THE ELUSVE 1us TC TARGET

[Detectorist#1]

Hi Tinkerer,

I think, for input clamping diode in PI MD is not sense to add term "reverse recovery". This term have to be applied only when at some moment reverse voltage is applied at input clamping diode.
So, the parameter "reverse recovery time" in datasheets is not applicable in our case.
In real situation, the current by clamping resistor discharges diffusion capacitance of the diode via the coil and the parallel resistor of the coil.
The voltage on diode decays and in one moment this voltage began to be lower than the necessary one for conducting via p-n junction, but diffusion capacitance connected in parallel in p-n junction continues
to be discharged via coil and the parallel resistor. At this moment summary damping resistor of the coil is changed of course.

[Detectorist#1]

The reason for changing of effective summary damping resistor is the fact that in the end of Tx pulse they are two currents via clamping diode- current via p-n junction and discharge current of diffusion capacitance.
When p-n junction stops to carry current, only capacitive discharge current remains and this changes summary value of the damping resistor.

[Detectorist#1]

To be more exact, the value of capacitance discharging current dеpends of the voltage on p-n junction. At the beginning of FET drain's voltage decay, the voltage
on p-n junction decay slowly and discharge current is relatively small. When p-n junction stops to carry current - the discharge current increases.

[Tec]

Attached is the response of a 25mm x 25mm square of light kitchen foil. compared to a aud 5c piece.

Note the peak of the 5c piece can be seen, and the kitchen foil peak looks like it happens within the first 4uSec when the coil is still discharging.

Coil: 10cm , 30 winds, (wire pvc insulated jaycar 3014), damping resistor not optimised (560 ohm)

Circuit Wombat PI

[Mr.Jaick]

I would much appreciate your input for trying to improve the setup. The idea is to start a discussion and to try all good ideas to improve the results.

This is the preamp used for the scope shot above.

Hello Tinkerer, my recommendation is to use Active Damping on the RX coil, I have done it and the results are very good, you see that large offset from 3 to 8us, we can get rid of that(and More), however I used it on a single ended pre-amp, you're using a floating one so maybe two damping devices needed, by the way what happened to the center taped coil idea for differential implementation? you don't use it in the AMX RX either
I couldn't follow the differential discussions so got me curious what made you ditch the center-taped coil.
now currently I'm using the RX in voltage mode + Rd, because I ran out of Timers in the DSP for the error control loop, currently migrating to a much more capable DSP, after that I will properly investigate the damping and what can be done with it.

[Tinkerer]

Mr.Jaick, thank you for your feedback.

The TX for this RX is bipolar. A bipolar square wave current TX. You can see the simulation here: https://www.geotech1.com/forums/foru...162#post419162

Since it is bipolar, that is, the current goes once in the clockwise direction around the coil, and once in the anti-clockwise direction.
To keep the input balanced, and ground referenced, I use R29 and R30. I don't know yet what the best resistance is for that. I tried 10k, 100k and others.
This is an induction balanced DD coil assembly with little current in the RX coil. However, the amount of current is proportional to the amount of resistance of the damping, ground reference and input resistors. low resistance draws more current into the RX coil. Higher (total) resistance draws less current that needs to be damped.
How much current will give the highest amplitude response? I set the lower limit of target at 1us. There are several reasons for that which I will explain in detail another time. It involves the smallest gold nugget of interest, but also involves a possibility of iron differentiation.
I am still trying to find the best compromise.
Any suggestions are welcome.

[Tinkerer]

Attached is the response of a 25mm x 25mm square of light kitchen foil. compared to a aud 5c piece.

Note the peak of the 5c piece can be seen, and the kitchen foil peak looks like it happens within the first 4uSec when the coil is still discharging.

Coil: 10cm , 30 winds, (wire pvc insulated jaycar 3014), damping resistor not optimised (560 ohm)

Circuit Wombat PI

Innovative and interesting.

[Mr.Jaick]

Mr.Jaick, thank you for your feedback.

The TX for this RX is bipolar. A bipolar square wave current TX. You can see the simulation here: https://www.geotech1.com/forums/foru...162#post419162

Correct, I thought you would be using the bipolar CCPI, but is it the reason you're not using center taped? I don't think it would depend on the way the TX is set up though.
with the differential system we want to eliminate any common mode signal / far field stuff coming our way, does this "floating" coil do the same? I can kind of see how this could work but not sure about the theory behind it without the center tap.

How much current will give the highest amplitude response? I set the lower limit of target at 1us.

I think the optimal RX load impedance is dependent on the way we measure the signal, are we measuring current or voltage defines that.

another idea; if you would use this preamp in non-inverting manner you would get much less noise, in that case no input resistors for thermal dance-around only the damping R parallel to the coil, and the fact that the signal gain will equal the noise-gain of the amp not anything less, especially important for such low gain settings.

There are several reasons for that which I will explain in detail another time. It involves the smallest gold nugget of interest, but also involves a possibility of iron differentiation.
I am still trying to find the best compromise.
Any suggestions are welcome.

I would love to learn more, So anytime you feel like it; will be greatly appreciated.

again from my experience, iron disc can be very unstable / ground dependent when sampling the fly-back X signal, also it gets very touchy on mechanical vibration even considering a "built like a Tank" DD coil, I rather sample the TXon-X signal, it's not super robust either but better than looking under the fly-back hood.

