> A friend of mine has asked me, in desperation, if the list > has any comments about his circuit. > > http://home.clear.net.nz/pages/joecolquitt/12vmotor.html Before commenting on the circuit I want to make it clear that I haven't chased down any of the part numbers so I will have to assume that max current ratings and the like have been checked. I will also assume that the 555 produces a PWM signal that is nearly rail to rail at pin 3 and will otherwise not look into that part of the circuit. That being said, parts of this circuit look rather confused to me. 1 - My gut reaction is that the snubber circuit should be more aggresive, but I haven't done calculations to verify that. 2 - I don't like the shottkey diode accross the motor. You do need a diode there to short out the flyback current, but unfortunately it needs to be both beafy (30A) and very fast recovery. Perhaps your BYV143-40 (-50?) can take the 30A, but what about its reverse recovery time? Unless you've already looked into this carefully, my top suspect is that the flyback diode is too slow and the FETs are taking a serious beating every time they turn on. This will either kill the FETs directly, or fry the diode so that it no longer clips the flyback pulses, which will kill the FETs by high voltage spikes. 3 - The diodes between the FET drains and gates don't belong there. They are probably contributing to the problem. I'm guessing that you are trying to use the FETs to clamp their own excessive drain voltages by turning them on. In my opinion the circuit has already failed if you get more than 28V on the drains, which I assume is the max voltage rating of the FETs. Trying to turn the FETs on partially at this case seems pointless and even harmful. 4 - I don't know what your PWM frequency is and what the apparent gate capacitances are, but I bet the FETs are spending way too much time in the transition region. The whole gate drive circuit looks pretty flaky. It can easily take 1A or more for a short time to properly switch the gate of a power FET. The 555 certainly isn't up for that, and the extra resistors will make the problem even worse. The slow switching time is my second suspect for killing the FETs. So here is my recommendation: Ditch everything from the 555 pin 3 on, except the FETs and the motor and start over by doing some real calculations. You could either use commercial FET driver chips, or a double emitter follower to provide the necessary current gain for the pin 3 signal. The double emitter follower consists of an NPN and a PNP transistor. The two bases are tied together, as are the two emitters. The collector of the NPN goes to the + supply, and the collector of the PNP to ground. Pin 3 drives the bases, and the emitters drive the FET gates. You will loose about 700mV of voltage swing on each end, but that should be fine here. In return, you get 100 times or so more gate drive current than the 555 can provide directly. This should switch the FETs much faster than they are being switched now, keeping them out of the deadly linear region as much as possible. There should be a pulldown resistor on pin 3, and another one on the FET gates. Don't put anything between gate and drain. Look into the snubber circuit. It probably needs to be heftier than what you showed. Now for the flyback diode. This part needs to be selected carefully. You have to assume that the motor looks like a pure inductor in the worst case. Actually this is pretty close to reality when the motor is stalled. If the max forward current is 30A, then that is also the max flyback current. Unless you can find (and pay for) a very fast recovery diode that can do 30A, you need to apply a little cleverness. As I said before, this is my prime suspect as to why the FETs are frying. I would put a resistor in series with the diode. Since the max flyback current is 30A, a 500mOhm resistor will develop 15V accross it worst case, which should still keep the drain voltage below 30V (which I assume is the FET rating). You still should get a fast recovery diode, but at least it won't look like a dead short when the FETs first turn on. 500mOhms is therefore the largest tolerable diode resistor, but there are good reasons form making this lower too. Depending on the PWM frequency, this could dissipate significant power. This could not only be cumbersome to cool, but would also negate the high efficiency of PWM drive in the first place. Therefore, you could use a lower resistor or multiple resistor-diode combinations in parallel. I like the latter better because it allows the use of lower current and therefore faster recovery diodes and spreads the heat dissipation. The series resistors will guarantee that the diodes will share the current reasonably evenly. Maybe four sets, each using 10A diodes. You'll have to work out the resistor tradeoffs yourself, but I'm guessing around a few 100 mOhms each. These may end up being properly measured lengths of wire of know thickness. ******************************************************************** Olin Lathrop, embedded systems consultant in Littleton Massachusetts (978) 742-9014, olin@embedinc.com, http://www.embedinc.com -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.