SPICE Simulation Challenge

[Qiaozhi]

Despite saying "That's All Folks!", I may have solved the problem after all ... at least to some extent, and possibly good enough for our purposes.

The attached simulation is for a VLF detector with a concentric (Tesoro) coil. With this setup, all targets give a phase-shift left, non-ferrous targets increase in amplitude, and ferrous give a decrease. There are three simulation runs comprising "no target", a "100us TC non-ferrous (eddy) target", and a "100us TC ferrous (magnetic) target". In order for the ferrous target to give the correct results, the coupling between the target and the RX loop is decreased by 25%, and the coupling coefficient between the nulling and RX loop is increased from 0.71 to 0.7135. As you can readily see, it only takes a small adjustment to account for the distortion of the magnetic field (due to absorption) and produce the necessary amplitude decrease below that of the no target response.

Having tested various targets with a 9" diameter concentric coil connected to a Raptor circuit, I can see that the simulation results are a reasonably close match with reality. I would have expected that the phase-shift in the RX loop would increase with higher TC targets, but this is not the case. In fact, the phase-shift seems to be unrelated to the TC, and is more affected by the distance of the target from the coil. Also, it is important not to confuse the RX phase-shift with the phase angle calculated by sampling the in-phase and quadrature components of the RX signal. In this case, the phase angle is defined by:



You also have to remember that the double-differentiating architecture of Tesoro-like analog detectors are processing the rate-of-change of the rate-of-change of the RX signal. When a VDI circuit is attached to the outputs of the GEB and DISC channels, the GEB channel will always go positive when any metal target passes over the coil. On the other hand the DISC channel goes positive for non-ferrous and negative for ferrous. In this latter case the value of R is negated, which shifts the calculated value of by 90 degrees. By measuring the R and X values in the simulation, I was able to calculate that the VDI value for the 100us non-ferrous target is 173, and the value for the 100us non-ferrous target is 23. These results align well with the expected values.

If you want to run the same simulation for a DD coil, it is necessary to adjust the TX-RX coupling coefficient to account for absorption with the ferrous target, instead of the nulling coil, which (of course) is no longer there. The target to RX loop coupling coefficient will also need to be adjusted accordingly.

In the attached simulation:
Green = no target
Blue = 100us TC non-ferrous (eddy) target
Red = 100us TC ferrous (magnetic) target

Even without running a similar simulation for a PI detector with a mono coil, it should be fairly obvious that the PI will be unable to distinguish between the ferrous and non-ferrous targets, as only changes in amplitude are available. Without a balanced loop system, there is no way to detect the effects of absorption produced by the ferrous target. All you see is an increase is RX signal for all targets.

Anyway, more food for thought...

[dbanner]

Ok, that is fine this is not what you are looking for.
The nail is actually average of 8 * 4 = 32 sampling periods.

I am in the "lets see what info" is in the decay curve and can we extract this info.

I am seriously thinking of a bi-polar tx pulse. This would simplify trying to do EFE.
I also plan to explore TX on sampling and analysis but haven't got there yet.

Thanks for questions and discussion.

I recall schematic of White's AF108 mine detector pulse induction has a bipolar TX. Don't know if it uses bipolar TX pulses though.

[godigit1]

just thought would this be Multiple Period sensing. dang!

[godigit1]

Just a thought about measuing on time. Im may be way off base here and this may not even be possible. But what if instead of trying for a super fast rise time it was controlled. I guess that would be constant current. But that is not what Im thinking Im thinking cliping the pulse to allow for a specific charge cycle or rise time the way a motor signal is clipped.
Say two fast pulses would be measured at on time then a third pulse to be sampled normally. it would have to be balanced coils and the off coils is sampled two amps two integrators then rise and fall can be averaged. So then this would be a bipolar pulse but at two to one??. Or A single pulse sampled in stages??
Am I having a craxy gold miners dream here.

[waltr]

log amplitude-linear time. Log-log is linear for viscous responses.

