Marx Generator

[Todd L. Peters]

Hello,

Is anyone able to help me with the design of a Marx Generator? I require a
fairly substantial unit to power a laboratory apparatus but am slightly put
off by the cost of commerical units, to say the least..

I wish to build a unit with a variable discharge voltage and electronically
triggered discharge. I understand that the discharge voltage can be varied
by adjusting the spark gap at the trigger. Could the same thing be achieved
by monitoring the potential difference and triggering the unit when the
desired voltage is achieved during charging?

My questions involve formula for the selection of capacitors and resistors,
and what overall needs to go into the selection of them as far as ratings.

I would also need to build a charging system, which seems like the complex
part.

Commercial units that I have seen are immersed in oil. Is this for safety
reasons? What are the advantages of doing this? How is the unit discharged
when it is submerged in oil (I assume that spark gaps can no longer be
used). It would seem to me that corona effects would be much less of an
issue in oil.

I am interested in building a SAFE unit that will allow me to perform
repeated, reproducible, calibrated discharges in a laboratory environment.
Fabrication of the unit in a quality manner is not a problem, it is the
design of the unit to be accurate and safe.

Any information is appreciated.

Sincerely,
Todd L. Peters

 

[John Larkin]

Hello,

Is anyone able to help me with the design of a Marx Generator? I require a
fairly substantial unit to power a laboratory apparatus but am slightly put
off by the cost of commerical units, to say the least..

I wish to build a unit with a variable discharge voltage and electronically
triggered discharge. I understand that the discharge voltage can be varied
by adjusting the spark gap at the trigger. Could the same thing be achieved
by monitoring the potential difference and triggering the unit when the
desired voltage is achieved during charging?

My questions involve formula for the selection of capacitors and resistors,
and what overall needs to go into the selection of them as far as ratings.

I would also need to build a charging system, which seems like the complex
part.

Commercial units that I have seen are immersed in oil. Is this for safety
reasons? What are the advantages of doing this? How is the unit discharged
when it is submerged in oil (I assume that spark gaps can no longer be
used). It would seem to me that corona effects would be much less of an
issue in oil.

I am interested in building a SAFE unit that will allow me to perform
repeated, reproducible, calibrated discharges in a laboratory environment.
Fabrication of the unit in a quality manner is not a problem, it is the
design of the unit to be accurate and safe.

Any information is appreciated.

Sincerely,
Todd L. Peters

Marx generators can vary from little PC board thingies with SOT-23
transistors up to megavolt/megajoule laser-triggered monsters. What
voltage, current, energy, reprate, and pulse shapes do you want? And
how much money do you have? What's the load?

John

 

[gwhite]

Is anyone able to help me with the design of a Marx Generator?

Make a statement like "the free market is best" and all manner of marxists will
come out of the woodwork and generate a loud cacophoney of marxist noise. Did
you really need to ask?

 

[Todd L. Peters]

John,

I'm looking for a discharge adjustable from 500kV to 2MV at a peak current
of 10k Amps. I am not too concerned about the waveform, but the more
uniform the better. The discharge time can be quite fast (or slow,
depending on your perspective), probably something like 10e-5 to 10e-4
seconds. A total energy of 10k J should suffice. I am thinking of a
charging voltage at somewhere between 50k and 100kV, or less, I suppose.

I think something of this size can be built with fairly inexpensive
components, can it not? When I say inexpensive, I mean probably for less
than one thousand dollars in material costs (and I'm not opposed to
scrounging for components). Commercially avialable units seem to be priced
in the vicinity of a nice three-bedroom house, hence my sudden interest in
constructing one for myself.

-Todd

 

[Michael A. Terrell]

Make a statement like "the free market is best" and all manner of marxists will
come out of the woodwork and generate a loud cacophoney of marxist noise. Did
you really need to ask?

"Marxist noise"? Harpo, and his horn! ;-)

 

[Dave VanHorn]

Make a statement like "the free market is best" and all manner of marxists will
come out of the woodwork and generate a loud cacophoney of marxist noise. Did
you really need to ask?

Ah, but will they be able to achieve succesful erection?

(no, not that..)

 

[Mike Harrison]

Hello,

Is anyone able to help me with the design of a Marx Generator? I require a
fairly substantial unit to power a laboratory apparatus but am slightly put
off by the cost of commerical units, to say the least..

I wish to build a unit with a variable discharge voltage and electronically
triggered discharge. I understand that the discharge voltage can be varied
by adjusting the spark gap at the trigger.

