High Voltage Multiplier Issue

This forum is for specialized infomation important to the construction and safe operation of the high voltage electrical supplies and related circuitry needed for fusor operation.
User avatar
Dennis P Brown
Posts: 1431
Joined: Sun May 20, 2012 2:46 pm
Real name: Dennis P Brown
Location: Glen Arm, MD

Re: High Voltage Multiplier Issue

Post by Dennis P Brown » Sun Oct 22, 2017 7:22 pm

Yes, in all likelihood you are correct relative to the 60 Hz, which is bad news for my final goal; still, will try and run a current measurement using the stainless steel accelerator globe as my current sink - likely, it appears, my VM will not to be enough. Still, nice creating the high voltage - oh, well.

Rex Allers
Posts: 221
Joined: Sun Dec 30, 2012 8:39 am
Real name:
Location: San Jose CA

Re: High Voltage Multiplier Issue

Post by Rex Allers » Mon Oct 23, 2017 2:10 am

Dennis,

Not trying to sound too negative, but sometimes your posts, especially electronics-related, leave me scratching my head.

I haven't followed all the details in this thread, maybe there was a more detail later, but I wonder if your NST2.pdf diagram is accurate? The two ground points would short out half of the transformer secondary. That can't be good.

But I'm writing mainly to discuss details of your HV measurement method.

In one post you wrote,
"... my high voltage probe, which directly reads kV on a display scale, is based on a Giga-ohm resistor."

Did you mean one resistor?

I think you wanted to have a supply in the 100 KV output vicinity. Most HV measurement resistive dividers I have seen for voltages above 10 KV use multiple resistors in series with each resistor being of a 100 - 300 Mohm value (all the same value but different designs could choose a particular value somewhere in that range).

I looked at several HV resistors I have around here. The highest voltage rating for one resistor was 30 KV and that resistor was about 3 inches long (7.6 cm). Comparing specs for a lot of HV rated resistors it seems that a reasonable rule of thumb for length vs. voltage rating is 1 inch per 10 KV. I've never seen any single resistor that was rated for 100 KV.

Some years ago I bought (on ebay) a HV divider board that was a part from some custom supply design that was making 100 KV or maybe a little more. This divider is a circuit board that has 10 200Mohm, 20KV rated, resistors in series. Each resistor is about 2" long. They are arranged in a slightly zig-zag pattern to make the board a little shorter than straight line but the board is still about 20 inches long.

So the 10 20KV-rated resistors would give the board a max 200 KV rating. I think in its original application it was immersed in oil too. I have used it several times in air to measure supplies up to around 60 KV.

This divider also has 500pF 20KV-rated capacitors in parallel with each resistor. I have seen this same general design in the voltage sensing dividers in Glassman supplies. That is, low-hundred Mohm resistors connected in series with HV caps in parallel with each.

I looked for a reference describing why dividers have the caps in parallel with each resistor but didn't find anything to quote. I'd guess it is to improve the sensing response to fast transients on the HV.

So, I wonder if your divider really has a practical voltage rating suitable for measuring around 100 KV?

I don't think I heard any discussion of issues I would expect at these voltages like corona discharge, either.


A thought on checking accuracy of HV divider:
-------------------------------------------
If you have a good DVM that you trust, most will measure up to 600 or 1000 V. They will usually also measure down to millivolts, maybe less.

If you can get a supply that will make a voltage in the 500 to 1 KV range you can use that to check your divider. I'll skip details but maybe a 12V out mains transformer, reversed and fed by a variac and then a diode and capacitor to make it DC. A nice commercial supply would be best but the preceding might work if you only put tiny load on it and don't go above 1KV out (500 V would be safer).

Use the meter on its high scale to set a nice high input voltage, say 500, 600, 900, 1000, depending on the upper range. Then measure the output voltage of the divider on the lowest appropriate meter range.

If you can live with a divide ratio of 10000 or less (that would be 10 V out for 100 KV measured) then if you put 500 V across the divider, you should see 50 mV out.

