EEVblog Electronics Community Forum
Electronics => Beginners => Topic started by: Cliff Matthews on August 18, 2017, 12:29:41 am
-
Digikey shows 1 amp Semtech devices weighing-in at $200 each. So this advice is just a setup for cascade failure, right?https://www.youtube.com/watch?v=Clt3ncng7A0 (https://www.youtube.com/watch?v=Clt3ncng7A0)(https://s12.postimg.org/meea3fc19/Cheap-_DIY-1amp-_HV-diode.jpg)
-
As a half-wave rectifier, I made the assumption (yes I know what they can do..) that immediately after conduction the last anode would briefly be at 9*Vf due to junction capacitance of the first 9 diodes and this would start cascade breakdown.
So reverse leakage negates that and causes all 10 to share Vr equivalent to the whole stack.
Also, would the slew-rate of higher AC frequencies (say 1Khz, not 50/60 hz) give any cause to de-rate Vr for the stack?
-
It is very common (and recommended) practice, to put equalization resistors across each diode. Usually 100-330K, depending on diode specs.
"rated to conduct say 10x the worst-case reverse leakage current across the volt/temperature range. Otherwise one diode leaks, and the other breaks down"
In some designs in the long past, we even added 10-20pF across them as well. This will keep all didoes closely matched in all conditions.
-
So would the extra resistors and caps likely be found inside the $200 Semtech part (for longevity or stability)?
**edit -Also even with a higher 470k across each part in the stack, 21-watts at 10Kv seems like a lot to loose (say, when running a generator and boost transformer to charge some oil capacitor bank).
-
Hi,
You might want to look at the diodes switching characteristics, Trr and Qrr, if you are using the part at frequencies higher than a few hundred kilohertz.
If the application is low current, you want to look at junction capacitance as well.
Regards,
Jay_Diddy_B
-
$200 sounds a lot for 10kV 1A.
http://hvstuff.com/1a-10kv-high-voltage-diode-hv-rectifier-tesla-coil-ham (http://hvstuff.com/1a-10kv-high-voltage-diode-hv-rectifier-tesla-coil-ham)
I've bought some of their 40kV 1A and 80kV 200mA diode stacks with acceptable results.
-
As a half-wave rectifier, I made the assumption (yes I know what they can do..) that immediately after conduction the last anode would briefly be at 9*Vf due to junction capacitance of the first 9 diodes and this would start cascade breakdown.
So reverse leakage negates that and causes all 10 to share Vr equivalent to the whole stack.
Also, would the slew-rate of higher AC frequencies (say 1Khz, not 50/60 hz) give any cause to de-rate Vr for the stack?
Assumptions can give you "sum pastion".
So when they are in series say 10; each one drops .6v for a total of 6 volts drop? Will they work like resistors if you put ten in series and powered a 3v led off 9 volts? Or would this blow up the led? That always confused me how diodes will lose straight .6 volts always but resistors voltage drop is based on current *and* voltage. Do they get hot when losing this .6 volts? How much current does the diode use? Can you measure ohms across it?
-
How about a microwave oven diode? They're not usually rated to 1A though, usually a few hundred mA.
https://www.eevblog.com/forum/beginners/circuit-with-op-amp-lm339/?action=dlattach;attach=190287;image (https://www.eevblog.com/forum/beginners/circuit-with-op-amp-lm339/?action=dlattach;attach=190287;image)
http://hvstuff.com/12kv-350ma-microwave-oven-high-voltage-diode-rectifier (http://hvstuff.com/12kv-350ma-microwave-oven-high-voltage-diode-rectifier)
-
When putting the 1N4007 close together like in the 1st picture, this will impede heat removal. So it might withstand 10 kV, but no more 1 A rating. I have seen 1N4007 in series been used at 40 kV (though less than 1 A).
Fast transients and thus reverse recovery currents could be a problem though, as in this case voltage sharing might no work well. At least it would be a good idea to have diodes from the same batch in this case, to have a close match in t_rr. Depending on the application one might want to have added capacitors or fast diodes.
-
You can definitely do it with 1N4007s but the diode of choice would be some type of avalanche rated rectifier which will not require leakage current and capacitance balancing. If it was handy, I would take a photograph and post it but I have a rectifier made this way which is like 3 feet long made from DO-4 packaged stud rectifiers on printed circuit boards sandwiched between fiberglass U-beams that is 10+ kilovolts at several amps.
-
The original TV high voltage rectifiers were just a stack of selenium discs placed in a tube, with a spring one end to provide contact pressure, and for decades all rectifiers were made from stacked rectifiers, either copper oxide or selenium, with a spring to provide clamping pressure.
Known to work, though I would for high voltage work buy a reel of diodes so all are at least from the same die, and derate them voltage wise to around half the rated, using more diodes. With the 1N4007 however they are very likely to be able to withstand 2000V, though this is not tested, they are only rated to withstand 1200V as this is tested during manufacture to bin them.
-
Two things not well noted yet in this thread:
1. Reverse recovery.
2. Matching.
Indeed, all EHV diodes are stacks. It's possible to make single junctions up to 6kV or so; recovery time is quite slow, though. (Example: http://www.vishay.com/docs/93177/vs-sd553cs50lseries.pdf (http://www.vishay.com/docs/93177/vs-sd553cs50lseries.pdf) note that 6us really is fast (compare 1N4007 at ~3us), for a diode of this voltage!) For both these reasons, a practical diode of, say, 10 or 20kV might be made of a half dozen dies, stacked.
Matching is critical, and that's why they make proper stacks.
At mains frequency, as mentioned above: recovery doesn't matter much, and the avalanche rating will take up the difference okay. 100k range resistors will do a good job of balancing the DC reverse voltage, by swamping the leakage currents (which won't match even under the best of circumstances :) ).
But as you go up in frequency, recovery-related losses (i.e., the avalanche that accompanies the earliest recovering diodes) come to dominate. You can try swamping this with capacitors, but your application might not allow much of that, either!
There's also dynamic recovery to worry about: this is better known (but still not very well known) with higher voltage BJTs, but it applies to diodes as well! The physics is this: the junction does not recover uniformly, but rather it's more that the charge sweeps out, like a bucket drains through a hole in the bottom: as the level falls, the surface stays parallel. That means the junction width, and therefore the instantaneous breakdown voltage, rises over time, until the charge is empty and it settles out to the normal value.
(When you see a "soft recovery" diode, it means that the recovery process has been tuned, for a particular setup condition (a few microseconds forward bias, then a modest dI/dt reverse recovery), to give a controlled breakdown voltage, causing the voltage to rise smoothly. Soft recovery costs more dissipation than an otherwise-identical diode made to be as fast as possible, but produces much less EMI. The difference might be, say, 20ns for a "ultrafast recovery" part, versus 40 or 50ns for a "soft recovery" part. Not a big deal at most switching frequencies.)
So the problem is that, the whole time a diode is becoming open circuit (during the voltage rise), it's burning power. And that voltage is clamped to a value that rises as recovery finishes, up to the rated voltage eventually. If you have diodes that have mismatched recovery times, in an application that is fast enough to be recovery limited: you're in for unhappy diodes.
What the manufacturers can do, then, is select dies by recovery time and charge, and junction capacitance, and use matched parts in the stacks.
Tim (has melted UF4007s by connecting just two in series and operating at 100kHz)
-
How about a microwave oven diode? They're not usually rated to 1A though, usually a few hundred mA.
https://www.eevblog.com/forum/beginners/circuit-with-op-amp-lm339/?action=dlattach;attach=190287;image (https://www.eevblog.com/forum/beginners/circuit-with-op-amp-lm339/?action=dlattach;attach=190287;image)
http://hvstuff.com/12kv-350ma-microwave-oven-high-voltage-diode-rectifier (http://hvstuff.com/12kv-350ma-microwave-oven-high-voltage-diode-rectifier)
The proper part only costs 2.15$? Why not just buy that? I was thinking it was some obscure hard to get expensive part.