MAGPI V4 -- CCPI MONOCOIL

[moodz]

I posted a bipolar current pulsing scheme with tilt correction for a monocoil TX a while back in the bipolar alternative H bridge thread

https://www.geotech1.com/forums/foru...628#post410628

So basically that circuit CCPI used current sensing in the TX coil to drive a feedback loop with considerable gain to correct "TILT" in the TX current pulses.
It would seem that this tilt is caused by losses from the TX coil energy and mostly due to energy lost during the flyback transition time .... which kinda makes sense if you think about it.

However it then follows that not only must tilt be corrected but also the amplitude ( which is also effected by energy loss ) of the current pulses in the TX coil.
The assumption being that either a change in Tilt and/or Amplitude will cause a change at a recieve coil.

Its easy to show ( using the CCPI ) that a target / ground losses in proximity of the TX coil will cause tilt and / or amplitude variation of the current pulses in the TX coil.

Observing the tilt correction feedback voltage it can be seen that the feedback voltage will vary with ( tilt / amplitude ) so what this means is that a target may be detected by monitoring the change in value of the feedback loop control voltage as the coil is moved near a target.

Thus this is a means of detecting targets and ground condition using only a monocoil with continuous bipolar constant current pulses. ( aka CCPI )

Likewise by varying the frequency and/or duty cycle of the bipolar pulses other information may be determined from the tilt loss ratios at different pulse / duration rates.

Dont worry I will also include a RX coil so in case the mono coil sensing does not work out.

moodz

[Olly]

I posted a bipolar current pulsing scheme with tilt correction for a monocoil TX a while back in the bipolar alternative H bridge thread

https://www.geotech1.com/forums/foru...628#post410628

So basically that circuit CCPI used current sensing in the TX coil to drive a feedback loop with considerable gain to correct "TILT" in the TX current pulses.
It would seem that this tilt is caused by losses from the TX coil energy and mostly due to energy lost during the flyback transition time .... which kinda makes sense if you think about it.

However it then follows that not only must tilt be corrected but also the amplitude ( which is also effected by energy loss ) of the current pulses in the TX coil.
The assumption being that either a change in Tilt and/or Amplitude will cause a change at a recieve coil.

Its easy to show ( using the CCPI ) that a target / ground losses in proximity of the TX coil will cause tilt and / or amplitude variation of the current pulses in the TX coil.

Observing the tilt correction feedback voltage it can be seen that the feedback voltage will vary with ( tilt / amplitude ) so what this means is that a target may be detected by monitoring the change in value of the feedback loop control voltage as the coil is moved near a target.

Thus this is a means of detecting targets and ground condition using only a monocoil with continuous bipolar constant current pulses. ( aka CCPI )

Likewise by varying the frequency and/or duty cycle of the bipolar pulses other information may be determined from the tilt loss ratios at different pulse / duration rates.

Dont worry I will also include a RX coil so in case the mono coil sensing does not work out.

moodz

I kind of like the thinking behind this approach, but how would we tell the difference between a genuine target and changing ground conditions?

[green]

I kind of like the thinking behind this approach, but how would we tell the difference between a genuine target and changing ground conditions?

Some thoughts, maybe correct maybe not.
Tilt problem not a problem with target detection, just target rejection. Tilt changes ground decay so two point subtraction GEB doesn't cancel ground signal. Don't know how high tilt needs to be to effect GEB. moodz other reply, 300uA tilt needs a time, maybe 3A/sec. Don't know if 3A/sec is close to being a problem or not. Amplitude doesn't change ground decay rate, thinking Carls or moodz method should both work. Fastest to correct maybe best?

[moodz]

I kind of like the thinking behind this approach, but how would we tell the difference between a genuine target and changing ground conditions?

Yes well that will be a question amongst many more ... as I have not explored this idea extensively. Maybe varying the pulse rate will give different tilts ( based on viscous ground effects and target decays ) .... dont know yet.
What was surprising ( to me anyway ) was that my loop gain for the tilt correction is about 100,000 and yet relatively small targets cause very strong changes in the feedback signal ... so I am thinking that this could be used for some form of target info possibly in addition to a recieve coil also.
Anyway I am making up a little single sided + ground CCPI widget board to experiment with ( not as complex as the AMX ) as I am sure I will let the smoke out more than once.

[moodz]

Some thoughts, maybe correct maybe not.
Tilt problem not a problem with target detection, just target rejection. Tilt changes ground decay so two point subtraction GEB doesn't cancel ground signal. Don't know how high tilt needs to be to effect GEB. moodz other reply, 300uA tilt needs a time, maybe 3A/sec. Don't know if 3A/sec is close to being a problem or not. Amplitude doesn't change ground decay rate, thinking Carls or moodz method should both work. Fastest to correct maybe best?

