Generic supply for detectors
Posted: Sat Feb 12, 2011 8:55 pm
I am working on adding a B10 lined neutron detector tube to my lashup, and needed yet another power supply for it. For those not familiar with these, they are old as the hills (maybe older than I) but they work fairly well. Ours are GE made, around the end of WW II or so. Basically it's just a big proportional type tube but lined with B10 metal to make much more signal when hit with neutrons. In practice, these are sensitive enough for fusor use, and in fact, better than you need. But they have one caveat. The signal is tiny, and riding on the DC supply power, so you need to mind your P's and Q's when it comes to powering one (I'll document the preamp elsewhere and cross-link the posts).
This turns out to not be a hard task with a CCFL inverter, my favorite being DigiKey's 289-1025, which is 12v nominal in, and 1kv nominal out. Of all the CCFLs I've tested, this one has the best "manners", draws the least no load power, and is hardest to fry, as well as making a very stiff output voltage that is tightly related to the input voltage, so a full closed loop regulator is very rarely needed, and that only if you're going to draw real power out of one (they'll make12w if you're careful). The manufacturer is JKL and their internal part number is BXA-12529.
I've tried a ton of CCFL inverters, BTW, and this on is hands down the best -- hard to fry, most power out (it's designed to drive two lamps) and lowest quiescent current by a rather fat margin over say, the TDK brand ones I've tried. It appears designing an efficient Royer oscillator is a little past most of the manufacturers, they don't use the right core material. This part number will run reliably all the way down to .7 volt inputs, which is nice too. Something to watch for in new CCFL's is that many are now uP controlled and do dimming via PWM. Those are utterly useless for this, as they won't run on lower than spec voltages to make the output adjustable (unless you happen to have multi henry filter inductors laying around doing nothing). They also tend to be real power hogs. This is another case like getting the HV stuff meant for old CRT TV's before it's all gone -- as the world rushes to flat panel displays and LED lighting for those, all this stuff is disappearing off the market fast. So stock up now!
For this case, I needed 650v on the dot, positive, with sub millivolt AC noise on it. Since the supply operates at 50-60khz, filtering is a snap -- a .01uf output capacitor is all that's needed.
In this case, we know the tube will never (we hope) draw much current at all -- it works out to millivolts across 10 megs, so we don't need a super output power or super stiffness under load either. I always run these with at least a volt doubler, as that means they waste less power running at a lower input voltage for a given output. In this case, I simply used one of the 27pf caps on the original board as the series cap in the doubler, as that's plenty and helps current limit the thing in a fault condition, and the usual two fast diodes and that nice ceramic .01 uF output cap.
To control the input, I used an LM 317, with a 200 ohm output to adj pin resistor, and a 2k 1k pot to ground for the adjustment, which in this case puts me in the middle of the pot when it's on bogey - perfect, and plenty of range should I need this for something else. The one special trick here is that I use the same bench 13.6 v supply to run all these, and you want to avoid ground loop issues in this sometimes noisy environment. So to help with that, I put a 120 ohm resistor in series with the output ground, and a 10k ohm resistor in series with the positive 650v output.
There is another .01 uf filter in the preamp to take care of any "antenna noise" and this way the real ground is provided by whatever audio amp or counter I connect it to, not the noisy 13.6v supply that has "antennas" all over it to power other detectors, the ion extraction power supply and so on (also CCFL's btw).
Almost doesn't deserve a picture, since it's so simple, but here's what it looks like on the inside. I put the trimpot on the other side of the board so I can tweak it through a hole in the front panel.
Now, these guys radiate quite a lot of 50khz stuff, so I grounded the box panel and put a bit of copper tape in there to shunt that to ground to keep the air around it quiet.
And that's it. If I do any more of these, oh please let me lay out a PCB for it -- I've made maybe 5-8 of them so far for various things; the basic design goes to 3kv fine with parts value adjustments, so a similar one is also running our 3He tube which has most of the same issues with noise, or worse. The reason I haven't is I don't have a stock of standard little boxes to put them in, you really do need a box, and so I wind up customizing each one to fit in whatever box comes out of my junk-box-box.
This particular one draws almost 20ma under the design load. I could about cut that in half with tweaking the lm 317 resistor values, where most of the actual power is going!
It'll run on 6v or so on up to whatever the lm 317 can take, 40v or so.
This turns out to not be a hard task with a CCFL inverter, my favorite being DigiKey's 289-1025, which is 12v nominal in, and 1kv nominal out. Of all the CCFLs I've tested, this one has the best "manners", draws the least no load power, and is hardest to fry, as well as making a very stiff output voltage that is tightly related to the input voltage, so a full closed loop regulator is very rarely needed, and that only if you're going to draw real power out of one (they'll make12w if you're careful). The manufacturer is JKL and their internal part number is BXA-12529.
I've tried a ton of CCFL inverters, BTW, and this on is hands down the best -- hard to fry, most power out (it's designed to drive two lamps) and lowest quiescent current by a rather fat margin over say, the TDK brand ones I've tried. It appears designing an efficient Royer oscillator is a little past most of the manufacturers, they don't use the right core material. This part number will run reliably all the way down to .7 volt inputs, which is nice too. Something to watch for in new CCFL's is that many are now uP controlled and do dimming via PWM. Those are utterly useless for this, as they won't run on lower than spec voltages to make the output adjustable (unless you happen to have multi henry filter inductors laying around doing nothing). They also tend to be real power hogs. This is another case like getting the HV stuff meant for old CRT TV's before it's all gone -- as the world rushes to flat panel displays and LED lighting for those, all this stuff is disappearing off the market fast. So stock up now!
For this case, I needed 650v on the dot, positive, with sub millivolt AC noise on it. Since the supply operates at 50-60khz, filtering is a snap -- a .01uf output capacitor is all that's needed.
In this case, we know the tube will never (we hope) draw much current at all -- it works out to millivolts across 10 megs, so we don't need a super output power or super stiffness under load either. I always run these with at least a volt doubler, as that means they waste less power running at a lower input voltage for a given output. In this case, I simply used one of the 27pf caps on the original board as the series cap in the doubler, as that's plenty and helps current limit the thing in a fault condition, and the usual two fast diodes and that nice ceramic .01 uF output cap.
To control the input, I used an LM 317, with a 200 ohm output to adj pin resistor, and a 2k 1k pot to ground for the adjustment, which in this case puts me in the middle of the pot when it's on bogey - perfect, and plenty of range should I need this for something else. The one special trick here is that I use the same bench 13.6 v supply to run all these, and you want to avoid ground loop issues in this sometimes noisy environment. So to help with that, I put a 120 ohm resistor in series with the output ground, and a 10k ohm resistor in series with the positive 650v output.
There is another .01 uf filter in the preamp to take care of any "antenna noise" and this way the real ground is provided by whatever audio amp or counter I connect it to, not the noisy 13.6v supply that has "antennas" all over it to power other detectors, the ion extraction power supply and so on (also CCFL's btw).
Almost doesn't deserve a picture, since it's so simple, but here's what it looks like on the inside. I put the trimpot on the other side of the board so I can tweak it through a hole in the front panel.
Now, these guys radiate quite a lot of 50khz stuff, so I grounded the box panel and put a bit of copper tape in there to shunt that to ground to keep the air around it quiet.
And that's it. If I do any more of these, oh please let me lay out a PCB for it -- I've made maybe 5-8 of them so far for various things; the basic design goes to 3kv fine with parts value adjustments, so a similar one is also running our 3He tube which has most of the same issues with noise, or worse. The reason I haven't is I don't have a stock of standard little boxes to put them in, you really do need a box, and so I wind up customizing each one to fit in whatever box comes out of my junk-box-box.
This particular one draws almost 20ma under the design load. I could about cut that in half with tweaking the lm 317 resistor values, where most of the actual power is going!
It'll run on 6v or so on up to whatever the lm 317 can take, 40v or so.