Double-Checking my MMC Calculations
KeeDx3, Mon Jul 10 2017, 04:48AMI recently picked up where I left off on a small SGTC I've been working on. I recently decided to use a different transformer, so I had to recalculate the MMC array.
The transformer I'm now using is an Allanson 12/30 NST.
According to Deep Fried Neon, with 12000v 30mA at 60HZ, I should be aiming for a total capacitance of 0.066uF.
According to what I have read in the past, the total voltage rating should be about twice the NST's rated voltage.
Using DFN's calculators, it seemed that the best match was the Cornell Dubilier 942C 942C20S22K-F rated at 2000vDC at 0.022uF. These would be arranged in an array of 4 strings of 13 caps giving the following specs:
Capacitance: 0.0067uF
Voltage Rating: 26000
Number of Caps: 52
I just thought I'd post this for review before spending almost $200 on caps. It was a little overwhelming choosing caps with different values and seeing how they interacted in the calculator. The model above seemed to give me the closest to target values with the lowest number of caps needed. Though I stand to be corrected, thus my posting here.
Re: Double-Checking my MMC Calculations
KeeDx3, Thu Jul 13 2017, 03:04AM
Well, I'll assume the crickets mean that I'm on the right track and at least not headed down the wrong path.
KeeDx3, Thu Jul 13 2017, 03:04AM
Well, I'll assume the crickets mean that I'm on the right track and at least not headed down the wrong path.
Re: Double-Checking my MMC Calculations
Sulaiman, Thu Jul 13 2017, 10:00AM
I guess the lack of replies is related to moral responsibility - not wanting to give the wrong advice.
You can have four strings of 13 caps in series 6n77, 26000 V (52 caps total)
or
three strings of ten caps in series, 6n66, 20 kV (30 caps total)
I believe that the voltage and current ratings of the caps are sufficient
and the cost is < 60% of the four string option.
However, 6.66 nF is the value of capacitance that will resonate with the output inductance of the NST at 60 Hz.
For maximum power it is common to increase the value of capacitance to larger than required for resonance, about 9 to10 nF
so go back to the calculator and catalogue to find a new cost effective combination of capacitors ;)
Sulaiman, Thu Jul 13 2017, 10:00AM
I guess the lack of replies is related to moral responsibility - not wanting to give the wrong advice.
You can have four strings of 13 caps in series 6n77, 26000 V (52 caps total)
or
three strings of ten caps in series, 6n66, 20 kV (30 caps total)
I believe that the voltage and current ratings of the caps are sufficient
and the cost is < 60% of the four string option.
However, 6.66 nF is the value of capacitance that will resonate with the output inductance of the NST at 60 Hz.
For maximum power it is common to increase the value of capacitance to larger than required for resonance, about 9 to10 nF
so go back to the calculator and catalogue to find a new cost effective combination of capacitors ;)
Re: Double-Checking my MMC Calculations
KeeDx3, Fri Jul 14 2017, 05:25AM
Thanks for this! That definitely helps me avoid some future grief.
And you're probably right about the advice. It's easy to forget that a simple post could mean life or death.
KeeDx3, Fri Jul 14 2017, 05:25AM
Sulaiman wrote ...
I guess the lack of replies is related to moral responsibility - not wanting to give the wrong advice.
You can have four strings of 13 caps in series 6n77, 26000 V (52 caps total)
or
three strings of ten caps in series, 6n66, 20 kV (30 caps total)
I believe that the voltage and current ratings of the caps are sufficient
and the cost is < 60% of the four string option.
However, 6.66 nF is the value of capacitance that will resonate with the output inductance of the NST at 60 Hz.
For maximum power it is common to increase the value of capacitance to larger than required for resonance, about 9 to10 nF
so go back to the calculator and catalogue to find a new cost effective combination of capacitors ;)
I guess the lack of replies is related to moral responsibility - not wanting to give the wrong advice.
You can have four strings of 13 caps in series 6n77, 26000 V (52 caps total)
or
three strings of ten caps in series, 6n66, 20 kV (30 caps total)
I believe that the voltage and current ratings of the caps are sufficient
and the cost is < 60% of the four string option.
However, 6.66 nF is the value of capacitance that will resonate with the output inductance of the NST at 60 Hz.
For maximum power it is common to increase the value of capacitance to larger than required for resonance, about 9 to10 nF
so go back to the calculator and catalogue to find a new cost effective combination of capacitors ;)
Thanks for this! That definitely helps me avoid some future grief.
And you're probably right about the advice. It's easy to forget that a simple post could mean life or death.
Re: Double-Checking my MMC Calculations
klugesmith, Sat Jul 15 2017, 02:29AM
Just a couple of comments from one non Tesla coiler.
1. In OP, K. says: aiming for a total capacitance of 0.066uF.
That's ten times more capacitance than the MMC arrays discussed in the rest of the thread. Typographical misplacement of the decimal point, right?
2. Compared to the original 13x4 solution, Sulaiman's 10x3 gets essentially the same net capacitance using 42% fewer capacitors. But in operation, the electrical stress on each individual 942C will be proportionally greater. Each cap will get 30% more RF voltage and 33% more RF current. Both of those factors might want to be squared (or used with some intermediate exponent), for analysis of reliability, temperature rise, etc.
So it comes down to a judgement call about how much design margin is appropriate. Tesla coiler wisdom is plentiful at this forum. Probably more relevant than guidelines from C-D customers whose designs have to operate continuously except for annual maintenance.
Someone said the MMC array should be rated for at least twice the nominal NST voltage. The ratios for our two candidates are 2.17 and 1.67. The MMC voltage ratings are 1.53 and 1.18 times the NST's peak V_open_circuit voltage, if my arithmetic is right.
There's an elephant in the room, about which I claim no expertise. Those 942C20S22K-F's are nominally 2000 volt devices. That's the DC rating; AC rating is only 500 volts.
klugesmith, Sat Jul 15 2017, 02:29AM
Just a couple of comments from one non Tesla coiler.
1. In OP, K. says: aiming for a total capacitance of 0.066uF.
That's ten times more capacitance than the MMC arrays discussed in the rest of the thread. Typographical misplacement of the decimal point, right?
2. Compared to the original 13x4 solution, Sulaiman's 10x3 gets essentially the same net capacitance using 42% fewer capacitors. But in operation, the electrical stress on each individual 942C will be proportionally greater. Each cap will get 30% more RF voltage and 33% more RF current. Both of those factors might want to be squared (or used with some intermediate exponent), for analysis of reliability, temperature rise, etc.
So it comes down to a judgement call about how much design margin is appropriate. Tesla coiler wisdom is plentiful at this forum. Probably more relevant than guidelines from C-D customers whose designs have to operate continuously except for annual maintenance.
Someone said the MMC array should be rated for at least twice the nominal NST voltage. The ratios for our two candidates are 2.17 and 1.67. The MMC voltage ratings are 1.53 and 1.18 times the NST's peak V_open_circuit voltage, if my arithmetic is right.
There's an elephant in the room, about which I claim no expertise. Those 942C20S22K-F's are nominally 2000 volt devices. That's the DC rating; AC rating is only 500 volts.
Re: Double-Checking my MMC Calculations
Sulaiman, Sat Jul 15 2017, 09:54AM
This is the difference between 'coiling' and 'engineering', safety margins,
a good part of the reason that no answers were given earlier.
This type of capacitor usually has three frequency operating regions;
1) Below F1 (varies with capacitor type,) a.c. voltage is determined by dielectric breakdown voltage
2) Above F2 (varies with capacitor type,) a.c. voltage is determined by rms current heating of the esr
3) Between F1 and F2 (typically 20 to 100 kHz for the Kemet A72 caps that I use) the voltage derating is, I believe, due to
3.1) an effective increase in the voltage stress on the dielectric due to the slow recovery speed of dielectric absorbtion and
3.2) dielectric heating due to loss tangent
Those are 'engineering' constraints, for safe, reliable continuous operation.
SGTCs have a very short duty cycle (typ. 1% to 5%) so the current handling capability can be up to 10x the continuous rms rating
The voltage rating of the capacitors is for continuous dc, but the capacitors usually have a surge rating of at least 1.5x rated dc voltage.
So,
you could 'engineer' an mmc to never receive more than rated voltage or current,
or as a 'coiler' you can stress your capacitors way beyond that in TC use,
how far can you push the capacitors ?
- that is where it becomes difficult to give honest recommendations.
e.g.
I am sure four strings of 13 series caps is adequate for a 12 kV 30 mA NST
I am fairly confident that three strings of ten series caps is ok
I have used single strings of A.72 caps.
but remember, greater than 6n7 would get the most from the NST.
Sulaiman, Sat Jul 15 2017, 09:54AM
This is the difference between 'coiling' and 'engineering', safety margins,
a good part of the reason that no answers were given earlier.
This type of capacitor usually has three frequency operating regions;
1) Below F1 (varies with capacitor type,) a.c. voltage is determined by dielectric breakdown voltage
2) Above F2 (varies with capacitor type,) a.c. voltage is determined by rms current heating of the esr
3) Between F1 and F2 (typically 20 to 100 kHz for the Kemet A72 caps that I use) the voltage derating is, I believe, due to
3.1) an effective increase in the voltage stress on the dielectric due to the slow recovery speed of dielectric absorbtion and
3.2) dielectric heating due to loss tangent
Those are 'engineering' constraints, for safe, reliable continuous operation.
SGTCs have a very short duty cycle (typ. 1% to 5%) so the current handling capability can be up to 10x the continuous rms rating
The voltage rating of the capacitors is for continuous dc, but the capacitors usually have a surge rating of at least 1.5x rated dc voltage.
So,
you could 'engineer' an mmc to never receive more than rated voltage or current,
or as a 'coiler' you can stress your capacitors way beyond that in TC use,
how far can you push the capacitors ?
- that is where it becomes difficult to give honest recommendations.
e.g.
I am sure four strings of 13 series caps is adequate for a 12 kV 30 mA NST
I am fairly confident that three strings of ten series caps is ok
I have used single strings of A.72 caps.
but remember, greater than 6n7 would get the most from the NST.
Re: Double-Checking my MMC Calculations
Uspring, Sat Jul 15 2017, 04:02PM
klugesmith wrote:
It shows, why the AC rating isn't proportional to the DC rating but lacks an answer about the frequency dependence.
Uspring, Sat Jul 15 2017, 04:02PM
klugesmith wrote:
There's an elephant in the room, about which I claim no expertise. Those 942C20S22K-F's are nominally 2000 volt devices. That's the DC rating; AC rating is only 500 volts.An explanation is here.
It shows, why the AC rating isn't proportional to the DC rating but lacks an answer about the frequency dependence.Print this page