TDR of a neutron detector tube
Posted: Wed Aug 08, 2018 10:57 pm
Did anyone else ever wonder how a thin-wire detector tube would behave as a coaxial transmission line? I found out in the lab the other day. The subject was my BF3 tube that we talked about here 19 months ago. viewtopic.php?f=46&t=11254&p=74207
At very high frequencies, it looked like an unterminated transmission line with a prop delay of 0.75 ns and a characteristic impedance of about 320 ohms. That's remarkably high for a coaxial geometry. It's consistent with a conductor diameter ratio of about 200. For example, 0.8 inches and 0.004 inches.
It was easy to measure by Time Domain Reflectometry, using an instrument that's ubiquitous in my line of work. The voltages are 10000 times smaller, and timescales 1000 times faster, than those of interest for neutron detection. TDR detects spatial variations in the impedance of an electrical transmission line. Here's one comprehensive tutorial: www.tek.com/dl/55W_14601_2.pdf
. Instrument in picture above is connected to a coaxial cable, SMA-to-BNC adapter, BNC-to-HN adapter (or so I was told here), and the gas-filled tube under test.
Last night's observations are adequately matched by a simple electrical model:
* Coax cable impedance of 50 ohms.
* The connector adapter stack amounts to about 0.2 ns of 50 Ω line.
* The tube is another 0.25 ns of 50 Ω line, then 0.75 ns at impedance of about 320 Ω, then open circuit.
I'd never even heard of impedance that high in a coaxial geometry. The vast majority of cables are designed for 50 Ω or 75 Ω. Value depends on the insulator's dielectric constant, and on logarithm of the conductor diameter ratio. It's the same as sqrt(L/C) per unit length of the cable medium. Here are some formulas and an online calculator: https://www.pasternack.com/t-calculator ... utoff.aspx
Guess there's time for a closer look at the display and its interpretation.
The screen shows two saved traces and one live trace. Horizontal scale is 2 ns/div. Vertical scale is 200 milli "rho" per division, referring to the ratio of reflected voltage to incident voltage.
Green trace is TDR waveform with nothing (open circuit) at the end of the probe cable. The instrumented end of the cable is connected to an oscilloscope channel and a voltage step generator. (A few hundred mV, with combined risetime < 50 ps, just like similar level instruments in the 1970's.) At the exposed end of the cable, the step reflects back with +100% amplitude and is sampled by the 'scope.
The red trace is waveform with both connector adapters in place, but no BF3 tube. Red step is later than green step by the amount of connector delay, round-trip. Red trace has some minor wiggles from impedance imperfections at the connections.
White trace shows what happens when the tube is connected. The 50 ohm enviroment, with no major reflections, continues for a while past the HN connector. At the beginning of the 320 ohm section, the voltage step is mostly reflected but partly transmitted, according to the rules for such things. The transmitted step is 100% reflected at the far end of 320 Ω section. At the 50 Ω junction on the return trip, it's partly transmitted to where we can see it, but mostly re-reflected with reversed voltage polarity. It rings back and forth in the 320 Ω section, which is very poorly matched on both ends.
For extra fun, here's a simulation produced with a freshly downloaded copy of LTSpice.
At very high frequencies, it looked like an unterminated transmission line with a prop delay of 0.75 ns and a characteristic impedance of about 320 ohms. That's remarkably high for a coaxial geometry. It's consistent with a conductor diameter ratio of about 200. For example, 0.8 inches and 0.004 inches.
It was easy to measure by Time Domain Reflectometry, using an instrument that's ubiquitous in my line of work. The voltages are 10000 times smaller, and timescales 1000 times faster, than those of interest for neutron detection. TDR detects spatial variations in the impedance of an electrical transmission line. Here's one comprehensive tutorial: www.tek.com/dl/55W_14601_2.pdf
. Instrument in picture above is connected to a coaxial cable, SMA-to-BNC adapter, BNC-to-HN adapter (or so I was told here), and the gas-filled tube under test.
Last night's observations are adequately matched by a simple electrical model:
* Coax cable impedance of 50 ohms.
* The connector adapter stack amounts to about 0.2 ns of 50 Ω line.
* The tube is another 0.25 ns of 50 Ω line, then 0.75 ns at impedance of about 320 Ω, then open circuit.
I'd never even heard of impedance that high in a coaxial geometry. The vast majority of cables are designed for 50 Ω or 75 Ω. Value depends on the insulator's dielectric constant, and on logarithm of the conductor diameter ratio. It's the same as sqrt(L/C) per unit length of the cable medium. Here are some formulas and an online calculator: https://www.pasternack.com/t-calculator ... utoff.aspx
Guess there's time for a closer look at the display and its interpretation.
The screen shows two saved traces and one live trace. Horizontal scale is 2 ns/div. Vertical scale is 200 milli "rho" per division, referring to the ratio of reflected voltage to incident voltage.
Green trace is TDR waveform with nothing (open circuit) at the end of the probe cable. The instrumented end of the cable is connected to an oscilloscope channel and a voltage step generator. (A few hundred mV, with combined risetime < 50 ps, just like similar level instruments in the 1970's.) At the exposed end of the cable, the step reflects back with +100% amplitude and is sampled by the 'scope.
The red trace is waveform with both connector adapters in place, but no BF3 tube. Red step is later than green step by the amount of connector delay, round-trip. Red trace has some minor wiggles from impedance imperfections at the connections.
White trace shows what happens when the tube is connected. The 50 ohm enviroment, with no major reflections, continues for a while past the HN connector. At the beginning of the 320 ohm section, the voltage step is mostly reflected but partly transmitted, according to the rules for such things. The transmitted step is 100% reflected at the far end of 320 Ω section. At the 50 Ω junction on the return trip, it's partly transmitted to where we can see it, but mostly re-reflected with reversed voltage polarity. It rings back and forth in the 320 Ω section, which is very poorly matched on both ends.
For extra fun, here's a simulation produced with a freshly downloaded copy of LTSpice.