# Winding transformer for -40KV power supply.

#### Lash432

##### New Member
Hello,
For use of researching plasma and other phenomena in a high vacuum chamber, I am constructing a DC negative hot power supply that will need to supply about -40kv at 20 milliamps. My current plan is to hook a variac up to a HV transformer and then rectifying the output to negative DC voltage.

I will admit, that I am inexperienced in building and working with HV power supplies. The only real experience I have is with making gas lasers, but usually the voltages there are only about 6kv. Safety is my number one priority, and I of course, want to learn as much as I can from doing this project.

With that said, I do not just want to go out and buy some power supply, plug it in, and call it a day, but I want to actually build the power supply so I am learn from it. I was thinking of winding my own transformer (a tedious and patience requiring task yes, but one I believe will be worth it). The catch is I want it to run on 60Hz, and I do not want to deal with any capacitors or CW multipliers (for safety reasons).

I figure that in order to do this I would need an Iron transformer core that is large (to accommodate what will most likely be very large coils), and that is rated for at least 1KVA (from doing some simple math I found the transformer at least needs to be able to withstand .8KVA), and then wind the transformer primary and secondary in a ratio of 10:10,000. This way for every 1 volt in, 1,000 come out. I also assume this transformer would need to be submerged in oil, and have a vacuum put on it to prevent arcing within the coils.

What do you all think? Could it work? Is there anything I am missing or should know?
All feedback, and criticism is welcome!

#### spec

##### Well-Known Member
Hi Lash,

You need -40Kv direct voltage at 20 mA direct current which equals 800 Watts of power.

And you want to generate that power with a transformer with a primary winding of 120V RMS at 60Hz (because you live in the USA).

Assuming that you use a bridge rectifier, the rectification efficiency would be around 70% and the transformer efficiency would be around 60% as a guess.

This means that the overall efficiency would be 42%, so the input power required would be 1905W.

Assuming you could tolerate a ripple voltage of 1% (40V) means that you would need a reservoir capacitor of 0.42 uF minimum with a working voltage of 44KV minimum.

You would also need 4 of 44 KV minimum 20mA minimum rectifiers.

The transformer turns ratio would be 236 to 1 approximately.

The transformer turns per volt would probably be 4 so the primary winding would be 120V * 4 = 480 turns

And the secondary turns would be 480 * 236 = 113,280 turns.

You would need to pay particular attention to the design of the secondary winding to avoid flash over and the transformer would probably need to be varnish impregnated.

My advice would be to wind the secondary in layers with each layer insulated from the other layers.

This transformer would not only be massive but it would also be a massive undertaking.

I hope the above gives you a good feel for what you are proposing.

spec

Last edited:

#### MikeMl

##### Well-Known Member
Nobody builds a 40kV supply this way... for good reason!

Early on, I worked in the semiconductor industry, and worked on some ion-implant accelerators that require 25 to 100kV. All of those used diode-capacitor stacks to generate the high voltage, starting from a transformer that had no more than a 5kV secondary...

#### Lash432

##### New Member
Hi Lash,

You need -40Kv direct voltage at 20 mA direct current which equals 800 Watts of power.

And you want to generate that power with a transformer with a primary winding of 120V RMS at 60Hz (because you live in the USA).

Assuming that you use a bridge rectifier, the rectification efficiency would be around 70% and the transformer efficiency would be around 60% as a guess.

This means that the overall efficiency would be 42%, so the input power required would be 1905W.

Assuming you could tolerate a ripple voltage of 1% (40V) means that you would need a reservoir capacitor of 0.42 uF minimum with a working voltage of 44KV minimum.

You would also need 4 of 44 KV minimum 20mA minimum rectifiers.

The transformer turns ratio would be 236 to 1 approximately.

The transformer turns per volt would probably be 4 so the primary winding would be 120V * 4 = 480 turns

And the secondary turns would be 480 * 236 = 113,280 turns.

You would need to pay particular attention to the design of the secondary winding to avoid flash over and the transformer would probably need to be varnish impregnated.

My advice would be to wind the secondary in layers with each layer insulated from the other layers.

This transformer would not only be massive but it would also be a massive undertaking.