I hope that idea makes sense. In my experience, just doing the math for the resistance on the low (output) end of the divider might be quite a bit off from actual measurements.
Rex Allers

User avatar
Richard Hull
Site Admin
Posts: 10728
Joined: Fri Jun 15, 2001 1:44 pm
Real name: Richard Hull

Re: High Voltage Multiplier Issue

Post by Richard Hull » Mon Oct 23, 2017 4:26 am

Assembling a good, reliable HV divider relies on experience with such voltages. One must come to realize that corona is a real issue above 40kv. At 100kv + Under oil, one needs to be concerned with the oil, if it is not extremely pure, can become part of a parallel divider network. measuring above 150kv will usually go back to air in a potted ladder ring divider with large field control torroids acting as equi-potential rings.

Simple bread boarding of chained resistors will be most dubious at 75kv without a calibrated reference to check one's work against allowing him to fine tune the effort.

We like to think of a high voltage supply as an electro-dynamic instrument but at the highest of voltages, accurate measurement of voltages and even currents become a battle with electrostatics.

Richard Hull
Progress may have been a good thing once, but it just went on too long. - Yogi Berra
Fusion is the energy of the future....and it always will be
Retired now...Doing only what I want and not what I should...every day is a saturday.

User avatar
Dennis P Brown
Posts: 1431
Joined: Sun May 20, 2012 2:46 pm
Real name: Dennis P Brown
Location: Glen Arm, MD

Re: High Voltage Multiplier Issue

Post by Dennis P Brown » Mon Oct 23, 2017 12:29 pm

Rex, my high voltage probe is a commercial unit; someone or a group with high voltage experience/knowledge designed/built these to operate accurately if one is careful in their use - more on that later in this post.

This HV probe is and must be a simple divider system (since it is analog); all I am referring to in the unit relative to resistors is the first (primary) resistor which is (as I recall) a 1.00 Giga-ohm monster (has a 1% accuracy.) It is a bit over seven inches long and a single piece - special made for the probe, I assume. I never paid the slightest attention to its built in meter (which, I assume, provided the remaining resistance that defines the system - I also assume the meter is a micro-amp meter that is calibrated so its read-off scale can be marked in "voltages", which is not what the meter is really reading, of course, but the current/voltage relationship does directly follow.) This unit only reads up to 50 kV and sorry if my posts didn't make that clear.

The reason I have discussed 100 kV (which I never directly measured) at all is I am extrapolating the voltage multiplier's (potential) performance once I get full power off the variac - that is, if the variac is providing a power usage of 50% of its rating and my voltage multiplier is reading 50 kV (and has show extremely linear response to that point) I assume the max the VM will reach when the variac is dialed up to 100% of its rated output is about 100 kV (not really true but should be close.) All this is done/measured completely UNDER oil - otherwise, corona issues (as Richard points out) would be extremely serious (leading to very false readings.) All values above 50 kV are extrapolations from the variac's scale. Real world measurements could be wildly off above my 50 kV readings once I get over this limit but if I am under oil, I am reasonably certain my extrapolations are not too badly off. I am not building a manned aircraft here so feel this is ok to assume.

As for possible shorting in the diagrams I posted relative to the transformers and voltage multiplier circuits - these circuit drawings are taken directly from the design of professional EE people. I do not claim to have knowledge of EE so can't defend those drawings (they are not mine.) From a logic stand-point, I too was EXTREMELY confused by those designs (relative to ground location and circuit performance) but I admit, after working with them and getting explanations, I feel they do the job but one must be careful about application - also, wire resistance/induction with a x-former core when an applied AC (60 Hz) excitation is used does not behave as a simple short as one might expect with these rather simplistic diagrams (for example: a NST 's high voltage secondary is, in fact, center grounded to the case and it works fine (but one gets out-of- phase +/- 7.5 kV, if I recall correctly); if one where to apply DC excitation to one of these circuits, then yes, short circuit big time in the x-formers to ground.)

For me, those ground points drove me to distraction since they had a huge effect upon final voltage performance of the VM and I had trouble understanding this effect.

I feel I do now, mostly, understand these circuits a good bit better (relative to grounding; the basic VM circuit is not at issue for me) but discovered that performance issues relative to where grounds are located for given transformer type is an issue one has to be very careful about - hence, my extensive posts here for others so they don't have to do what I've done. That is why I did these experiments and posted so extensively - I have seen that the literature is sorely lacking in addressing these very issues - that is, explaining why a given ground is located where it is and what this causes for the VM circuit is not explicit in most treatments on VM stacks.

