PI SAMPLE DELAY QUESTION

[Carl-NC]

The nature of large copper coins puts their TC in the 100us range so that a TX pulse of about 300us fully charges the mass of the coin's metal to release eddy currents when the TX current is suddenly turned off. The quick discharge of the TX charge current releases the eddy current energy stored during the TX pulse that is detected in the RX window right after the delay time set on the PI delay setting.

As Qiaozhi notes above, this is a misconception. In normal PI the TX pulse doesn't charge the target, and stored target energy isn't released at turn-off. Eddy currents are only generated by a dB/dt, if the TX field is steady-state the target eddies decay to zero.

At the start of the TX pulse the coil current exponentially rises with a tau of L/Rs and 'forward' eddy currents are generated in the target. If you leave the TX current on for more than ~3 tau it will reach a fairly steady-state value and the 'forward' eddy currents in the target cease to flow. Then when the TX current is shut off, the resulting collapse of the magnetic field causes 'reverse' eddy currents to start flowing in the target. Ideally they start from zero, rise to a peak, then exponentially decay according to the tau of the target. The response is a t*e^-t/tau curve. Meanwhile, the coil flyback is damped with a tau of L/Rd, and you pray that you've gotten the coil damping tau faster than the desired target tau.

In most PI detectors the TX current is shut off way before it reaches steady-state, so there are still 'forward' eddy currents flowing in the target. These subtract directly from the desired 'reverse' eddy currents and reduce sensitivity, but not by much since the dB/dt's are so different. Maybe 2-3% typically.

You can, of course, design PI detectors to also use the TX-on response so then you might be tempted to say, "I'm charging the target during the TX pulse." But it's still a t*e^-t/tau response convolved with a 1-e^-t/tau excitation, and the result is still a decaying signal. If you consider the TX pulse to be an ideal current pulse (instant turn-on and instant turn-off) the mental picture is easier to see.

- Carl

[Davor]

Most probably for the viscous grounds the "forward" response is responsible for differences in the power exponent of the observed ground in different equipment. Provided the mechanism is mostly linear.

[green]

[can you explain in more detail about time constants cause im very confused by this.] reply #11

I charted a nickel, dime and quarter decay curve at two coil on times, 25 usec and 145 usec. The coins are plotted flat and on edge to the coil. The nickel plotted the same at the two on times. The dime and quarter plot a straighter curve and a higher time constant with the 145 usec on time. The coins on edge have a shorter time constant. The plot is amplifier out vs time (target signal recording - no target signal recording) The dime on edge shows a decay curve similar to flat at a lower amplitude, why I don't know, maybe not centered?

[daverave]

i have now removed two turns from my coil and the inductance has dropped to 230uh...is this too low ???? the coil capacitance is about 42pf without any screening...i really want to make a fast coil as im very dissapointed with my last coil made with cat6 wire..im still confused with getting the damping resistor correct as there will be ringing in the back emf to get a steep slope..so maybe to get 10us
delay you have to sample in the ringing....am i right or wrong in my thinking on this ????

[Carl-NC]

There is really no inductance "too low" for PI, I was recently working on a design that used 100uH. You can even do a single turn loop if you like.

[daverave]

surely less turns means weaker field with less depth

[Silver Dollar]

No less turns means a stronger TX as it goes by amp turns but the
RX gets weaker as less coil to induce voltage into...

[daverave]

i see your point about RX return being weaker due to less turns !!! i guess its all swings and roundabouts...im going to experiment more by trial and error with the hammerhead pcb and with the litz coil and critical damping and see what happens !!!

[Davor]

The limiting factor at the end is Rx noise performance. Equivalent resistance of 1nV/sqrt(Hz) is ~60ohm, which means that Rx coil resistance below 60ohm will not spoil noise performance of Rx front end with 1nV/sqrt(Hz) preamp. Reversing the things a bit, a more common 3ohm monocoil would not spoil noise performance of a front end with 0.25nV/sqrt(Hz), and the only problem is that we do not have such front ends at hand.
The way around it would be using a separate Rx with coil separated in sections, of which each section would be damped separately. That way we could reach some serious transformation ratio, and a much better noise performance with nowadays low noise components. Extreme voltages could be avoided by induction balancing of such a coil.

[green]

i have now removed two turns from my coil and the inductance has dropped to 230uh...is this too low ???? the coil capacitance is about 42pf without any screening...i really want to make a fast coil as im very dissapointed with my last coil made with cat6 wire..im still confused with getting the damping resistor correct as there will be ringing in the back emf to get a steep slope..so maybe to get 10us
delay you have to sample in the ringing....am i right or wrong in my thinking on this ????

Including a scope picture. Coil volts and amplifier out (2 stage, gain 300) The coil is shielded with aluminum foil. Get some ringing if not shielded. Get some overshoot with a 1k damping resistor (under damped). The circuit should critical damp with a 1k resistor, it does with LT spice. Changing R damping to 500 ohms still looks under damped and takes the same time to go to zero. I've read about adding a target to balance a induction balanced coil. Can add a target to reduce the overshoot. Don't know if it matters. Maybe not the same problem as daverave but still a problem. I would be interested in seeing some scope pictures of what the no target decay should look like and how important it is to be zero volts. I've posted the overshoot problem before, sometimes thinking I've solved it. I haven't, any suggestions appreciated. The scope probe across the coil lowered the resonance a little but didn't seem to effect the problem.