I tried Carls method and must be missing something as I had Amplitude modulation of the corrected pulses ( which modulates target responses ) and the "correction" caused distortion ( rolloff ) of the current switching edges ).

As to what works ... I guess we wont know till we know ( a known unknown ).

[moodz]

...TX test board scheme. ( with tilt control mosfet )
[ATTACH]n412635[/ATTACH]

.. PCB layout
[ATTACH]n412636[/ATTACH]

.. 5 minutes later from the resin printer

moodz

[moodz]

.. and the built up board. Its all single sided with a ground plane for simple fab.
The tilt control FET is not fitted ( thats on the other board )

I will smoke it tomorrow ...

[moodz]

... managed not to let the smoke out.

1 amp peak to peak and 650 volt flybacks with about 700 nanosecond transition.

The coil is a fast basketweave litz wire 0.5 ohm mono.

Slight Tilt but nothing to write home about .... I am starting to think DSP will take care of it

[moodz]

Good news ...

the CCPI circuit as it stands has a balanced coil and you cant use a coax fed DD coil or similiar or you would have one side grounded or flybacks on the shield outer of the Coax.

Re-adapted the frontend to to drive a typical 0.5 ohm 300 uH DD coil and it seems we can achieve 2.4 amps peak to peak and "proper" bipolar operation ( before the SIC mosfets cant handle it ).

A switch and a cap converts the TX from PI mode to a pure sine wave mode ( or combination wave ) ... So here is a TX circuit that can convert from a PI to a VLF TX in on the fly.

So you can search in PI mode then switch to Discriminate mode to decide if you want to dig or not.

The RX will of course handle PI and Frequency Domain signals. ( DSP etc )

Below is the PI waveform on the DD coil TX.

now I have to go and make some circuits ....

moodz

[moodz]

.. and here is the same circuit TX waveforms in discrim mode ... no need to change coils !!


and the spectrum of same ... the CCPI produces a very clean signal

[moodz]

...slaving over a hot soldering iron all day has resulted in the intial results for the transmit circuit and so far it is matching or exceeding the simulation results.

with moderate settings ( to avoid smoke )

In PI mode ...

0.5 ohm TX coil / 300 uH ( your fairly standard PI TX coil )

1 amp peak to peak with 3 microsecond transitions and power supply = 2.5 volts @ 500 ma = 1.25 watts consumption.

Yellow trace = gate drive
Pink trace is flyback at TX coil
Blue trace is 0.1 ohm current sense resistor.

Notice that tilt correction is applied ( very simple solution ).

next I will be testing the VLF mode



...

[moodz]

... and VLF mode TX ...

Same coil but this time crank up the voltage to 5 volts to see what happens !

We end up with a sinus current waveform in the TX coil of 8.62 Khz ( other frequencies can be set .. )

The power supply is 5 volts @ 500 ma = 2.5 watts.

The TX coil current is 4 amps peak to peak @ 60 volts peak to peak ... very nice.

Yelllow trace is the gate drive.
Pink trace is the coil voltage.
Blue trace is the coil current sensor ( 0.1 ohm resistor ) 100mv / amp.

[eclipse]

Is it possible to combine VLF+PI mode in alternating cycles. Because repetition rate is so high you could have 50% VLF 50% PI of the total time but is it possible to switch in and out the capacitor to achieve this?

[moodz]

Is it possible to combine VLF+PI mode in alternating cycles. Because repetition rate is so high you could have 50% VLF 50% PI of the total time but is it possible to switch in and out the capacitor to achieve this?

It certainly is possible ... I was exploring it ... not sure if a mechanical relay is suitable as it would be buzzing and short lived ... an electronic switch is required should be relatively simple.

I have a sim that switches does 5 milliseconds of VFL then 5 milliseconds of PI ....
Its all synchronous so the RX should be able to track OK its just the first couple of milliseconds where the waveform is "winding up" in each burst ...
but I figure even this will be useful information.

The effective "sampling" rate is 100 Hertz ...( 50 bursts of PI and VLF per second ) should be fast enough for reasonable swing speeds ?

I worked on pi recievers for years and now I realise I should have been working on the TX circuit ...

Here is the PI VLF ELSTER-SPIEGEL mode !!!

TX coil voltage and current ... 0.5 ohm / 300 uH TX coil in DD setup.

[moodz]

I have designed a solid state switch to alternate between VLF and PI TX modes and it worked first time ... better than I thought it would.
Tweaked the component values so we are at about the peak voltage limit of the SIC mosfets internally at 1.1 KV rated at 1.2 KV ( for a +500 to -500 swing at the TX coil in Pi mode ).

We get 2.77 amps peak to peak in PI mode and more than 4 amps peak to peak in VLF mode. The power supply to achieve this is set to 3.3 volts @ 1.5 amps = 4.95 watts ... I can wear 5 watts for this amount of TX goodness.

waveforms at the TX coil below