Not Quite As Simple CSA

This area is for discussions involving any fusion related radiation metrology issues. Neutrons are the key signature of fusion, but other radiations are of interest to the amateur fusioneer as well.
User avatar
Carl Willis
Posts: 2841
Joined: Thu Jul 26, 2001 11:33 pm
Real name: Carl Willis
Location: Albuquerque, New Mexico, USA
Contact:

Re: Not Quite As Simple CSA

Post by Carl Willis » Wed Mar 03, 2010 11:02 pm

Hi Richard,

Thanks for your efforts with this design, and your previous contributions. I'll have to give it a try.

If I had one suggestion from just looking at your design, it would be to increase the decay time constant. Charge collection in proportional counters takes a long time, probably on average a few microseconds if the response is to remain linear in energy. So I'd advocate 30+ uS for the product of R1*C2 for that reason. An unrelated reason to make the decay time longer is that the pole-zero adjustment on commercial NIM amplifiers typically works between 50 microseconds and infinity. I think the only downside of using a few dozen MOhm would be the lack of availability at Radio Shack or in most junk boxes...what do you think?

-Carl
Carl Willis
http://carlwillis.wordpress.com/
TEL: +1-505-412-3277

Richard Hester
Posts: 1458
Joined: Thu Jul 05, 2001 4:07 am
Real name:

Re: Not Quite As Simple CSA

Post by Richard Hester » Thu Mar 04, 2010 2:35 am

I have some extra intelligence on that, Carl - I was looking up data on Boron-lined counters, and a group in Washington state is evaluating them as a replacement for He3 tubes in portal monitors due to the He3 shortage (sound familiar?). They were talking about using a short time constant to aid in gamma discrimination. It's something that could be played with, as we don't expect a lot of gamma action with the fusor, and it doesn't take a lot of lead to screen out X-rays and such.The paper is one of the first things that pops out if you do a search for "B10 counter tube" or "B10 proportional tube" or something similar. Anyway, it's easy to tailor the time constant by playing with the discharge resistor. The larger you make that resistor, though, the more chance you have of leakage starting to affect your results. I also don't trust normal -type resistors too much over 10M or so. I'd keep the charge storage cap small because of the low amount of picocoulombs involved. When you have a big, fat pulse like from a PMT, then you can splurge on the picofarads.

Richard Hester
Posts: 1458
Joined: Thu Jul 05, 2001 4:07 am
Real name:

Re: Not Quite As Simple CSA

Post by Richard Hester » Thu Mar 04, 2010 2:40 am

To Doug - I really dont care, either. I messed with CSAs for upwards of a year developing various circuits and know what I've seen. The circuit posted is a simplified version of a CSA available from Bicron - I even posted the schematic in files.

nemesistech
Posts: 64
Joined: Thu Jul 26, 2007 2:17 pm
Real name:

Re: Not Quite As Simple CSA

Post by nemesistech » Thu Mar 04, 2010 3:06 am

How bout I ask simple questions that I am sure most of you can answer with ease. I am still in the learning stages with building circuits, so forgive me if these questions bore you.
First, should I assume you would want to use metail film resistors instead of carbon?
Second, on Q3 transistor, is that one piggybacked on the other, if so, what is the purpose of this?

User avatar
Doug Coulter
Posts: 1312
Joined: Sun May 27, 2007 7:18 pm
Real name: Doug Coulter
Location: Floyd, VA, USA
Contact:

Re: Not Quite As Simple CSA

Post by Doug Coulter » Thu Mar 04, 2010 4:30 am

Richard, sorry, bad day at the office, you know. And for once it's Carl being polite and me abrasive ;~) Not really called for....I do what I do and see what I see too, but when that conflicts with predictions and book learnin, I try and find out why -- I suspected it may have been something not obvious in that case as I do also have a lot of experience with things like this. I agree with Carl on the time constants here -- this stuff is on the way slow side compared to things like PMT's.

To Mike , that last transistor is two in one package, in the Darlington configuration, used as an impedance reducing device (eg it will drive more current with this than without it). In common collector mode, it's just a gain of approximately one for voltage, but has a ton of current gain. This way the design can use that 82k resistor in the pnp for good voltage gain, without having the load you put on the preamp affect that (the output of the pnp is a current which the resistor changes to a voltage), and it will drive a hefty load (which helps with noise pickup downstream, keeps the gain stable, and other good things like that).

So think of it as a gain of one (for voltage) buffer that doesn't need much current to drive it, but will put out far more current than it takes to drive it. You can do this with two transistors in separate packages too, and the net current gain is that of the two transistors multiplied -- so the current gain gets big quickly. Darlingtons are usually slow, particularly in turn off, as there is no resistor to discharge the base-emitter capacity of the second transistor when the first turns off (or maybe there's one inside the package and just not shown in this case, I'm not familiar with that particular number).
Why guess when you can know? Measure!

