PI Monocoil - sampling under 4us almost impossible?

[Tinkerer]

Hi Simon ... instead of Bifilar coil I used 8 turns single layer of RG58C/U but connected like Bifilar on 20 cm form. Coil differential voltage was 1000 volts across damping resistor ( 1K ) with 12 volt supply. MOSFET sees half ... these ones are IRFP460 and avalanche at 500 volts so the coil could probably go higher than 1000 volts with higher volt mosfet. This coil achieved just above 5 uS. There must have been a fair bit of peak current because my steel target started 'singing' ... didn't measure it though.

Regards

moodz.

Hi moodz,

to drive the Flyback to a thousand and more Volt, certainly would shorten the decay time. I also like the differential preamp idea. I am considering a differential input as option for the TINKERERS IB-PI at: http://www.geotech1.com/forums/showt...ewpost&t=15470

Please take a look at it, I would appreciate your advise.

Tinkerer

[moodz]

I had a conjecture that a single-layer wound coil (or flat spiral wound) would have less capacitance than jumble wound because less voltage difference between adjacent wires.

Has anyone tested that or know about it?

-SB

Hi Simon ... instead of Bifilar coil I used 8 turns single layer of RG58C/U but connected like Bifilar on 20 cm form. Coil differential voltage was 1000 volts across damping resistor ( 1K ) with 12 volt supply. MOSFET sees half ... these ones are IRFP460 and avalanche at 500 volts so the coil could probably go higher than 1000 volts with higher volt mosfet. This coil achieved just above 5 uS. There must have been a fair bit of peak current because my steel target started 'singing' ... didn't measure it though.

Regards

moodz.

[Carl-NC]

Gary (Chemelec) has also done a lot with spiral and conical coils.

[Tinkerer]

I had a conjecture that a single-layer wound coil (or flat spiral wound) would have less capacitance than jumble wound because less voltage difference between adjacent wires.

Has anyone tested that or know about it?

-SB

I have built a few spiral wound coils. They ave very good depth and sensitivity, but a very narrow footprint.

Tinkerer

[simonbaker]

Hi Aziz,

I've seen Paul's circuit at http://www.geotech1.com/forums/showt...t=15371&page=4 and I think it is a great idea, it's worth pursuing.
It has the advantage of obtaining a double voltage at the end of the coils, while having just half the voltage on the transistor.

About your advice, of increasing the current through the pulse width, I am not sure that works (I run a simulation and the dead-time is at least as large). You are right about lower capacitance of the coil, that helps.

Regards,
Nicolae

I had a conjecture that a single-layer wound coil (or flat spiral wound) would have less capacitance than jumble wound because less voltage difference between adjacent wires.

Has anyone tested that or know about it?

-SB

[nick_f]

Hi Nicolae,

there is a way of making the dead-time slightly lower with the anti-parallel diodes: You have to burn the energy quickly into heat in the damping resistors Rd.

This can be achieved by increasing the flyback voltage. Take a high voltage rated mosfet (higher avalanche breakdown voltage). But then, you have to sample earlier not to loose signal.

To increase the flyback voltage, increase either the transmit pulse width and/or decrease the coil capacitance (+parasitics). The latter one would lead to increase of dampening resistor Rd.

Rd = sqrt(L/(4*C)), L=coil inductance, C=coil capacitance (+parasitics)

You also could take a bifilar wound balanced coil (like center-tapped coil) to decouple the other half of the high flyback voltage to the mosfet. You might double the flyback voltage with this method. Then a differential signal processing would be available too. Paul (moodz) has already showed some circuits here.

Aziz

Hi Aziz,

I've seen Paul's circuit at http://www.geotech1.com/forums/showt...t=15371&page=4 and I think it is a great idea, it's worth pursuing.
It has the advantage of obtaining a double voltage at the end of the coils, while having just half the voltage on the transistor.

About your advice, of increasing the current through the pulse width, I am not sure that works (I run a simulation and the dead-time is at least as large). You are right about lower capacitance of the coil, that helps.

Regards,
Nicolae

[Aziz]

Hi Nicolae,

there is a way of making the dead-time slightly lower with the anti-parallel diodes: You have to burn the energy quickly into heat in the damping resistors Rd.

This can be achieved by increasing the flyback voltage. Take a high voltage rated mosfet (higher avalanche breakdown voltage). But then, you have to sample earlier not to loose signal.

To increase the flyback voltage, increase either the transmit pulse width and/or decrease the coil capacitance (+parasitics). The latter one would lead to increase of dampening resistor Rd.

Rd = sqrt(L/(4*C)), L=coil inductance, C=coil capacitance (+parasitics)

You also could take a bifilar wound balanced coil (like center-tapped coil) to decouple the other half of the high flyback voltage to the mosfet. You might double the flyback voltage with this method. Then a differential signal processing would be available too. Paul (moodz) has already showed some circuits here.

