Multi-kilovolt DC power supply... Where to start?

[jimdeane]

I need a H.v. power supply for some plasma physics experiments and demos. It needs to have an output range of at least 50v-2000v, at a max of about 10ma. I would prefer to be axle to adjust voltage and have an adjustable current limit.

I've thought about using a Cockroft-Walton H.v. amplifier, but I'm not sure how to calculate maximum current capability of the system.  I also don't know how stable the output is, I might want to run at a threshold value and wouldn't want the voltage to be varying widely with time. This would potentially run at maximum voltage and current indefinitely.

If you could point me in a direction, I'm eager to learn. If I'm barking up the wrong tree with the C-W idea, please let me know.

[Berni]

What you probably want is a neon sign transformer plugged in to a variac. They will give you up to 10 to 15kV and are designed to current limit them self to about 5 to 50mA even when shorted out. This is a good solution if your priority is current limiting since the output voltage of these transformers will not be very stable when the load on it changes. They are designed to do that so that they can provide a fairly constant current at a wide range of voltages.

If you really want a proper lab like supply where you can exactly dial in the current and voltage and have it regulate it within under 1% then you probably need to build your own switchmode converter. Wind a high voltage coil with a few 1000 turns and put it on a ferite core then drive a 10 turn coil using a mosfet thats driven by PWM from some regulator circuit that monitors the output voltage and current.

But whatever you do be careful with these voltages, introducing capacitors to them CAN make them lethal!

[mikeselectricstuff]

NSTs aren't ideal as they are centre-grounded, though you could use a half-wave rectifier.
They are also not designed for continuous operation at their open-circuit voltage, and insulation can break down over time.

High frequency is the way to go so you don't need big caps for smoothing. e.g. flyback converter, maybe with a CW doubler on the output.

For current limiting, do it on the primary side - much easier! 

[mikeselectricstuff]

Take a look on ebay for  electrophoresis power supplies - these are typically adjustable for voltage, current and power and in the sort of range you are looking for.

[Marco]

This is near the range of PMT power supplies (slightly higher power). There's a lot of designs for them using a linear regulator to modulate the input voltage of a CCFL inverter with a subsequent multiplier.

PS. for primary side current limiting you might need some inductance on the secondary side so it doesn't arc before it kicks in.

[eneuro]

For current limiting, do it on the primary side - much easier!

Yep, I've made some research and looked for ignition coil datasheets, where we can find that output voltages 20kV and ~1:100 ratio between transformer primary (200 turns or so) and secondary-it outputs from 12V battery current still dangereous at the levels of 100mA.
Not sure but maybe they have builtin HV diodes to dump oscilations and allow high frequency ignition, eg. in a car with engine speed 3000 RPM?


I wonder, if we provide lets say 100mA input 12V high frequency to such ignition car coil primary, than we should get from turn ratio ~1mA HV output current?

The same what about reusing MOT transformer with the same precautions-very small primary current, so at 1:100 ratio 100mA primary should be quite safe (unless we have HV capacitor which can accumulate those small currents and create dangereus current levels HV) , since we get only 1mA at output in the case of good magnetic coupling?

[MagicSmoker]

Take a look on ebay for  electrophoresis power supplies - these are typically adjustable for voltage, current and power and in the sort of range you are looking for.

I strongly second this recommendation. Definitely the best/cheapest/easiest route to go. I bought a Fisher "FB 600" off of ebay a few years ago for around $100 while I was looking for something else because I could resist getting a 4kV/200mA max CC/CV/CP power supply for so cheap (NB - looks like they are going for $150 these days - still a bargain)! That said, I have actually used it to non-destructively test the breakdown/avalanche voltage of abused but seemingly-still-functional power semiconductors, so a win all the way around.

[helius]

Are electrophoresis HV leads a standard item? I tried searching for them and didn't get anywhere.