Starfire
Posts: 1482
Joined: Wed Oct 24, 2001 6:14 pm
Real name:

Re: Not Quite As Simple CSA

Post by Starfire » Thu Mar 04, 2010 12:19 pm

Less talk more build - suck it and see - general comment not direct at any one in particular

User avatar
Carl Willis
Posts: 2841
Joined: Thu Jul 26, 2001 11:33 pm
Real name: Carl Willis
Location: Albuquerque, New Mexico, USA
Contact:

Re: Not Quite As Simple CSA

Post by Carl Willis » Thu Mar 04, 2010 7:07 pm

Hi Richard,

I have a preamp at home, derived from probably the same Bicron preamp you mention, on which I use a regular 10M resistor and a 5.6-pF, NPO-grade ATC cap. The cap and resistor, as well as the protection diodes, form a self-supporting "dead-bug" structure. It works well with proportional tubes in my experience, but I cannot comment on its temperature stability or noise figure or anything like that. I do know that the observed time constant--when fed from a charge-terminated pulser--is not any different from what it should be ideally.

Thanks for clarifying your thinking about the short time constant. I had no idea B-lined tubes were making a comeback.

To your point about construction technique, I'll just add that I think it matters a lot. My feeling is that stray capacitance is probably what causes observable differences from the ideal when building the input stage solely on PCB pads, but it could be surface leakage as well, or some of both. I am in the habit of building the rest of the boards by "Dremel etching" like a total amateur, but the input stage is indeed unpredictable if built this way. Ground loops are a huge issue with preamps. Microphonics are a problem also. Power supply filtration is an issue. Cleanliness is a huge issue especially if the load resistor and HV components are in the preamp also. Leaky BNCs...leaky Victoreen resistors...rosin threads...humidity...all very relevant issues, not easy to avoid. I wash with methanol ("Heet") dried over zeolite pellets to get rid of rosin and humidity. Does not always effect a cure, though.

-Carl
Carl Willis
http://carlwillis.wordpress.com/
TEL: +1-505-412-3277

User avatar
Richard Hull
Moderator
Posts: 12855
Joined: Fri Jun 15, 2001 1:44 pm
Real name: Richard Hull

Re: Not Quite As Simple CSA

Post by Richard Hull » Thu Mar 04, 2010 8:26 pm

As I first noted above, I just got used to cleaning and floating gate stuff in electrostatic/electrometer systems I have built and worked with and have seen it done on virtually every single charge preamp from GeLi preamps to 3He and BF3 NIM preamps. I'll do it reflexively until the day I die. Once the faraday case or shell closes, it should be a sleeping dog for a long time and require no revisiting.

I do hope to construct this thing soon, just 'fer grins and googlers.

Richard Hull
Progress may have been a good thing once, but it just went on too long. - Yogi Berra
Fusion is the energy of the future....and it always will be
Retired now...Doing only what I want and not what I should...every day is a saturday.

DaveC
Posts: 2346
Joined: Sat Jun 30, 2001 5:13 am
Real name:

Re: Not Quite As Simple CSA

Post by DaveC » Thu Mar 04, 2010 10:54 pm

In response to Doug's earlier comments on characteristic impedances.

Hey - I said "75 ohms or thereabouts"


Z0 depends on the Log of the diameter ratios, not ratio directly... thus pretty insensitive to size, right?

Further, within a factor 2 or 3 is "Exact" for physics discussions, and not even bad for arm waving electronics.

So what's your beef, eh?


Dave Cooper

User avatar
Doug Coulter
Posts: 1312
Joined: Sun May 27, 2007 7:18 pm
Real name: Doug Coulter
Location: Floyd, VA, USA
Contact:

Re: Not Quite As Simple CSA

Post by Doug Coulter » Sat Mar 06, 2010 1:09 am

No beef at all Dave -- you are absolutely correct on that one, and here it's just not an issue, as we are far shorter than any length that constitutes transmission line behavior for the risetimes involved anyway.

My apologies, I guess I should more often calm down before posting sometimes.
On email, they can't see you grinning and armwaving, so I get in trouble.

Ever notice that stuff that comes off as mere agitated excitement in person sounds pretty bad (or can be taken as such) on email/print? This one bites me now and again. People who have met me in person know it's just me spouting off, but others can't know. Sorry 'bout that.

I have a two transistor circuit I'm using on the 3He tube (and in a lot of other places) I like better myself, and it doesn't need all the capacitors, which is nice both for complexity and for recovery times on overload. This circuit in that use also has builtin adjustable threshold for gamma discrimination.
As I have two B10 tubes, I may try a version optimized for them (that's a somewhat tougher signal) and let everyone know.