Aziz

[nick_f]

Green is trigger pulse at turn off, Blue is Inductor voltage, Red is Inductor Amps.
Is this what your simulating?

Hi B^C,
I was simulating the waveform (voltage) on the antiparallel diodes where the preamplifier is connected. See the attached image. The delay you can get there is the best delay we can obtain, considering that all the other stages following this point will only add additional delays.
I run simulations with various MOSFET transistors, various values for C1, R1, R2, D1, D2+D3, etc and I can't go much lower under 4us. I wanted to see if other people can get a simulation with better deadtime than 4us.
For example, in the waveform I just posted, you can see the deadtime is actually 5.5us (the damping resistor is not optimised, should be a bit lower). If we amplify that signal 1000x, you'll see it doesn't look very nice.
We can improve this circuit with lower capacity transistor, lower capacity coil, lower capacity D1, etc.

Regards,
Nicolae

[B^C]

Not sure if this is what you have been simulating with LTSpice as i haven't had the time to load your models to run them.

I had a bit of a play around this morning with what i think your testing & have attached a screen shot of the result.

Green is trigger pulse at turn off, Blue is Inductor voltage, Red is Inductor Amps.

Is this what your simulating?

[B^C]

Nicolae,

Yep i'm an Aussie, sure we can arrange a prospecting trip if you don't mind walking a lot of Km's & digging a lot of holes. It is an adventure as you put it but it is hard hot work so there will be lots of blisters on the feet & hands.

I usually go away for long periods at a time, last time was nearly a year but we can organise something, i am home for a while at the moment & will be until i get this modification sorted & understood.

I'll let you know when i am about to make a move & we can sort something out, you will get hooked on it so be prepared, it will be a while until i make another trip but not to long hopefully.

I'll look forward to the trip.

PS,

I hope you have a sense of humour because mine sucks hahaha!!

[B^C]

Another graph.

[B^C]

Hi B^C,

That's one of the best insights I read about metal detectors and Minelab characteristics lately. Thank you very much for sharing your knowledge.
I have some comments about the air tests. I consider it is very true that they don't accurately reflect the capacity of the detector of working in bad ground conditions. However, I think they are still very useful up to a stage. For example, without air tests, how would I know if I improved the sensitivity of my circuit? And if a detector can't detect a coin at 5cm in the air, don't even bother going outside with it.
Also, with the discrimination, if a detector can't make the difference between iron and copper in the air, I doubt there is any chance to do that in the ground.
Therefore I see the air tests as the first stage of testing a detector. The second stage would be testing targets in the ground (or bad ground if available ) .
You're in Australia, aren't you? Maybe one day you'll take me with you for some prospecting. Never been searching for gold before and I am looking forward to the first adventure


Regards,
Nicolae

Gday Nicolae,


Firstly we can test at home, more on that a bit later.

I absolutely agree that we can air test so please don't get me wrong, the thing is we have to know what it is we should be testing. Once we know this then we can air test, but like everything there's a secret that at the moment only ML know what it is, this is why like it or not Mr Candy is a smart sucker.

I think everybody will see from the attached pictures that there is more to this than meets the eye.

This is an International standard test with the results on various top of the line detectors--the big boys & there toys.

Have a look at the comparison from air testing to ground testing, only ML have it right, this is what i have been working on for some time.

You will also have to remember that these tests are done with the soil that was available at the location of testing--foreign soil to Australia.
Now bring there detectors here & ML win again by a broader margin.

[simonbaker]

Hi Simon,
It's very good you were able to run the simulation.
I also had to change the prefix to X. I actually mentioned in the crash course we have to change it to X (I was in a hurry and I didn't specify what was the original prefix MN).

("Next, you check the prefix. We use X when we want to use our imported (or created) circuits, not part of the LTSpice libraries. Also, the value field has to contain the part we want, in this case: XIRF740")

Regards,
Nicolae

Here is the updated file with the X prefix as convenience.

Regards,

-SB

[nick_f]

Hi Nicolae:

I was able to run your circuit after doing this:

I right-click on the U1 XIRF720 part, and change the Prefix from "MN" to "X".

Does yours work without that?
-SB

Hi Simon,
It's very good you were able to run the simulation.
I also had to change the prefix to X. I actually mentioned in the crash course we have to change it to X (I was in a hurry and I didn't specify what was the original prefix MN).

("Next, you check the prefix. We use X when we want to use our imported (or created) circuits, not part of the LTSpice libraries. Also, the value field has to contain the part we want, in this case: XIRF740")

Regards,
Nicolae

[simonbaker]

Thanks porkluvr,

I also found some new stuff. Thanks for posting. I will also include the model in the symbol, I think it's more convenient this way.

Regards,
Nicolae

Hi Nicolae:

I was able to run your circuit after doing this:

I right-click on the U1 XIRF720 part, and change the Prefix from "MN" to "X".

Does yours work without that?

Thanks everyone for all LTSpice advice, I always have problems with configuring it, so always welcome tips.

Regards,

-SB