Ideas to tinker

[Tinkerer]

Zero to 500mA, -500mA to zero. Both halves, increasing slope.

I can agree with that.

[Tinkerer]

Now I look at it from a different angle:

During half of the cycle the current flows into the coil.
The other half of the cycle the current flows out of the coil.

[green]

Attached is the simulation with a few changes to make it fit the Mosfet, without avalanching.

With the desired design parameters, we could design it specifically.

Thanks for the reply. My simulation uses a coil that I have. Using my coil specs, reduced PS to 6V and added a 10nf capacitor across coil similar to what you did and it seems to work.

[Tinkerer]

Thanks for the reply. My simulation uses a coil that I have. Using my coil specs, reduced PS to 6V and added a 10nf capacitor across coil similar to what you did and it seems to work.

The capacitor rating needs to be AC, and considerably higher Voltage than the Flyback. Film works good, but Ceramic also works, but needs a footprint for high Voltage.

Any amount of inductance works. Then adjust the timing and/or the capacitor.

Bundle wound magnet wire works fine.

You can use high repetition rate or low.

[Carl-NC]

While it is AC and arguably bipolar it is not bipolar-symmetric. There is a slow ramp dB/dt that is uniformly positive, and a fast ramp dB/dt that is uniformly negative. You will normally sample either ramp response in an additive manner, meaning that Earth field effect is not naturally canceled. I suppose there is a way to combine a response sample from the fast negative and the slow positive to cancel EFE but that imposes a severe restriction on how you can do target processing.

Normally the answer would be to add a late EFE sample and subtract it. In this case the TX is always active (exponential) and so the target response is also always active. However, if you consider the slow ramp to be perfectly linear then a fast target will probably be substantially decayed near the end of the ramp and EFE subtraction will work OK. But a high conductor response will be substantially canceled.

Disclaimer: Having never built this circuit or even done much investigation on it, I will admit that this is speculation. Feel free to shoot me down.

[Tinkerer]

While it is AC and arguably bipolar it is not bipolar-symmetric. There is a slow ramp dB/dt that is uniformly positive, and a fast ramp dB/dt that is uniformly negative. You will normally sample either ramp response in an additive manner, meaning that Earth field effect is not naturally canceled. I suppose there is a way to combine a response sample from the fast negative and the slow positive to cancel EFE but that imposes a severe restriction on how you can do target processing.

Normally the answer would be to add a late EFE sample and subtract it. In this case the TX is always active (exponential) and so the target response is also always active. However, if you consider the slow ramp to be perfectly linear then a fast target will probably be substantially decayed near the end of the ramp and EFE subtraction will work OK. But a high conductor response will be substantially canceled.

Disclaimer: Having never built this circuit or even done much investigation on it, I will admit that this is speculation. Feel free to shoot me down.

Thank you for your help, Carl.
Your analysis is very helpful and you have pointed out many features. I can agree with most.

This waveform is however very deceptive in it?s simplicity.

I believe the waveform can be made to be much more symmetric and to look more linear, but it will always stay exponential. There are in fact 4 exponentials in this waveform.
The target will add at least one more exponential superposed on each of the 4 exponentials.

As the TX current is constantly changing, it induces eddy currents in the target at all times.

The good thing about that, is that it is possible to sample the waveform at all times, during the positive half cycle and during the negative half cycle.
At different times, different information about the target can be gleaned. The bad thing is that there are so many choices to be made of where to look. One really needs to decide on a specific purpose to know what to look for.

And then there are the dead spots. The Zero crossings. Again, they are not symmetric. More complications.
We will go down this rabbit hole another time.

[eclipse]

.

Trying to layout 10 sample window signal paths in analog finally got so complicated that I searched for a digital solution.
Could not find any help for DSP. DSP needed real programming skill, whereas I just barely managed to setup a sequential timing schedule for an 8bit PIC.

I'm thinking if it's possible to construct a FPGA software that does the signal sampling and complex math, and a Windows based graphical app where you
can add sampling windows at specific times; adjust sampling gain; do calculations like add/substract S1+S2 - S3 etc. (multiple of these) then set weight of each calculation
0 - 100 so if formula overlaps with another (and another) the FPGA can decide which one is highest priority. When done with the formulas just click on generate and load in the FPGA.

At the end output:
- Do GB subtraction
- Do we have a signal, how strong is it
- Is it gold; Is it iron; Is it both

[waltr]

Yes, this could be done with an FPGA. Could even be done with a PIC32MZ EF or even an STM32.

