QCW Tank Capacitor Choice

[L_Mangold]

Hello everyone!

I have a question specific to some recommended capacitor choices for the MMC on a QCW set up. Because the MMC may be powered for more than 1000 rf cycles, Arms ratings and low losses are some of the top considerations. Some example coils like those built by Eric Goodchild and Goa Guangyan utilize silver mica transmitting caps. It was to my understanding that ceramic doorknob transmitting caps also have really decent Arms ratings, but other than a few VTTC example coils I've never seen ceramic doorknob capacitors in any SSTC set up. Is there a disadvantage to those doorknob capacitors that I'm unaware of? Any input is greatly appreciated!

[davekni]

I recommend 0.33uF 1200VDC, 630VAC Chinese induction cooker capacitors.  Here's the EBay link for the specific ones I use and tested:
    https://www.ebay.com/itm/10PCS-New-BM-Capacitor-MKPH-0-33uF-630VAC-1200VDC-for-Induction-cooker-P-30-5/272271654633?ssPageName=STRK%3AMEBIDX%3AIT&_trksid=p2060353.m2749.l2649

I've purchased a couple other versions (one with a blue plastic housing, another low-profile), and they appear to be about the same (except slightly higher ESR for the low-profile version).  I've used them for Royer oscillators and for my DRSSTC.  (They're also used in the Chinese induction heating kits that have been reviewed elsewhere here.)  My abuse-testing was aimed at DRSSTC use, but I think they will work well for SSTC too.

My DRSSTC initially used 6KV 0.022uF WIMA FKP1 pulse capacitors, but they failed quite quickly at +-4kV.  Before building a new MMC, I started with an abuse test of a pair of the 1200VDC 0.33uF parts, running at +-1200V 75kHz, 1% duty cycle bursts.  After this ran fine for 5 days straight, I increased the voltage/current to +-1700V.  Ran a couple more days with little sign of degradation!  Increased again to +-2050V.  After a day there, their capacitance had dropped about 5%, which is likely getting close to catastrophic runaway failure.  That's at 170% of rated peak voltage!  The 160-cap array is working quite well in my DRSSTC, at least so far.  Barely noticeable temperature rise w/o any intentional air flow.

[Hydron]

I'm using a massive (8s*32p) array of 6.2nF 1.6kV EPCOS metalised polypropylene caps, 256 in total. Sounds expensive, but I got them for something like 35 GBP total from Farnell, who are not know for sharp pricing (they have gone up though, now more like 50GBP+tax). These caps (B32672L series) are actually rated for short term low duty-cycle over-voltage, which is perfect for a tesla coil!

All that said, I think you'll find that there are many cheap sources of suitable capacitors, as the required total capacitance/voltage specs (and thus energy) are pretty modest in a QCW. I'd look for the ability to use a few strings in parallel (you likely _don't_ need 32 like me though!) to get a decent RMS current rating, and go for metalised polypropylene film rather than foil or foil/film (you don't need foil's pulse current capability so much in a QCW, but the self-healing of metalised film is very handy if you're pushing the voltage rating). All in all the MMC should be one of the cheaper parts of a QCW coil, even if you don't take advantage of a bulk lot or similar on ebay.

The last thing to consider is time and methods for construction. Given how cheap PCBs are these days I built my MMC on 8 PCBs, with SMD HV resistors as individual bleeders across each cap. With 2 resistors per cap this did take quite a while to solder (768 total parts!), but the PCB mounted caps did save a lot of mucking around with mechanical mounting considerations etc. I'd certainly suggest looking at PCBs, especially if you end up with many capacitors in the MMC.

Edit: I'd chill out about the whole "you need a mica cap" thing - if you have sufficient total RMS current capability there is no reason a metalised polypropylene MMC shouldn't work just as well, and is much cheaper and easier to obtain.
Edit 2: Check out the following links describing one of Steve Ward's coils:
https://4hv.org/e107_plugins/forum/forum_viewtopic.php?167363 - specs
https://www.flickr.com/photos/kickermagnet/15052290263/in/photostream/ - see MMC on pipe in corner, made of commodity polypropylene capacitors
https://www.flickr.com/photos/kickermagnet/15705193211/in/photostream/ - results - if you can do that with poly caps then you don't need mica!

[Weston]

Doorknob capacitors are aesthetically pleasing and come with nice screw terminals (which I suspect are the real reasons why they are used so often), but there is no need for them in a QCW coil. A MMC of correctly chosen film capacitors is a lot cheaper, allows you to tweak the capacitance value, and is a lot easier to source.