[vindoline]

Occasionally used HV Power Designs supplies come up on eBay at reasonable prices. I have two 3K10B's that I got for under $100 shipped. They ooooze quality.  These are high end lab grade supplies typically used in nuclear physics labs. If you see one with a high price, try a lowball offer. It worked for me

[vindoline]

This is near the range of PMT power supplies (slightly higher power). There's a lot of designs for them using a linear regulator to modulate the input voltage of a CCFL inverter with a subsequent multiplier.

This is an excellent low cost solution. For example I've tested the attached circuit and it works well for it's intended use, powering photomultiplier tubes. I don't know if it will source 10 mA or not, DIY PMT setups are generally pretty low current - under 1 mA. I'm pretty sure most (or all) of the DIY Geiger counter HV supplies will NOT be able to source the amount of current you need. Most Geiger tubes use uA of current.

One thing to watch out for in DIY PMT and Geiger supplies is that you can't measure the HV with a regular DMM. The 10 Meg input impedance is much too low in this case and the meter will load down the circuit and give a false low reading. Have fun!

[MagicSmoker]

Are electrophoresis HV leads a standard item? I tried searching for them and didn't get anywhere.

The FB 600 uses standard female banana jacks for the output.

[Marco]

Yep, I've made some research and looked for ignition coil datasheets

Only problem is that they are autotransformers, better to use CRT flybacks for the isolation (as actual flybacks, not for a Royer with the multiplier diodes removed/blown).

[eneuro]

Yep, I've made some research and looked for ignition coil datasheets

Only problem is that they are autotransformers, better to use CRT flybacks for the isolation (as actual flybacks, not for a Royer with the multiplier diodes removed/blown).

I've already salvaged one (FB-694) from Mag CRT monitor or something older (not sure) and trying to figureout howto connect this thing to eg. ZVS flyback driver ?


What is purpose of those three wires turned 2-3 times over bead ferrite and this black one turn over FB main core I guess?


Bottom PCB looks like this (I didn't desoldered it yet).


Tried find any connection from bottom PCB ground to this main HV thick red wire, but noway-maybe didn't broke PCB bottom insulaltion with multimeter probe

However, it looks like it is quite tricky to find flyback secondary ground and maybe this method will work from video below?
Flyback Driver Completely From Television Tutorial

Update: Probbaly more complete video and... a few cm nice blue plasma 


Google says nothing about FB-694-we'll see after desoldering this thing from PCB, however there is GND1 description on bottom PCb shown above, so maybe only more tricky way needed to verify this FB secondary GND 

[mikeselectricstuff]

Are electrophoresis HV leads a standard item? I tried searching for them and didn't get anywhere.

The FB 600 uses standard female banana jacks for the output.

some do have interlocks - I once spent an afternoon fautltfinding one only to find the plug I used wasn't quite the right shape to  enable the interlock on the socket.
And BTW don't forget
THESE THINGS CAN KILL YOU

[eneuro]

And BTW don't forget
THESE THINGS CAN KILL YOU

It looks like Marx generators with choosen very small capacitors (pF ?) and a few kV PSU limited to lets say 0.1mA very low capacity output might be an option for HV experiments, when we know what we are doing, since we can get huge output voltage, adjust switchng frequency (output current? ) but when we switch off power source than the only power available during discharge will be from energy stored in those those tiny capacitors (this can be calculated), but we have to remember that they are at high voltage, so how to figure out what can be safe capacitance at lets say 10kV-100kV?

Maybe academic experiments could help... by estiamting amount of charge stored onVan de Graaff generator metal balls like in those high school experiments?
Electrostatic Potential and Electric Energy | 8.02 Electricity and Magnetism (Walter Lewin)

Or something else-use human body capacitance as some estiamtion of capacitance acceptable, but of course not for very HV...

Yep, it is quite complicated stuff, but human body is charged and discharged when we touch door many times and sparks happen, so... it is a lot of kV for sure....