[Carl-NC]

A center-tapped RX coil does not eliminate far-field signals, it has exactly the same response as the floating coil. To cancel far-field signals you have to reverse one of the RX coils, and then it cannot be concentric with the other. Figure-8, butterfly, and DOD coils are examples.

On noise, there is almost no difference between non-inverting and inverting types if the resistor values are the same, like this:


The signal induced in the RX coil is always a voltage (ε = -N*A*dB/dt) and can tolerate a fairly low resistive load, depending on the resistance of the coil itself. You could probably go as low as 100Ω on most coils with little degradation.

[Mr.Jaick]

A center-tapped RX coil does not eliminate far-field signals, it has exactly the same response as the floating coil. To cancel far-field signals you have to reverse one of the RX coils, and then it cannot be concentric with the other. Figure-8, butterfly, and DOD coils are examples.

Thank you for the clarification.

that's interesting, So that assumes far-field noise does not produce a common-mode signal on a center-tap or floating coil but a differential signal, which now that I think of it makes sense because for example if we have two identical coils separate from each other but in the same physical space and we measure their signal/noise we will get totally the same signal in both channels, now if we put them in series we just add those signals at coil level, but since we are measuring the other ends of each coil we get differential signals, nothing canceled.


it's also interesting that Paul Moody claimed this topology cancels environmental EMI, not sure what he meant by that...
so what are the benefits then?

On noise, there is almost no difference between non-inverting and inverting types if the resistor values are the same, like this:

Take a look at this app note:
https://www.analog.com/en/analog-dia...oise-gain.html
But I get it it's not a huge deal considering the EMI that comes through.

The signal induced in the RX coil is always a voltage (ε = -N*A*dB/dt) and can tolerate a fairly low resistive load, depending on the resistance of the coil itself. You could probably go as low as 100Ω on most coils with little degradation.

True but in my explanation above I assumed the coil was a lumped element, so like a black box that gives you some power( V * I ), we can't have just one of these two parameters, so seems like the saying of voltage or current alone is not mature enough, we should say "we measure power".

[Carl-NC]

it's also interesting that Paul Moody claimed this topology cancels environmental EMI, not sure what he meant by that...
so what are the benefits then?

Honestly, I don't know. I've never given much thought to a center-tapped RX coil.

Take a look at this app note:
https://www.analog.com/en/analog-dia...oise-gain.html
But I get it it's not a huge deal considering the EMI that comes through.

You can see that the noise gains for the two are identical, but the signal gain for the non-inverting is very slightly higher. So non-inverting can offer very slightly better SNR. However, when you include the clamping resistor like this:


then the inverting opamp has lower noise because the feedforward resistor doubles as the clamping resistor.

True but in my explanation above I assumed the coil was a lumped element, so like a black box that gives you some power( V * I ), we can't have just one of these two parameters, so seems like the saying of voltage or current alone is not mature enough, we should say "we measure power".

Think of the RX coil as a voltage source with an output resistance equal to the R of the coil winding. Power delivered depends on the load resistance. We usually don't care about getting power from the RX coil or matching its impedance to anything, so more often than not it's loaded lightly with a large resistance, maybe 10k, or in a transient system whatever is needed to properly dampen it.

[Mr.Jaick]

Honestly, I don't know. I've never given much thought to a center-tapped RX coil.

fair enough.
now Thinkere might get upset with me if I ask any more questions about another topic in his 1us challenge so I'll talk about possible solutions for that next.

You can see that the noise gains for the two are identical, but the signal gain for the non-inverting is very slightly higher. So non-inverting can offer very slightly better SNR. However, when you include the clamping resistor like this:


then the inverting opamp has lower noise because the feedforward resistor doubles as the clamping resistor.

I never use that current limiting resistor for the RX coil, only the diodes, and never had any issues.

Think of the RX coil as a voltage source with an output resistance equal to the R of the coil winding. Power delivered depends on the load resistance. We usually don't care about getting power from the RX coil or matching its impedance to anything, so more often than not it's loaded lightly with a large resistance, maybe 10k, or in a transient system whatever is needed to properly dampen it.

that's a very good way of looking at it, A voltage source namely L with a source resistance of R, one must load it.

[Mr.Jaick]

one way to get into the fly-back regime sampling is to electronically null our RX coil, and I think we don't strictly need 24-bit codec chips to get this done, however, it would be nice.
if we are direct sampling then we can use a 12bit DAC to get into the ballpark so we can see the signal after amplification, a 12bit DAC with a 1V reference(or external attenuation) gives us 244uV of resolution with a sampling rate of let's say 5 to 10Msps(STM F4 series) we can get the job done pretty good.

[Tinkerer]

I have a reasonable sensitivity now, about 2V for the 25x25mm alu foil. About 1us TC.
The delay allows sampling now at 3.2us.
There are still noise problems to solve.
It may be a good time to look at the induction balance of the coil now. With the damping and gain getting closer to a workable situation, it is probable that we can improve on the null.
How close to Null should we aim for? With the Fyback at about 1000V we have an input of about 60V on the RX now.
The coil assembly is still open so I can try to improve on the physical Null. Then we can look at the possibilities of an electronic Null.
The TX coil has an inductance of 2.04mH, 2.8R and a SFR of about 180kHz, in circuit, with cable. 380pF distributed capacitance for coil, shielding, cable, Mosfets etc. seems acceptable.
The RX coil 300uH I still need to measure the SFR .

[Tinkerer]

[ATTACH]n419232[/ATTACH]