You really want a 'manual' trigger (mechanical, HV relay or triggered gap) - relying on the
breakdown voltage of a trigger gap is rather unreliable. When you use forced triggering, the
requirement on the gaps is that they fire at Vin*2 but don't fire at Vin, giving a reasonable input
( and therefore output) voltage adjustment without adjusting gaps.

Could the same thing be achieved
by monitoring the potential difference and triggering the unit when the
desired voltage is achieved during charging?

Yes, you could do this, or have a charging circuit which stops when it reaches the required voltage.

My questions involve formula for the selection of capacitors and resistors,
and what overall needs to go into the selection of them as far as ratings.

Resistor selection is down to the compromise between required charge rate and the extent to which
they drain the output. You might also want to look at inductive charging. Finding suitably rated
resistors at a sensible price can be a major problem.
Of course you will also need to find a source of suitable capacitors.

I would also need to build a charging system, which seems like the complex
part.

Depends how quick you want to charge. A TV flyback based PSU could be used for low charge-rates.
The next step up would probably be something like a neon sign transformer with a voltage multiplier.
For higher input voltages (40kv+) maybe look for surplus X-Ray PSUs.
HV PSUs also appear occasionally on the surplus market - keep an eye on ebay.

Commercial units that I have seen are immersed in oil. Is this for safety
reasons? What are the advantages of doing this? How is the unit discharged
when it is submerged in oil (I assume that spark gaps can no longer be
used). It would seem to me that corona effects would be much less of an
issue in oil.

oil immersion is for insulation and corona prevention - without it things need to be BIG to achieve
enough clearance, and sufficient radius of curvature to reduce corona loss.

I am interested in building a SAFE unit that will allow me to perform
repeated, reproducible, calibrated discharges in a laboratory environment.
Fabrication of the unit in a quality manner is not a problem, it is the
design of the unit to be accurate and safe.

The repeatability requirement measn that corona prevention will probably be very important - I doubt
you will be able to get away with not using oil or encapsulation. Pressurised gas may be an
alternative,

There are many practical issues that you will need to overcome - this will undoubtedly be a learning
experience!

Take a look at my homebrew 500KV marxgen : www.electricstuff.co.uk/marxthree.html

 

[Todd L. Peters]

Mike-

Actually, I had seen your web page already and made me a believer that this
could be done.

I'd like to see adjustable output of between 500kV and 2MV at a peak current
of 10kA. Is perhaps the best way to do this with four 500kV towers that can
be switched in and out?

Long charging times are fine with me (I think you said yours charged within
a few seconds, I don't care if it takes a few minutes) so I could go with
the TV flyback PSU.

What is the first step in design (I really am not an electrical guy)? Do I
have to know the output of the PSU first so that I can select the capacitor
and resistors for the charging rate?

As far as selecting the capacitors, they are limited by the potential
difference they can handle, correct? So if I had a ten stage 1MV unit each
capacitor would have to be rated for 100kV, and then do I just select the
capacitance based on a capacitor network in parrellel or do I need to take
into account the entire RC circuit? Seems like a bit of repetive design to
find the number of stages that work with the available capacitor rating and
storage ability that are available. I did look at Tony Welsh's website and
he seems like a good source for the caps.

Upon reading further, I see from your use of Micro-Cap that selecting the
values for these things properly is not exactly trivial which is perhaps why
I was a bit confused as to where to start.

Where are you getting your resistors? Is there a relatively easy way to
size those?

If I immerse the unit in oil, is there a special oil that's used? Does the
unit still discharge across it's spark gaps even though it's immersed in the
oil?

Inert gas pressurization might be easier and would certainly make less of a
mess of things.

Have you had any concerns with the unit accidentally discharging by arcing
onto itself or yourself or shorting? Or is this completely avoided by
properly grounding the unit?

-Todd

You really want a 'manual' trigger (mechanical, HV relay or triggered gap) - relying on the
breakdown voltage of a trigger gap is rather unreliable. When you use forced triggering, the
requirement on the gaps is that they fire at Vin*2 but don't fire at Vin, giving a reasonable input
( and therefore output) voltage adjustment without adjusting gaps.


Yes, you could do this, or have a charging circuit which stops when it reaches the required voltage.

ratings.

Resistor selection is down to the compromise between required charge rate and the extent to which
they drain the output. You might also want to look at inductive charging. Finding suitably rated
resistors at a sensible price can be a major problem.
Of course you will also need to find a source of suitable capacitors.