I hope the above gives you a good feel for what you are proposing.

spec
Thank you for your response. I knew the transformer would be large. The only type of transformers that I know of that are actually able to supply this current and voltage are x ray transformers (and most of the old transformers are at least in 200 lb. steel tanks). As for size, is there any way to calculate the core size needed, or do you know of any cores large enough to do this job? Also I take it the core would probably be in an EE style.

With regards to insulating each layer on the secondary, what would I use to insulate it?

#### spec

##### Well-Known Member
No probs Lash
I knew the transformer would be large.
That's good because it will be large.
The only type of transformers that I know of that are actually able to supply this current and voltage are x ray transformers (and most of the old transformers are at least in 200 lb. steel tanks).
The transformer will not weigh as much as 200 lb. Off the top of my head, probably the weight would be around 12 lbs and the transformer would measure, 9" x 9" x 9"
As for size, is there any way to calculate the core size needed.
Yes. I would guess a center core of around 2.5" x 2.5".
or do you know of any cores large enough to do this job?
No but a suitable core will be available. One approach is to canablise an existing high-power transformer
Also I take it the core would probably be in an EE style.
Not sure at this stage. Do you mean an E and I laminated core.
With regards to insulating each layer on the secondary, what would I use to insulate it?
There are special transformer materials: insulating sheets and tapes for the job.

The physical layout of the secondary and the secondary insulation requirements will dominate the size with the transformer you propose.

I was expecting you to reconsider your approach after my post #2. For example, you could half the number of turns on the secondary by using two half-wave rectifiers, one negative and one positive, but the best approach may be Mike's diode stacks mentioned in post #3.

The overriding concern with a 40 KV secondary winding is arcing and, in view of the number of turns, the high output impedance which would degrade the regulation.

Are you dead set on a transformer approach?

spec

Last edited:

#### Lash432

##### New Member
No probs Lash

Are you dead set on a transformer approach?

spec
Well, I was just thinking it would be the simplest. The voltage multipliers using capacitors, from my understanding, require high frequency. Also, i'm not sure that I want to have all those HV capacitors around (I'm not the largest fan of the possibility of lethal amount of power being stored even when i'm not powering the unit). honestly though, I would love a cost effective and simple way to get that amount voltage and current, but the options seem to limited to either using a transformer, or using voltage multipliers... I will admit though, the thought of making a transformer with a coil of 113,260 turns does seem a bit daunting.

#### spec

##### Well-Known Member
Well, I was just thinking it would be the simplest. The voltage multipliers using capacitors, from my understanding, require high frequency.
No, that is not true; a voltage multiplier can be used at any frequency, but the higher the frequency the smaller the components can be. For example a reservoir capacitor may need to be 0.42uf at 120Hz (twice 60Hz) but, at 600Hz, the reservoir capacitor would only need to be 0.042uF for the same ripple voltage. The mass of the transformer is also inversely proportional to frequency, for a perfect transformer, that is.
Also, I'm not sure that I want to have all those HV capacitors around (I'm not the largest fan of the possibility of lethal amount of power being stored even when I'm not powering the unit).
Very wise to take safety into account but, providing the HV generator is designed to comply with national safety standards (not difficult), there will not be a hazard.

Discharge resistors could also be connected across the capacitors if you wished to ensure that the capacitors are discharged when the mains is turned off.

You are bound to have at least one capacitor with all approaches anyway.
I would love a cost effective and simple way to get that amount voltage and current, but the options seem to limited to either using a transformer, or using voltage multipliers. It is not as cut and dried as you think.
One approach would be to make a 10KV mains transformer and multiply that by a four voltage doubler stages. Eight diodes and eight capacitors would be required and the whole voltage multiplier stage could be put in an earthed metal case for safety.
I will admit though, the thought of making a transformer with a coil of 113,260 turns does seem a bit daunting.
Not only would the transformer be a pain to wind, but there would be a high risk that the transformer would not perform as required.

spec

#### Lash432

##### New Member
One approach would be to make a 10KV mains transformer and multiply that by a four voltage doubler stages. Eight diodes and eight capacitors would be required and the whole voltage multiplier stage could be put in an earthed metal case for safety.

spec
Hmm. These are some good points. I think I shall do a bit more research into using multipliers then. Four more questions about this.

