DRSSTC MMC Sanity Check
thedatastream, Tue May 22 2007, 10:03AMHi folks,
I'm designing the MMC for my DRSSTC and would appreciate a sanity check on my design. Here's an overview of the relevant design information for the coil:
Number of turns = 5.3 // tap point
Primary Inductance (uH) = 8.23
Primary Resonant Capacitor (nF) = 352.96 // resonant with primary at same frequency as secondary
Frequency check (kHz) = 93.38
Surge impedance (ohms) = 4.83 // a little low?
Guesstimate peak current (amps) = 300 // reasonable for TO-247
Estimated capacitor voltage (kV) = 1.45
On Time (us) = 350.00 // worst case on time
PRF (Hz) = 100.00
Duty Cycle = 0.035
MMC RMS Current (A) = 46.96 // worst case RMS current
The capacitors I'm using are Arcotronics A72-KP series caps, 1000V 470nF. Oddly, this value isn't listed on the datasheet --
-- but based on this info I'm working on a max dV/dt of 2000 V/usQuestions
1) For the pulse current, I'm assuming that I = C * dV/dt holds true. However, for the RMS current there is no rating on the capacitors. Is there a way of guesstimating / calculating the RMS current rating for these caps? Does anyone have a similar cap that has an RMS rating?
2) I have two options for my MMC (given the number of caps I have at the moment = 12)
I could to go with the first configuration because the lower capacitances mean I can use larger primary inductances (more turns) which will be handy for a first run up and test. However, as I increase the capacitance to increase turns ratio (and therefore output voltage) I'm concerned it may result in too much current through the caps.
I can re-configure my cap bank later, but I'd rather settle on a single value of capacitor bank. With this in mind, I'm probably going for 3 strings of 4 caps in series to get 352nF @ 4kVdc.
Is this voltage rating enough? My calcs indicate it would only get up to 1.45kV which should be sufficient given that you should derate the voltage when operating at high frequencies.
3) Does all this sound reasonable? Does anyone have any suggestions on a correct course of action?
Thanks in advance,
James
Re: DRSSTC MMC Sanity Check
Steve Ward, Tue May 22 2007, 05:10PM
Hi James,
Im usually a big supporter of using really big caps, but i think this weekend i found out (through experiment) that a cap thats too big simple isnt "current effective". What i mean by that is, you *may* get longer sparks, but the increase in primary current to get even modest results will be higher than before, and then to exceed this spark length requires much higher currents resulting in dead IGBTs
.
In any case, if you are aiming for only 300Apk, you will certainly want a higher Zsurge. My coils usually are around 6 ohms. Worst case for your current setup (assuming 400VDC, full bridge), you would hit 83A the first half cycle, then 248A on the next half cycle (that is, after only 1 cycle of operation!). The next half cycle would peak to ~414A. Of course, this is assuming no secondary in place, but in general this trend tends to hold. BTW, that trend is that the first half cycle is Ipk(1)= VDC/Zsurge, then that Ipk(n) = Ipk(n-1)+2*Ipk(1) -- it simply addes Ipk(1)*2 on every half cycle. Again, this assumes a first order LC (no R, and no secondary) driven ZCS from a voltage source. I find that with secondary loading, it will typically follow this trend for about 3 cycles and then it begins to plateau or drop off from secondary loading. This only applies to the way i tune my coils, though... so as usual, YMMV.
BTW, id assume your caps can do about 12A RMS by comparison to the CDE version of your cap (942C series). You might only really need 2 strings of caps, particularly if you are aiming for only 300Apk currents (lower peak usually implies lower RMS). If i had to guess at what size tank cap you should have, id say something around 100nF actually. The closest you could do is 2 strings of 6 for 157nF. By comparison to my DRSSTC3, which used a 100nF cap and a half-bridge at 350V, it ran 400Apk to produce 3 foot sparks. Modifying the tuning made it able to lower the peak currents but also shortened the sparks. Anywho, my coil used only 2 parallel strings of caps (rated 13ARMS each), and they did ok for rather long runs.
Steve Ward, Tue May 22 2007, 05:10PM
Hi James,
Im usually a big supporter of using really big caps, but i think this weekend i found out (through experiment) that a cap thats too big simple isnt "current effective". What i mean by that is, you *may* get longer sparks, but the increase in primary current to get even modest results will be higher than before, and then to exceed this spark length requires much higher currents resulting in dead IGBTs
.In any case, if you are aiming for only 300Apk, you will certainly want a higher Zsurge. My coils usually are around 6 ohms. Worst case for your current setup (assuming 400VDC, full bridge), you would hit 83A the first half cycle, then 248A on the next half cycle (that is, after only 1 cycle of operation!). The next half cycle would peak to ~414A. Of course, this is assuming no secondary in place, but in general this trend tends to hold. BTW, that trend is that the first half cycle is Ipk(1)= VDC/Zsurge, then that Ipk(n) = Ipk(n-1)+2*Ipk(1) -- it simply addes Ipk(1)*2 on every half cycle. Again, this assumes a first order LC (no R, and no secondary) driven ZCS from a voltage source. I find that with secondary loading, it will typically follow this trend for about 3 cycles and then it begins to plateau or drop off from secondary loading. This only applies to the way i tune my coils, though... so as usual, YMMV.
BTW, id assume your caps can do about 12A RMS by comparison to the CDE version of your cap (942C series). You might only really need 2 strings of caps, particularly if you are aiming for only 300Apk currents (lower peak usually implies lower RMS). If i had to guess at what size tank cap you should have, id say something around 100nF actually. The closest you could do is 2 strings of 6 for 157nF. By comparison to my DRSSTC3, which used a 100nF cap and a half-bridge at 350V, it ran 400Apk to produce 3 foot sparks. Modifying the tuning made it able to lower the peak currents but also shortened the sparks. Anywho, my coil used only 2 parallel strings of caps (rated 13ARMS each), and they did ok for rather long runs.
Re: DRSSTC MMC Sanity Check
thedatastream, Wed May 23 2007, 07:30AM
Hi Steve, thanks for all the info.
Interesting what you say about Zsurge and the rate of current rise, it wasn't something I'd considered. Sounds like I need to re-do my sums for a surge impedance of about 8-12 ohms or so - a larger diameter primary may well be in order.
Also, based on what you are saying, I think I will re-design my cap bank for two parallel strings and stick some more series caps in there to get the total capacitance down. Plus it will give me some more flexibility in selecting my capacitance.
IIRC, you tune your "small" coils to ringup until loaded by the secondary whereby the streamer loading brings it into tune, hence your sentence "...begins to plateau or drop off from secondary loading...". Correct?
James
thedatastream, Wed May 23 2007, 07:30AM
Hi Steve, thanks for all the info.
Interesting what you say about Zsurge and the rate of current rise, it wasn't something I'd considered. Sounds like I need to re-do my sums for a surge impedance of about 8-12 ohms or so - a larger diameter primary may well be in order.
Also, based on what you are saying, I think I will re-design my cap bank for two parallel strings and stick some more series caps in there to get the total capacitance down. Plus it will give me some more flexibility in selecting my capacitance.
IIRC, you tune your "small" coils to ringup until loaded by the secondary whereby the streamer loading brings it into tune, hence your sentence "...begins to plateau or drop off from secondary loading...". Correct?
James
Re: DRSSTC MMC Sanity Check
Steve Conner, Wed May 23 2007, 11:32AM
Well, my DRSSTC was originally built with a 50nF tank capacitor, which Steve Ward and everyone else said was way too small. I tried 100nF with extra topload, this made the sparks bigger and the primary current somewhat higher, but also increased the risk of flashovers. 50nF seemed fine for normal use.
I am currently rebuilding it with a MMC because the red top Hivotronics cap I used overheats badly. It was leaking molten wax by the end of the last demo day I did. I plan to use two strings of 10 caps of the exact same kind that James is using, to give 100nF, but keep the same small topload. If it provs to be too fierce, I can run on just one string, although that is a bit ridiculous in terms of RMS current.
From my experiments, I believe there is a good deal of leeway in everything, and no real magic values, so I expect to get a usable coil with either 50 or 100nF.
PS, it ran at 220kHz.
Steve Conner, Wed May 23 2007, 11:32AM
Well, my DRSSTC was originally built with a 50nF tank capacitor, which Steve Ward and everyone else said was way too small. I tried 100nF with extra topload, this made the sparks bigger and the primary current somewhat higher, but also increased the risk of flashovers. 50nF seemed fine for normal use.
I am currently rebuilding it with a MMC because the red top Hivotronics cap I used overheats badly. It was leaking molten wax by the end of the last demo day I did. I plan to use two strings of 10 caps of the exact same kind that James is using, to give 100nF, but keep the same small topload. If it provs to be too fierce, I can run on just one string, although that is a bit ridiculous in terms of RMS current.
From my experiments, I believe there is a good deal of leeway in everything, and no real magic values, so I expect to get a usable coil with either 50 or 100nF.
PS, it ran at 220kHz.
Re: DRSSTC MMC Sanity Check
thedatastream, Thu Jun 14 2007, 08:10PM
I now have 2 parallel strings of 9 caps to form a 100nF, 9kV capacitor using copper clad board to make the connections.
Thanks for the advice
James
thedatastream, Thu Jun 14 2007, 08:10PM
I now have 2 parallel strings of 9 caps to form a 100nF, 9kV capacitor using copper clad board to make the connections.
Thanks for the advice
James
Print this page