Depends how quick you want to charge. A TV flyback based PSU could be used for low charge-rates.
The next step up would probably be something like a neon sign transformer with a voltage multiplier.
For higher input voltages (40kv+) maybe look for surplus X-Ray PSUs.
HV PSUs also appear occasionally on the surplus market - keep an eye on ebay.

oil immersion is for insulation and corona prevention - without it things need to be BIG to achieve
enough clearance, and sufficient radius of curvature to reduce corona loss.

The repeatability requirement measn that corona prevention will probably be very important - I doubt
you will be able to get away with not using oil or encapsulation. Pressurised gas may be an
alternative,

There are many practical issues that you will need to overcome - this will undoubtedly be a learning
experience!

Take a look at my homebrew 500KV marxgen :

www.electricstuff.co.uk/marxthree.html

 

[John Larkin]

Resistor selection is down to the compromise between required charge rate and the extent to which
they drain the output. You might also want to look at inductive charging. Finding suitably rated
resistors at a sensible price can be a major problem.

I saw a biggish (maybe 30 feet high, open-air vertical stack) Marx out
at Livermore a while back. They used laser-triggered spark gaps and,
most cool, water-filled clear plastic hoses as the charging resistors.
Just tweak the ion level in the water to trim resistance!

John

 

[Chris Carlen]

John,

I'm looking for a discharge adjustable from 500kV to 2MV at a peak current
of 10k Amps. I am not too concerned about the waveform, but the more
uniform the better. The discharge time can be quite fast (or slow,
depending on your perspective), probably something like 10e-5 to 10e-4
seconds. A total energy of 10k J should suffice. I am thinking of a
charging voltage at somewhere between 50k and 100kV, or less, I suppose.

I think something of this size can be built with fairly inexpensive
components, can it not? When I say inexpensive, I mean probably for less
than one thousand dollars in material costs (and I'm not opposed to
scrounging for components). Commercially avialable units seem to be priced
in the vicinity of a nice three-bedroom house, hence my sudden interest in
constructing one for myself.

-Todd

You are talking about massive energy levels. Consider this, the Marx
bank shown by Mike Harrison contains about 20 0.0022uF 25kV caps.
That's only 13.8J . I just bought a lot of 95 of the same caps, to make
a larger 1.2MV (2 x 600kV) bipolar Marx bank pair. These cost about
$240, for a total energy of 65J.

These caps were surplus, very rare, and a lucky find. Any set of caps
with the required voltage levels and energies you are talking about, and
able to withstand repeated 10kA pulsing, will be very expensive.

You would be lucky to get just the caps needed for 2MV, 10kJ for about
$5000-$20000. New, a set of 20x 100kV 0.1uF pulse rated caps would be
about $20000 minimum, I'd guess.

Then when you factor in the other components, and the high voltage
design considerations like corona mitigation that would need to be
addressed to make something work at this voltage level with labaratory
grade repeatability, this would be a large undertaking.

The commercial prices don't really sound so bad.


Good day.


--
____________________________________
Christopher R. Carlen
Principal Laser/Optical Technologist
Sandia National Laboratories CA USA
[email protected]

 

[Todd L. Peters]

Chris,

Where are good sources to look for these kind of caps?

What if I drop the energy requirement to a smaller scale and go with
something like 100kV x 25nF caps?

-Todd

 

[Rene Tschaggelar]

Mike-

Actually, I had seen your web page already and made me a believer that this
could be done.

I'd like to see adjustable output of between 500kV and 2MV at a peak current
of 10kA. Is perhaps the best way to do this with four 500kV towers that can
be switched in and out?

Long charging times are fine with me (I think you said yours charged within
a few seconds, I don't care if it takes a few minutes) so I could go with
the TV flyback PSU.

What is the first step in design (I really am not an electrical guy)? Do I
have to know the output of the PSU first so that I can select the capacitor
and resistors for the charging rate?

not_the electric_guy,
are sure you want to risk your pension for some sparks ?

Unfortunately there are these websites such as the one previously
mentioned that let the stuff appear as nice toys.
However this is not the case, ..., you should get the safety
guidelines for working with HV first.

Rene

 

[Robert Baer]

Hello,

Is anyone able to help me with the design of a Marx Generator? I require a
fairly substantial unit to power a laboratory apparatus but am slightly put
off by the cost of commerical units, to say the least..

I wish to build a unit with a variable discharge voltage and electronically
triggered discharge. I understand that the discharge voltage can be varied
by adjusting the spark gap at the trigger. Could the same thing be achieved
by monitoring the potential difference and triggering the unit when the
desired voltage is achieved during charging?

My questions involve formula for the selection of capacitors and resistors,
and what overall needs to go into the selection of them as far as ratings.