I apologize for my own initial confusion relative to these issues since I, too, was "scratching my head" wondering why these circuits worked (i.e. gave a specific voltage value) for a given x-former/ground configuration. This made my total thread very confusing, I think; again, sorry but this is a place to learn and ask questions so someone like me, can figure out the why's of a given experiment. In summation, I decided to plow through various iterations, post these results with questions and then get to the "bottom" of the problem. Not necessarily how this should have been done but what occurred.

Aside: digital voltmeters are not something I have any knowledge of and as a consequence, can't comment upon. How caps/resistors and so forth are used to get accurate results is, for me, a black box (lol.) Analog meters (using simple voltage dividers) are easy to understand and I often make my own (for custom voltage ranges) besides using the high end one I have.

User avatar
Dennis P Brown
Posts: 1431
Joined: Sun May 20, 2012 2:46 pm
Real name: Dennis P Brown
Location: Glen Arm, MD

Re: High Voltage Multiplier Issue

Post by Dennis P Brown » Mon Oct 23, 2017 3:32 pm

Back to my original goal (exciter for the deuteron gun.) In that retrospect, John Futter is likely correct and I should (have) build (built) a full wave voltage multiplier to start.

In any case, I will convert my current ten stage (mixed caps) half-wave system back into the original seven stages (all door knob caps. Gives me the max current for my acceptable voltage.) Then measure the actual current at max voltage and see what this yields. If marginally within the range required, a full wave VM would be the next step since this will double the current making my application more likely to succeed. If this current VM exceeds what I would likely need, then a full test would then make sense using my accelerator tube.

However, if the seven stage half wave creates too low of a current, then best to abandon the effort.

Regardless, of these outcomes still glad I did this to see what making a 200 kV system involves - even if the current is not sufficient.

User avatar
Dennis P Brown
Posts: 1431
Joined: Sun May 20, 2012 2:46 pm
Real name: Dennis P Brown
Location: Glen Arm, MD

Re: High Voltage Multiplier Issue

Post by Dennis P Brown » Tue Oct 24, 2017 6:00 pm

I reworked/reduced the voltage multiplier (VM: returned it to 8 stages again, all door-knob caps; and I added a far higher current tolerant diode to the first stage.) This has created a dramatic increase in the overall performance of the VM.

The current performance (directly from the VM output) looks fairly good thru a 1.00 G-ohm resistor; with the variac at 18% load, the voltage is 50 kV and the current is about 48 micro-amps (read by a 50 micro-amp meter; I think the curent would increase much further if the voltage is further increased - that measurement will require another meter.) I've noticed that both the current and voltage is fairly linear under this light load (not likely to remain linear above 50 kV or so) on the VM.

Aside: I think this pretty much shows my HV probe is based on a 50 micro-amp meter (good to know. Might build a 0 - 5 kV meter now that I know some performance data for 60 Hz systems.) Also, ignoring RMS, this means each stage in the VM is giving slightly more (2.3) than a doubling of voltage: obviously, if the RMS voltage is used, that value is not really more than twice; so it appears, to first order calculation, my stages are operating close to theoretical values. For no real loss/load on the VM, this result isn't exactly surprising but is reassuring that the circuit performance is close to idealized values.

Using no resistor and exciting the globe of my accelerator with the VM, the current from the VM appears too great (the VM can handle it but it is at a dangerous level for me): that is, at just 5-10 kV, the VM is outputting a full 1 milliamp! (measured)

So, I hooked up a 5 G-ohm resistor to the accelerator metal globe (the current meter is at ground and a wire leads from that to the other side of the 5 G-ohm resistor that is attached to the globe), and at about 140 kV (about 45% load on the variac), I am drawing somewhere around 30 - 50 micro-amps thru the resistor (hard to accuracy read on the milli-amp meter I used for this test.) However, the sphere is being highly charged (and carrying a possibly lethal charge, apparently (the caps in the VM are not being fully discharged by that minor loss thru the resistor and static discharge by the globe (which is really charging the area nearby.)))

