Current and Voltage Metering for a HV Supply

[userjjb]

I know this has been covered several times but I'm unclear on a couple things...

See the diagram below (we are using a bridge rectification setup with the positive output from the bridge grounded because our transformer doesn't have a center tap)

Will a DMM in series with the grounded positive terminal from the bridge rectifier read the proper current? I've seen current shunts mentioned, but am hazy on what exactly these are/do/consist of?

I've also seen a need for current limiting mentioned, currently we have none. I've seen both inductive ballasts (MOTs or NSTs with shorted secondaries and the primaries in series) and resistive ballasts (1 ohm per volt or thereabouts?) mentioned. Which current limiting method is better, or should both be used in conjunction?

Finally how should voltage output be measured? I've seen voltage dividers mentioned and after some research think the scheme drawn in the diagram is appropriate. The DMM put in parallel with R2 should read R2/R1 fraction of the output voltage, correct? What values are appropriate for R2 and R1? I've seen ratios of 1000:1 cited so the DMM will read in kV, but what about actual values for R1? I'd imagine R1 should be sufficiently high so it doesn't divert an appreciable amount of current, but not infeasibly high.

Thanks,

-Josh

Edit: I originally and erroneously showed the MOT ballast as being after the HV transformer in the diagram which of course is wrong, a little editing in paint fixed this.

[DaveC]

Josh -

Your circuit looks good to me... except that your ballast... (the MOT) should connected with the SECONDARY in series with HV... it's the winding that has the insulation for HV, with the primary shorted... or better with some resistance across it.

[ Edit: - You edit fixed it. Looks good, now.!!]

The voltage divider ratio of 1000:1 is very conventient. The design aspect of the divider is simply how much current you it to draw. The Divider circuit typically should not take more than a few percent of the HVDC output... so a 20 kV 10mA supply (puts out 200 watts) could use a divider string that consumes maybe 4 watts. Using the formula P = V^2/R, gives a nominal R around 100 Meg Ohms...


For a 1000:1 ratio, near this value, with a 10 Mohm input DMM., you could use a 95 M Ohm string for your R1 and a 10 M ohm resistor for R2 (shunted with the meter's 10 M ohms input iresistance). The 95 M Ohm string.. could be 9 10 meg resistors or 18 - 5 meg resistors plus a 5 M ohm resistor in a little bit of perf board. Most carbon or metal film resistors can;t be used at such high voltages, as a single element.. and finding very high values resistors takes to just a few mfrs. and they get pricey and long delivery times. So stringing them together can save you some money and time.

This kind of divider, assembly CAN operate in air at up to about 20 kV if you build it with enough spacing and nice smooth solder ball type joints. It works a lot better in oil, though. Use 1% resistors and you will have a system that's as accurate as any commercial HV divider.

When you get if working, you can also check for ripple levels, by just switching your meter to AC. If the ripple voltage is more than about 10% of the DC, put a 1uf 50 V polyester film cap across R2 for some filtering.

Just be careful... HV can bite!

Dave Cooper

[userjjb]

Thanks a lot Dave! Up until now we've been pretty sparse in the way of instrumentation as far as output power goes. Flying by the seat of your pants when it comes to HV is never a good idea. We do have an electrostatic voltmeter that reads up to 20 kV, unfortunately it only reads positive so the only time we've been able to get some quantitative idea of output voltage is when we had a reverse polarity setup and were just measuring open circuit potential.

We have 20 HV resistors from a cannibalized power supply rated at 50 M Ohm, I'm not sure of the power rating but they're fairly chunky, approx. 5 cm long and maybe 2 cm in diameter, which I'm guessing should do the trick. If I stick 4 of those in series for my R1, and have a max operating voltage of say 50kV, they'd each dissipate 3.125 W of power which I think (and will verify) is within their operating specs. Then if I had a DMM with 10 M Ohm impedance, I could pick a R2 of 200 k Ohm and have close to a 1000:1 ratio. Do you think I could substitute a 200 k Ohm potentiometer in place of R2 so I could calibrate the divider using the electrostatic voltmeter as a comparison?