1. I am glad you brought up grounding, because this was something else I was wondering about. I plan on using a "star grounding" method. I was planning on doing this by having the chassis of the power supply, the steel vacuum chamber, and all other pumps and electronics grounded (about equidistant) to a screw tapped into a metal plate, that would have a wire then going to the a wall outlet ground. Does this seem acceptable?

2. With voltage multipliers the polarity of the DC current will just depend on which way the diodes and capacitors are wired? (If so I'll probably ask in a different post if it is possible to add a switch to reverse the polarity).

3. I take it since I am planning having this connected to a variac that is then connected to a wall outlet, that I will need to add a fuse and some other measures before the transformer to keep it from drawing too much current? (I know this may seem trivial but I find that when working with specific things like this, sometimes things can get a bit odd or finicky, so I would rather ask than not ask and destroy a breaker box).

4. Also, I know that when measuring the voltage you would take measure it from the main output (by adding something like a 100 Mega ohm resistor and using a regular voltmeter and then have it hooked to the ground). But with current you would take it from where the PSU is grounded correct?

Anyways, thank you very much for responding and being very specific and informative with your answers. I greatly appreciate it!

#### spec

##### Well-Known Member
Hmm. These are some good points. I think I shall do a bit more research into using multipliers then.
Good move. In general, it is best with any new project to consider all approaches, never mind how unlikely, and do paper designs to highlight any benefits/major problems. This applies particularly to challenging projects.

The project you are considering is quite demanding, because of the high voltage and high power required. 20mA does not sound much, but at 40KV it is (800W).

One problem that immediately stands out is the availability/size/cost of the reservoir capacitor. It is important at this early stage to define what ripple voltage you can accept, because the acceptable ripple voltage, along with the frequency, defines the size of the reservoir capacitor. For example, if you double the acceptable ripple voltage, you half the reservoir capacitor value required.
1. I am glad you brought up grounding, because this was something else I was wondering about. I plan on using a "star grounding" method. I was planning on doing this by having the chassis of the power supply, the steel vacuum chamber, and all other pumps and electronics grounded (about equidistant) to a screw tapped into a metal plate, that would have a wire then going to the a wall outlet ground. Does this seem acceptable?
Excellent- a Rolls Royce approach. Grounding conductors, tags, wire etc, should always be substantial to handle any fault currents.

Grounding arrangements should always be mechanically big and reliable. The principle being that if all else fails in a unit, the grounding circuit is always there.

You also need to attach a standard extra high voltage hazard label to the outside of the unit and inside the unit.

Can you describe how this unit will be used, because there may be some safety issues there.
2. With voltage multipliers the polarity of the DC current will just depend on which way the diodes and capacitors are wired? (If so I'll probably ask in a different post if it is possible to add a switch to reverse the polarity).
You can have any polarity you like- it is not an issue. But the high voltage will mean that a large and expensive double pole change-over switch/relay may be required to change polarity. The polarity changing method can be sorted later, when you decide what approach to use and a few trial designs have been considered.
3. I take it since I am planning having this connected to a Variac that is then connected to a wall outlet, that I will need to add a fuse and some other measures before the transformer to keep it from drawing too much current? (I know this may seem trivial but I find that when working with specific things like this, sometimes things can get a bit odd or finicky, so I would rather ask than not ask and destroy a breaker box).
That is correct, and not a big issue.

It is also advisable to have a live/neutral current imbalance trip (standard unit) included in the mains supply path.
4. Also, I know that when measuring the voltage you would take measure it from the main output (by adding something like a 100 Mega ohm resistor and using a regular voltmeter and then have it hooked to the ground).
That is correct. But it would be best to consider having the voltage measuring device built-in. Meters are available quite cheaply ($5US). But with current you would take it from where the PSU is grounded correct? If you are talking about the current consumed by the load, yes. Measuring the current at the earth end would be a very wise approach. Again, best to consider having the current monitoring built in. (about$5US again).

Just one question, what accuracy do you need for the voltage and current measurement. Also do you want digital or analog meters.

Anyways, thank you very much for responding and being very specific and informative with your answers. I greatly appreciate it!
My pleasure Lash. As has been said before on ETO, it is nice to talk to someone who is informative, explicit, prompt and answers our questions.

spec

Last edited:

#### spec

##### Well-Known Member
That's what we have have been taking about Eric.

When I was a learning about electronics, I just could not understand how you could possibly get more voltage out than you put in without using a transformer.

