Ed Nisley's Blog: Shop notes, electronics, firmware, machinery, 3D printing, laser cuttery, and curiosities. Contents: 100% human thinking, 0% AI slop.
They’re glass electrometer resistors from late in the Cold War:
Russian 100 G electrometer resistor
That one presents 100 GΩ between its lead wires, which would count as open in any other circuit I’ve ever built.
The assortment arrived much richer than advertised, although I’d be even happier with a few more 10 GΩ and a few less 100 MΩ resistors. The 1000 GΩ = 1 TΩ resistor in the upper right seems absurd on the face of it, but there it sits.
I have no way to measure these, other than to build an electrometer amp and see what happens…
Given the ionization chamber’s tiny currents and the huge resistors required to turn them into voltages, reviewing the thermal noise I generally ignore seems in order…
The RMS noise voltage of an ordinary resistor:
vn = √ (4 kB T R Δf)
The constants:
kB – Boltzman’s Constant = 1.38×10-23 J/K
T – temperature in kelvin = 300 K (close enough)
Mashing them together:
vn = √ (16.6x10-21 R Δf)
vn = 129x10-12 √ (R Δf)
For a (generous) pulse current of 20 fA, a 10 GΩ resistor produces a mere 200 μV, so wrap a gain of 100 around the op amp to get 20 mV. An LMC6081 has a GBW just over 1 MHz, giving a 10 kHz bandwidth:
vn = 129x10-12 √ (10x109 10x103) = 1.3 mV
Which says the noise will be loud, but not deafening.
A 100 GΩ resistor increases the voltage by a factor of 10, so you can decrease the gain by a factor of ten for the same 20 mV output, which increases the bandwidth by a factor of ten, which increases the noise by a factor of … ten.
Ouch.
With the same gain of 100 (and therefore 10 kHz bandwidth) after the 100 GΩ resistor, the output increases by a factor of ten to 200 mV, but the noise increases by only √10 to 4 mV.
The LMC6081 has 22 nV/√Hz and 0.2 fA/√Hz input-referred noise, neither of which will rise above the grass from the resistor.
It’s a good thing I have a pretty deep parts stock, as one of the caps didn’t fit into its holes at all.
The Russian CI-3BG glass tube, according to the datasheet and discussion on MightyOhm, is sensitive to gamma and beta radiation, so it should serve as a simple cross-check on my ionization chamber results. It’s not clear the C8600 is applying the correct voltage to the CI-3BG tube, but it probably doesn’t make much difference; the supply is so feeble that there’s no way to actually measure the results.
A closer look at the CI-3BG suggests the active volume lies inside that spiral-wrapped section between the white insulators:
Russian CI-3BG Glass Geiger Tube – detail
In round numbers, that section is 6 mm long and 3 mm OD. Figuring the ID at 2.5 mm, that’s a volume of 30 mm3 = 0.030 cm3. That’s maybe 1/7300 of the ionization chamber volume, so, (handwaving) assuming roughly equal sensitivity, the chamber should report three orders of magnitude more pulses than this little thing.
It’s mildly sensitive to a radium-dial watch and perks up when a watch hand lines up along the spiral-wrapped volume. Given that the radium decay sequence spits out betas and no gammas, the (scaled) count may be a bit higher than the ionization chamber produces, but there are so many other imponderables that it might not matter in the least.
Feeding the output voltage into the ‘scope, with AC coupling to strip off the DC bias, produces this:
Darlington 12k load – multiple
Those cute little spikes seem to be gamma ray ionization events: they are always positive-going, there are no similar negative-going pulses, they occur irregularly at a few per second with occasional clusters, and generally seem about like random radioactive events. The picture shows a particularly busy interval; mostly, nothing happens and the baseline voltage wobbles around in a low frequency rumble.
For what it’s worth, the shielding around the circuit completely eliminates not only 60 Hz interference, but everything else, too: astonishingly good results from a fairly simple layout.
Taking a closer look at one pulse:
Darl 12k – single detail
(Vigorous handwaving begins)
The tallest spikes are typically 20 mV above the baseline, corresponding to peak output current of 20 mV / 12 kΩ = 1.5 µA and a chamber current of 1.5 µA / 100×106 = 15 fA.
