openinverter forum

I scoped all six gate driver PWM signals a few weeks ago. The signals didnt look like your YouTube instructions, so I tested again recently. I was probing around the gate driver board and suddenly the signal started looking strange. It is possible this happened when I was measuring with two probes on different oscilloscope channels and was flipping switches and eexploring/learning how to use the oscilloscope. The gate board measured 4.6V across the 5V Vcc/gnd input pins while the other two gate driver boards measure 5V. The next day, the main board was pulling 320mV and that faulty gate driver board measured 4.9V across the Vcc/Gnd pins but the gate driver input PWM signals show up on the scope intermittently for all three gate driver boards. Every few seconds, the PWM signal will appear on the scope for less than a second (maybe < 1/10 sec) then disappears. The signal looks recognizable. I may have shorted something on the gate driver board but wasnt probing the main board. I had test probe ground attached to the GND pin and was probing the ETop Ebot Gtop Gbot and the input PWM pins. Is it possible one of the web interface parameters is interfering with a clean duty cycle pwm? Did I damage something with two probes grounded to GND pin and probing PWM input and PWM output pins simultaneously? Where to start?

After 9hrs of rest, i turned inverter on and picked up a duty cycle waveform that held, but after a few minutes, it started to disappear intermittently like this morning. Eventually i couldnt see a pwm signal. When it did appear, the pwm looked the same as the first round of successful testing...the duty seems to constantly fluxuate. It appears that the duty slowly fluxuates from 25-75% pauses and then back down to 25% which takes roughly 1 sec to cycle up and down.

Ran the oscilloscope for over 10min and signal eventually dropped out. Will try running 50/50 again later to see if it might be an issue with the old scope.

Hm... How did you test this? Can you command 50% duty?
You start brain in manual mode like here... https://openinverter.org/wiki/Schematic ... te_drivers

Can you scope signal lines from the main board? If there signals are good then you have to check drivers for fault in solder jount.
Alternatively you may want to put LED with 300R resistor or similar load on each driver output and command them 1Hz. Those LEDs should be lighting up and you can see where the fault should be.

Well since before i started to play with protected drivers there could be several issues....
1. Broken driver chip... try measureing the DCDC converters You should get +15V/-15V or +15V/-8V on theoutputs depends which ones you use. If there is anything different you have an issue. What i did, i replaced Recoms with Murata MGJ2D051509SC which is capable of surviving short indefinitely. So if a driver died on me, it didnt took DCDC (hard to purchase) and hopefully IGBT (expensive) with it.

2. Broken or wrong value TVS diode... This diode grounds the output in case output voltage is too high. If TVS would be too low it would start to conduct at 15V instead of 17V and you would loose signal power everytime.

3. Incorrect or broken capacitors... again dielectric short draws some power from functioning circuit.

Finally have a good looking pwm signal, clean and steady. After uploading all .bin files and restoring default parameters, I unexpectedly got a good 50% duty pwm. I suspect the problem was caused by me initially configuring my voltage (udcnom, etc) then disconnecting the voltage source during pwm testing. With no voltage source, the pwm signal looks normal when I set udcnom < 14. When udcnom is set to 14 or higher, the pwm signal fluxuates. On to the next problem, bad GBOT signal on one of the gate driver boards...

Garagehermit wrote: ↑Sat Oct 19, 2019 5:16 pm Finally have a good looking pwm signal, clean and steady. After uploading all .bin files and restoring default parameters, I unexpectedly got a good 50% duty pwm. I suspect the problem was caused by me initially configuring my voltage (udcnom, etc) then disconnecting the voltage source during pwm testing. With no voltage source, the pwm signal looks normal when I set udcnom < 14. When udcnom is set to 14 or higher, the pwm signal fluxuates. On to the next problem, bad GBOT signal on one of the gate driver boards...

DOH! Like i said check TVS diodes first. If diode is bad it is constantly pulling signal to gnd.
Then see if you use 10K resistor between G and E. If it is less it can interfere with signal.
Next observe driver boards power drain. Use a good bench PSU and just supply driver board with 5V and observe what happens when you send 5V to each drive line. Of course use 1K resistor on the signal line just because...
If there is significant draw difference among boards i would measure voltage n DCDC and its caps. If it is strange i would remove driver chips and observe DCDC voltage. Which DCDCs you use? Mine were Murata MGJ2D051509SC and they can survive short indefinitely. If you use Recom DCDC and your driver is shorted DCDC is probably damaged.

This is why i recommend people to use protected driver boards...