So, it appears I will need to feed the VM's output thru a high resistance resistor (Value?) so the current from the VM is tightly limited (needs to be non-lethal); all that remains is to determine a resistance to yield 100 - 150 micro-amps for the max voltage (say 50 - 60% load on the variac); the VM's voltage and current are extremely high for my application - apparently, the VM can deliver a number of milliamps in the 50 kV range with no load.

This VM is working too well. I am guessing that a resistor in the 750 to 1000 M-ohm mounted as an output source for the VM will supply all the current I need regardless of corona losses by the sphere (apparently, just need to match my voltage probe resistance since that appears to yield my required current ... . ) So for the globe to be elevated to the range of 135 kV (I have weak confidence in that number since it can't be measured) adding an output resistor of a Giga-ohm should do it - my only concern is making certain the VM will operate in a safe, current limited, range for the 120+kV that will be obtained on the exposed accelerator sphere. I will try some values and test with a current meter on that device to be certain of the final performance.

So much for my worrying that 60 Hz wouldn't cut it for supplying enough current from that VM to excite my accelerator ... my only real worry, it turns out, is preventing it from supplying a lethal current to the operator (my accelerator's main sphere (contains deuterium supply/ion gun and its power supply) is rather large and exposed to an operator.)

Aside: if memory serves, low milli-amps is the cross over for kV systems to be lethal (60 Hz); if others have more exact figures, do weigh in.

User avatar
Dennis P Brown
Posts: 1431
Joined: Sun May 20, 2012 2:46 pm
Real name: Dennis P Brown
Location: Glen Arm, MD

Re: High Voltage Multiplier Issue

Post by Dennis P Brown » Wed Oct 25, 2017 11:21 am

Well, for proper operation/testing of the voltage multiplier system for the accelerator it looks like I need to locate a 1 G-ohm extra long (4-5 inches) HV resistor. Until then, I can't really do much. Still, made a lot of good progress (90% of which simply consists of getting a better understanding VM's.)

If I can procure the resistor, I will put the deuterium accelerator back into running condition: that is, fill the small secondary deuterium tank; repair the sphere's internal mounts; check the operation of the deuterium ionization gun in the sphere, and retest the vacuum system's performance (might have to bake out the main system again.) Also, I will place silver on the target holder (for a final proof of operation test via Ag activation.) Finally, re-install the anti-electron back streaming device in the target head to suppress x-ray generation (which can, in this device, be significant.)

So, now that I feel that I have achieved the operational VM I need to excite my deuterium accelerator, and unless someone wants to ask questions or address issues relative to the VM system, this thread is done. I will, once I get the affore mentioned issues properly addressed (locate and install the required resistor), re-start my "Ion gun" thread for this accelerator project as, and hopefully, it then moves forward again after its long hiatus.

User avatar
Bob Reite
Posts: 268
Joined: Mon Aug 26, 2013 1:03 am
Real name: Bob Reite
Location: Wilkes Barre/Scranton area

Re: High Voltage Multiplier Issue

Post by Bob Reite » Sun Oct 29, 2017 5:54 am

Get 10 100 M HV rated resistors and put them in series for your 1 Gigohm resistor.
The more reactive the materials, the more spectacular the failures.
The testing isn't over until the prototype is destroyed.

Silviu Tamasdan
Posts: 147
Joined: Wed Jun 28, 2006 11:17 pm
Real name: Silviu Tamasdan
Location: Connecticut

Re: High Voltage Multiplier Issue

Post by Silviu Tamasdan » Sun Oct 29, 2017 12:59 pm

I recently bought a couple of 1.5G resistors on ebay. They're supposedly rated for 35kV. They come from a Russian seller but unfortunately those were the last pieces he had. I don't have them yet but from the pictures they're about 5 inches long each. Nice feature is that they have threaded terminal ends so you can screw them together in series.

Keep looking, you'll find what you need.
There _is_ madness to my method.

User avatar
Dennis P Brown
Posts: 1431
Joined: Sun May 20, 2012 2:46 pm
Real name: Dennis P Brown
Location: Glen Arm, MD

Re: High Voltage Multiplier Issue

Post by Dennis P Brown » Sun Oct 29, 2017 5:01 pm

Thanks for the comments: A longer string of resistors (in air) is always a good idea but mine would be under oil and are fairly long so not too worried on that score.

Post Reply