On Thu, Nov 26, 2015 at 9:14 PM, Neil wrote: > > > On 11/25/2015 2:13 AM, rubenjonsson@bredband.net wrote: > > If it gets cooler when you ad an external diode it could mean that you > > have a lot of ringing when you switch the transistor and the peak > > voltage > > from that may be higher than the supply voltage so the body diode > > starts > > to conduct. This diode is not very good and will produce significant > > heat > > when conducting. > > I haven't tried without an external flyback diode... with a ~30V 5W > zener, the zener got very hot and smoked. But the MOSFET didn't get > very hot. With a gen purpose 10A rectifier, the rectifier only got warm > but the MOSFET got very hot. So this points to the internal body diode > conducting...? > Instead a zenner, use a fast schottky diode connected with the anode at plus and the cathode on the load and drain. Zenner smoked because limits the overvoltage in the drain. Since there is no limiting current resistor it ends by burning. > > > The ringing may also affect the gate voltage in such a way that the > > transistor is not switched on as sharply as you have calculated > > (capacitance between drain/source and gate). > > > > To get rid of the ringing you can slow down the switch on time with a > > resistor in series with the gate. Too big resistor will mean too slow > > switch on time which instead will produce heat in RDSOn while Vgs is > > rising. Another option is to put a snubber (capacitor in series with a > > resistor) in parallel with the transistor. The R and C should be > > selected > > based on the ringing frequency. The resistor may have to be a high > > power > > type (2+ Watts and could get hot in itself) and will contribute to > > power > > losses. > I'm going to try a gate resistor. I'm reading up on it now, but not > seeing a way to calculate a fair starting value yet. > almost sure will be no difference, but you should spend your time as you wish... > > > > Ringing may be caused by inductive load and/or stray capacitances and > > inductances in the switch circuit and long wire traces with impedance > > mismatching between the driver and the transistor or between the micro > > and > > the driver. A 1 ns rise/fall time equals around 6 inches/150 mm in a > > PCB > > and if the traces are more than say 1/4 of that, they should be treated > > as > > critical and may have to be impedance matched on the receiver end. > What do you mean by "A 1 ns rise/fall time equals around 6 inches/150 mm > in a PCB" ? I've got about 1.25" of trace length between my driver > output and gate. So you're suggesting that that would result in a much > quicker rise/fall (at the gate), and hence be more prone to ringing? > > > > /Ruben > > > > -- > http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive > View/change your membership options at > http://mailman.mit.edu/mailman/listinfo/piclist > --=20 http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist .