Need to make inexpensive HV resistor for voltage multiplier

[MRMILSTAR]

I need a better (higher power rating) current-limiting resistor for my CW voltage multiplier. I need a resistance of around 5 Mohms. The value isn't critical. It needs to have a power handling capability of about 200 watts. The voltage rating needs to be about 300 KV.

Stringing commercial HV resistors is too expensive. I don't think a water resistor would work very well because of electrolysis caused by the DC current. Does anyone have any ideas?

[klugesmith]

I have an unfinished project to make inexpensive HV dummy load resistor using long string of 2W axial lead resistors.
Found them to be lowest price  per watt -- a whole reel from Digikey cost less than $50.
I think they are 47K ohms, so each 22 resistors add up to 1 megohm 44 watts 11,000 volts rating.
For you to get 300 kV rating without using 600 resistors, you'd need to use HV resistors.

The challenge is physical mounting of the resistors for easy soldering, good air circulation, and attention to HV issues like corona.
One version leaves them on the paper tapes, as they come from the reel.  Electrically connected to make a serpentine current path that can be rolled back up on the reel.   I practiced ways to make & solder the jumper wires quickly.
Another is cordwood style: 3D block of parallel resistors, with the ends paired up at holes in perf-board plates on each end of the block.

[MRMILSTAR]

My current plan is to use some sort of HV resistor string depending on cost and availability. Something like ten in series rated for 30 KV each and 10 watts power dissipation each. Mount them in a sort of staggered "Pan's pipe" configuration in a PVC tube filled with mineral oil.

[MRMILSTAR]

I'm wondering if I even need a load resistance to protect the diodes of my 14-stage multiplier. At full load, the design is only supposed to generate 1 ma of current. My diodes are rated for 30 ma. Thoughts?

[davekni]

I expect you need some resistance.  Without resistance you will get hard sparks like from a Marx generator with 100+ amps peak.  Do the diodes have a peak current rating?  Such peak ratings are usually for a much longer time (1ms or 10ms) than discharging your CW multiplier, so you could likely get away with well higher than even the diode's peak current rating.

If you have an extra diode or two, you could set up a single diode with a cap and resistor and spark-gap and run stress tests.

A string of inductors is another option instead of resistors.  That is what I did for my Marx generator charging ladder.  Less lost energy (more into sparks).  Diode current will last longer, however, so don't go as far above peak current rating as you might with resistors.

[MRMILSTAR]

The diodes are 2CL20KV. I have them in strings of 3 for a 60 KV voltage rating. The data sheet says the surge current is 1 amp for 10 milliseconds. It sounds like I need some sort of resistive load. I have a resistive load in my original build but the resistors were under-rated for power dissipation and started to burn so I have to replace them with something beefier. I was hoping that the oil would provide enough additional cooling for the under-rated resistors, but no.

In the current build, the resistors are located in the same tube as the multiplier stack and they barely fit. I will have to locate any new resistive load in the adjacent discharge tower just for the space to mount the components.

[Twospoons]

How about using some 1k carbon fiber roving, wound on a large PVC pipe and coated with epoxy?  Not sure how much you'd need - might need to split the roving to fewer fibers for more R per metre.

[davekni]

With 1A surge capability you could go down to 150k ohms (2A peak, with half through diodes), and likely even lower given the short surge duration.  Doesn't help with the need to dissipate 200W still.

Inductors end up being possible, but not too practical.  For 150k resonant impedance with 2 * 1.7nF / 14 = 243pF, inductance needs to be 5.5H.  550 small 10mH inductors in series would work.  (My Marx generator charging ladder has 24 inductors, each consisting of a series string of 133 1mH inductors, 3192 total inductors.)

[MRMILSTAR]

After a lot of thought and investigation I came to the conclusion that there is no cheap way out that will perform reliably. Consequently, I ordered 21 Ohmite HV resistors with the following specs:

* Part number: MOX-5-122504F
* Resistance: 2.5M ohms
* Maximum voltage: 48 KV
* Maximum power dissipation: 12.5 watts

At least they were attractively priced on E-Bay as NOS. My plan is to connect them in parallel-series for an equivalent resistance with the following specs:

* Resistance: 5.83 Mohms
* Maximum voltage: 336 KV
* Maximum power dissipation: 262 watts

The only good news is that I shouldn't have to place them under oil since corona loss from joints won't be a concern and I am allocating about 38" of height to mount the resistor strings.

That brings up another question that I have. For the 21 resistors is it better to have a 3 x 7 configuration or a 7 x 3 configuration? The 3 x 7 configuration would consist of 7 groups of resistors in series where each group would consist of 3 resistors in parallel. The 7 x 3 configuration would consist of 3 groups of resistors in parallel where each group would consist of 7 resistors in series. For low voltage applications, either topology would work. For HV applications though I am thinking that one may be better than the other. From a construction point of view, I much prefer the 7 x 3 configuration but I suspect that the 3 x 7 is preferred from a performance point of view.

[davekni]

Is there some physical construction implied by 3x7 or 7x3?  I'm picturing an array that is 3 resistors wide and 7 resistors tall with resistors vertical.  The only difference would then be whether there are six horizontal connections between resistor ends at a given height.  They are theoretically at the same potential, so horizontal connections wouldn't matter if there or not.  Capacitive coupling might make a slight potential difference, but I still doubt it would matter if there are horizontal connections or not.

[klugesmith]

Good move, buying HV-rated resistors after finding a good price on Ebay.

As I was writing when Dave's post appeared, the question seems to be whether or not to join all 6 pins at each equipotential level.  Why not? Then there are only 6 intermediate-voltage nodes to be mechanically supported, with no worry about high voltage between pairs if one resistor connection failed.  Perhaps solder pins to a metal lug or perforated strip that's attached to insulating backbone for the whole array.  How about little anti-corona toroids at each intermediate level?

Are the resistor voltage and power requirements based on peak or average stress?
Can the Marx bank run continuously (with some pulse repetition rate) while the average capacitor-charging power is more than 200 watts?