AnsweredAssumed Answered

L6390 odd overcurrent problem

Question asked by tiers.jerrold on Mar 12, 2014
A new application is producing problems with very high spike


We are driving 4    STGP20H60DF using two L6390 in two half bridge

setups, for a specialty single phase motor (not a standard motor, I

can't talk about it, but inrush is not an issue).

230V input, very standard virtually app note circuit with 0.025 ohm

sense resistor, common to both half bridges.  Gate resistors in the

47 to 100 ohm area for turn on, 22 ohm turn-off.    About 10A rms,

similar to starting currents in prior designs to drive  PSC type  

single phase motors. This one won't drive those motors, despite very

similar circuit design and layout.

The IGBTs are low gate charge, and should be driveable by this chip,

although they are a very fast type.

We get very high current spikes of very short duration at switching

(80A and 100-200 ns), that we have not seen in any prior design.  The

spikes seem to actually shut down the chip, although we cannot see

any voltage changes or pulses on the overcurrent input.  In fact we

have shorted the current sensing, and still see the problem.

UVLO is not the problem, at least the 15V input is very stable with

reference to the chip common pin, and VBOOT is also stable.

Current sense has even been shorted at the pin, and no change.

The chip heats up a bit more when the problem occurs. and the problem

is current sensitive, more current = more problem.

The only things which have made any large difference are freezing the

chip, which fixes it, and also inserting a fairly large resistor in

the 15V supply to the chip, in the 100 ohm area, which substantially

reduces it.

Older date code devices seemed to be far less sensitive, not showing

the problem until much higher currents.

The problem is noticed on the lower output, which is normally on for

a half cycle (upper is PWM'd). It shuts off for several ms partway

into the half cycle, then comes back on.   WAY shorter than the /SD

timing.  And the /SD pin is not affected.  With time, the shutdown

becomes shorter, but does not go away.

Very low switching frequency... around 5 kHz, audible sound is not an

issue, and there is just no need for anything higher.

It seems that we have managed to do something undesirable, but it is

not at all obvious exactly what.   I have never seen this before,

with this or any other IC driver.

Anyone seen this problem?  Have a solution?