Another effort to understand grounding

[IQChallenged]

So, I guess, in summation, if one overlooks the dynamic nature of the motor (as opposed something like a light bulb), you could reject the notion that a blown fuse could lead to a blown driver, and you'd be wrong. But if said driver happens to have the intrinsic diode then you'd be right again even if you didn't know why.

[Mariss Freimanis]

Fuses blow when something has gone very wrong. Because the fuse takes a very long time to blow (milliseconds), whatever went very wrong has already damaged the drive. There are a few happy exceptions where the fuse blows before electronics damage has occurred. In that instance, it would be a shame if "the operation was successful but the patient died".

1N4004 diodes cost a few pennies. Putting them in only increases the odds slightly damage won't occur to the drive and it's nearly a tossup for being worth it.

The most common drive (non-G203V) failure is an inadvertent external short to ground on a motor phase output. The top half-bridge MOSFET on that output 'sees' a very large current and promply fails 100uS later and fails as a dead-short itself (all silicon devices fail shorted). Next the bottom half-bridge MOSFET 'sees' a dead short for a load (dead top MOSFET), and oftentimes it dies shorted. It lives a little longer because the drive turns it off 20,000 times a second when it detects high currents. Both dead MOSFETs are in series across the power supply, shorting it. A long time later the fuse wakes up and blows.

Sometimes we 3 or 4 drives back from a client. Testing shows identical damage on the same phase output. We ask what happened and hear "The drive blew a fuse. I replaced the drive with another one and the fuse blew again". Most stop after the second blown fuse; our current record holder did in 12 drives.:-)

OK, we just had a big earthquake... Gotta go.

Mariss

[neilw20]

Mariss.

Have you ever looked at the highside like si9910 which monitor the voltage across the hi side device and turn it off if it does not saturate, or comes out of saturation due to overload. 2uS is quick enough to save hardware, but does not fault too often.
I slow the hi-side gate drive down till it faults, then back off a bit.
The slow turn on allows adequate reverse recovery time when low side flywheel diode(s) happens to be conducting, getting rid of many shoot through type faults, and removing excessive filtering from current sense circuits.
With 4 quadrant regenerative stuff most of the hazards go away.

DO need any work done??
Neil.

[tauntdesigns]

Originally Posted by Mariss Freimanis OK, we just had a big earthquake... Gotta go. Mariss

5.8

I hope everybody is OK!

No baby Geckos got hurt, did they?

Cheers,
Jack

[IQChallenged]

Originally Posted by neilw20 Mariss. Have you ever looked at the highside like si9910 which monitor the voltage across the hi side device and turn it off if it does not saturate, or comes out of saturation due to overload. 2uS is quick enough to save hardware, but does not fault too often. I slow the hi-side gate drive down till it faults, then back off a bit. The slow turn on allows adequate reverse recovery time when low side flywheel diode(s) happens to be conducting, getting rid of many shoot through type faults, and removing excessive filtering from current sense circuits. With 4 quadrant regenerative stuff most of the hazards go away.

Every time I do that the spikes from the residual capacitor action mitigate the amplitude negating the electron flow. So I usually just tweak the oscillation a few farads and let it go.

[Al_The_Man]

Originally Posted by IQChallenged So I usually just tweak the oscillation a few farads and let it go.

This should ONLY be attempted after a very stiff scotch and soda