They’re generally 5 ms wide, which is orders of magnitude longer than the actual ion generation time, but the area under that spike should be more-or-less proportional to the area under the actual impulse.
If you grant that and agree those pulses look mostly triangular, their integral is:
1/2 x 15 fA x 5 ms = 40 fA·ms = 40 aC
That’s “a” for “atto” =10-18 = a billionth of a billionth = hardly anything at all.
Indeed, seeing as how one coulomb contains 6.2×1018 electron charges, that pulse represents 250 ion pairs, at least assuming a zero-current baseline.
Gamma rays arrive with various energies, produce ionization trails of various lengths, and don’t necessarily traverse the entire chamber, so the pulses have various heights & widths; you can see smaller pulses sticking up out of the grass in the first scope shot. Assuming all those average out to five “big” pulses every second, the chamber collector electrode passes 200 aC/s into the transistor base → 200 aA → 0.20 fA. At 1 fA per 100 µR/h, that’s 20 µR/h of gamma background.
Working from the other end of the scale, a bit of searching shows that 1 R produces 2.08×109 ion pairs in 1 cm3 of dry air at STP. The ionization chamber dimensions give the can’s volume:
π x 4.52 x 3.5 = 220 cm3
So assuming a somewhat unreasonably large pure-gamma dose of 10 µR/h in that volume will produce:
10x10-6 x 2.08x109 x 220 = 4600x103 ion pairs/h = 1300 ion pairs/s
That’s about five “big pulses” per second, under the stack of assumptions thus far, and seems absurdly close.
An old NIST report on Calibration of X-Ray and Gamma-Ray Measuring Instruments says that 1 R/s (that’s per second, not per hour) produces a current of 300 pA/cm3 in an “ideal ionization chamber”. Scaling that down to 10 µR/h and up to the chamber volume gives an average current of 180 aA. That’s absurdly close, too.
Note bene: Because 1 C = 6.241×1018 ion pairs, 2.08×109 ion pairs is 333×10-12 C and, if you do that in one second, you get 333 pA of current from your ideal 1 cm3 ionization chamber. Those two approaches should be equally close.
(Vigorous handwaving ends)
Again, I don’t trust any of the values to within an order of magnitude and surely made a major blunder in running some of the numbers, but the results seem encouraging.
The coaxial cable’s capacitance could explain why the pulses look like triangles: the capacitance integrates a rectangular current pulse into a voltage ramp. The cable measures 200 pF and the scope input adds 13 pF, but let’s call it 200 pF across the 12 kΩ emitter resistor. Raising the voltage across that capacitance by 20 mV in 2 ms requires a current of:
200x10-12 x (20 mV / 2 ms) = 2 nA
Dividing that by 100×106 gives a chamber current pulse of 20×10-18 = 20 aA: three orders of magnitude less than the original guesstimate. That suggests the (handwaved) 15 fA chamber current, amplified by the absurd gain of two stacked Darlingtons, easily drives the cable capacitance. Something else causes the ramp.
The chamber itself has 10 pF capacitance, but it’s not clear to me how (or if) that enters into the proceedings. The entire collection of ions appears in mid-air, as if by magic, whereupon the +24 V chamber bias voltage draws them (well, the positive ones, anyway) to the transistor base without appreciable voltage change.
Perhaps the triangle represents the actual arrival of the ions: a few at first from the near side of the trail, a big bunch from the main trail, stragglers from the far side, then tapering off back to the baseline.
That’s definitely more than anyone should infer from a glitch produced by a pair of transistors…
The Victoreen 710-104 ionization chamber specs say it produces 1 pA in a 100 mR/h gamma environment, which suggests the actual current will be much, much lower in the Basement Laboratory. In fact, I’m hoping to spot individual gamma rays, rather than the overall radiation background current, which will involve counting groups of electrons as they march by.
