Diodes and HV specs
IamSmooth, Tue Dec 26 2006, 07:30PMIf I have a diode that can withstand 1000v DC and I put two together in series can they take 2000v as a unit?
Re: Diodes and HV specs
Marko, Tue Dec 26 2006, 07:46PM
Yes, but for high voltage rectifying, one must be really generous with voltage rating, as much as 2x peak voltage (for 2000V MOT for example, you need ~6kV diode string).
As voltage goes up, number of seriesed diodes can reach hundreds. 1N4007's are cheap, though
Some also use high value resistors paralell with each diode in string, but I don't believe it's very critical, especially if avalanche rated diodes are used.
Marko, Tue Dec 26 2006, 07:46PM
Yes, but for high voltage rectifying, one must be really generous with voltage rating, as much as 2x peak voltage (for 2000V MOT for example, you need ~6kV diode string).
As voltage goes up, number of seriesed diodes can reach hundreds. 1N4007's are cheap, though
Some also use high value resistors paralell with each diode in string, but I don't believe it's very critical, especially if avalanche rated diodes are used.
Re: Diodes and HV specs
Colin 99, Tue Dec 26 2006, 10:05PM
Depending on the diode, they could have different reverse resistance (leakage) in each diode and this can cause different voltage drops across each diode, sometimes over their rating. I found this out the hard way one time, it gave me and coworkers a good spark show
I used 1N5408 diode at the time and put 4.7 Mohm resistors across each one to corrent it.
Apperently with newer versions of diodes, they (the manufacturer) fixed this problem.
Shaun
Colin 99, Tue Dec 26 2006, 10:05PM
Depending on the diode, they could have different reverse resistance (leakage) in each diode and this can cause different voltage drops across each diode, sometimes over their rating. I found this out the hard way one time, it gave me and coworkers a good spark show
I used 1N5408 diode at the time and put 4.7 Mohm resistors across each one to corrent it.
Apperently with newer versions of diodes, they (the manufacturer) fixed this problem.
Shaun
Re: Diodes and HV specs
Dr. Slack, Wed Dec 27 2006, 10:16AM
If the connection point of the two diodes stays at the average of the two terminal voltages then, yes indeed, 1000+1000 = 2000.
However, there are several mechanisms which will disturb that desired state of affairs.
a) if the reverse voltage rises rapidly from 0 to PIV, then it will distribute itself between the two diodes in inverse ratio to their effective reverse capacitiance. For two diodes in the same batch, it's likely to be similar, but not identical. On the good side, capaciticance is a function of the dfdevice geometry, so you might hope that it is fairly repeatable from device to device.
b) If the reverse voltage stays there for an etended period of time, the difference in leakage currents will charge the mid point. As leakage is a function of diode impurity, this will be less well matched between even diodes from the same batch.
c) The real killer though is turn-off time. Switch rapidly from conducting to blocking, say one diode takes 10nS longer than the other to run out of stored charge. That's 10nS the reverse recovery current is charging the few pFs of mid point junction capacitance.
d) And then there's geometry of the circuit. If the mid point has highest stray capacitance to the grounds, or maybe to the HV AC terminal, that will upset the diode stray matching.
Diodes from the same batch, and severe under-rating (>50%) is one way to get diodes strings to survive.
A parallel C and R across each diode to swamp the stray mismatches allows you to get away with moderate (<50%) derating.
Avalanche rated diodes, which don't die immediately they get over-volted, are also good.
Dr. Slack, Wed Dec 27 2006, 10:16AM
If the connection point of the two diodes stays at the average of the two terminal voltages then, yes indeed, 1000+1000 = 2000.
However, there are several mechanisms which will disturb that desired state of affairs.
a) if the reverse voltage rises rapidly from 0 to PIV, then it will distribute itself between the two diodes in inverse ratio to their effective reverse capacitiance. For two diodes in the same batch, it's likely to be similar, but not identical. On the good side, capaciticance is a function of the dfdevice geometry, so you might hope that it is fairly repeatable from device to device.
b) If the reverse voltage stays there for an etended period of time, the difference in leakage currents will charge the mid point. As leakage is a function of diode impurity, this will be less well matched between even diodes from the same batch.
c) The real killer though is turn-off time. Switch rapidly from conducting to blocking, say one diode takes 10nS longer than the other to run out of stored charge. That's 10nS the reverse recovery current is charging the few pFs of mid point junction capacitance.
d) And then there's geometry of the circuit. If the mid point has highest stray capacitance to the grounds, or maybe to the HV AC terminal, that will upset the diode stray matching.
Diodes from the same batch, and severe under-rating (>50%) is one way to get diodes strings to survive.
A parallel C and R across each diode to swamp the stray mismatches allows you to get away with moderate (<50%) derating.
Avalanche rated diodes, which don't die immediately they get over-volted, are also good.
Re: Diodes and HV specs
Tesladownunder, Wed Dec 27 2006, 11:46AM
Here are a couple of HV strings of diodes.
First is mine using 300 1N4007's to rectify a 30kV dental transformer.
Second is Syd Klinge and his 100KVA coil which is a DC coil with 3 phase rectification and about 1000 1N4007's in batches of 5 parallelled.
Peter
Tesladownunder, Wed Dec 27 2006, 11:46AM
Here are a couple of HV strings of diodes.
First is mine using 300 1N4007's to rectify a 30kV dental transformer.
Second is Syd Klinge and his 100KVA coil which is a DC coil with 3 phase rectification and about 1000 1N4007's in batches of 5 parallelled.
Peter
Print this page