Electrolytic capacitors for voltage multiplier / High Voltage / Forums

Forums

4hv.org :: Forums :: High Voltage

« Previous topic | Next topic »

Electrolytic capacitors for voltage multiplier

1 2 next

Move Thread

LAN_403

Flachzange

Fri Feb 15 2019, 07:09PM

Registered Member #61569 Joined: Sat Apr 15 2017, 05:12PM
Location: Germany
Posts: 28

Hello high voltage community.
We are planning on the construction of a voltage multiplier for metrology purposes.
The output current is in the order of a few mA. At most a maximum of 5mA.
The overall output voltage should be 1000 kV and if every thing works fine we aim for 2 MV.
The most critical task is the ripple. We could buy a similar power supply for less than 500.000 â‚¬ but since they all have a ripple factor of < 3% and best 1% it is not suitable for high voltage metrology. Our aim is 1*10^-5, so 10ppm. Furthermore the stability is dependent on the main power grid since no active voltage regulation is available for systems < 500kV. High frequency, regulated DC power supplies with these requirements are simply not available above 500kV. Below < 300 kV we have good experience with Heinzinger, FuG and Glassman.
So our plan is to build the power supply by our selfes with the following requirements:
- Voltage at least 1000 kV (maybe 2000 kV)
- output current 5 mA
- ripple leass than 10ppm
- stability over one hour better than 100 ppm
- modular design
- number of stages (for 1000 kV) 5, so 200 kV per capacitor and diode
- height for 1MV: 5 m
- we would like to go with full wave rectification but for the modular design it would incorporate too many additional pieces, so half wave rectification is the way to go

So for the main question:
For the smaller voltage multipliers we used cornell dubilier 3000 V 100 nF capacitors with 33 MOhm in parallel.
However for this multiplier we would like to go with a higher capacitance which would easily add up to 1000 capacitors per stage to achieve the low ripple.
So we thought about electrolytic capacitors but we have not found any references where they are used for voltage multipliers.
Has anyone used them so far for high voltage applications?
The datasheet of most electrolytics mention that they start to leak close to the nominal voltage, but with at least 20 % safety margin that would not be a problem.

The transformer is almost finished and it produces 100 kV peak at 25 kHz. At these switching frequencies the ESR of the electrolytics should not be a problem.

The main problem will be the discharge and safety.
For instance if we would use 450 V 470 ÂµF capacitors:
For one stage (200 kV [247 kV as a safety margin]) we estimate 550 capacitors
--> total capacitance 854 nF
With a parallel resistor of 2 GOhm (0.1 mA) to equalize the capacitors we get a time constant of 1700 s!
Even if we use 200 MOhm for the resistance it takes almost one hour before the multiplier is fully discharged. Also the stored energy within one capacitor is 17 kJ!
A discharge band which gets cranked through the multiplier discharging everything is also not acceptable since the discharges might harm the sensitive measurement equipment (we have already killed too many 3458A).
So we dont see any easy way to reduce the voltage in a fast way.

If you have any suggestions please feel free to comment.

Patrick

Sat Feb 16 2019, 05:08AM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

I think you are asking alot.

Flachzange

Sat Feb 16 2019, 08:33AM

Registered Member #61569 Joined: Sat Apr 15 2017, 05:12PM
Location: Germany
Posts: 28

Yeah, i know but i jst wanted to make sure i listed all the details.

The main question is if there is something to consider when using electrolytic caps other than the ESR and the polarity.

Sulaiman

Sat Feb 16 2019, 10:37AM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

I don't know but I suspect that electrolytic capacitors may be unsuitable for long term oil immersion,
or at least the oil and/or capacitors should be chosen carefully.

