Diode Chain Balance Resistors? / High Voltage / Forums

Forums

4hv.org :: Forums :: High Voltage

« Previous topic | Next topic »

Diode Chain Balance Resistors?

Move Thread

LAN_403

si2030

Fri Jul 11 2008, 02:39AM

Registered Member #1571 Joined: Wed Jul 02 2008, 03:26AM
Location: Bendigo Victoria Australia
Posts: 44

Hi There,

I am building a Cockcroft Walton Multiplier and am still learning. I picked up some rather large caps 120kV 500pF and I have diodes - from China - 2CLG100/15 - 15kV each in strings of 9. (See Picture)

My intention is to study the field of the HV potential rather than to produce sparks. So the cascade will not need to be discharged except to shut it down at the end.

The total multiplier will sit in oil - canola is the best as I understand. It will be driven via a resonant secondary air coil to 50kV to cancel out the cacade's capacitave reactance. (spice modelling has been undertaken using LT Spice)

Each diode has a Vfm Max of 35Volts meaning the diode string has a Vfm of 315Volts.

It has struck me just recently I may well need to add resistance to each diode as part of the string.

My question is, what resitance should each resistor be? How should you calculate it?

Simon

Dr. Slack

Fri Jul 11 2008, 11:18AM

Registered Member #72 Joined: Thu Feb 09 2006, 08:29AM
Location: UK St. Albans
Posts: 1659

Generally, diodes will have a high enough Isurge to withstand powering up into an uncharged multiplier, if not, then reducing the AC side capacitors in the CW multiplier will current limit the charging current. It's onl;y if neither of these work do you need series resistance.

Generally, diodes will have a low enough leakage, and a high enough junction capacitance, that they will not overvolt the others in the string, because they get discharged and reset back to 0v when they conduct every half cycle, so their charging/drift time is only at worst 20mS, so you don't need parallel resistance for that.

Parallel resistors might be useful if you turn the drive off but leave the capacitors charged for a long time, then the drift time may be long enough for a low leakage diode to get stressed by its high leakage string-mates. The value should be to swamp (so 10x) the expected leakage current, but you need to use many 10s of Mohms anyway to avoid excess dissipation. As you need discharge resistors for safety for the whole setup anyway, you might as well put them across the diodes and kill two birds with one stone.

You might want to consider parallel capacitors, to swamp the effect of the difference in charge storage and hence recovery time between the diodes, when the one that turns off early gets to see the full reverse voltage of the string. However, if you have enough series diodes in the string, so rated PIV = 2x actual PIV, then diodes that fail due to overvoltage will usually fail short, and the string will keep running OK without balancing components.

good luck

si2030

Sat Jul 12 2008, 01:41AM

Registered Member #1571 Joined: Wed Jul 02 2008, 03:26AM
Location: Bendigo Victoria Australia
Posts: 44

Hi Neil

Much appreciated. I have done some cursory pricing of 100Meg resistors and they are expensive. It does occur to me that I need more diodes in the string though. At the moment the string stands at PIV = 1.2x actual PIV which is not high enough. At the time I put these strings together I tested them for similar reverse current leakage with a friends designed leakage detector.

As for luck Yep.. need fair dose.

Cheers
Simon

Proud Mary

Sat Jul 12 2008, 08:46AM

Registered Member #543 Joined: Tue Feb 20 2007, 04:26PM
Location: UK
Posts: 4992

Many older C&W designs using semiconductors plump for equalization, but I've seen it said that technology has improved so much in the last thirty years that equalization of series diodes is no longer necessary.

My own approach? To de-rate the diodes by 50%, since its simpler and cheaper to have more diodes than to start paying out for costly HV resistors and capacitors, which take up space, generate heat, and contribute to output 'sagging'.

si2030

Sun Jul 13 2008, 02:42AM

Registered Member #1571 Joined: Wed Jul 02 2008, 03:26AM
Location: Bendigo Victoria Australia
Posts: 44

Hi guys

Turns out when I was testing these diodes they had a reverse leakage of around 1nA - 1.8nA at close to its maximum voltage of 15kV. I used the following apparatus (designed and build by a good friend of mine) to test the reverse current on all diodes used and then matched them to each other so each diode string had diodes with roughly the same diode leakage.

I am intending to revamp the strings to a point close to PIV = 1.8x actual PIV. For those that get this idea alot better than I do... is this enough or should I have the full 2x PIV given the reverse leakage?

Simon

Herr Zapp

Wed Jul 16 2008, 03:57AM

Registered Member #480 Joined: Thu Jul 06 2006, 07:08PM
Location: North America
Posts: 644

Hello Simon -

Could you provide us with the details of the reverse leakage-current test setup? I'm about to assemble several HV diode series strings, and would like to take advantage of as many many factors as possible to optimize reliability. Matching diodes for reverse leakage current won't guarantee anything, but certainly can't hurt.

Best regards,
Herr Zapp

Dr. Slack

Wed Jul 16 2008, 07:18AM

Registered Member #72 Joined: Thu Feb 09 2006, 08:29AM
Location: UK St. Albans
Posts: 1659

I am intending to revamp the strings to a point close to PIV = 1.8x actual PIV. For those that get this idea alot better than I do... is this enough or should I have the full 2x PIV given the reverse leakage?

Bear in mind this is engineering here, not measurement science, so 1.8 = 2, to the same sort of accuracy as is used for the rated PIV of the diodes, the use conditions (CW is a lot "tamer" than some other uses which might generate big spikes), and the fact that the diodes are from China (hmm, better reckon that the actual usable PIV of each diode is 50% of the figure in the data sheet, am I joking? it might be worth testing one of your diodes to destruction just to make sure). It's certainly a lot better than 1.2x.

quicksilver

Wed Jul 16 2008, 02:57PM

Registered Member #1408 Joined: Fri Mar 21 2008, 03:49PM
Location: Oracle, AZ
Posts: 679

Firms that specialize in the deign and industrial use of CW and related multipliers publish "screening guides" that you folks may find interesting;

Harry had mentioned that over-engineering of a specialized design may be a better idea and in the light of industrial design and use that makes a lot of sense. The only issue with that such screening guides is that you need to take a step deeper and understand what the identified standards mean as they are frequently subject change (military specifications, etc).

An interesting off-shoot of such things are the "medical" standards of electronic engineering; when something HAS to work. That's a fascinating concept but you can quickly identify why such designs cost VERY large amounts of money.

si2030

Thu Jul 17 2008, 05:02AM

Registered Member #1571 Joined: Wed Jul 02 2008, 03:26AM
Location: Bendigo Victoria Australia
Posts: 44

Hi Guys,

I would be the first to admit my limitations with this and if there is anything I have learnt from this type of electronics its that, from the unanitiated, many many things can go awry with these projects. I guess the first thing you do is not run the system close to any single compnents specified limit. Ie have lots of headroom. So with regard to the diodes, I intend to re-build the strings with almost 50% PIV and make sure the extra diodes have been measured as best I can as I am sure this is a disaster waiting to happen in its present state.

As for the diode tester Scott, I have placed a description of this into the Projects section of this site. It does have a working simulation for LTspice but the file types allowed dont include a spice simulation extension. (perhaps they should or a file section for spice simulations etc)

Anyway I am happy to take any help offered with this as you never know what you might have missed.

Cheers

Simon

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