I'm afraid you don't. There is a bending of a nail response past 120us, yet most probably it is just noise that plays with your expectations. If the same bending persists after averaging 16 or more curves you'll be able to say it is truly there.
In this sample period it is apparent that the slope is not nearly as steep as non-Fe target. Perhaps it is OK to go with a shorter period, and decide the nature of a target according to the slope alone, just as the manufacturer-whose-name-must-not-be-said has.
Perhaps it will be possible to see the bend at much earlier samples in a response of a ferrite ring core, but there is no practical meaning in knowing its response - there are no ring cores laying about in nature.
Also please note how the gold ring and US cent response cross each other at about 55us. So say I want to have a PI with high repetition rate, and Tx pulse duration = total sample period. It only means I can't expect EF to do its role in presence of Fe, and I should go for bipolar pulsing.

Ok, that is fine this is not what you are looking for.
The nail is actually average of 8 * 4 = 32 sampling periods.

I am in the "lets see what info" is in the decay curve and can we extract this info.

I am seriously thinking of a bi-polar tx pulse. This would simplify trying to do EFE.
I also plan to explore TX on sampling and analysis but haven't got there yet.

Thanks for questions and discussion.

[Davor]

...there are no ring cores laying about in nature...

But when I think about it, there are hot rocks that do lay about in nature, and provide similar response...
So there is a practical meaning to it after all.

[Davor]

log amplitude-linear time or log amplitude-log time?

log amplitude-linear time. Log-log is linear for viscous responses.

I think I see this 'knee' on the nail's curve.

I'm afraid you don't. There is a bending of a nail response past 120us, yet most probably it is just noise that plays with your expectations. If the same bending persists after averaging 16 or more curves you'll be able to say it is truly there.
In this sample period it is apparent that the slope is not nearly as steep as non-Fe target. Perhaps it is OK to go with a shorter period, and decide the nature of a target according to the slope alone, just as the manufacturer-whose-name-must-not-be-said has.
Perhaps it will be possible to see the bend at much earlier samples in a response of a ferrite ring core, but there is no practical meaning in knowing its response - there are no ring cores laying about in nature.
Also please note how the gold ring and US cent response cross each other at about 55us. So say I want to have a PI with high repetition rate, and Tx pulse duration = total sample period. It only means I can't expect EF to do its role in presence of Fe, and I should go for bipolar pulsing.

[waltr]

I'd like to see the rightmost part with less noise. But if you summed the samples already, I'm afraid it is nearly the best possible.
I'm seeking the knees in ferrous response.

Interesting but the values at the right end of my plots are about 4 uV at the coil or 4mV at the ADC input. Targets are 3 inch above coil center. I can get stronger signals if targets are closer to coil.
I ran this again. Captured and averaged 4 curves of no target and a nail sitting on the coil. put 1 curve of gold ring and US cent .
I think I see this 'knee' on the nail's curve.
Note: X axis is point number so multiple by 2usec for time.

[green]

Well, having a few different specimens' responses in log amplitude should do. There is a growing number of Fe responses in frequency domain in trade/scientific papers, but not so much in time domain PI.

log amplitude-linear time or log amplitude-log time?

[Davor]

Well, having a few different specimens' responses in log amplitude should do. There is a growing number of Fe responses in frequency domain in trade/scientific papers, but not so much in time domain PI.

[green]

I expect to see some bending of the ferrous responses in log amplitude scale.

Ferrous targets don't have straight line decay, log amplitude-linear time. Thicker non ferrous targets don't have straight line decay, log amplitude-linear time.

[Davor]

I expect to see some bending of the ferrous responses in log amplitude scale.

[green]

I'd like to see the rightmost part with less noise. But if you summed the samples already, I'm afraid it is nearly the best possible.
I'm seeking the knees in ferrous response.

What do you mean by a knee. Some scope pictures of a screw and washer. Rx_(2)200mm round fig8 IB, Tx surrounds Rx. Labeled centered on coil, should be centered over one of the Rx coils.

[Davor]

Which targets would you like to see?

I'd like to see the rightmost part with less noise. But if you summed the samples already, I'm afraid it is nearly the best possible.
I'm seeking the knees in ferrous response.

[waltr]

Thinking US nickel should have a TC close to 10us anywhere in the decay it's tested.

My data/calc shows that the TC drops closer to 10us when a Nickle is closer to the coil - gives a stronger signal and become a larger part of the exp sums.