Let's say I'm rectifiying a 408-0-408 PT with discrete diodes. With 1kV parts like 1n4007 or 1n5408, I need to double up, because PIV is about 1.14kV. I'm also using a series pair of reverse-biased diodes as a half-wave rectifier for bias supply.
There's a 1.4kV@2A bridge rectifier IC, the DBL209G. I could use that in a FW configuration (i.e., use the positive output for B+ and the negative output to feed the bias supply, and ground the PT CT, with no connection from the bridge to ground).
It's cheap (65 cents), small, and much simpler - takes the place of six discrete diodes. Is there any reason not to use the IC?
https://www.mouser.com/datasheet/2/395/DBL201G_SERIES_J15-1918085.pdf
There's a 1.4kV@2A bridge rectifier IC, the DBL209G. I could use that in a FW configuration (i.e., use the positive output for B+ and the negative output to feed the bias supply, and ground the PT CT, with no connection from the bridge to ground).
It's cheap (65 cents), small, and much simpler - takes the place of six discrete diodes. Is there any reason not to use the IC?
https://www.mouser.com/datasheet/2/395/DBL201G_SERIES_J15-1918085.pdf
Last edited:
You haven't said anything about application current.
Most people would use a fast diode (eg. like two UF4007 in series for your application), either because they just use fast diodes nowadays by default, or a design was forcing a high peak to average diode current waveform and the transformer had significant leakage inductance.
Most people would use a fast diode (eg. like two UF4007 in series for your application), either because they just use fast diodes nowadays by default, or a design was forcing a high peak to average diode current waveform and the transformer had significant leakage inductance.
Less than 10% of the bridge's rated 2A. It idles at 100mA (75mA for plates, 5mA for screens, and 20mA for the preamp). At max output, around 200mA (140mA plates, 35mA screens, and 20mA for the preamp).
Gotcha. This is a guitar amp that will be heavily distorted most of the time, so I figured that building it out with standard-recovery diodes would suffice (and that's what I had on hand - built it out with 5408s in prototype). I'll lay in some fast diodes for the next project.
When using series rectifier diodes, be sure reverse leakage current is the same (at voltage!) in the series-connected diodes, or else the one with the higher leakage will tend to drop less voltage and possibly over stress the companion diode to failure. Sometimes high value resistors, in the meg-ohms, are placed across the diodes to facilitate this voltage sharing.
I've always read that ultra-fast diodes cause nastier 120 hz interferance because the more vertical shut off edge has higher frequency components that transmit via RF better. Read fourrier transform mathematics. Thus the standard recovery diode is superior, edge can be suppressed by a simple .01 uf cap from + to - of bridge.
You are correct, 1n4007 etc are too short for long life near the rated PIV voltage. They can build up dust on the outside and will short across in a decade or two. Unless the cabinet is sealed, which is unlikely with hot vacuum tubes.
You are correct, 1n4007 etc are too short for long life near the rated PIV voltage. They can build up dust on the outside and will short across in a decade or two. Unless the cabinet is sealed, which is unlikely with hot vacuum tubes.
Last edited:
Imho, this is not required for modern batch parts. Yes one could do a simple 'diode test' to confirm any new diode is not blatantly faulty before inserting in to a circuit. The next step in pre-checking would require an Insulation Resistance tester, and even if you had one of those it would likely only do a 1kVdc test, and wouldn't be able to test at a higher level at the onset of where leakage current would be measurable. The leakage levels of modern 1kV diodes is very very low, and the voltage at which leakage starts to rise is significantly higher than 1kV, and given the diodes are in series, then they both would happily support in excess of 1kV PIV each with close matching.
That's why soft-recovery is desirable in a fast rectifier diode. Ultra-fast rectifiers were made for switch-mode at high frequencies and they can make a lot of RF noise - in a good SMPS this is mainly filtered out and the supply is in a faraday cage, and the interference is typically ultrasonic.
If you don't want to add filtering and cage to a linear mains supply, don't use fast diodes - there's no need for a fast diode at these timescales! Or at least use soft-recovery diodes, and you can snub the diodes too.