I made a small-ish QCW coil last year and spent some time sorting through digikey for the best QCW MMC capacitors, ultimately compiling this spreadsheet comparing all the relevant capacitors I could find: https://docs.google.com/spreadsheets/d/142e_12Ojj_ahDvq699jDEijQQiF_3dN0BCUUxEYnwjM/edit?usp=sharing

I ended up using B32642B0333J capacitors for my MMC as they were the most cost effective I found https://www.digikey.com/products/en?keywords=B32642B0333J . The whole B32642B series compares pretty favorably to other lines and around 33nf per capacitor seemed to be the sweet spot of maximum reactive power at 400KHz within the voltage / current ratings for most varieties of capacitors I looked at.

The resulting MMC was 30 capacitors, which cost $18.  Its 10KV 10nF and rated for 15Arms with a 10 degree C temp rise based on the datasheet for the caps. Extrapolating from that, you get 30Arms with a 40 degree temp rise. The peak current rating based on the capacitor dv/dt limit is 300A. The 30Arms is a continuous current rating so you would have to be running at a pretty high bps to exceed that.

You can see a picture of the MMC here: https://www.flickr.com/photos/63341195@N08/47089852332/ .  I was using it at 80 Amp tank current with no issues. You would most likely have thermal issues with your switches before having thermal issues with the MMC.

Gerbers for the MMC are here if anyone wants to use it: https://github.com/westonb/cute-qcw . It fits on a 100 x 100mm two layer PCB.

[L_Mangold]

This is all excellent to hear, thank you very much everyone!

[Mads Barnkob]

Very nice list you made there Weston, thank you for sharing!

I also made a MMC similar to Steve Wards, round around a rod, if you scroll down to the following chapter, you can see pictures, here: http://kaizerpowerelectronics.dk/tesla-coils/kaizer-drsstc-iv/

6th October 2015

Construction of a very cheap MMC from capacitors that was bought from a 10$ ebay auction for 100x Kemet R474N247000A1K, rated for 900VDC 28 A peak and 4 A rms. A easy and uniform construction, with current sharing in mind, is to construct it around a round piece of wood or plastic tube.

The resulting MMC has a capacitance of 0.047 uF at 9000 VDC, 280 A peak and 40 A rms rating. Which is spot on for this coil to be running with design goal primary inductance of around 100 uH, 300 A peak, 5000 uS on-time and maximum 10 BPS.

[shrad]

I would also consider the russian surplus capacitors, either film or doorknobs, because by definition they are made to be robust

[flyglas]

Hello Weston,

thank you very much for your speadsheet. I tried to understand your choise B32642B0333J.

I ended up using B32642B0333J capacitors for my MMC as they were the most cost effective I found https://www.digikey.com/products/en?keywords=B32642B0333J

In the datasheet https://www.tdk-electronics.tdk.com/inf/20/20/ds/MMKP_B32641B_B32643B.pdf I can not find 70V rms at 400 kHz.
If I check the graph on page 15 of 36 i only can read 35V at 400 kHz (33nF, 1000V, 15mm lead spacing).
Can you please help me?

[Weston]

Good catch, it looks like there are two versions of the datasheet online. I used the one archived on digikey, which was published in 2015 https://media.digikey.com/pdf/Data%20Sheets/TDK%20PDFs/B32641B-B32642B.pdf My values are correct for the 2015 version (table on page 10), you observation is correct for the updated datasheet published in 2018.

I really wonder what went wrong there, with TDK changing the datasheet values so drastically. Either way, the capacitors seem to work well. I wonder how conservative the updated ratings are. The only reason for the change I can think of other than massive error in the original is that the updated datasheet lists the capacitors as automotive qualified, which may have required some derating to meet the more rigorous specifications. If I have time I might have to set up some resonant circuit and take a thermal camera to one of the capacitors.

Given that discrepancy, the B32672L1333K000 (sold on mouser, not digikey) is probably the next best capacitor, costing ~$30 for a similar MMC. Its the same lead spacing and only slightly wider than the B32642B0333J and has a higher maximum temperature and slightly higher RMS current / voltage rating than the values I originally had for the B32642B0333J. 

[flyglas]

Thank you Weston for your explanation. The two datasheets are indeed very different.
A test with a resonant circuit to find the usable limit would be very interesting.

I have build a 20p11s MMC out of capacitors Kemet R76TI1680SE30H. I bought them for 0,19 EUR each.
The MMC is for a small QCW coil. At the moment the project is delayed due to other hobbies.