Banned
Joined 2002
The best thing to do is buy yourself a Xantrex "60-5" (60V/5A) lab supply or two.
Building one of these things in a linear form would be the easiest and by far the cheapest thing to do, but it would require an obscene amount of heatsinking. Building a switching power supply that's stable at both 0A and 5A loading over an enormous voltage range is best left to the professionals...
Building one of these things in a linear form would be the easiest and by far the cheapest thing to do, but it would require an obscene amount of heatsinking. Building a switching power supply that's stable at both 0A and 5A loading over an enormous voltage range is best left to the professionals...
Banned
Joined 2002
Banned
Joined 2002
Forgive me for the "sorry, should buy one" reply. My post here came shortly after I responded to a post on another forum from a guy who decided he was going to build his own computer motherboard 
I lack schematics for a high voltage linear power supply, but here's a basic concept of one... First, you'll need an unregulated 50 volt supply - use a big toroid and a bank of caps. If you want to pull high currents at low voltages out of the supply, see if you can get a transformer with multiple taps - this will keep the power dissipation in your pass transistors reasonable.
Look up the datasheet and some app notes on the TI/Unitrode UC3834 chip; I've used this before in a couple of designs. These chips only do 40V unfortunately; you'll need to use a zener diode or something similar to drop the unregulated voltage from your transformer down to something that the chip can handle.
Get a few big PNP transistors, rated for at least 60 volts and lots of power dissipation, to use as your pass transistors. You'll want a big heatsink to bolt these to. Depending on the hFE of the transistors involved, you might want to use an extra one of these to form a darlington configuration, since the 3834 can only sink 250mA if memory serves me correct... eg for 10 amps output, the PNPs need a hFE greater than 40.
You'll also want to put a zener diode in series with the base of your PNP transistor, to subtract a few volts and keep the output driver of the 3834 within a safe voltage range. Watch the power dissipation in this zener, though...
Then design an appropriate control loop (I think it can't be too hard to steal one out of an appnote or something) and you should be good to go.
I'll draw up a rough schematic or something if you're interested.
I lack schematics for a high voltage linear power supply, but here's a basic concept of one... First, you'll need an unregulated 50 volt supply - use a big toroid and a bank of caps. If you want to pull high currents at low voltages out of the supply, see if you can get a transformer with multiple taps - this will keep the power dissipation in your pass transistors reasonable.
Look up the datasheet and some app notes on the TI/Unitrode UC3834 chip; I've used this before in a couple of designs. These chips only do 40V unfortunately; you'll need to use a zener diode or something similar to drop the unregulated voltage from your transformer down to something that the chip can handle.
Get a few big PNP transistors, rated for at least 60 volts and lots of power dissipation, to use as your pass transistors. You'll want a big heatsink to bolt these to. Depending on the hFE of the transistors involved, you might want to use an extra one of these to form a darlington configuration, since the 3834 can only sink 250mA if memory serves me correct... eg for 10 amps output, the PNPs need a hFE greater than 40.
You'll also want to put a zener diode in series with the base of your PNP transistor, to subtract a few volts and keep the output driver of the 3834 within a safe voltage range. Watch the power dissipation in this zener, though...
Then design an appropriate control loop (I think it can't be too hard to steal one out of an appnote or something) and you should be good to go.
I'll draw up a rough schematic or something if you're interested.
One simple way to do it would be to buy a variac from a cheap source such as All Electronics ( www.allelectronics.com ) or Marlin Jones ( www.mpja.com ), and to build yourself a nice simple powersupply using the variac, a transformer, two bridges and a few caps. Using the variac to vary the input voltage would allow you get any voltage you want and you would always get a rated capacity of approximately 5A with a 500W variac. For safety and regulation purposes, you should add a transformer between the bridges and the variac. Then, the ratio between the ouput of the circuit and the input would be the one between the primary voltage and the secondary voltage of the transformer. (A good transformer would be a 500-1000VA 35V-0-35V)
The scripted schmatic would be as follow:
wall->variac->transformer->bridge->caps->load
For the readout part, you could always use some cheap RadioShack DMM.
Hope this helps!
Sébastien
The scripted schmatic would be as follow:
wall->variac->transformer->bridge->caps->load
For the readout part, you could always use some cheap RadioShack DMM.
Hope this helps!
Sébastien
Banned
Joined 2002
Banned
Joined 2002
if you want to digitally control the power supply, instead of a fixed VRef for the error amplifier, use a 10 or 12-bit DAC with a 4.096 or 5.000 volt reference. I have used the Linear Tech LTC1446 and the Maxim 5250 for stuff like this. The Linear Tech part has an incorporated Voltage Reference and it's extremely easy to use, just "Data", "Enable" and "Clock".
It's then a matter of creating a simple Voltage Amplifier, followed by a Current stage.
I have worked with BJT's and HexFets and if it is a one-off project would recommend HexFets for their linearity. One digital high voltage power supply I just finished 2 months ago tracks 0.1% ---
Power supplies are still cheap on EBay, but the Telcom bust now seems to be waning -- the Sorensen, Trygon supplies are really excellent. Many of the programmable supplies are going to require a GP-IB or HP-IB card from National Instruments.
It's then a matter of creating a simple Voltage Amplifier, followed by a Current stage.
I have worked with BJT's and HexFets and if it is a one-off project would recommend HexFets for their linearity. One digital high voltage power supply I just finished 2 months ago tracks 0.1% ---
Power supplies are still cheap on EBay, but the Telcom bust now seems to be waning -- the Sorensen, Trygon supplies are really excellent. Many of the programmable supplies are going to require a GP-IB or HP-IB card from National Instruments.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.