"Ideal" coil inductance for PI

[Searchy]

From what i have seen so far, most coils for PI are at around 300...400uH.
Is there a somewhat "ideal" coil inductance?

If i e.g. half the inductance, i also reduce the capacitance, the resistance and thus the TC.
Ofcourse i get less sensitivity, but i have a longer integration time and can start at an earlier point of the target decay which should give me even higher voltage to be integrated (at least at the beginning)?
I'm sure i did a mistake here, but which?

[Kev]

Gid day Searchy,
I'll have a go. I think it was a compromise between battery power consumption, the time constant of a typical coin or ring, and adequate inductance to intercept the targets response, and possibly more things I can't draw down.

If you use lower inductance and more power you will need to wait longer for the eddys produced from field initiation to die down before pulse shutdown and sampling. Cutting your pulse short while the current was still rising will cause opposing eddys in the target to cancel part of the total response.

I think to take advantage of lower inductance TX, you need to use high voltage initiated Constant Current, where you achieve quick steady state, and thus less time for eddys to die down before off-time sampling (and sample the on time decay as well to double signal response) and use a separate RX coil of 400uH or so, then you overcome the compromises of the 300uH.

Cheers
Kev.

[Elliot]

H Searchy, what first time to sample are you trying to achieve?

[Kev]

Searchy my explanation was a bit esoteric, perhaps the attached images will explain it better.
100uH coil in both scenarios, but one is the traditional PI and the other is CC or Constant Current, similar to that used on the Minelab GPZ, although they use bi-polar pulsing. It will detect a nickel at 2 feet in an air test not even maxed out, this is one reason why.

The Green Trace is the current and the white trace is a simulated signal as seen by a target with a 25uS time constant.
Firstly look at the traditional PI with 100uH coil, you will notice that the coil current is still rising when the pulse ends, also notice that the target signal is negative at this time, this negative value will subtract from the positive value when you go to sample. Any changing magnetic field will cause eddys in your target, positive or negative.

Looking at the CC PI you can see that the response in the target has almost returned to 0V when the pulse ends, now look at the value of the positive response, in effect you get nearly twice the signal for a similar amount of current. Plus look at the response just after turn-on, you can also sample the turn-on decay as well and cut you pulse rate in half and still receive almost twice the bang for buck.

For clarity there's the 300uH traditional PI also to compare, ideally you would design it to have steady state current, where it flattens out before terminating the pulse.

Cheers
Kev.

[green]

Searchy my explanation was a bit esoteric, perhaps the attached images will explain it better.
100uH coil in both scenarios, but one is the traditional PI and the other is CC or Constant Current, similar to that used on the Minelab GPZ, although they use bi-polar pulsing. It will detect a nickel at 2 feet in an air test not even maxed out, this is one reason why.

The Green Trace is the current and the white trace is a simulated signal as seen by a target with a 25uS time constant.
Firstly look at the traditional PI with 100uH coil, you will notice that the coil current is still rising when the pulse ends, also notice that the target signal is negative at this time, this negative value will subtract from the positive value when you go to sample. Any changing magnetic field will cause eddys in your target, positive or negative.

Looking at the CC PI you can see that the response in the target has almost returned to 0V when the pulse ends, now look at the value of the positive response, in effect you get nearly twice the signal for a similar amount of current. Plus look at the response just after turn-on, you can also sample the turn-on decay as well and cut you pulse rate in half and still receive almost twice the bang for buck.

For clarity there's the 300uH traditional PI also to compare, ideally you would design it to have steady state current, where it flattens out before terminating the pulse.

Cheers
Kev.

[ATTACH]50762[/ATTACH]
[ATTACH]50763[/ATTACH]
[ATTACH]50764[/ATTACH]

Why is the target decay slope for the CC different than the standard decay? After Tx off.

[Kev]

How do you mean different Green?
It begins from a higher amplitude, the corollary of the on-time decay.

