Optimizing PI Coils

[Carl-NC]

I'm not sure I understand your question. However, I keep forgetting that you are using a Figure-8 coil. Did you dampen each individual RX coil separately or once after they are joined?

[green]

I'm not sure I understand your question. However, I keep forgetting that you are using a Figure-8 coil. Did you dampen each individual RX coil separately or once after they are joined?

I have been after you suggested I should damp each Rx coil separately in another thread. Not sure about this coil. Need to start listing coil specs when I make them, (before shielding, without leads) Tx_ inductance and resonance, Rx_ inductance and resonance for each Rx coil, added internal Rd across each coil. (After shielding with leads) Tx inductance and resonance. Any others? Including another decay chart. Closer target looks worse.

Including a spice analysis that I think should simulate to my bench circuit. Coil volts goes to almost 800V at coil off (2.5million A/sec) for about .5us. When the coil has decayed to 80V (250,000 A/sec), is the target signal being effected? If so, how low does the coil voltage need to drop to not effect target signal eddy decay? Maybe a stupid question?

Tx volts=coil volts/1000 on the charts.

[green]

Appears initial balance and maybe being a little under damped helps.

[green]

More testing this morning. Very little difference with Tx (Rd)=1000 or 1400. Appears the problem has been coil balance. Tx and the two Rx coils were wrapped with AWG28 magnet wire are detected if not balanced. Hoping a 133mm figure8 Rx made with AWG30 and a two layer Tx AWG30 connected in parallel (same as AWG7) will be less critical to balance. Another option, wind spiral Rx coils with AWG 30 or 34 magnet wire. Wind spiral Tx with AWG26 (7x34) stranded PVC coated copper wire. Stranded wire is a lot thicker so Tx would be about 5 times wider. Would it be better to wind same number of turns or same inductance as Tx wound with AWG30 magnet wire, both same mean diameter.

[Carl-NC]

I would probably keep N constant so the ampere-turns remains close. However, the peak amps at turn-off depends on inductance so it may take an added turn to really keep it constant.

[Elliot]

What first sample time are people trying to achieve here? I sample at 6uS, which I find fast enough. Is the goal 4uS or something? I haven't dabbled with IB coils yet, but that would be my next step to sample faster.

[green]

What first sample time are people trying to achieve here? I sample at 6uS, which I find fast enough. Is the goal 4uS or something? I haven't dabbled with IB coils yet, but that would be my next step to sample faster.

For me it's a learning experience to see what is needed to sample as low as possible. A lot of targets it's not necessary. For short time constant targets (targets<10 us) reducing the delay .7 times the time constant doubles the signal. For #9 lead shot (.7us time constant) that's .5us, one of the nuggets I have is close to that.

For a US nickel (10us time constant) 2us wouldn't matter, for #9 lead shot 16 times the signal does.

[Olly]

Here's a front-end I've been playing with, the idea is to sample with a 16-Bit 1MHz ADC (MCP33131-10).





optipulse.zip

The ADC is single-ended so D5 and D6 are used to force unity gain for the half cycle that I don't want.
The LT6201 has very low voltage noise but quite high current noise so the LT6202 actually gives better noise performance but at slightly reduced speed.
The LT6201 also has quite high input bias currents hence the matching networks on the non-inverting inputs for improved DC offset.
R13 has been added as the LT6201 has built-in protection diodes which may have slightly lower forward voltages than the 1N4148.

As you can see it is just about possible to start sampling at 4uSec and then it is still possible to extract the #9 lead shot curve difference at 5, 6 & 7 uSec.

Damping is fairly critical and R3 is tweaked for ever so slight underdamping.
L1 has the characteristics of one of my 3DSS coils. (323uH, 3.2R and 114pF).

[green]

Here's a front-end I've been playing with, the idea is to sample with a 16-Bit 1MHz ADC (MCP33131-10).

[ATTACH]56935[/ATTACH]

[ATTACH]56936[/ATTACH]

[ATTACH]56937[/ATTACH]

The ADC is single-ended so D5 and D6 are used to force unity gain for the half cycle that I don't want.
The LT6201 has very low voltage noise but quite high current noise so the LT6202 actually gives better noise performance but at slightly reduced speed.
The LT6201 also has quite high input bias currents hence the matching networks on the non-inverting inputs for improved DC offset.
R13 has been added as the LT6201 has built-in protection diodes which may have slightly lower forward voltages than the 1N4148.

As you can see it is just about possible to start sampling at 4uSec and then it is still possible to extract the #9 lead shot curve difference at 5, 6 & 7 uSec.