The simplest possible electrometer amplifier, an MPSA14 NPN Darlington, produced 25 nA of current that, assuming a gain of 10 k, corresponds to an unrealistically high 2.5 pA of chamber current and is, realistically, entirely leakage current.
Adding an MPSA65 PNP Darlington boosts the overall gain to maybe (10 k)2 = 100 x 106:
Current Amp – Dual Darlington – Schematic
Granted, there’s not much to like about that circuit (“Any sufficiently sensitive instrument is indistinguishable from a thermometer”) and stuffing that much gain into a pair of inverters is basically crazy talk, but it looks like this:
Electrometer amp – circuitry
The blue trimpot in the foreground drives the base of a duplicate pair of transistors in a misguided attempt to make a differential amp that would balance out some thermal effects. Turned out to be not worth the effort, due to the adjustment’s fiddly nature, but also not worth unsoldering the parts.
The black lump covers two RG-174 coaxial cables that run off to the oscilloscope; they already had BNC connectors on the end and were small enough for the job.
Some DC measurements:
The output idles at 6.5 VDC → 550 μA of Q101 collector current → 6 pA of Q102 base current. Yeah, right.
Grounding the base of Q102 → 5.9 V output → 500 µA → 50 nA leakage into Q102’s collector. Maybe.
Shorting Q101’s base to its emitter produces 350 mV at the output → 30 µA of output current. Huh.
After restoring the status quo ante, the output idled at 10.2 V. See previous comment about thermometers, modulo soldering transistor leads.
So, given the predictably absurd temperature sensitivity of this whole lashup, it’s reasonable to say the entire DC output current comes from leakage, which also agrees with the fact that I’m not dying of gamma exposure. In point of fact, an ancient CDV-715 Radiological Survey Meter with a similar ionization chamber (which, at this late date, passes its “circuit test” function and seems to be perfectly happy) reports exactly zero background on its most sensitive 500 mR/h scale.
Lining the shield support box with copper foil tape turned out to be surprisingly easy:
Electrometer amp – shield – end view
The flat surface is two overlapping strips of 2 inch wide copper tape. I traced the exterior of the support box on the tape, cut neatly along the lines, slit the corners, bent the edges upward, peeled off the backing paper, stuck the tape into the box, pressed the edges into the corners, and didn’t cut myself once.
Applying 1 inch wide tape to the wall went just as smoothly, after I realized that I should cut it into strips just slightly longer than the hexagon’s sides.
The tape along the rim is adhesive copper mesh that’s springy enough to make contact all around the edge. I cut the 1 inch wide tape in half, which was just barely wide enough to reach::
Electrometer amp – shield – mesh soldering
Although you’re supposed to join the entire length of each seam for best RF-proofing, I tacked the corners and the middle of the long edge, then hoped for the best. The copper mesh seems to be plated on plastic threads that requires a fast hand to solder without melting, but I’m getting better at it. The adhesive is said to be conductive, but I loves me some good solder blob action.
The resistance from the flat bottom to the side panels and the fabric on the edge started out at a few ohms before soldering and dropped to 0.0 Ω after soldering, so I’ll call it a success. Didn’t even melt the outside of the PETG box, but I admit I didn’t take it apart to see what the copper-to-PETG surface looks like.
Covering the foil on the sides with 1 inch Kapton tape completed the decoration. I didn’t bother to cover the flat surface, because none of the circuitry should reach that far, and didn’t worry about covering the fabric tape for similar reasons. As madbodger pointed out, this violates the no-plastic-on-the-inside rule, but I’m still hoping for better results than having the entire plastic structure with all its charges on the inside.
A strip of horribly clashing orange plastic tape (which might be splicing tape for reel-to-reel recording tape) covers the outside edges of the fabric, prevents fraying, and gives the black electrical tape that holds the box down a solid grip:
Electrometer amp – shield – exterior
Yeah, like you’d notice mismatched colors around here.
Using black tape as an anchor seemed easier and better than messing with nesting pins & sockets. The copper fabric tape makes good contact with the rim of the PCB all the way around the perimeter and the black tape holds it firmly in place.
Early reports suggest the shield works pretty well…
The Smell of Molten Projects in the Morning 