bidirectional PWM controller

but

and

The most obvious problem is that this circuit does not have protection, wheeling or ring-off diodes to keep the inductive kick from the motor from damaging the FETs when switched off. Any inductor, including a motor, when switched off will try to maintain the flow of current into the off circuit. This creates a high voltage spike that is the probable reason for FET failure.

The FETs have an internal body diode that is totally unsuited for this function. They are too slow and retain too much charge.

Place four fast diodes, reversed biased, from M1 and M2 to the plus voltage rail and to ground. This should protect the FETs from inductive spikes.

Secondly, insure that at no time are the FETs of either side of the bridge EVER on together. This situation creates a dead short from the power rail to ground and will blow FETs.

Driving the FETs are LM324 op-amps used as comparators. These are lousy amps and a worse comparators. They may not turn off fast enough to insure that no on time overlap occurs. It's hard to find this kind of stuff without good test equipment like a fast storage scope.

Try the diode fix first, then dig deeper if necessary.

I allready did place shotky on every mosfet, still burning, also I did try with some other gate resistors and so....problem is in dead time then....but this design end then.

Does anybody have something similar to this?

That's a stunningly bad circuit, but then most circuits are stunningly bad.

These work nicely:

but I'd add some clamp schottkies for luck.

John

Um, this crcuit PWMs at 270Hz.

As long as there's no actual shoot-through or gate overstress failure, and as long as the 324 can actually shut the Pchannel parts off, theres no reason why the circuit shouldn't work below 20V. It obviously does, to a certain extent.

Some kind of ripple current, though.

If all you're doing is introducing better drivers, you could probably retain the electronics to generate the control signals, providing that they're troubleshot before applying juice.

RL

It sure could use some ceramic low esr caps across the mosfet bridge. The mosfets have some antiparellel diodes in them already albeit mediocre ones. when the freewheeling currents attempt to go back into the bus you're getting a spike on the bus. A couple of caps with low ESR and ESL (think x7r ceramic) with a couple hundred uF electrlytic in parallel as a bulk capacitor and you might be in business. These go diretly across the mosfets (drain of the high side to the source of the low side.

That said hte circuit design is poor at best. The 10K gate resistors will be SLOW turnon and turn off causing heat and shootthrouh.

I'd scrap that and find a Brushed motor driver PWN IC (TI, onsemi, fairchild) that had real mosfet drivers in it. you'll be much better off.

You mean cap to paralel on each MOSFET?

I try to put 70r also but is same.....

I tought, but I need it for about 10-15A here, if someone know some IC which can go inreverse maybee?

Don't top post or you'll get chastised by this group. :) (friendly warning)

I would put say a .1uF X7R ceramic caps directly from pin S on Q3 to Pin S on Q6 and two more from pin S on Q5 to pin S on Q4. Then add a large (several of the C2 (100uF) caps from V+ and to Q4,6 S pin.

The reason whoy is the freewheeling currents have to go somewhere.

Imaging an inductor between M1 an M2 and Q3 and Q4 are on. Current flows V+ through Q3, from M1 to M2 then through Q6 to ground. When Q3 and Q6 turnoff say to stop the motor. the inductive currents in the motor will not stop flowing instantaneously. they will continue to flow by M1 reducing its voltage to a slightly negative value and M2 increasing its voltage to a posive value. When M1 is 1 diode drop below ground the antiparellel diode built into Q4 will turn on. Same thing happens when the M2 voltage goes above V+, the antiparellel diode turn on and current now flows from ground, throgh Q4 (diode) then M1 to M2 then Q5 (diode) to V+. Now the indictance is pumping current into v+. if that current does not have a low impedance path back to ground, the V+ voltage will rise. It it spikes higher than ~30V, you blow the Q3 and Q5 gate insulation (rated for

That said this isa poor nonprotected design nad definately not a robust way to do what your are attepmting to do.

Are you sure the fets are not overheating? do you have them attached ot a large heatsink (of they are to-220 they need isolation from the sink and mounting screw)

Some problems I see are:

1. There should be a large low ESR capacitor from V+ to GND to handle the initial surge when the MOSFETs turn on.

1. There should be a base to emitter resistor (about 10k) on Q1 and Q2 to make sure it turns off when the output of the LM324 is low (unless it really drives within a couple hundred mV to negative rail).

2. A better quad op-amp with rail-to-rail output would be better.

You might want to build a microcontroller based PWM circuit like the one shown here:

more tools and information:

I have a simple motor control kit from Microchip, called a PICtail Daughter Board, that is very inexpensive and versatile.

Paul

that pig again.

there was a detailed discussion about that circuit a year or two ago. None of the skilled SED regulars had anything nice to say about it.

Dont bother repairing it, it will never work properly.

Cheers Terry