[time-nuts] 4 KV Power Supply Recommendations

Lux, Jim (337C) james.p.lux at jpl.nasa.gov
Mon Jan 18 16:26:30 UTC 2010




On 1/18/10 2:15 AM, "Bruce Griffiths" <bruce.griffiths at xtra.co.nz> wrote:

> Mark Sims wrote:
>> A lot of the Tesla coil, etc people insulate their wires with silicone rubber
>> tubing (available at hobby shops as fuel lines).   Make sure it is silicone
>> rubber and not PVC, etc.
>> 
>> I have seen diode strings built up out of lots of cheap diodes (like 1000V
>> 1N4007s).  The diodes are strung together and then crammed into the silicone
>> tubing.  The ends are sealed and the tubing is then filled with mineral oil
>> (some people use vegetable oil).
>> 
>> ----------------
>> power supply diode and resistor strings covered in heat shrink but don't
>> touch as it is not good at HV.
>>   
> Such series strings of non avalanche rated diodes like 1N4007's without
> suitably proportioned parallel  RC networks almost inevitably fail.
>


Actually no. In fact, the typical recommendation these days is to not use
the RC networks because they actually increase the failure probability.
Today, the parts are sufficiently well matched (coming off the same spool
typically) that reverse recovery time is pretty much the same.

The typical problem in these strings is that one diode turns off a bit
faster than the others, and so, takes more of the voltage until the other
diodes turn off  In a "rectify from sine waves" application, though, the
reverse voltage doesn't rise instantaneously, so there's plenty of time for
them all to turn off.  This doesn't necessarily apply for "flyback" type
power supplies, which can have fairly high dv/dt.

 The other thing you want to do is use a LOT of parts in series.  If you're
using 1000 PIV parts and you want a 5 kV diode, use 10 diodes.

Where people get into trouble is when they use 5 1000 volt parts, randomly
selected from the surplus bin with 70xx date codes, to build a 4800V power
supply.

The other thing to watch out for on Cockroft Walton type circuits is current
limiting the output if there's a fault.  Short the output and you discharge
all those capacitors through the diodes, cooking them.  You can use a series
inductor if your load is constant and you don't want to foul up the (already
crummy) regulation.


Also, as a note, there are actual HV rectifiers from philips available
widely in the surplus world.  They're often 6kV PIV rating at a few hundred
mA, but MUST be potted or oil immersed, because the package length is too
short to run them in air at full voltage.  They're designed to be seriesed
or used in voltage multiplier applications, so they have soft recovery
that's very consistent, and decent reverse breakdown properties, too.




More information about the time-nuts mailing list