In this space, visitors are invited to post any comments, questions, or skeptical observations about Philo T. Farnsworth's contributions to the field of Nuclear Fusion research.
Subject: Re: Diodes?
Date: Sep 23, 3:41 pm
Poster: Jim Lux
On Sep 23, 3:41 pm, Jim Lux wrote:
>I almost forgot. If the diodes are really slow and have a wide divergence in cutoffs, placing capacitors of about .001ufd across each diode and resistor will help to shunt the vlotage rise across an early turned off diode.
If you put enough diodes in the string, you don't care if some turn off slower.. Say you put 10 diodes in the string at 6kv each. Now, if you run the string at a max of 30 kV, they normally only have 3 kV across them in reverse (plus some variation due to different reverse impedances). Now assume that two of them turn off much slower. Now you have 30 kV across only 8 diodes, 3.75kV each, still way less than the rated max of 6. At $0.50 each, doubling the number of diodes is a whole bunch cheaper than equalizing resistors.
As for the caps, they have to be really well matched, otherwise they act as a voltage divider and can cause the reverse voltage to be divided unevenly, particularly for transients. As an extreme, imagine that one of the caps goes open, while the others are still good. A fast transient comes in, for which the caps have low impedance (that's why they are there). So, now the entire transient appears across the failed cap (and it's diode) killing the diode.
As a practical matter, the diodes will be much better matched than any resistors or capacitors will be, so rather than protecting the string, the caps will kill it faster. If you want to bypass HF transients, use a single cap across the whole string (all those caps in series will do), but I think a good choke might be a better choice.
And, of particular interest to fusorites, who are likely just rectifying 60Hz, the fact that the input waveform is sinusoidal helps a lot.. It doesn't instantly apply the full reverse voltage. It gradually switches polarity and comes up, so the diodes all have time to "turn off". Those of you building high frequency switching supplies, though, had better watch out.