Use PSUDII, depict your transformer, rectification & capacitor filter components there, simulate the reg draw as 50mA CCS load at the end of the chain to ground. 50mA is good for 32mA load. Then insert a resistive element either between bridge and cap or between caps if you got two, tweak value until it will give you 395VDC with smooth rise curve. Take care that the physical resistor will be 10W.
OK, I see what you mean. I've already got a well-filtered DC supply coming in, so I just need a resistor to drop the voltage, not an RC filter. Ok, thanks. Will run that in LT Spice just to make sure. I guess and additional filter cap wouldn't hurt, though.
Gary
Isn't there any "overhead" from the rest of the circuit though? Or is it fairly trivial?
Regulator "Latching" ?
Salas,
I've got a 12K load on the output. 400V input, with 375V out. R4 set to measure 500mV across test points. I'm using a variac to bring the voltage up to 400V in.
However, when I switch the power supply on from 0, the regulator seems to latch up. The voltage is around 10V, with current still showing around 50mA as set.
I have a similar setup for the 350V out, 75mA load and it's behaving the same.
Any suggestions as to what might cause this?
Salas,
I've got a 12K load on the output. 400V input, with 375V out. R4 set to measure 500mV across test points. I'm using a variac to bring the voltage up to 400V in.
However, when I switch the power supply on from 0, the regulator seems to latch up. The voltage is around 10V, with current still showing around 50mA as set.
I have a similar setup for the 350V out, 75mA load and it's behaving the same.
Any suggestions as to what might cause this?
After building any HV board, I always use a variac for initial testing; this one was no different. After doing all the adjustments, I then bypass the variac and just plug in directly. It wasn't until then I noticed this was happening.
I have another question - what should the drain to source voltage across Q2 be, with 400V in?
Thanks for taking the time to answer my questions, I'll keep you posted on what I find.
Salas,
To make sure everything was 100%, I replaced every semi device. I verified all resistor values.
Regulator behaves perfectly when bring voltage up from 0V using the variac. However, when I adjust the variac to the AC line voltage I need to get 390V out of the DC supply, and just switch it on, it latches up the regulator. In fact, it doesn't seem to matter what the DC voltage into the regulator is - even 100V switched on instantly latches it up.
Once again, resistive load is 12K, DC voltage set at 375V, voltage across R5 setpoint is 500mV.
The last regulator I built was tested on the same setup (using the same DC input supply), so this is very strange. It's installed in a piece of equipment I don't want to take apart to retest.
Any suggestions...ideas?
Thanks...
OK, I tried that, and it did not help.
I then switched to a higher resistive load - 20K -thinking that maybe reducing load current might cause it to behave differently. No go.
I then built another DC filter supply. Where my bench supply has a CLC filter, on this one just diodes and a filter cap, and I used a smaller 230VAC secondary transformer. Under load this should provide around 300VDC secondary, but good enough to see if this behavior changes. It does not change, exactly the same. Any surge into the regulator causes this latch-up at about 10V.
However I one other test that may suggest something. I bring the voltage up so I have, let's say 280V at the regulator output. Then, I switch off the power and watch the voltage decay. If I switch (I mean switch instantly) the supply input back on before the output gets below about 70V, it goes back to full output with no latch up.
There is C2 and D3. When we power up, C2 charges with Q6's constant current (2.2mA for say 300VDC out) up to full Vo. D3 keeps 12V across Q6 so it won't see a big voltage even for very short time while C2 comes up. D4,R10,R9 are the voltage reference items that translate Q6's constant current to voltage and bias the Q4, Q5 cascode. I don't have a variac, but I can try see in LT Spice if something holds under a threshold when the reg's Vref is programmed for a certain current via VR11, thus voltage output.
Salas,
After thinking carefully about what I was seeing, and knowing I'd made an identical regulator on the same test setup, I realized there had to be something wrong with the board. I discovered that capacitor C1 was essentially not in the circuit. I used a Wima 0.33uF/630V, with 22mm lead spacing. When I soldered it in place, it was across the same side of the circuit, not bridging across. A bit sloppy, the multiple holes caught me sleeping. It now works perfect at full voltage in on start-up.
Also, I have successfully used a different MOSFET in the Q1 position. I chose this based on similar specifications, and although not identical it is fairly close with slight differences in capacitance.
IXTP08N50D2 IXYS | IXTP08N50D2-ND | DigiKey
I thought it would be worth trying as I really need to have that safety margin on the input voltage. Using a series dropping resistor may be OK, but with voltage surges it could easily get over 400V and any higher dropping resistor would not leave me the 20V margin for the regulator at 375V out.
I ran the regulator with 450V in for about 2 hours and there is no issues, and no noise or oscillation I also switched it on and off about 30 times, to 435V in, about where it will be operating, with no issues.
If possible, can you have a look at this FET and offer your feedback on whether or not this should be stable over the long term? Also, it's a fairly new device, and may not have been available yet when you designed the SSHV2.
Thanks for your help,
Gary
About the alternative 500V IXYS Mosfet, it looks good, even has bit more Vgs than the Supertex one under it its possible, that is good for stability in the CCS. Needs at least 20V across it for its Crss to get low region. Don't prefer to put it in the lower position that sets the current also, it may demand other trimmer values etc. and it will show more reverse capacitance at that low Vds position than the Supertex.