[Tinkerer]

I'm thinking if it's possible to construct a FPGA software that does the signal sampling and complex math, and a Windows based graphical app where you
can add sampling windows at specific times; adjust sampling gain; do calculations like add/substract S1+S2 - S3 etc. (multiple of these) then set weight of each calculation
0 - 100 so if formula overlaps with another (and another) the FPGA can decide which one is highest priority. When done with the formulas just click on generate and load in the FPGA.

At the end output:
- Do GB subtraction
- Do we have a signal, how strong is it
- Is it gold; Is it iron; Is it both

This is a very good idea.
I strongly believe the TEM TX method makes a lot of target information available. However the complexity of the 4 different exponentials demands also a demodulation capable to unravel this complexity.
To do this in analog, was beyond my capabilities.

I did observe good target separation by TC of the targets and good magnetic/non magnetic separation at times. But changing something means changing a lot of other things too. A digital system can do that.

[Tinkerer]

Symmetrical TEM

Here is the current waveform tweaked a little bit to make it more symmetrical.

I separated the cycle with a long "dead period" to make the view easier to understand.

[eclipse]

So the curious (at least to me) questions next are: how do you do the GB sampling?
Do you take GB sample on the top of the half sine pulse (described in US9285496) or you do something else for doing GB?
How do you do the RX to null and sample during TX ON? Do you apply the method you described here(link) or you rely on your method with the P mosfet and the small nulling coil in reverse or no special care to null the coils (essentially usage of any commercial coil would be possible in this case)?

Thanks!

[Tinkerer]

So the curious (at least to me) questions next are: how do you do the GB sampling?
Do you take GB sample on the top of the half sine pulse (described in US9285496) or you do something else for doing GB?
How do you do the RX to null and sample during TX ON? Do you apply the method you described here(link) or you rely on your method with the P mosfet and the small nulling coil in reverse or no special care to null the coils (essentially usage of any commercial coil would be possible in this case)?

Thanks!

I can not remember to ever have used a half sine current pulse.

Looking at the link you gave described here(link), I don't remember what it was about, just vaguely remember having given simulation a first try. I would have to go back and read the whole thread, for which I do not have the time right now.
What I can say, in the 10 years since then I have continued to work on the original ideas and these ideas have evolved a long way. I just kept trying and trying and over the years got to understand better and better of how these things work. Even today, it happens that I make a break through and then realize that I had been there many years ago, but at the time did not understand what happened.

What exactly the status is now, I can not say, because I sold my soul to the "The commercial".

Ah, the "nulling coil" If I remember right, Dave Emery posted an article giving all the details on IB coil construction. Generally we call it a bucking coil. The resistor in parallel with the bucking coil, serves as damping resistor, but also allows to bypass some of the coil current, which gives some degree of compensation for aggressive ground that would saturate the preamp. Ha, see, if prodded, sometimes I manage to remember something.

[eclipse]

I own a concentric commercial coil just wandered what you are using these days in your designs. I meant if you sample GB at the top of the half sine TEM pulse (voltage) but probably you cant talk about that in deeper details. Keep up the good work and keep us updated with some cool ideas we can play with. Cheers

[Tinkerer]

I own a concentric commercial coil just wandered what you are using these days in your designs. I meant if you sample GB at the top of the half sine TEM pulse (voltage) but probably you cant talk about that in deeper details. Keep up the good work and keep us updated with some cool ideas we can play with. Cheers

Concentric coil:
This could be an induction balanced coil with outer TX coil and inner RX coil and a bucking coil near the RX coil or wound on top of the RX coil.
It could also be a (usually outer) TX coil, an RX coil and a passive compensation coil.
Or it could be an outer TX coil and an inner RX coil that is not induction balanced. This gives you the possibility to look at the RX response while the TX is ON.

Once you get involved in ON time and OFF time signal demodulation, You should also look at:
ZVS, zero voltage switching.
And ZCS, zero current switching.

You keep mentioning GB. Don't you like the "Eric Foster Ground Balance Method". Why? Because of the target hole?
Yes, there are other methods to do GB, but like everything, nothing is perfect. It always ends up with a compromise between desired features and undesired features.

[eclipse]

The concentric coil I have I don't how it's made its for the SD/GPX series which I just obtained but intend to put into some good work.
Never liked the GB method (target hole) of the TDI.
I think they tried to advertise it like a "discrimination" feature (*some dealers/or just users which aren't aware of how this turns out to work)
but anyway it's what it is - a drawback. I guess you're right it's hard to beat the game and come winning every-time.
Cheers!