Maybe @T3sl4co1l have some interesting theory behind this, howto estimate those HV caps capacitance to get safe discharge levels when someone touches such Marx generator output by accident? 
What about assume safe energy levels, then calculate capacitance at given HV from 1/2*C*V^2, so when we connect in bad case sceanario where human body at 300 Ohm resistance shorts such HV cap, claculate time in which it will discharge those HV caps, so we can calculate discharge currents and time 
Than recalculate for better case 1000 Ohm resistance.

Kitchen ignition toys are also available.... with decent sparks... maybe teardown of those things could be interesting, while they at least should be designed to do not kill anyone?

[mikeselectricstuff]

Marx gens are only good for producing impulses

[jimdeane]

Some very good ideas here, thanks. I hadn't thought of electrophoresis supplies, I'll have to see what comes up. The cheapest new commercial supply I found is from a British company and would be around $300. Since the grant funding dried up, I'm trying to do this on a much more limited budget.

[eneuro]

I mean that Marx generators can be maybe easier to control-when we stop running input AC, we can estimate amount of energy, so we can store in bigger output capacitor calculated amount of energy based on how many AC pulses were allowed from last discharge.
So, many impulses at high frequency stored in output capacitor can be discharged later, so it can be some kind of HV PSU... and dangerous too, when many small pulses will be stored in bigger HV capacitor for longer period of time...

Anyway wiki says: https://en.wikipedia.org/wiki/Body_capacitance

The Human Body Model for capacitance, as defined by the Electrostatic Discharge Association (ESDA) is a 100pF capacitor in series with a 1.5k? resistor.

Of course it will depend on environment, but.. lets try calculate for Marx generator 10 stages of  100pF each at 10kV  and what energy levels  we get at 100kV or better put it into spice? 

[Someone]

when we switch off power source than the only power available during discharge will be from energy stored in those those tiny capacitors (this can be calculated), but we have to remember that they are at high voltage, so how to figure out what can be safe capacitance at lets say 10kV-100kV?

The various international 61010 standards have limits for safe capacitance in high voltage equipment, just the connected cabling can make a system dangerous.

[dexters_lab]

Some very good ideas here, thanks. I hadn't thought of electrophoresis supplies, I'll have to see what comes up. The cheapest new commercial supply I found is from a British company and would be around $300. Since the grant funding dried up, I'm trying to do this on a much more limited budget.

there should be plenty on ebay, they wont always be listed as electrophoresis supplies though, better to search for the brand names and the model 'range' name.

There is a recyclers near me that regulary has them for sale and usually sell for less then £50, i would have bought one by now but i am trying to finish some projects at the moment rather than starting new ones! 

[Marco]

I've already salvaged one (FB-694) from Mag CRT monitor or something older (not sure) and trying to figureout howto connect this thing to eg. ZVS flyback driver ?

Royer ZVS is good at getting a lot of Watts out of it, which is why the people who like big arcs use it.

For a low power regulated supply I'd use actual flyback though, can be regulated much easier, has less stored energy and is less likely to kill you.

[eneuro]

I've already salvaged one (FB-694) from Mag CRT monitor or something older (not sure) and trying to figureout howto connect this thing to eg. ZVS flyback driver ?

Royer ZVS is good at getting a lot of Watts out of it, which is why the people who like big arcs use it.

I've simulated in simple circuit simulator one of such ZVS drivers used by folks playing with plasma generated using FBs  and I've already did my custom version of this thing-soldering today etched double layer PCB  shown in other thread

Now desoldered PCB from this CRT monitor FB-694 and trying to figureout pinout,which is not easy, since when probed with multimeter it doesn't fit to descriptions from YT videos -I've resistance below 1 Ohm on all connected pins and I've no idea what primary  or secondary could be 


Anyway, I think, I will do it in different way: I will add 10 turns around flyback core and connect 100kHz-400kHz 5VAC at low current ~100mAmax and I will solder two LEDs in antiparallel configuration to HV wires salvaged from this thing (main red one is rated 40KVDC based on description on it), to short output current, so I think it could be good for testing  HV output current-if such trafo will  those leds due to too high current it will mean it is not safe anymore.


Thinking only which can be minimum If current at which typical 3mm green/yellow LED will  fail, but it is easy to test it  with variable PSU and resistor

Datasheets says about 20mA-30mA typical If, but probably no chance to survive 100mA or even less 50mA or so for lets say 1 hour test?

This will not be of course any safety testing tool, but just estiamtion what current level is present in transformer secondary AC (HV) output without touching anything or probing with HV scope etc.-just see if anti-parelled LEDs fail or not while it clamps voltage and short-circuit

BTW: My plan is reuse this quite nice FB ferrite core to make custom transformer with a few turns on primary and thick secondary to provide HV high frequency low current component to help ignite arc in transformer welder, so testing those FB pins is only for fun just to see if those 3mm LEDs will surrvive HV assuming they will clamp this HV AC voltage in both directions at low current-it shoule be below 10mA from tranforemer estimated ratios 

[Morgoroth]

old tv and microwave transformers works well to start,  the first one comes with diode,  anyway to control the volage level you must control the charge in primary inductance,  and thats mean pwm at 50% at some frecuency with some mosfet and cheap arduino.

the simulation in ltspice is ready in 5 mins and show what I'm talking about.

carefull with this things,  take away all yours rf stuffs,  I burn my expensive bluetooth keyboard playing with this.

[eneuro]

anyway to control the volage level you must control the charge in primary inductance,  and thats mean pwm at 50% at some frecuency with some mosfet and cheap arduino.

ATTiny85 probably cheaper than arduino 
However, it doesn't matter if we use push-pull or full-wave DC-DC, but did you used FB oryginal primary windings or custom and what was switching frequency?
I have no data about this FB core ferromagnetic properties, so even if I know its dimensions, no way to calculate any flux, etc 

I've added 10T as shown below on FB core and driven this thing with ~4VAC at 400kHz using low current h-bridge via 54 Ohm series resistance (2 x 27R) and 2x 100nF ceramic caps.


No sparks, so far but using my HV LEDs tester shown above it looks like pins 1-2 are shorted (no light from LEDs) and the same for 7-8, but they have resistance close to short-circuit.
LEDs had ligth maximum between pins 3-6, and the same but lower light beetween (1-2)-3 and (1-2)-6, so it looks like it might FB primary coils.

NOTE: I've connected now this "HV tester" to my full-wave ~4VAC "SMPS" directly (2x27R and 2x100nF) in series and... of course those LEDs @ 400kHz are not  burned, so a very little input current it is-theya not even bright-similar light was on FB primary coil 

The only way to get any light between main red thick HV wire (0) were between (1-2) pins and light was significant low (probably less than 1mA equivalent-no supprise if there were 1:100 or more ratio between my 10T and this FB HV output coil), so it looks like pins (1-2) might be ground, while on PCB there were screws visible, however.... no sparks of course due to very limited input current and not great magnetic coupling of this added 10T to FB core 

BTW: I've tested the same LEds tester with random ferrite core with similar 8T primary and ~20T secondary and LEDs output similar not too much light @ 400kHz using such current limited "SMPS", so probably will try also connect now to FB internal primary (to pins 3-6 where I've got greatest light response), but maybe someone knows those FB has splited this internal primary to be able drive it in push-pull mode during normal operation in CRT monitor/TV ? 

[bktemp]

Look at the PCB where the big transistor originally was connected. From the spacing between the pins my guess is pin 1 and 2 is the primary.

If you are wondering about the multiple connections to the anode, here is a typical schematic of a flyback transformer:
http://www.hrdiemen.com/reparation/flyback/scheme/8951
Depending on the actual type the capacitor and the resistors can have seperate pins or all share the same gnd pin.
The white and gray wire are probably F1 and F2.
If the resistor has a seperate pin, it can be used for a voltage divider for feedback.