I would also need to build a charging system, which seems like the complex
part.

Commercial units that I have seen are immersed in oil. Is this for safety
reasons? What are the advantages of doing this? How is the unit discharged
when it is submerged in oil (I assume that spark gaps can no longer be
used). It would seem to me that corona effects would be much less of an
issue in oil.

I am interested in building a SAFE unit that will allow me to perform
repeated, reproducible, calibrated discharges in a laboratory environment.
Fabrication of the unit in a quality manner is not a problem, it is the
design of the unit to be accurate and safe.

Any information is appreciated.

Sincerely,
Todd L. Peters

The output voltage of a Marx Generator is *not* a function of the
trigger voltage.
Theoretically, the output voltage is N times the input (or charging)
voltage, where N is the number of capacitors that will discharge in
series.
One can calculate the actual output a little closer, by subtracting
about 32 volts for each connecting gap.
Now your idea of monitoring the charging voltage and triggering at the
proper time is a possible "trick" to get a desired output voltage. One
may need a realtively long charge time to achieve good "resolution".
However, then one has the problems of (a) limited repetition rate, (b)
variable repetition rate related to varying "desired" output voltage,
and (c) difficulty of achieving "one-shot" or very low reprate output.

Resistor and capacitor values will be determined by the energies you
want, and the repetition rate you want; and the average power will be
limited by the supply used.
Energy in Joules equals one-half the product of the capacitance tomes
the square of the discahrge voltage.
Capacitors should have a *pulse* (or AC) rating that is at least 20
percent higher than actual stress, resistors have their own voltage
rating, and i suggest you design for no more than 75% of rating across
them. Also, be aware of power rating of the resistors, and displacement
current ratings of the capacitors.

Charging is simple: power supply has series resistor (string of Rs
will be needed for voltage rating considerations) to charge each
capacitor; the resistors from PS to each C and from C to "ground" are
configured to charge them in parallel.
The gaps, when fired, connect the capacitors in series.
To increase the rise time, use a light pipe from the triggering spark
gap to illuminate each and every other spark gap; this could get you
into the nanosecond region (maybe better). The inductance of the
capacitors and wiring will be a limiting factor.

Oil imay be used as an insulator for higher voltage units, it has
nothing to do with safety, and a lot to do with corona from use of large
input voltages.

 

[Robert Baer]

John,

I'm looking for a discharge adjustable from 500kV to 2MV at a peak current
of 10k Amps. I am not too concerned about the waveform, but the more
uniform the better. The discharge time can be quite fast (or slow,
depending on your perspective), probably something like 10e-5 to 10e-4
seconds. A total energy of 10k J should suffice. I am thinking of a
charging voltage at somewhere between 50k and 100kV, or less, I suppose.

I think something of this size can be built with fairly inexpensive
components, can it not? When I say inexpensive, I mean probably for less
than one thousand dollars in material costs (and I'm not opposed to
scrounging for components). Commercially avialable units seem to be priced
in the vicinity of a nice three-bedroom house, hence my sudden interest in
constructing one for myself.

-Todd

For that kind of current, you need very special *LOW* inductance
capacitors and they ain't cheap!!!
You will have to do a *lot* of looking around to find any on the
surplus market, as they aer rare to begin with, and some surplus dealers
may toss them out of ignorance or lack of a market.
And if you "educate" them, the price will go up one or two orders of
magnitude.
With a constant load, the waveform will be rather uniform, as the R, L
and C components will not change too much *IF* the Rs and Cs are not
stressed to their voltage, current, power and pulse ratings. This
assumes air (not iron or ferrite) for the inductors.

 

[Robert Baer]

Make a statement like "the free market is best" and all manner of marxists will
come out of the woodwork and generate a loud cacophoney of marxist noise. Did
you really need to ask?

...since it is a *Marx* generator, maybe that is appropiate????
(just joking)

 

[Robert Baer]

You really want a 'manual' trigger (mechanical, HV relay or triggered gap) - relying on the
breakdown voltage of a trigger gap is rather unreliable. When you use forced triggering, the
requirement on the gaps is that they fire at Vin*2 but don't fire at Vin, giving a reasonable input
( and therefore output) voltage adjustment without adjusting gaps.


Yes, you could do this, or have a charging circuit which stops when it reaches the required voltage.


Resistor selection is down to the compromise between required charge rate and the extent to which
they drain the output. You might also want to look at inductive charging. Finding suitably rated
resistors at a sensible price can be a major problem.
Of course you will also need to find a source of suitable capacitors.


