thank you, i am looking for the solution that is least expensive and least labour intensive.
would you think then that perhaps i should replace the existing 4 diodes with diodes that have a higher voltage rating, would that be the better solution, as compared to assisting the underrated diodes by using caps?
or maybe a microwave rectifier or something already made for this job?

 

Neon transformers are different. Open circuit voltage is usually higher than claimed. The transformers are pretty much designed to withstand a short circuit.

Your diodes also need to be mounted to a heat sink with heat sink grease.

Never did like working on the 15 kV DC 1.5 A or the 100 kV at 0.1 A power supplies at work.

 

interesting that it is only that one diode that has an issue (it is away from conductors) and is in mid air for cooling, and it fries a bit immediately then it still works for hours when i reduce the load by reducing the spark gap distance

If you limit the current, the "failure" is not catastrophic, so it makes perfect sense.

EDIT: The peak inverse voltage is basically a sloppily controlled Zener diode. I know of the avalanche and Zener mechanisms. let's just call it Zener.

 

The only things for (almost) sure is that you need to increase reverse voltage rating of your rectifier.

 

Hello,

Have a look at these pages:
https://www.nutsvolts.com/questions-and-answers/diodes-in-series
https://www.daenotes.com/electronics/basic-electronics/diodes-in-series

Bertus

 

The only things for (almost) sure is that you need to increase reverse voltage rating of your rectifier.

and so by doubling up in series the diodes that "almost" work as is, do you think that would increase the reverse voltage rating of the rectifier. that is, each diode would be in direct connection in series with the same type diode --->I--->I--- nearly doubling the reverse voltage rating, correct?

 

Generally the reasons may be several:
1) too bad thermal contact to diode`s radiator. TO220 or other diodes body must be fixed to cooler with good thermal pasta and grovered screw. If not, over 100 mA may happen burn-out. Thermal camera well serve for diagnosticing such kind of problem.
2) too high initial current while cap is zero, then diode takes near short-circuit currrent. Solution - lade-up via resistor what goes relay-shortened when capacitor is up at last.
3) too much different diode i(reverse), what manifests in very different vltages in serial switched diode series. Older diodes are very prone to that, in 60~ies each handbook told the diodes MUST be paralelled by appropriate multi-MOhm resistor matrix, yet nowadays diodes are cooked up rather identical thus the problem is coped without of resistors. Yet measure up is worth to.
4) With HV measurements sometimes are VERY inaccurate figures, thus reading 15 kV probably there are 10 or maybe 22 kV. Thus, measurements must be verified between different instruments, especially if higher frequencies persist, like 1...100 kHz.
5) The spikes, every previous staying Royer cascade or flashback cascade may give a rere voltage spikes of multitude the normal voltage. If so shows a memory-type of oscillo, the solution is RC integration unit, or better the Zener.

 

Generally the reasons may be several:
1) too bad thermal contact to diode`s radiator. TO220 or other diodes body must be fixed to cooler with good thermal pasta and grovered screw. If not, over 100 mA may happen burn-out. Thermal camera well serve for diagnosticing such kind of problem.
2) too high initial current while cap is zero, then diode takes near short-circuit currrent. Solution - lade-up via resistor what goes relay-shortened when capacitor is up at last.
3) too much different diode i(reverse), what manifests in very different vltages in serial switched diode series. Older diodes are very prone to that, in 60~ies each handbook told the diodes MUST be paralelled by appropriate multi-MOhm resistor matrix, yet nowadays diodes are cooked up rather identical thus the problem is coped without of resistors. Yet measure up is worth to.
4) With HV measurements sometimes are VERY inaccurate figures, thus reading 15 kV probably there are 10 or maybe 22 kV. Thus, measurements must be verified between different instruments, especially if higher frequencies persist, like 1...100 kHz.
5) The spikes, every previous staying Royer cascade or flashback cascade may give a rere voltage spikes of multitude the normal voltage. If so shows a memory-type of oscillo, the solution is RC integration unit, or better the Zener.

Thank you so much for taking the time to explain to me the likely problems that are causing the diode to burn.

can I ask you if all of the issues that you brought up would they not be fixed by having 2 diodes in series to replace each one diode in the rectifier?

because it seems that since this device already (nearly) works as is, so it must be very close to being able to sustain the required load but it is just a little bit short of being able to withstand the full load of the fully separated spark gap terminals, so then it seems that if the diodes are doubled up then all of the issues that you mentioned will be almost halved per individual diode as compared to the current set up witch comes very close to coping with the load.

and so I would like to ask if in your opinion; if just simply doubling up on each diode in series would resolve each of the issues that you think could be causing this? (picture attached with transformer load) diode

• High Voltage Rectifier Diode;Model : PRHVP2A-20
• Max. Repetitive Peak Reverse Voltage : 20000V;
• Max. Average Forward Rectified Current : 2A