[DaveC]

Josh --

Those components sound like just what you need.

You've got the idea about the voltage divider.. you can just either just ignore the small error caused by the DVM loading or... calculate it out and make a 1 v= xxx kV chart or list. A lot of the time that's all I do. But if you've got a good stock of 1% resistors in the precision resistance ranges... (finer intervals of resistance)... you can probably just sort through the values and get something that very close exact. DVM's internal resistances are also NOT exactly 10M or whatever.

I use a 6-1/2 digit DVM to read all these values, so you see the variations all the time.

I didn't think the Electrostatic Voltmeter was polarity sensitive. Those things work on the reduction of energy principle... and the force is proportional to E^2 , So they actually should be indifferent to polarity and.... to AC or DC. They are almost exactly like a old air-spaced parallel plate tuning capacitor, like those that used to be in every radio... except that the movable plates had geater spacing and were pivoted like a meter movement.

I'll dig mine out and double check.... (Ask your E&M prof..about energy conservation principles applied to this sort of device.... but be prepared for some "extra credit " homework to show its characteristics.)


The potentiometer is also good. Just be sure to use the outer ends of the resistance element for the connections to HV resistor and ground. The DVM goes on the slider and to ground. This way you don't have to worry about slider contacts getting intermittent and routing the HV in strange and wonderful directions.

Dave Cooper

[MSimon]

Current shunt:

Assume your meter has a 200 mV movement and you want to read out 200 mA Max.

A 1 ohm resistor in place of the Amp meter will do the trick. Then just read the voltage across it. A 1/4 W resistor should give you plenty of margin.

Other values could be used if direct reading is not critical.

You can also check for high frequencies by putting an O-Scope across the shunt.

If you want a higher voltage for measurement purposes (2V range) Make your resistor 10X larger.

In this application current shunt voltages up to 20V will not have a serious regulation effect on the power supplied to the fusor.

[MSimon]

Another way to handle read out is to use a 110 Mohm resistor for dropping most of the voltage. Put 1 M ohm in series with that. add 100 K plus 10K in series with that.

Read the voltage across the 100 K. That should put you within 2% with a 10 M meter shunting the 100 K. If you want to get within the error of your resistors add a series resistor of 10K to the string and do your read out across the 10K.

Do the math. It is very interesting.

That kind of voltage divider means you can use standard decade values rather than having to find say a 999 M ohm series resistor plus a 1 M ohm or some such unobtanium.

That is a pretty slick trick I worked out when trying to figure out how to make a divider from standard parts.

[John Futter]

Josh

all the afore mentioned advice is right, but you will need to take some extra precautions around the DMM readout device.
The input to your final readout needs transil diodes across the inputs to supply rails.
This is protect the readout devices is case of a sudden breakdown /arc /whatever that will fool your divider (dV/dT, dI/dT) and input a nasty transient into the readout device.

Transils across the input = rating = to the supply volts for the readout

GDT across the divider min voltage available =90 volts but it is very fast

After the GDT across the divier put aircored inductors to slow the voltage spike for the Transil to react across the readout device input terminals.

may your electronics live long


Or if you can get it a sensitive analog meter with apropriate scaling resistors (with back to back diodes across the movement to protect it)

[MSimon]

John,

Good advice. However the device you want is called a Tranzorb. A least in the USA. They break down in 1E-9 sec or less. If you want to look at RF across the current shunt get some low capacitance devices. You can get devices with shunt capacitance as low as .5pF. Do an RC time constant calculation .16/(shunt resistance * transorb capacitance) to get the 3dB frequency.

A tranzorb in the 5V to 10 V range should be OK - expect peak voltage of around 2X the rated voltage during a portion of the breakdown event. A voltmeter should be able to handle 50V on the .200 V range so your meter should be protected.

[SteveZ]

One minor issue with this design is that with the voltage monitoring circuit on the fusor side of the ampmeter, you will also be reading the current flowing through the resistor divider and voltmeter. While the current will be fairly small, so is the current through the fusor. I'd move the voltage sense circuit across the bridge rectifier to reduce the error factor on tour current sensing. Just a thought.

