Would these SIC diodes have controlled avalanche breakdown?
https://assets.wolfspeed.com/uploads/2023/11/Wolfspeed_C3D02060F_data_sheet.pdf
https://assets.wolfspeed.com/uploads/2023/11/Wolfspeed_C3D02060F_data_sheet.pdf
Controlled avalanche ratings would appear in the data sheet, if they apply. But they are usually aimed at fast switching converters where brief defined-energy overvoltage events take place in circuits with L values in the nH to uH region.
Such ratings would not apply to the possibility of overvoltage caused by mH level leakage inductance from our power transformers!
Such ratings would not apply to the possibility of overvoltage caused by mH level leakage inductance from our power transformers!
I tend to design with lots of margin for reliability.
I once used 1kV PRV diodes in a full wave B+ circuit (not a bridge, so more of a potential problem).
350VAC x 1.414 = 495V Peak.
495V x 2 = 990V Peak Reverse Volts across the diode
Very little margin for error.
That worked just fine for a few power-ups. Then, one day, opening the power switch at a particular phase angle of mains power, the resultant transient took out both diodes.
My problem with semiconductors over the years is not the fact that they do not meet specifications.
My problem is when some are near to their specifications, and some are much better than their specifications.
IDSS 3:1 or even worse
PRV that can be at spec, or much better.
Etc.
Temperature can degrade PRV.
When you rely on equal leakage current, you may be reducing reliability.
The same goes for series electrolytics that do not have equalizing resistors.
YMMV
Enjoyable Listening is impossible if the amplifier is busted.
I once used 1kV PRV diodes in a full wave B+ circuit (not a bridge, so more of a potential problem).
350VAC x 1.414 = 495V Peak.
495V x 2 = 990V Peak Reverse Volts across the diode
Very little margin for error.
That worked just fine for a few power-ups. Then, one day, opening the power switch at a particular phase angle of mains power, the resultant transient took out both diodes.
My problem with semiconductors over the years is not the fact that they do not meet specifications.
My problem is when some are near to their specifications, and some are much better than their specifications.
IDSS 3:1 or even worse
PRV that can be at spec, or much better.
Etc.
Temperature can degrade PRV.
When you rely on equal leakage current, you may be reducing reliability.
The same goes for series electrolytics that do not have equalizing resistors.
YMMV
Enjoyable Listening is impossible if the amplifier is busted.
Another problem is line voltage variation. I measured the voltage in the garage at my new house the other day and it was 270Vac. I expect all my amps will now have higher B+ since my line voltage here is only peaks out at 260Vac. I need to walk the power line and see who else is on my line. I believe there is only one more house above me, and a barn.
And then there is the B+ voltage before the output tubes are warm and loading B+, the voltage drops in the primary DCR, secondary DCR, rectifier R, and choke DCR, become insignificant . . . B+ at its highest voltage without those voltage drops.
The rectifiers will see their highest PRV at that time.
The rectifiers will see their highest PRV at that time.
Going back to the original question of this thread, do these diodes in series need equalising resistors? Do we have a difference of opinion here?
Or is the recommendation to use 1200V diodes and forget 2x 600V in series?
Or is the recommendation to use 1200V diodes and forget 2x 600V in series?
600V diodes should be good enough. No series pairs needed. The screenshot shows LTspice analysis. I guessed at the transformer and capacitor parameters in the absence of better information, That mostly affects the diode peak current waveform but the voltage part should be accurate.
andyjevans, perhaps a good idea to read the datasheet for those 600V C3D2060F, especially the max levels for repetitive and continuous forward currents. SiC diodes are poor performers compared to the humble Si diode. They also need other heavy derating for temperature, so heatsinking is likely another practical headache.
Wrt reverse voltage capability and leakage for series connection, then like a Si diode, although there may be static voltage imbalance for the same level of leakage current, the two diodes in series will not see a higher level of leakage current than allowed by the 'worst' diode. So even if mains voltage did push one diode to have to withstand a peak voltage that would normally cause its leakage current to noticeably rise into the knee/avalanche region, the other diode will suppress that rise in leakage current, so no avalanche current level would be reached until both diodes had excessive PIV across each of them. So imho, the huge hassle of adding balancing resistors with acceptable voltage ratings is mute, given suitable derating margins.
There are similar issues with dynamic balancing of voltage across series connected diodes. Although the junction capacitance is higher for these SiC, it is the batch difference in capacitance when inverse voltage is out past 100V, where device capacitance is circa 10pF. It would be a concern if your devices were from different batches/manufacturers, of if you knew that high dV/dt transients were making it through the secondary winding, or you were using choke-input filtering without transient protection. Also note that adding balancing capacitors may be just as much hassle as with resistors, as the Vac rating of the cap needs to be acceptable, and the capacitance tolerance needs to very tight or you go to the effort to match capacitance of all those added parts.
Wrt reverse voltage capability and leakage for series connection, then like a Si diode, although there may be static voltage imbalance for the same level of leakage current, the two diodes in series will not see a higher level of leakage current than allowed by the 'worst' diode. So even if mains voltage did push one diode to have to withstand a peak voltage that would normally cause its leakage current to noticeably rise into the knee/avalanche region, the other diode will suppress that rise in leakage current, so no avalanche current level would be reached until both diodes had excessive PIV across each of them. So imho, the huge hassle of adding balancing resistors with acceptable voltage ratings is mute, given suitable derating margins.
There are similar issues with dynamic balancing of voltage across series connected diodes. Although the junction capacitance is higher for these SiC, it is the batch difference in capacitance when inverse voltage is out past 100V, where device capacitance is circa 10pF. It would be a concern if your devices were from different batches/manufacturers, of if you knew that high dV/dt transients were making it through the secondary winding, or you were using choke-input filtering without transient protection. Also note that adding balancing capacitors may be just as much hassle as with resistors, as the Vac rating of the cap needs to be acceptable, and the capacitance tolerance needs to very tight or you go to the effort to match capacitance of all those added parts.
trobbins - thank you! I have not even a fraction of your knowledge of these devices. So am I right in thinking I can leave out the resistors and capacitors in these series resistors? I would want for safety reasons to use 2 diodes in series, as per 6A3sUMMER's advice. The B+ is around 350V and current anything from 20mA to 150mA.
Perhaps better to identify the power transformer, with specs/data, and identify the load circuitry. Do you have the PT, or is this just getting a feel for what is needed, or are you following a clone/original circuit/amp design ?
I have various power transformers. In fact I just need these diodes for the input stages of my 2 stage SE amps, so current would be up to 50mA but more like 25mA in general. Power transformers are generally around 0-230V or 0-250V. Occasionally something like 280-0-280V. Some vintage off ebay, some new toroids 0-230V 100mA. I design my own circuits. So it's just a basic voltage stage with resistor load, cathode resistor and bypass cap.
This link goes to discussing the design margins that may be appropriate for diode PIV. P.14 indicates above 280-0-280 is reasonable to go 2x 1kV PIV, so pro-rata for 600V PIV, and double for full-bridge (so say 340Vac).
Even though you may have a low load current, the turn-on surge depends on the first filter capacitance and transformer, which means you should imho still do a PSUD2 type assessment of what happens in first 10ms or a bit longer, and check against diode datasheet limits. Perhaps best to choose a power transformer and measure its winding resistances, and set up a PSUD2 model.
https://www.dalmura.com.au/static/Power supply issues for tube amps.pdf
Even though you may have a low load current, the turn-on surge depends on the first filter capacitance and transformer, which means you should imho still do a PSUD2 type assessment of what happens in first 10ms or a bit longer, and check against diode datasheet limits. Perhaps best to choose a power transformer and measure its winding resistances, and set up a PSUD2 model.
https://www.dalmura.com.au/static/Power supply issues for tube amps.pdf
I have a Mac and haven't succeeded in installing PSUD. I know it's theoretically possible. First cap would not be over 28uF, followed by a choke.
That's a very long and complex attachment - it would take me a while to understand it. I don't have an EE background - I just know tube circuits.
Hence I really appreciate guidance from those who know this stuff well.
That's a very long and complex attachment - it would take me a while to understand it. I don't have an EE background - I just know tube circuits.
Hence I really appreciate guidance from those who know this stuff well.