I once built a voltage multiplier with ceramic disk capacitors on the input/pumping side and electrolytics on the output side,
(e.g. C1, C3 = ceramic, C2, C4 = electrolytic in the first schematic

)
the small energy transfer per cycle of the ceramic caps is absorbed by the electrolytics to give low ripple.
The leakage current of electrolytics could be a problem in very low power multipliers,
I used a TVS across each electrolytic rather than a resistor to ensure voltage distribution / no over voltage on any capacitor.
I never did work out a quick safe method of discharging the capacitors on power down - that did not consume excessive power during operation.

jpsmith123

Sun Feb 17 2019, 01:34AM

Registered Member #1321 Joined: Sat Feb 16 2008, 03:22AM
Location:
Posts: 843

What about if you were use a "dynamitron" (parallel fed) type HV power supply design? Maybe it would be easier, cheaper and less technically risky.

Flachzange

Sun Feb 17 2019, 08:58AM

Registered Member #61569 Joined: Sat Apr 15 2017, 05:12PM
Location: Germany
Posts: 28

Good idea about the Dynamitron, but since they require a pressure vessel it is not possible to use it, since the high voltage bushing or a cable for these voltage levels would be a lot more expensive than the multiplier itself, if they are even available.
Unfortunatly 99 % of the calibrations are done with direct high voltage connection by big tubes and not HVDC cables. This is the reason why we can not used a dead tank design.

Sulaiman

Sun Feb 17 2019, 10:55AM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

I consider one of your challenges to be voltage measurement and regulation as I doubt that you could rely on the multiplication factor remaining constant with a variable or varying load.
So no matter what method of generation is used, a reliable voltage measurement will be required.
Chains of resistive elements, field mill, etc.

When I was working at Harwell in the '70's there was a tall slim building housing a very large VDG generator powering a linear accelerator that produced >1MeV particles.(>2MeV using a little 'trickery' I was informed)
So >1 MVdc is certainly do-able, I guess that in internet search would produce even higher voltage VDGs
A bit old-school but possibly an economically viable route ?

Flachzange

Sun Feb 17 2019, 11:27AM

Registered Member #61569 Joined: Sat Apr 15 2017, 05:12PM
Location: Germany
Posts: 28

Well, we have a modular voltage divider which has 400 kV modules and can be stacked to achieve the desired voltage level. We have tested it up to 800 kV and the overall expanded uncertainty we get is below 100 ppm. It could be better but further measurements are needed. But we plan to use this divider as the feedback divider for the active voltage regulation and therefore achieve the stable output voltage.
The publication for this divider is the following, unfortunatly i noticed that i have not uploaded the paper on researchgate yet. I will do this in the next days when i am back in the office.

In an other department they scrapped a VDG for 4 MeV. It was filled up to 10 Bar with SF6. But unfortunatly the bands for the generator where not available anymore. However when it was running the output was not stable to more than 1 %.

jpsmith123

Sun Feb 17 2019, 05:09PM

Registered Member #1321 Joined: Sat Feb 16 2008, 03:22AM
Location:
Posts: 843

I was thinking, being that you're not going to accelerate a particle beam with it, you won't need access to a beam tube, and unlike the accelerator case you want the HV terminal exposed, perhaps the design can be modified for your application?

(One reason I suggested this approach is that I happened to come across an abstract to an IEEE paper where someone claimed to be getting the kind of performance you're seeking).

Also IIRC, the Russian "ELV" type particle accelerators are similar to a dynamitron but they use inductive rather than capacitive coupling. Maybe the company in Russia that manufactures them can modify one for use as a power supply at a lower cost than what it would be to develop your own? I'm just thinking out loud here...

Flachzange wrote ...

Good idea about the Dynamitron, but since they require a pressure vessel it is not possible to use it, since the high voltage bushing or a cable for these voltage levels would be a lot more expensive than the multiplier itself, if they are even available.
Unfortunatly 99 % of the calibrations are done with direct high voltage connection by big tubes and not HVDC cables. This is the reason why we can not used a dead tank design.

johnf

Sun Feb 17 2019, 06:28PM

Registered Member #230 Joined: Tue Feb 21 2006, 08:01PM
Location: Gracefield lower Hutt
Posts: 284

most people use a field mill to measure very high voltages. We use these on our accelerators at work
NEC 0.5MeV and a Van der Graff KN 3MeV to get the particle beam around the bending magnet the voltage has to be measured to much better than 100ppm as 100ppm is more than the edge to edge of the beam energy

1 2 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob