by Doug Coulter » Mon Nov 05, 2012 10:43 am
Notes on use and tweaking this design.
The top and bottom ground planes are tied. You can bolt this right to a metal grounded base, solder it down, whatever. You can do this by having a hole in the mounting plate to solder through, (be careful about too much heat) or a bunch of little short wires to the edges of the ground plane on top. I've also drilled a hole in a corner and bolted it down.
Ground is the negative side of your 9v battery as well. I put a switch in series with the positive side of the battery - while battery life tends to be decent, if the preamp is tuned for much drop across the output 220 ohm resistor, it will still eat batteries. Gain or bias changes can tune or retune for this - you only want a fraction of a volt (say about 1-2 ma total drain on the 9v).
The pads are on the tiny side, with small clearances to ground so they are a little hard to solder. My technique is to put a tiny blob of solder on the pad first, tin the wire (perhaps with a very short 90 degree bend on the end) and then heat it and stick it in the blob. Make sure (magnifying glass) there are no shorts to ground after this. I usually mount the board first, then put on the wires, so anything "fat" like say, the lead of the input coupling capacitor (you supply that for most situations) which is stiff, is not under stress when soldered down. Eg you do your lead bending first, not after, soldering.
To tweak gain, change the resistor that is sideways (the only one). On a stock unit, this is a 15k ohm resistor (R6). I've used up to 47k with good results on the weaker output tubes. You may want to re-bias to keep the headroom right and the battery draw low. This is done by altering the r2, and r3 resistors. Making them larger (usually both at once) reduces the current drain or drop across that output resistor, R10, which is what you want to get the biggest output signal headroom. If things are drawing too much current, you'll want to do this, it also reduces recovery time in any coupling capacitors in the output, and this has one.
Choose your input coupling capacitor according to the pulse width from your detector. For slow tubes, I use .01uf, rated at a couple of KV higher than the actual voltage it sees, to reduce corona and leakage noise. If you've got a real fast pulse situation, drop that to 1/10 or so (1000 pf or less).
This is designed for negative-going pulses in and out. You may need a pullup on your output to keep the "resting" voltage at a binary 1 for your logic. This will drive TTL or CMOS, but likes CMOS better since it draws less current on lows and therefore doesn't put much charge on our output coupling capacitor. This is pretty important when count rates get high. There is provision for pullup and pulldown resistors on the board if you can handle soldering 0802 size components, but you can do this externally to your logic supply just as well if not better.
I use the DC coupled input with an external coupling capacitor for most things. The one place you don't need a capacitor here is with a phototube input with a floating anode and negative high voltage on the cathode and dividers. This works great if the dark current is very low (and it'd better be or you have a light leak somewhere - but measure that before hooking it up). This is also the time you use that pad labled "coax rtn" which is floating up at near 9v, since this is the input "signal ground" so this return won't capacitively couple noise into the input. Only connect one end of the coax shield - the board end.
For the output, ground is ground and it doesn't matter so much after all that gain.
This isn't purely DC coupled due to R4 and C3 - it's not meant to pass huge many-millisecond pulses, those components provide a bit of a high-pass action, which also reduces the 1/f noise in the transistors.
I try to put the battery right inside the enclosure with this and the sensor. This is just a noise pickup situation, though we do have a bit of a filter on the input power.
Beware corona noise from your HV detector supply and related parts, it can be as large as the desired signal! Rounded edges and glyptal or HV "dope" are the rule there.
These are so sensitive, the way I test them is to hang a scope on the output, about 1v/div, and simply hook a 1" wire to the input. touching or merely bringing your hand near it should produce very visible output. If there's a CCFL or LED lamp nearby, you should get 7-8 volt signal out from the noise it generates at the switching frequency of the bulb electronics. Obviously, if you're using a CCFL inverter for your detector HV, it'd better be in a separate box or a very well shielded compartment in the same box. I've used soldered copper flashing to make boxes here - works a charm if you solder all the overlapping joints tightly.
Small aluminum boxes work well if the power supply has its own box and is a few inches away (use coax to move the HV to the detector, and put 1k ohms or so in each wire and a bypass at the detector end) - this will eliminate ground loop noise and at the HV, where current is minimal, you can get away with large series resistors in these connections so your output coax to the counter provides the real system signal ground.
Choose your input coupling cap based on pulse width from the detector. For the Russian tubes I use .01 uf, rated at least 1 kv higher than what it really sees. This is to reduce noisy leakage across the capacitor. If you're got a really fast (narrow) pulse, cut that to 1000 pf or less (like from a phototube and fast scintillator where you might have only 10 ns pulse width).
As noted above in red, the tiny 680pf cap on the ac coupled input will not stand off HV - 50v or so is the limit. You really can't stand off much voltage at those tiny PCB spacings even if you could find a capacitor rated for that.
Clean is the rule - clean all insulators with alcohol or something similar, and don't touch them after that. This would include the coupling cap you use, any shielding around the HV series resistor and so on.
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.