0
\$\begingroup\$

enter image description here I have designed around the LM5106MM/NOPB gate driver. Used two IPB017N10N5ATMA1 N-CHANNEL MOSFETS. This circuit has actually been used in another design in production by a engineering group beyond my time. Works great as an LED driver and is in production. Now mimicking the same circuit for another product that will drive both LEDs and resistive heaters.

I did make some alterations from the original circuit, such as adding in D7, D9, R40, R43, R50, R53 and using a higher current MOSFET and inductor (part numbers previously mentioned). Other circuit also used +10V whereas I am using +9V (can source 500mA). Sounded like I was good to go!

Now the board is in and I am testing things out, and with very different results than what I was expecting. When I place +3.3V on the IN pin of the gate driver, with no load attached, there is self-oscillation. Measuring the HO pin, I see a spike of about 40V and then quickly goes to ~22V and then exponentially decays to nearly zero, and repeats all over again.

One thing I should have realized is that placing R43 was a particularly a bad idea. Through muscle memory (and shear clumsy-ness), I placed that resistor there to bleed off the gate charge. But the driver does this and instead I made a bad feedback path. No biggie, I have 3 other channels and tested it with that resistor removed.

Similar behavior. Tried to remove both capacitors as maybe the delayed turn on set by R47 perhaps wasn't long enough to fully discharge C31 and C32. Nope, still giving similar-ish effects. Even got desperate and tried to short the inductor out. Circuit gremlins still in control. I also tried to remove C32 and place the + on +36V_CH1 and - on C31's +. Stopped the spiking but still oscillating.

I am now back to searching the internet on those who have taken the same path but I couldn't find an issue similar enough. Has anyone else had an issue like this?

\$\endgroup\$

1 Answer 1

4
\$\begingroup\$

The upper driver inside the chip that drives the top FET gate is powered by the bootstrap cap C29. Since this only stores enough energy to power the driver for a short amount of time, this kind of driver can't keep the top FET ON continuously, it has to pulse the bottom FET ON to recharge the bootstrap cap.

When trying to keep the top FET ON continuously by holding pin 1 to logic 1, what you're seeing is expected. First the top FET will turn ON, then after a delay the bootstrap cap will run out of charge and the top FET will turn off. On some drivers it turns off slowly, on others it turns off abruptly.

Measuring the HO pin, I see a spike of about 40V and then quickly goes to ~22V and then exponentially decays to nearly zero, and repeats all over again.

When output voltage decays to zero, the bootstrap cap recharges via the diode, so it is again able to turn on the top FET, and the cycle starts again.

R43 would discharge the bootstrap cap quicker, so it should be removed.

\$\endgroup\$
6
  • \$\begingroup\$ So this oscillating behavior might actually be normal after all. A higher valued bootstrap capacitor would yield better results in keeping the output on longer? The oscillating behavior isn't useful for providing initial current to a heater on startup. Though the circuit still does not seem to behave correctly as the initial waveforms seemed very noisy and erratic. I am going to try a fresh board and re-measure somethings to see if I keep seeing this oscillating. It might be worth looking at the older production circuit as well to see if does he same thing with a continuous high on the IN pin. \$\endgroup\$
    – joe
    Dec 15, 2020 at 20:08
  • \$\begingroup\$ You can try a larger value bootstrap cap. If the voltage decay and oscillation frequency are in inverse proportion to bootstrap cap value then the problem is bootstrap cap discharging. It could be something else, but this way you can test the hypothesis. You could show your layout, maybe there's a problem there... \$\endgroup\$
    – bobflux
    Dec 15, 2020 at 20:21
  • \$\begingroup\$ I just stumbled across this lecture & it has been helping me understand the mechanism a lot better. youtube.com/watch?v=q43QI3peWKM It also confirms what you are saying about the oscillations and how the bootstrap cap is working. Initially, a uC has been placing a high the entire time. Sounds like I need to have it keep switching in order to have any useful effect. It sounds like a higher gate resistor would ease the turn on time easing switching transients. Your input & the video will help me further investigate After this, I can show a layout if nothing seems to improve. Thanks! \$\endgroup\$
    – joe
    Dec 15, 2020 at 20:38
  • \$\begingroup\$ Looks good. btw, it you use it to PWM a heater, why use a full buck circuit with inductor and output caps? You could use just a low side FET... or, the bottom FET can be turned on continuously, so you can put the heater between +36V and the output instead of between the output and GND... \$\endgroup\$
    – bobflux
    Dec 15, 2020 at 22:29
  • \$\begingroup\$ Tried injecting my own square wave, and found the 11kHz was enough to be smooth enough for my application. Frequency range was 2kHz to 22kHz, other than that, it goes crazy again and messed up the bootstrap cap. Thank you so much for your helpful hint. So I was tasked to make this circuit very general for many devices. It may be a peltier, heater (resistive or inductive), LEDs, etc. So I figured it was just less risk to place the LC filter on the output of a MOSFET totem pole. \$\endgroup\$
    – joe
    Dec 16, 2020 at 15:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.