Depends how quick you want to charge. A TV flyback based PSU could be used for low charge-rates.
The next step up would probably be something like a neon sign transformer with a voltage multiplier.
For higher input voltages (40kv+) maybe look for surplus X-Ray PSUs.
HV PSUs also appear occasionally on the surplus market - keep an eye on ebay.


oil immersion is for insulation and corona prevention - without it things need to be BIG to achieve
enough clearance, and sufficient radius of curvature to reduce corona loss.


The repeatability requirement measn that corona prevention will probably be very important - I doubt
you will be able to get away with not using oil or encapsulation. Pressurised gas may be an
alternative,

There are many practical issues that you will need to overcome - this will undoubtedly be a learning
experience!

Take a look at my homebrew 500KV marxgen : www.electricstuff.co.uk/marxthree.html

Trigger gaps can be very reliable; standoff of the charging voltage is
the major requirement, breakdown can be helped by using the light pipes
i mentioned.

 

[Robert Baer]

Mike-

Actually, I had seen your web page already and made me a believer that this
could be done.

I'd like to see adjustable output of between 500kV and 2MV at a peak current
of 10kA. Is perhaps the best way to do this with four 500kV towers that can
be switched in and out?

Long charging times are fine with me (I think you said yours charged within
a few seconds, I don't care if it takes a few minutes) so I could go with
the TV flyback PSU.

What is the first step in design (I really am not an electrical guy)? Do I
have to know the output of the PSU first so that I can select the capacitor
and resistors for the charging rate?

As far as selecting the capacitors, they are limited by the potential
difference they can handle, correct? So if I had a ten stage 1MV unit each
capacitor would have to be rated for 100kV, and then do I just select the
capacitance based on a capacitor network in parrellel or do I need to take
into account the entire RC circuit? Seems like a bit of repetive design to
find the number of stages that work with the available capacitor rating and
storage ability that are available. I did look at Tony Welsh's website and
he seems like a good source for the caps.

Upon reading further, I see from your use of Micro-Cap that selecting the
values for these things properly is not exactly trivial which is perhaps why
I was a bit confused as to where to start.

Where are you getting your resistors? Is there a relatively easy way to
size those?

If I immerse the unit in oil, is there a special oil that's used? Does the
unit still discharge across it's spark gaps even though it's immersed in the
oil?

Inert gas pressurization might be easier and would certainly make less of a
mess of things.

Have you had any concerns with the unit accidentally discharging by arcing
onto itself or yourself or shorting? Or is this completely avoided by
properly grounding the unit?

-Todd

www.electricstuff.co.uk/marxthree.html

I think you would be best served by not using oil.
If you enclose the generator and use argon, the light fromthe spark
gaps will be violet and UV, which may be incrementally better with light
pipes that using air.
But i do not think it would be worth the effort and cost for the small
gain.

 

[Marc H.Popek]

Todd,

We have a 450KV Marx Generator in the EM_Lab@UNLV. The unit had its genesis
@NRL, then to Cornell, and now UNLV. High-quality cast offs

As the Marx Bank proper is where the DC power and the Pulsed power meet, and
we value the longevity of our DC power supplies, the spark gaps in the Marx
Bank are not F@#$%d with to vary the pulsed power voltage. Not a good idea!

The Marx Bank is connected to a pulsed charging line and a Blumlien Pulse
forming network, terminating in a 10^-9 Torr Vacuum Diode. Material&Energy
interactions are observed in the diode and a significant amplitude and
surprisingly crisp (50 nS Pulse Width @ 450KV@60KA = Photons.) radiographic
field is also present. There is a second gap between the Marx Bank and the
PFN, and that where you may vary pulse coupling and thefore the peak
amplitude.

The Oil... Increases the absolute voltage breakdown level between things.
While you are in the charging mode, breakdown can be very bad for your
primary DC power supplies. It never happened to us mind you ;~) The added
bonus is once you size the C.V and set the energy level, you construct the
oil chamber to double as a EO containment vessel as capacitors do fail. One
failure mode is incendiary...

Two things drive a pulsed power system....

The window of performance that defines the energy to the load, (Vp, PW, PRF,
Zl)

Safety

Why not stop by the pulse power lab next time you are in Las Vegas?

Marc

 

[Marc H.Popek]

DVH,

That is the correct terminology! You Erect the Marx Bank!

Marco

 

[Marc H.Popek]

Hello John,

Those water resistors have a twinge of copper sulfate to tune the
resistance. And measuring a water resistor is weird, as they act like they
are alive from the water's polar activity.

Marc