Bleed R's for MMC caps ?

[pentode]

Looking at many MMC constructions on the net some have bleed R's across each cap but most
have none.

So, to bleed or not to bleed, that is the question?
 ..............Across caps rated at a few kV each!


 

[Mads Barnkob]

I seem to recall that it is a legacy from SGTC to balance out the MMC if the capacitance of the capacitors is not that great, something that has improved a lot with modern film power capacitors.

As to bleed off excess energy, for large DRSSTCs using a UD2.1 or higher version, a inverter bleeder resistor across the inverter output helps against instability issues due to energy residue in the MMC between pulses, so that kinds of removes the need for single capacitor bleeder resistors and the power rating / voltage rating is also much lower for a 10K power resistor across the inverter, compared to the usually used 10M metal film resistors on the single caps.

Any others have a comment on this?

[Hydron]

Individual bleed resistors would eliminate the danger of getting a zap off a MMC cap due to uneven stored charge, whereas across the bridge gives less guarantees. Not sure how much of an issue this is in reality though.

I would definitely recommend a resistor across the bridge as Mads said, especially if you are using a UD2.x style driver. Only needs to withstand the bus voltage at your max duty cycle so it's easy and cheap to add.

[pentode]

Thanks both for that.
Incidentally Max,on your excellent article on MMC
http://kaizerpowerelectronics.dk/tesla-coils/drsstc-design-guide/mmc-tank-capacitor/
you show photos of MMC caps both with and without individual bleeders.
I daresay many are photos going back some time though.

[Hydron]

Thinking about it more, I personally would add the bleed resistors if only for safety when working on the MMC/coil. It is unfortunately quite annoying to solder them all (I once put together a MMC with 512 bleed resistors - two in series for each cap!).

You will need to try and find resistors with a suitable "overload" voltage or put two in series - their normal rated voltage for steady state is often significantly lower than the MMC capacitor rating but we are not running them in steady state operation.

If you have PCB design experience then you could maybe get something made by a cheap chinese PCB supplier to hold the caps and bleed resistors - they are extremely cheap, especially if you only have 2 layers and 1oz copper, and you can always use both layers (and even stack a couple of identical boards) to get the amount of copper up. See https://github.com/Hydron-4hv/QCW_MMC_PCB/raw/master/Reference%20Outputs/Schematic%20%26%20BoM/QCW_MMC_PCBA-(NO_WE_PRESSFIT).PDF for one I did (I built up 8 of these boards!).

Finally I would not worry so much about the current sharing bit in Mad's guide - this is important for sharing between low impedance parts such as bus capacitors or semiconductors, but the impedance of the MMC capacitors is high enough that they should share pretty much perfectly regardless of connection point unless you do something wrong. Do however space the caps out a bit, and try not to have the resistors touching the caps too.

[pentode]

512 resistors. Wow!
Unlike your huge array I only have 18  poly 942  caps to fit so it's easy to put them in
my small 6 x 3 array.

Does your M10.2 R's take a long time to discharge caps?
The bleed current will be very low, at 1000 v being only
some 100 micro amps !
I would wait some time before getting hold of them

Thanks for suggestions though.

[Hydron]

Actually it's very quick to discharge - the caps are only 6.2nF so the time constant is only 63ms - enough to render the array safe in well under a second, and even enough to perform the job of the inverter bleed resistor too given the low pulse rep rate of the QCW coil this MMC has been built for. From a safety perspective a 10M resistor should be good enough for even capacitors into the low-uF range if you give them a minute or two to discharge.

For 18 caps I would 100% put them in, even if you have to put two in series 36 resistors will not take long to solder up.

[pentode]

Ooops sorry, I read your nano caps as Micro farads !

Whilst on bleeders I have a 6,800 microfarad filter cap  working at 360 volts
to bleed.
Using  8k still leaves 40v after 2 minutes !

[Hydron]

Yes I know about that well - I have 4x6800uF for my qcw coil and an even larger bleed resistor - takes a long time to drop to a safe level. I do however have a high efficiency led in series with the bleeder R to tell me when there is voltage present - stays obviously lit until well after its safe. May also add a neon indicator to be doubly sure.

[pentode]

 FOUR 6800's. Gosh

That's a nice idea. I may put one on my 6800.
Better safe than sorry.
But putting something in series with bleeders puts
  a lot of faith in their continuity, don't you think? . 

[Hydron]

yeah that did occur to me - hence the planned addition of a neon indicator (which should run on DC OK) as a separate safety feature.

So far I have only had it running up to ~100V in testing, though I did manage to short it out at ~50V or so with a test probe (which promptly disappeared into vapour!).

I'd rather not run so much capacitance, but for a QCW I need to have enough to avoid the bus voltage drooping too much over a single pulse and a thumping great capacitor bank was the easiest (and cheapest, given I got them free) way to do it.