Damping is fairly critical and R3 is tweaked for ever so slight underdamping.
L1 has the characteristics of one of my 3DSS coils. (323uH, 3.2R and 114pF).

Tried your zip. Wasn't what I would expect. Looked at target signal, again not what I expected. Lot different than I get with my circuits but I don't see why.

[Olly]

I've just downloaded your circuit and the initial difference that I see is the target inductor polarity, just flipped it on mine so now the target signal is positve going as yours is, but overall output at V(ADC) remains unchanged...

[Olly]

Your target vs mine with polarity change on my target inductor and some stepping added to yours.



Your voltages are higher as my pulse width is only 50uS and yours are faster with the lower coil capacitance.

[green]

Your target vs mine with polarity change on my target inductor and some stepping added to yours.

[ATTACH]56948[/ATTACH]

Your voltages are higher as my pulse width is only 50uS and yours are faster with the lower coil capacitance.

You are correct. I was being stupid. Your statement reply #39 As you can see it is just about possible to start sampling at 4uSec and then it is still possible to extract the #9 lead shot curve difference at 5, 6 & 7 uSec had me thinking wrong. I can just see #9 lead shot with a 4us delay with my 133mm figure 8 coil. Your analysis looked like it would be easy to see it, mine too. Trying to find something that wasn't there.

[Olly]

You are correct. I was being stupid. Your statement reply #39 As you can see it is just about possible to start sampling at 4uSec and then it is still possible to extract the #9 lead shot curve difference at 5, 6 & 7 uSec had me thinking wrong. I can just see #9 lead shot with a 4us delay with my 133mm figure 8 coil. Your analysis looked like it would be easy to see it, mine too. Trying to find something that wasn't there.

No silver bullet I'm afraid .
I'm trying to put together a useful direct sampling system with the Raspberry Pi Pico and the ADC I mentioned earlier and was intrigued by your post about the #9 lead shot which of course led to me to trying out whether it could be usefully detected with my preamp configuration. It should be possible as long as noise can be kept to a minimum and therein lies the rub...

Most of the noise originates in the 1K input resistor (4nV/Rt Hz) but this can't really be dropped much lower as it affects the damping adversely.
The LT6200 has only 0.95nV/Rt Hz voltage noise but it's current noise through the 1K input resistor gives around 3.5uV/Rt Hz so using it's slightly slower sibling the LT6202 with slightly more voltage noise but considerably less current noise gives an overall better noise figure but sadly its reduced bandwidth (100MHz vs 165MHz) makes detecting the #9 lead shot more difficult.

Everything's a trade-off I'm afraid but that's the fun of the game.

Cheers,

[green]

More testing this morning. Very little difference with Tx (Rd)=1000 or 1400. Appears the problem has been coil balance. Tx and the two Rx coils were wrapped with AWG28 magnet wire are detected if not balanced. Hoping a 133mm figure8 Rx made with AWG30 and a two layer Tx AWG30 connected in parallel (same as AWG7) will be less critical to balance. Another option, wind spiral Rx coils with AWG 30 or 34 magnet wire. Wind spiral Tx with AWG26 (7x34) stranded PVC coated copper wire. Stranded wire is a lot thicker so Tx would be about 5 times wider. Would it be better to wind same number of turns or same inductance as Tx wound with AWG30 magnet wire, both same mean diameter.

Thinking of trying the smaller wire to minimize it being detected. Rx, AWG30 magnet wire. Tx, AWG30 magnet wire or AWG26 PVC coated. Wind the same mean and number of turns as my AWG28(Rx and Tx) coil for comparison. Would need to order the #26 stranded wire. The #26 stranded would make a wider coil. Any reason to believe coil resonance and signal amplitude vs target distance would be much different between the different coils?

[bbsailor]

Thinking of trying the smaller wire to minimize it being detected. Rx, AWG30 magnet wire. Tx, AWG30 magnet wire or AWG26 PVC coated. Wind the same mean and number of turns as my AWG28(Rx and Tx) coil for comparison. Would need to order the #26 stranded wire. The #26 stranded would make a wider coil. Any reason to believe coil resonance and signal amplitude vs target distance would be much different between the different coils?

Green,

What is the inductance of the TX coil.
What is the total resistance of the TX coil drive circuit, including:
1. Coil resistance
2. MOSFET coil driver on resistance
3. Any series resistance in the coil circuit

With this information we can see the coil current rise Time Constant.

Thanks
Joseph J. Rogowski