One of the problems with voltage multipliers is the output impedance, which defines the regulation.

The only way, that I know, to reduce the output impedance (and ripple voltage) is to increase the capacitor values, but that gets a bit impractical at high voltages because of the size and cost of the capacitors, or increase the frequency.

spec

Last edited:

#### Les Jones

##### Well-Known Member
Another thing to consider is a way to discharge the energy stores in the capacitors when the power source is switched off. Even constructing a suitable resistor chain for that voltage would need a lot of care. Another possibility would be to construct a special switch that would arc before it finaly closed.

Les.

#### spec

##### Well-Known Member
Hi Lash,

An important parameter that controls the design is the voltage regulation, which means the amount that the output voltage will drop from a zero current load to a 20mA load. Incidentally the ripple voltage will increase from practically zero at zero current load, to maximum at 20mA current load.

Can you specify the values that you need for both regulation and ripple voltage.

spec

Last edited:

#### ericgibbs

##### Well-Known Member
hi L,
As you say, you have no experience in designing or working with HV circuits, I would advise you not to do attempt to construct a 40kV 20mA.
Any mistake you make could prove lethal to you or others.

Contact the Electrical department of a local College or University and ask them for guidance/assistance.

Asking for advice on a on-line Forum is IMHO , is not the way to go.

Eric

#### spec

##### Well-Known Member
Another thing to consider is a way to discharge the energy stores in the capacitors when the power source is switched off. Even constructing a suitable resistor chain for that voltage would need a lot of care. Another possibility would be to construct a special switch that would arc before it finaly closed.

Les.
Yes everything gets difficult, and expensive, with extra high tension (EHT).

I am not sure that the capacitors do need to be discharged though. As far as I know, it is not a safety requirement.

On some of the RADAR systems, for example, they have a safety notice about the danger of charged capacitors and some systems even have a 'debollocking rod', that you could use to discharge capacitors with.

spec

Last edited:

#### KeepItSimpleStupid

##### Well-Known Member
Thank you for your response. I knew the transformer would be large. The only type of transformers that I know of that are actually able to supply this current and voltage are x ray transformers (and most of the old transformers are at least in 200 lb. steel tanks)
Probably not 200 lbs, but they are huge, The one we had was about 2' in diameter and 1.5 feet tall. I don't think it was metal. The oil had to be replaced once. Output of the PS was 100 kV max at 0.1 Amps. Indeed, it used a Variac as the input.
I rebuilt the supply once.

Here http://adammunich.com/x-ray-transformers/ is an interesting article.

Here's a company that still makes them: http://www.mttcorp.us/mtt_xray.html

Last edited:

#### Lash432

##### New Member
Good move. In general, it is best with any new project to consider all approaches, never mind how unlikely, and do paper designs to highlight any benefits/major problems. This applies particularly to challenging projects.

spec
Of Course! I always like to keep an open mind. The transformer project does still sound interesting to me, but it does sound like a large project on its own maybe to be attempted at a later time (the power supply is part of a much larger project, and not necessarily the main focus itself).

That is correct. But it would be best to consider having the voltage measuring device built-in. Meters are available quite cheaply ($5US). If you are talking about the current consumed by the load, yes. Measuring the current at the earth end would be a very wise approach. Again, best to consider having the current monitoring built in. (about$5US again).

Just one question, what accuracy do you need for the voltage and current measurement. Also do you want digital or analog meters.

spec
Buying a self-powered digital voltmeter was the plan, but almost no voltmeter that I know of (besides very specialized multimeters), can handle that voltage to be measured (hence why I would attach the resistor before the voltmeter). As for current, current is the trickiest part to me. Apparently you can do some things with resistors to get the voltmeter to read 1 volt per milliamp, but I am wondering if there is a better way. The plan with the measurement is simply to get the meters to read 1 volt per KV and for current 1 volt per mA.

An important parameter that controls the design is the voltage regulation, which means the amount that the output voltage will drop from a zero current load to a 20mA load. Incidentally the ripple voltage will increase from practically zero at zero current load, to maximum at 20mA current load.

