I have an old ailing power supply stepping up the voltage on my electrostatic speakers. It's potted in wax and given the time/effort required to get the wax out, create a jig for the connectors, then return the wax and set the connectors properly, I'm looking to make some upgrades/replacement of old parts while the wax is out.
The Selenium diode on one of the supplies is failing. The repair, using readily available parts, is a chain of three 1N4007 diodes. What, if any, would be be superior alternative to this three diode chain? Cree Schottkey diodes, hexafreds, etc?
Any suggestions or ideas are appreciated. Thank you.
The Selenium diode on one of the supplies is failing. The repair, using readily available parts, is a chain of three 1N4007 diodes. What, if any, would be be superior alternative to this three diode chain? Cree Schottkey diodes, hexafreds, etc?
Any suggestions or ideas are appreciated. Thank you.
in this case the comparsion between 1N5406 and BYW27 are of interest:
http://home.arcor.de/dauner765/7293/
go to the box on right hand to follow buttons:
MESS-1N5406.PDF
MESS-BYW29.PDF
http://home.arcor.de/dauner765/7293/
go to the box on right hand to follow buttons:
MESS-1N5406.PDF
MESS-BYW29.PDF
Last edited:
If you replace a single rectifier with a string of three, make sure that you place equalising resistors across each diode. Diode leakage varies wildly between units and without the resistors you may end up with one diode of the string trying to block more than half the reverse voltage and going poof!
Something like 1meg to 5meg usually works fine.
jan
Something like 1meg to 5meg usually works fine.
jan
Ordinary diodes work like zena diodes at voltages well above there ratings.
Try a 1N4001 in series with 100K 2W on an HT transformer with a rectifier and Variac.
Don't power it for too long as the diode will dissipate a what or so at only a few ma.
As long as the total reverse voltage is well above the supply voltage it should be fine.
Look at those selenium stacks that used to be used in old TV sets for the HT. They never had resistors at all.
Try a 1N4001 in series with 100K 2W on an HT transformer with a rectifier and Variac.
Don't power it for too long as the diode will dissipate a what or so at only a few ma.
As long as the total reverse voltage is well above the supply voltage it should be fine.
Look at those selenium stacks that used to be used in old TV sets for the HT. They never had resistors at all.
Is this what happens?
I would think that the leakage current through the string will determine the Vfr of each diode in the string.
That same leakage current will give rise to different Vfr across each individual diode, but I don't think that equates to going "poof".
The leakage increases.
If only one diode is being used then that diode will eventually avalanche into total failure.
Where a string of two or more diodes are being used the leakage does not initially avalanche. The other diodes have to pass the increased leakage and they each increase their reverse voltage and so prevent the avalanche effect until a much higher total voltage is applied.
I would not be surprised if one diode could be around 20%, or much more, above it's rated voltage and yet still not avalanche into totally collapsing.
I wonder if junction dissipation could be used to predict when the diode fails?
If only one diode is being used then that diode will eventually avalanche into total failure.
Where a string of two or more diodes are being used the leakage does not initially avalanche. The other diodes have to pass the increased leakage and they each increase their reverse voltage and so prevent the avalanche effect until a much higher total voltage is applied.
I would not be surprised if one diode could be around 20%, or much more, above it's rated voltage and yet still not avalanche into totally collapsing.
I wonder if junction dissipation could be used to predict when the diode fails?
Hmmm yes I think that could be what happens initially. But eventually the leakage and thus dissipation increases and at a certain point something DOES go poof. No free lunch.
jan