The circuit under discussion seems to be derived from the second one here, also in the same book.
(1980 linear apps handbook)
There are some interesting variations of it that provide for almost zero input capacity, due to bootstrapping the fet G-S cap, and eliminating the G-D miller cap by use of the cascode -- the drain basically is driving an emitter, and there's almost no signal volts there -- it's all current.

Which made it a great place for the inductor -- since that's a low impedance node, you don't need one as large, and using an inductor protects the rest of the circuit from the wild variations in Idss most fets tend to have from unit to unit (if not from moment to moment, and temperature to temperature).

Joe Sousa and I try to eliminate all L's and C's where we can and it makes sense. Old habits -- he's a chip designer where you can't have much of either (but where taking a few transistors to make a current source is nothing), and I'm an old cheapskate with a weird idea of what constitutes "elegant", and who used to have to fix up behind designers of mass market stuff, and guess what made all the troubles -- caps, which at least we could get, and inductors, which in some random TV set or stereo, just try to find out what it even *was* so you know what to replace it with, and more often than not the Q and self resonance was as important as the inductance to the thing working right. Nightmare for a tech, which is what I was, coming up in the '60s. I later graduated to EE, then Scientist, but I didn't forget some fantastic lessons I learned about repeatability and reliability as a service tech. We saw what really worked and of course, what didn't -- many times a day.

I first saw this circuit in a very good early solid state Marantz reciever, used in the tone controls, then some years later another version was in the National Semi apps book used at the other extreme of impedances for a moving coil phone cartridge preamp that was near noise theoretic at room temp.

By changing some speeds and feeds, it's in use on my 3He tubes, my BF3 tubes, and some phototubes, all just by changing some values. Very simple, low noise if tuned right for the impedances involved, cleanly closed loop system that can also be an integrator of the type wanted here if you add a cap across the NFB R. Works like a champ, and is not real sensitive to power supply issues. You can use a fet for the input device if you make some small changes for bias issues.
I don't usually have to, with the current gain of the transistor pair with NFB, the input impedance can be plenty high with plain old bipolars (and therefore more rugged, and in some eyes, more elegant).

Here's a scan from the National book. In most other uses, you don't use their fancy low impedance transistor pair, but the good old 2n3904/3906 kind of thing, much lower currents, and in the case of the tubes that put out a negative pulse, use a pnp in the input, then npn (power supply reversed) since the tube will turn on the pnp better with the negative pulse.... and then you can fool with the bias to put a threshold on there -- no parts. In my practice here, I do that via a simple pot across the signal input (after the input cap) to adjust the tube net gain to match tyhe Vbe threshold with no other input bas at all. (I run my tubes higher volt than spec to get into a good gas-gain region and they drive this fine -- out come 5v pulses at decently low impedance).

So basically, all the R's in the national circuit are much higher value, and for some things I do the NFB differently -- sometimes none if I just want a count/no-count situation and am using input bias (or the lack of it) to set the threshold.

In the third one we see the bootstrapping, which can only work if you have a gain == 1 node someplace to play with (however you get it, I've seen people use dividers off a gainy circuit to make that happen). Pretty cool, actually, and this is close to what we wound up with to take EEGs with pure capacitive input coupling -- no shaved heads and no goo needed.
Same class of problem but at a very different passband.

I have a bias toward real pc board building, since it's really not hard these days, and when I did this for a living, of course that was all there was. I use a freeware software tool (Protel TraxEdit, very old dos program, and IMO better than their current multithousand buck offering by far -- simple), a laser printer, and pre-sensitized PCB (available all over). Even if I'm not real sure the thing will work, it's easier to cut the odd track or add the odd wire, change parts etc, and get going.
And once it's right -- it's right forever, every one you make is the same re all the leakage R's, parasitic C's and interwire coupling....worth the effort for me, because once you get set up, it's not much effort at all. Then I can send the same files to AP circuits in Canada and get them made really nice, pretty cheap, and very fast if I need a bunch. I have zip troubles with leakage unless I'm getting way out there on something. Real high volts and real high R's I air-wire as required -- there it doesn't take much conductance to make trouble, but at 5 volts, nah. In this case, the tube supply resistor wouldn't live on the PCB at all, of course, and neither would the input coupling cap as one end is very hot. At the normal solid state volts, up to about 10 megs, no issues whatever with PCB leakages, even if you need a pretty precise 10 megs. Above that, well then -- first find good enough parts at all, then yes, you have issues especially at high volts where tiny amounts of corona play hob with SNR.
Attachments
NatBS.gif
NatBS.gif (6.83 KiB) Viewed 2204 times
NatPre.gif
NatPre.gif (47.55 KiB) Viewed 2204 times
NatCascode.gif
NatCascode.gif (7.67 KiB) Viewed 2204 times
Why guess when you can know? Measure!

Post Reply