[AFW]

Josh- I think the transformer secondary will have about half the DC output superimposed on its AC, which might cause breakdown unless the insulation is very good.

Tony webb

[MSimon]

There will be no DC on the secondary.

It is a full wave bridge circuit.

[AFW]

Yes, but suppose all the diodes have the same reverse current: that puts each end of the secondary half-way up a potentiometer connecting the (+) and (-) terminals. Since the (+) terminal is grounded ,wouldn't that put half the output voltage on the winding, superimposed on the AC?

Tony Webb.

[MSimon]

No. The reverse current is small and the winding resistance is relatively low. In addition it is swamped out by the AC current.

[AFW]

I still think I'm right here. I'm not talking about DC flowing through the winding, but a voltage superimposed on the existing waveform. Imagine the winding had a centre-tap, and only half the bridge was connected (the two diodes with their cathodes joined), The junction of the cathodes would come to a (positive) peak voltage with respect to the tap, equal to half the peak voltage across the winding (this is a common full- wave rectifier circuit) . So if this junction were grounded, the tap would come to this voltage (negative) with respect to Earth. If we now add the remaining two diodes, and ignore the tap, we have the regular diode- bridge circuit, giving a (negative) peak voltage equal to the peak voltage across the winding.
The fact that there is no tap is irrelevant. Actually, the insulation will probably stand it, as there is usually leeway in this area, but I thought I ought to mention it.

Tony Webb

[userjjb]

While we're on the subject of the transformer, I have another question.

Pictured below is the transformer that we're using. The other people on our team found it somewhere. I'm not sure if it was pulled out of some equipment or if it was just by itself in the shelves of random stuff in the lab. In either case however they said it wassn't immersed in oil or anything.

My question is whether it should be put in oil or not. I guess it wasn't in it's previous life, but would it be useful? I'm not sure if transformer oil could harm it in anyway or if it's even worth the trouble to have the added complexity of fabbing a tank, getting oil etc.

You can see the full-bridge rectifier in the background as well. I think it'd be a good idea to put that and our voltage divider under oil, so while I'm at it maybe I should just put the transformer in too.

-Josh

[Jon Rosenstiel]

Josh,

You’re also going to need somewhere around 40 kohm of resistive ballast in the HV circuit. If possible, connect your voltage metering divider to the fusor side of the ballast, that way your meter will indicate actual cathode voltage.

I’ve found that both the inductive and resistive ballasts have a large effect on fusor stability… too little of either make the fusor a bear to control. I suggest you initially start out with the MOT’s secondary open, that way you’ll have maximum current limiting in case something goes wrong. (Not unheard of when dealing with HV).

Transformer oil will attack adhesives, so putting it under oil may not be the wise thing to do.

Jon Rosenstiel

[AllenWallace]

I strongly agree using a Transzorb and a current shunt. This keeps the high voltage away from the meters!

Please ask yourself, what is the worst thing which can happen if any one component shorts out or opens up (bad connection).

In your original schematic, several faults would put high voltage into the meter movements which would jump out through the glass into the observer if conditions were right.

My son's fusor used back to back diodes across each meter movement. His circuit used microamp meters, so the voltage drop across each meter was smaller than the 0.6 V drop of the protection diodes.

The idea is to keep the HV away from meters during a fault condition.

[Carl Willis]

Hi Josh,

Look at the side of the secondary coil. Is the coil potted in anything solid? Something like epoxy or wax? If so, you're right that oil is unnecessary. If not, it was meant to be in oil. I can't tell from the photo.

Returning to the metering issue, note that resistors have voltage ratings as well as power ratings. Most garden-variety carbon-comp resistors are rated between 330 and 500 V, and power resistors are seldom designed to handle more than a few kV, regardless of the value and the power rating! When you are building dividers for voltage measurement, you must take this into account or the resistors will flash over or violently disintegrate. Some companies (e.g. Caddock) make HV resistors. If you make your own, get a whole mess o' 1/2 watt carbon composition resistors (like maybe 100 of the 10 M value) and series them up in a ladder configuration on a long piece of perf board or drilled acrylic. At a hobby store like Michaels, look for silver-plated brass spherical beads (1/8", 1/4"). You get a bag with ~20 of them for $2.50. Use these to make solder joints that will hold off HV discharges. Finish the thing by dropping the perf board into an oil-filled pipe section. I have a divider made this way that I have used above 100 kV. You can bridge the resistors with capacitors if you want good high-frequency or pulsed response.

