the flux. The core limit is regardless of polarity, so X gauss in either direction will saturate you. With AC, you can drive from -X to +X, but with DC you can only drive to +X or -X, not 2X. There is a trick that will let you get to 2X, but it's evil, and hard to use in a pot-core or toroid. > >a CS3844 or some such would > >be a lot more predictable, plus operating nicely at 300Khz. It also > >monitors the core current, and will not let you saturate the core, as > >long as your sense resistor is picked properly. > > now that could make life easier. who makes this device? do you have > any preferred devices? Cherry, Ericsson, and Unitrode all make these, 3842,43,44,45 The differences are what startup voltage, and wether or not they have a max duty cycle limit. They also have a nice beefy output drive for turning on a FET, IIRC it's about an amp in both directions. Plan on adding a resistive divider in the gate drive, this will let you control the gate ramp. In my applications, it turned the FET on and off so fast that I was getting a fair bit of EMI, (from the rapid dV/dT) Slowing it down slightly made the fet a little warmer, but much quieter I prefer the cherry parts, the internal zener is a bit more robust, but they all work well.