I nearly crapped myself - CNCdrive.com braking module circuit diagram

Power circuit isolation:

The power circuit in my system is supposed to be isolated on account of the power transformer (which was originally a site / safety isolation transformer), the servo drives (which use opto isolation between the logic circuits and the power circuits) and the various relays etc that span the barrier between the mains, power circuit and logic / PC section. But it's probably not helpful to have the power circuit truly floating in my system. Industrial drives often do this so that they can withstand a single short circuit to ground somewhere on the high voltage circuit - this allows them to power down in a controlled fashion. 

So I fitted a link between the chassis and the -ve rail of the DC bus, right next to the bridge rectumfrier. Bad mistake. Although I'd done a very basic insulation check between the DC bus and the grounded baseplate, I don't have a proper insultation tester like a 500V Megger. 

Oops:

Anyway, it seems the braking module wasn't capable of withstanding the 140VDC that happens between the DC bus +ve and its heatsink / baseplate. The entire DC bus discharged itself through something in the braking module, blowing a track off the board. Simultaneously I almost discharged myself into my underwear.

WTF??? Even with an unexpected connection somewhere in the baseplate wiring, there was no obvious means to generate a genuine overvoltage on the bus. The braking module is only supposed to start clamping at around 180VDC. So it sounded more like a simple insulation / airgap breakdown.

Time to investigate....

Sure enough, the breakdown was across the mounting screw of a TO-220 device. The tab is connected to DC bus +ve and the screw is grounded. So, by grounding the DC bus -ve, I put the 140VDC between the screw and the tab. There appears to have been a top hat insultor bush there but most of it had gone west in the thermal event. Interestingly though, whoever assembled the module in the factory(?) had dropped a top hat bush and it had become stuck in the elephant spunk (thermal compound). That was quite helpful, as it clearly illustrated part of the issue. 

The TO-220 device is partly insulated by means of a conventional Silpad-type film insultor. However, if the hole in the device and the hole in the baseplate are the same diameter, there will typically be a near-zero clearance distance between them - nominally the same distance as the thickness of the Silpad. This is an old chestnut, aka the sort of mistake a complete novice would make. The correct approach is to either space the device off from the heatsink by means of a thick aluminium oxide insulator or to counterbore the hole in the baseplate to give a controllable clearance distance between the heatsink and the nearest point on the tab. 

Fix:

Yawn. Found a top hat bush that actually went right through the tab and came out the other side. Then drilled out the hole to 5.5mm. Fitted a 2.5mm screw (was 3mm) and finally a layer of Kapton tape to give a finite clearance between the external edge of the tab and the heatsink - another area where there was fuck all other than the thickness of the Silpad.


Of course, the cloud of plasma that resulted from all of that energy vaporising the track resulted in damage to the control circuit too, so now I have a fucked controller to look into.

The braking module control circuit:

Took a bit of messing about, Googling and desoldering a few of the parts. But the circuit is actually very simple - and rather crude. It wouldn't be seen in any sort of industrial product and certainly wouldn't get past any circuit review involving me. However, it is what it is and hopefully it will work adequately once it's mended.

Key parts:

• MJE13007 NPN (On Semi) used as a linear regulator (buggered)
• UA7812ACP (TI)
• LM393 dual comparator
• MCP1416 (Microchip) gate driver / buffer device (I suspect it wasn't damaged)
• IRFP260 (main switching FET)

Comments:

• This is a simple comparator. If the input is above the threshold, the resistors are switched in. 
• There is no hysteresis used on the comparator.
• The second half of the dual comparator is simply used to drive the LED. 
• The 7812 is used as the voltage reference
• The threshold works out at a nominal 169VDC. I haven't bothered to estimate a tolerance on that.

The gate drive circuit looks pretty basic. Not convinced many of the components around the FET gate actually achieve anything.

I'll need to order up most of these parts......