-Carl

[DaveC]

Carl -

That's a real clever idea - using the silver plated brass beads. I've built more than a few HV boards, dividers and multipliers with carefully done "positive meniscus" solder beads... tedious to say the least, and not always perfectly round.

Thanks for that little tip.


Dave Cooper

[MSimon]

I have been designing line powered DC supplies for 50 years as an amateur and a professional.

If there is such an effect I have never seen it mentioned in the literature nor in the design considerations.

Which means the effect is less than 1% if it exists at all. i.e. not significant given normal component tolerances.

There is a back feed effect when you use equalizing resistors and capacitors across a diode string. Normally that is kept to 1% of full supply current (less if possible). Those currents are taken into account.

[MSimon]

When using capacitance across the resistors for HF response you need a variable cap across the Low voltage resistor at the bottom of the string to get the proper HF response.

Adjust the Variable cap so that a square wave into the sting looks like a square wave at the scope.

[bpaddock]

Wallace, Allen wrote:

> The idea is to keep the HV away from meters during a fault condition.

Has anyone considered, or used, optical effects like Brag,
Faraday, Kerr or Pockels to measure high current/voltage? I know
AB&B uses/used Pockles in some of their high-tension switching
gear.

For sake of example, as it would be tough in practice to get this
to work finding the correct fiber cable material etc, you could
wrap some Fiber Optic cable around a high voltage wire. You put
a light source of a known polarization into one end of the fiber,
and the polarization of the light is changed in proportional the
high voltage potential, which is measured on the other end of the
fiber.

There is also the more mundane use of fiber optics where you put a
micro and A/D at the high voltage end, and use fiber or IR/IrDA to link
back to the instrument panel. You don't have to worry about
meter faults turning in to electrocution hazards.

[David D Speck MD]

If plastic optical fiber would work, let me know. I have significant quantities of surplus plastic optical fiber approx 0.004" diameter hundreds of yards long, and limited quantities of glass in short lengths (approx 0.001" up to 10 feet long). Precise chemical composition of the fibers is not known (if this is relevant), but the fiber was intended for illumination applications, not data transmission.

Dave

> Has anyone considered, or used, optical effects like Brag,
> Faraday, Kerr or Pockels to measure high current/voltage? I know
> AB&B uses/used Pockles in some of their high-tension switching
> gear.
>
> For sake of example, as it would be tough in practice to get this
> to work finding the correct fiber cable material etc, you could
> wrap some Fiber Optic cable around a high voltage wire. You put
> a light source of a known polarization into one end of the fiber,
> and the polarization of the light is changed in proportional the
> high voltage potential, which is measured on the other end of the
> fiber.
>
> There is also the more mundane use of fiber optics where you put a
> micro and A/D at the high voltage end, and use fiber or IR/IrDA to link
> back to the instrument panel. You don\\'t have to worry about
> meter faults turning in to electrocution hazards.

[bpaddock]

> Precise chemical composition of the fibers is not known (if this is relevant),

Unfortunately the material does mater. Some of it is quite toxic for the Kerr
type, that is generally why expensive Pockles crystals are used.

> but the fiber was intended for illumination applications, not data transmission.

Still might work A/D isolation type.

[David D Speck MD]

Bob Paddock wrote:
> > Precise chemical composition of the fibers is not known (if this is relevant),
>
> Unfortunately the material does mater. Some of it is quite toxic for the Kerr
> type, that is generally why expensive Pockles crystals are used.
>
> > but the fiber was intended for illumination applications, not data transmission.
>
> Still might work A/D isolation type.

Bob,

If anyone wants to try some, eMail me privately, and I'll send you some. I have a lot of material stashed away, but tomorrow is the last day, as the factory will move to Juarez, MX, to take advantage of lower labor costs. Sigh....

Dave