[Qiaozhi]

Get some overshoot with a 1k damping resistor (under damped). The circuit should critical damp with a 1k resistor, it does with LT spice. Changing R damping to 500 ohms still looks under damped and takes the same time to go to zero.

Try the trick of connecting a 10k pot in series with a 220R resistor, and put both in parallel with a 1k2 resistor. Fit this network in place of the damping resistor, and adjust for critical damping. Then disconnect and measure the total resistance, and fit a damping resistor of the nearest preferred value.

[daverave]

i remember a electronics engineer who makes detectors many years ago saying that you can sample in the ringing to get more sensitivity.....im not sure if this is correct cause no one so far has made any comment on this....my detectors use a 390 ohm resistor to get a clean back emf curve with no ringing or over shoot...maybe my resistor value is too low !!!! maybe some ringing and over shoot is good to sample early ???? but im out of my depth here and im not really sure !!!

[Qiaozhi]

i remember a electronics engineer who makes detectors many years ago saying that you can sample in the ringing to get more sensitivity.....im not sure if this is correct cause no one so far has made any comment on this....my detectors use a 390 ohm resistor to get a clean back emf curve with no ringing or over shoot...maybe my resistor value is too low !!!! maybe some ringing and over shoot is good to sample early ???? but im out of my depth here and im not really sure !!!

With the resistor network trick it's easy to test this theory and determine how much ringing is ok.
In the past I've found that a small amount of ringing does increase sensitivity, but too much has the reverse effect. Setting up a PI correctly is all about compromise.

[deemon]

As Qiaozhi notes above, this is a misconception. In normal PI the TX pulse doesn't charge the target, and stored target energy isn't released at turn-off. Eddy currents are only generated by a dB/dt, if the TX field is steady-state the target eddies decay to zero.

At the start of the TX pulse the coil current exponentially rises with a tau of L/Rs and 'forward' eddy currents are generated in the target. If you leave the TX current on for more than ~3 tau it will reach a fairly steady-state value and the 'forward' eddy currents in the target cease to flow. Then when the TX current is shut off, the resulting collapse of the magnetic field causes 'reverse' eddy currents to start flowing in the target. Ideally they start from zero, rise to a peak, then exponentially decay according to the tau of the target. The response is a t*e^-t/tau curve. Meanwhile, the coil flyback is damped with a tau of L/Rd, and you pray that you've gotten the coil damping tau faster than the desired target tau.

In most PI detectors the TX current is shut off way before it reaches steady-state, so there are still 'forward' eddy currents flowing in the target. These subtract directly from the desired 'reverse' eddy currents and reduce sensitivity, but not by much since the dB/dt's are so different. Maybe 2-3% typically.

You can, of course, design PI detectors to also use the TX-on response so then you might be tempted to say, "I'm charging the target during the TX pulse." But it's still a t*e^-t/tau response convolved with a 1-e^-t/tau excitation, and the result is still a decaying signal. If you consider the TX pulse to be an ideal current pulse (instant turn-on and instant turn-off) the mental picture is easier to see.

- Carl

As for me , I cannot completely agree with you , Carl . The situation looks like some "charging" or "energy pumping" of the target really takes place ... You see , when we turn on the coil ( connecting it to the power supply ) - the current starts to increase in the coil , and then reaches some steady state , as you told before . But if the target does have a large time constant ( TC ) , the eddy current in this target doesn't finish its decay at this moment , and don't forget that this eddy current produces its own field that works in opposite polarity to the transmitter field ( being subtracted from it , according to Lenz's law ) , so it looks like the transmitter coil field cannot completely "penetrate" into the target . And when we stop the transmitter "on" pulse at this moment - we'll have a "reduced" reply pulse amplitude .... for example , if we have a half of the maximum field in the target before current stopping - we'll loose a half of response amplitude on the receiver preamp , not 2-3% of course . But if we hold the steady current in the coil , waiting till the target eddy currents disappeared completely ( or almost completely ) - it would mean that the field reaches its maximum in the target body at this moment , and we'll have a full 100% response from this target , so now our device will work with maximum efficiency . Response magnitude is proportional to the target field magnitude just before the current stopping moment , because it starts to decay from this point to zero , and we cannot receive more . Of course , this must be true if we shut down the coil current fast enough , we must not forget about this ...

In another words , it looks like that some portion of energy is really "stored" in the target , being released and received just after the transmitter current stopping ( and transmitter field collapsing ) . And it really looks like we need a TIME to pump this energy into the target during the ON state of the transmitter circuit .... so the whole case is very similar to a well-familiar process of the capacitor recharging - it needs time to charge and needs time to discharge , and so on . And this is why the idea of "target charging" isn't a nonsense anyhow , as I think .

[daverave]

ive just measured my cable and its 25pf and my coil is the same at 25 pf without shielding....my coil housing is only about 5mm thick so i have to lay the turns of wire side by side and use graphite shielding but not sure if epoxy resin will damage the graphite coating....i will experiment with the resistor network and see if i can improve and get this new coil to go faster !!!