Can you specify the values that you need for both regulation and ripple voltage.

spec
For what it is being used for I do not believe ripple is that large of an issue.

Another thing to consider is a way to discharge the energy stores in the capacitors when the power source is switched off. Even constructing a suitable resistor chain for that voltage would need a lot of care. Another possibility would be to construct a special switch that would arc before it finaly closed.

Les.

If I do use any capacitors I would like to have some bleeder resistors added. Having all that stored energy around is not something I am too comfortable with, for my safety and others.

Can you describe how this unit will be used, because there may be some safety issues there.

spec

The unit is going to be used to supply power to a grid within a high vacuum chamber to create a large enough potential difference between the grid and the chamber to Ionize the gas. I am wanting to look at and research some of the effects of the plasmas of different gases at different voltages and pressures. I understand in order to get a plasma much lower voltages than -40KV are needed. However, I do plan on using this supply and chamber for eventual experiments with such thing as IEC. This power supply also would be something can be used in most other projects I plan to do with HV.

hi L,
As you say, you have no experience in designing or working with HV circuits, I would advise you not to do attempt to construct a 40kV 20mA.
Any mistake you make could prove lethal to you or others.

Contact the Electrical department of a local College or University and ask them for guidance/assistance.

Asking for advice on a on-line Forum is IMHO , is not the way to go.

Eric

I agree, safety for me, and especially those around is the most important factor to consider. I do actually have people who are going to help me this. I have been recently talking to a guy who does work with x-ray machines, and also does a lot of electronics projects. I also have been talking to the dean of engineering at my local university, and he said he is going to put me in touch with some people who can help with certain aspects of the project.
As to asking on this forum, I absolutely was not going to use this as my only recourse, but it does help me learn some information, get advice, and put things into perspective on what I am trying to do, while also introducing new ideas.

#### spec

##### Well-Known Member
Of Course! I always like to keep an open mind. The transformer project does still sound interesting to me, but it does sound like a large project on its own maybe to be attempted at a later time (the power supply is part of a much larger project, and not necessarily the main focus itself).
As I see it, you are taking a very considered approach and gathering all the information you can about the various issues, including safety. This is commendable.

Buying a self-powered digital voltmeter was the plan, but almost no voltmeter that I know of (besides very specialized multimeters), can handle that voltage to be measured (hence why I would attach the resistor before the voltmeter). As for current, current is the trickiest part to me. Apparently you can do some things with resistors to get the voltmeter to read 1 volt per milliamp, but I am wondering if there is a better way. The plan with the measurement is simply to get the meters to read 1 volt per KV and for current 1 volt per mA.
I would recommend buying a couple of digital panel meters, and a small mains to 15V power supply. The 15v power supply line would power all your auxiliary functions, including ditital meters, and the two digital panel meters would indicate voltage and current respectively. Yes, you would use high value resistors to divide the high voltage down. Now that we know what you want, this area is not really a technical or cost driver and can be detailed further down the line.
For what it is being used for I do not believe ripple is that large of an issue.
Would a full wave rectified, 40Kv peak signal be acceptable? If so the reservoir capacitor can be eliminated.
If I do use any capacitors I would like to have some bleeder resistors added. Having all that stored energy around is not something I am too comfortable with, for my safety and others.
Noted
The unit is going to be used to supply power to a grid within a high vacuum chamber to create a large enough potential difference between the grid and the chamber to Ionize the gas. I am wanting to look at and research some of the effects of the plasmas of different gases at different voltages and pressures. I understand in order to get a plasma much lower voltages than -40KV are needed. However, I do plan on using this supply and chamber for eventual experiments with such thing as IEC. This power supply also would be something can be used in most other projects I plan to do with HV.
Thanks for information- sounds like interesting research.

spec

Last edited:

#### spec

##### Well-Known Member
Just a bit of background information to show the main components of a 15KV switching inverter- small cheap & easy: http://www.ebay.co.uk/itm/High-Volt...172975?hash=item2cb7521d2f:g:JIIAAOSwA3dYUmks

Notice the way that the high voltage winding is on a former with separate insulated sections to eliminate flash-over.

The transformer core is made from two 'C' sections which is common on high voltage transformers.

The primary and secondary are also separate which also simplifies the transformer design and construction.

spec

Last edited: