0 - 1000V power supply project

[mysticalchemist]

I've finally been able to set up my physical workbench again after a pause of five years, and have quite a backlog of projects to work on. One future project is to build a 0-1000V DC power supply for trying some of the ideas presented by "Spherics" and "Grumpy".
I've considered buying a used one of Ebay, but anything reasonable seems to be pretty expensive, and very expensive to freight to Australia. Also you miss out on the fun of actually building it yourself if you just buy one.

I've set my minimum spec at 0- 1000V, 50mA. (1100V, 100mA is the ideal goal if it works out that way.) and preferably an all-linear supply.

For the regulator section, I plan to use a floating 723 or similar circuit, but the real problem so far is the input stage. High voltage transformers are quite scarce these days, so I've had to look at other options using easy to get components, like back to back common low voltage mains transformers.

Option 1./  Use 2 back to back transformers, say 230V to 50V then 50V to 230V, followed by a voltage quadrupler stage to give around 1200VDC   (not my favorite method due to the very poor regulation out of the multiplier stage, but only uses 2 large transformers.)

Option 2./ Use 4 pairs of back to back transformer say 230V to 30V then 30V to 230V, followed by a full wave bridge, and the four stages operate in series to give around 1280VDC. (This is currently my favorite. The 8 small transformers cost very little, but do come with quite a high loss budget. All the parts are very common and easy to get. It offers the possibility of fine tuning the final output voltage by changing the low voltage taps on one of the transformer pairs.)

Option 3./Use one large 230V to 30V transfomer, and 4 smaller 30V to 230V feeding seperate bridge rectifiers as above. (Down to 5 transformers, and a reduction in transformer losses, but loose the ability to fine tune the output level.)

Option 4./ Strip the secondary off a standard 230 to 30V transformer and rewind it to suit. (It's tempting to go down this path, and if it was not such a high voltage, it is the way I would first choose, but I'm not so keen to wind to the voltage level required.)

Option 5./ Find a used HV transformer and build the unit around it.  (Not favoured as an old transformer could fail and then it becomes a real chore to find or wind a replacement.)

Option 6./ Give up trying to keep it all analog and build a switching regulator.

I'll keep working on the options while searching for suitable parts, hoping to get started with the actual construction in a couple of weeks.

Peter

[Grumpy]

I use a little flyback dc-dc converter and it can be varied from about 2kv to 4kv and around 20ma.  Then I use several hv diode bridge and several hv caps.  I have another dc-dc converter that is variable 0 to 1kv, but it is only 1ma. 

[mysticalchemist]

Thanks for that info Grumpy. 
Early this morning I just bought a partly faulty Fluke 412B High Voltage Power Supply cheaply from ebay, so when I fix it I'll have 0-2000V from that.
Still going to continue with making one myself though.

Next step is a HV pulse source.   

[Peterae]

I had a pretty cheap build going here
http://www.overunityresearch.com/index.php?topic=307.msg3878#msg3878

and further down the thread it works 
http://www.overunityresearch.com/index.php?topic=307.msg4025#msg4025

[ion]

Don't discount buying a used electrophoresis power supply on ebay. They can be had quite cheaply.

Most don't know that these are well designed power supplies at a fraction of the cost of a "normal" supply.

The word "electrophoresis" throws a lot of people off this good purchase.

Lots of bang for the buck, and lots of nice features. Very robust supplies with massive transformers, SCR regulated drive. Constant voltage, constant current or constant power all adjustable with nice metering.

e.g. http://www.ebay.com/itm/BRL-4000-DC-electrophoresis-POWER-SUPPLY-4000V-200W-200MA-/251097352347?pt=LH_DefaultDomain_0&hash=item3a7691889b

I have about  eight various types in my lab, wouldn't part with any of them.

While building your own power supply can be fun and a good learning experience (I've built many in the past), to build a really good one is time consuming and I'd rather spend my time doing research.

Don't discount using an MOT / Bridge / Capacitor and a Variac on the input as a nice variable HT supply. Use resistors on the output to limit peak current.

[Grumpy]

The supplies Ion mentions are great because you can vary the current and the voltage and both are controlled.

Here are some of the DC-DC converters I mentioned:

http://www.ebay.com/sch/i.html?_nkw=high+voltage+emco&_sacat=0&_odkw=high+voltage+supply&_osacat=0

I checked ebay in AU, but only saw a 500v unit and smaller 1.5w units.  I like the 10w proportional ones.

You won't be sorry with any of them, but I recommend a minimum of 1500v, but you only need a few ma.

[mysticalchemist]

OK. I've bitten the bullet and bought an EC600. It will be quite a few weeks before it arrives, and I'll still have to sort out the mains power supply, so now I can concentrate on building a pulser stage.

Peter

[Grumpy]

OK. I've bitten the bullet and bought an EC600. It will be quite a few weeks before it arrives, and I'll still have to sort out the mains power supply, so now I can concentrate on building a pulser stage.

Peter

I have an EC600 and and EC650 (and an older EC600 that is for parts).  They still scare the Hell out of me when you hear that click and power is applied - I think of old Frankenstein movies!  Voltage and current are controlled but they can deliver a lot more.

Glad to see someone else interested in pulsers.  They are the only way to ride!

Since you have the EC600, and 4kv at your disposal, what voltage do you want to pulse at?

[mysticalchemist]

Since you have the EC600, and 4kv at your disposal, what voltage do you want to pulse at?

That's a real good question. Originally I was just looking at trying 1kV to get a feel of what happens, but now with more voltage, I'm not sure.

1.5kV seems a good point as far as component selection goes, as I've noted that 1.2kV seems the min for positive results.

I'm certainly open to suggestion as to where to aim for, and what circuits to use, but at the moment I intend to try a 1000V avalanche transistor circuit using 2N5551's to get a feel for things and to practice measuring around the circuit.

I've got an old HP  8013B  pulse gen coming as a startup drive source.

Peter

[Grumpy]

I posted several papers on avalanche transistors here somewhere - it's been a while...

Order a bunch of 2N5551's if you can.  This way they will be of the same lot and same properties for the most part.  You can expect an attrition rate of about 10%.  I test each one individually for avalanche properties and look at the output on a scope.  I also fire a neon and see how bright it is.  If the output is attenuated, noisy, if they go into Schotky conduction mode, or if the neon is not bright, then I reject them and do not use them in a pulser stack.

I keep the current as low as I can too.  They are easily heated to failure when they avalanche.

You have to put resistors across each one in the stack to distribute the voltage.  I also use a 100pf HV cap across each one to make sure it avalanches.

Once you start with these, then try others and high voltage ones.  I have some that avalanche at about 1200v - sweet!  I have tried all kinds to vary degrees of success. 

Diodes are another avenue of investigation and some of them make a damn sweet pulse.  Again, you must distribute the voltage or else the first one fires and then they all fire.  Distribute according to the voltage percentage if the avalanche voltages are different when mixing devices.

[mysticalchemist]

I posted several papers on avalanche transistors here somewhere - it's been a while...

Order a bunch of 2N5551's if you can.

I have had some of your circuits on my disk for a while now. I got in a batch of 2N5551's a fair while back ready, and do plan on making that little test rig to check each one. I'll try to get that going on the weekend to make a start.

I can remember a photo you posted way back of one of your switches, with all those blue caps. (Havn't been able to find the pic lately though)

How do you find those blue ceramic caps on pulse?   I used to replace a lot of them fried in Tv horivontal outputs, and don't have a high regard of them now. I know they are cheap and easy to get though.

What sort of resistor value do you use across the transisitors -- 1M ?


Peter

[Grumpy]

1 meg resistors, but I have seen a few pulser circutis with 2 meg resistors.

The caps are 1 kv or 2kv 100 picofarad (I'll verify)

Some links here:

Avalanche Transistors:
http://www.overunityresearch.com/index.php?topic=378.0

[Grumpy]

EC600 Users Manual attached.


There is a minimum amount of current to keep it latched, but I don't recall what it is.  I think it is 1mA, but I might be totally wrong.

You will need to use a 100k to 1 meg resistor across the output for it to latch on (solder to banana plug in second socket - thanks Ion!)  Once it latches on (Frankenstein Mode) it will stay on until the resistance exceeds 150 meg ohms, so it will stay on as long as your resistor is there or you turn it off. 

Use a HV Resistor.

[ion]

mysticalchemist:

EC Apparatus privately labelled their units under a few other names such as BRL , American BioNuclear, Integrated Separation Systems, Thermo, Fisher, VWR Scientific and others.

The link I supplied earlier is definitely an EC Apparatus unit.

EC Apparatus units  use large transformers designed for continuous duty over days and SCR regulation circuitry.

They are heavy units compared to the newer switchmode designs, but are "bullet proof".

Which unit did you buy?

Interestingly a lot of folk that sell these units on ebay sell them "as is" because they do not know the trick of loading the output with the resistor before it can pull in the main output relay. A digital meter often has too high an input resistance to pull it in so they think they are non working units.


I bought a nice smaller unit sold "for parts" for $40 that worked perfectly upon arrival. Their lack of knowledge is our gain.

Please note that these units will require an isolation diode on the output if you have a capacitive load  or battery connected as they do not like to see any stored energy (a safety feature) and will not turn on under that condition.


I prefer the analog metered units as opposed to digital readout (which can bounce around a lot).

Here is nice 1000 Volt unit, if the price doesn't go too high.

http://www.ebay.com/itm/E-C-EC-Apparatus-EC-400-Electrophoresis-Power-Supply-/230836957049?pt=LH_DefaultDomain_0&hash=item35bef46f79

[tao]

ION, Grumpy...

I just got two of these, and I must say, LOL, it does make a 'frankenstein' type loud movie-like large-contraption startup noise, you have to hear it to believe it folks, ION and Grumpy know ROFL!

Thanks for the comments on relay latching, as I thought too that the units might be busted when I got them, but threw in a 4.7M Ohm for a quick test and both stay on steady .

[giantkiller]

Wow...

I have a varistor / mot supply to 2kv.
I also have a EC600. Click! It's alive!

@mystical-dude, Did you get your ec600 yet?