I need to add another caveat, the coils I am simulating are Minelab coils, they have a resistance less than 1 Ohm.
Here is the same setup with a coil of 6 Ohms.....more sensible current...but a 1 Amp CC device would kick its' butt.

[Kev]

Green I'm using a 180V leading edge pulse of about 5us duration, then maintaining it with 8.4V

Cheers
Kev.

[green]

I'm thinking the decay after Tx off should be target decay, 25us TC. The one with CC appears to be different. It would be easier to see decay slope if Y axis was log.

[Carl-NC]

From what i have seen so far, most coils for PI are at around 300...400uH.
Is there a somewhat "ideal" coil inductance?

If i e.g. half the inductance, i also reduce the capacitance, the resistance and thus the TC.
Ofcourse i get less sensitivity, but i have a longer integration time and can start at an earlier point of the target decay which should give me even higher voltage to be integrated (at least at the beginning)?
I'm sure i did a mistake here, but which?

Assuming a mono coil, there are a lot of factors. On the TX side fewer turns produces a stronger magnetic field (the higher amps outruns the fewer turns) and has lower C. But on the RX side more turns produce a stronger EMF, but also increase the C which pushes out the sample delay. Increasing the turns helps the RX EMF more than it hurts the TX field, so you go as high as you can to achieve the optimum balance between TX field, RX EMF, and sample delay. 300uH seems to be what the industry settled on some time ago, and it may or may not be optimum depending on the application.

[Kev]

If you look at the target response just after pulse initiation the voltage is about -62nV then given the TC of the target is 25uS after one TC the voltage should be about 63% less i.e. ~23nV, and I recon that's about what it is, and after 100us, the pulse width and 4 TCs, it looks to be about 1.2nV and after 5 TC if the pulse had been 125uS it would be ~430pV.
It's not the slope that excites me, it's the amplitude.

[green]

I'm wondering about decay after Tx off. Can you post the sim you are using?

[Kev]

Sorry Green it has some features I don't want to disclose at this time.
I've provided enough that you should be able to replicate it.
Additionally though, use a high voltage PNP to switch in the 180V to initiate the pulse, it doesn't need to handle high current. Depending upon how you regulate the current 4 or 5uS on-time is all that's required, a resistor is enough for sims. Use diodes to isolate the HV from the low voltage part and place an 82 to 100uF 400V tank cap in the HV supply. The flyback can be diode steered back to the cap to recover some of the power. For application a modified Nixie PSU can provide the HV 30 to 50mA required.



Any variance in decay shape may stem from non-critical damping in one or more of the sims.
I didn't check the Voltage of each sim, I just wanted to quickly knock out the coil current and target response to demonstrate my point for Searchy

[green]

My attempt. Target TC=25us, peak coil current=1A, Tx on time=100us. Coil TC same as Tx on time. Don't know if IDEAL. Don't know a good way to make a CC driver so went the easy way and used a current source(problem is I can't find one in the catalog)maybe someone could suggest a circuit. Target signal is 15% higher and average coil current is 60% higher with CC. Target decay after Tx off same for both methods. Replace the MOSFET with a current source for CC sim.

[Kev]

Yes you are right Green, I get the same results, I must have a bug in my sim somewhere, even so the sims do not always transfer to reality, the reality is not as bad as described, and likely I have exaggerated something somewhere.

Cheers
Kev.

[Kev]

I've found the problem I think. When I uncouple the HV driver from my sim circuit I get 42% less max signal strength. So in your CC example the response would be 6.56mV rather than 3.8mV if the pulse had a HV leading edge.
My first attempt at CC I simulated an H-Bridge CC circuit, it worked really well virtually, but when I built it, it failed to work, I didn't use a HV pulse former to blast through all the parasitic C and reluctance etc.
The off-time is a corollary of the on-time within several TCs of the coil at least, it appears that what happens to the target at current-on affects what happens at current-off.

I hope I've got this right otherwise my whole circuit is mush.