NEW !!! Full differential PI front end.

[moodz]

....well I found another spare 20 minutes and it seems there has been bit of discussion going ... all good stuff.

Anyway to answer quickly the questions about the 4066 driver cct ...
It does actually work

below is the output of channel 1 with about a 40 uS delay after first turn off the green trace is the input to the 4066 ... white trace is the sampled output ... the timing is set to intercept the decay slope at about 75 %.



and here is the breadboard with 60 turn 'boss' differential coil. Note he circuit here is built with a balanced FET driver. Because of the number of turns in the coil it is very sensitive to Earth's magnetic field. However it is also very sensitive to targets even very small ones though the sample is out at 40uS.
... Hey Aziz when I get more time we proceed OK ...

Cheers from moodz

[Aziz]

Hi moodz,

I am just waiting to finish the list of needed expensive parts. But I have two samples of the following analog switches/muxes till now.
MAX313
MAX392
MAX333A
I will test the MAX313 and MAX392 (pin compatible). It makes the driving really simple and reduces circuit complexity.

I probably need several month to afford the rest of the parts list. So it's going really slowly.

Just keep on the good work.
Aziz

[gadgetmaker]

Parts Help for Aziz

Aziz,

Please list part numbers and quantities you need and I will get them for you.

Karl

[Carl-NC]

Anyway to answer quickly the questions about the 4066 driver cct ...
It does actually work

I'm surprised! You may want to check the margins, it may be just barely working.

[bklein]

Can someone please expand upon what Msysta has posted? I am most interested in the raw basics - the coil is energized and then the energy is cancelled and he says the target eddy current interracts with the flyback decay.
But what if there were a separate receive coil (perhaps at some distance away) that was shunted during the transmit pulse of the traditional PI circuit - would it then pick up the eddy current in a more "pristine" and manageable fashion? Could we do more with this signal as it would be better normalized for receive circuitry?
Kind of seems we are making little EMP devices and receptors...

I am about to find out more on this with a Surfmaster PI I have discussed in that forum section. The drive transistor was a TIP32B. That has a lower Vce rating than a TIP32C part for example, and it had a truncated flyback peak maybe because of this. But why did White's choose the B part then?

[Tinkerer]

Can someone please expand upon what Msysta has posted? I am most interested in the raw basics - the coil is energized and then the energy is cancelled and he says the target eddy current interracts with the flyback decay.
But what if there were a separate receive coil (perhaps at some distance away) that was shunted during the transmit pulse of the traditional PI circuit - would it then pick up the eddy current in a more "pristine" and manageable fashion? Could we do more with this signal as it would be better normalized for receive circuitry?
Kind of seems we are making little EMP devices and receptors...

I am about to find out more on this with a Surfmaster PI I have discussed in that forum section. The drive transistor was a TIP32B. That has a lower Vce rating than a TIP32C part for example, and it had a truncated flyback peak maybe because of this. But why did White's choose the B part then?

bklein,

At the thread below, you can find the circuit of the TINKERERS_V1, discriminating PI, that uses a separate coil and differential input preamplifier.

A spark gap to dampen the Flyback would be very noisy.


http://www.geotech1.com/forums/showp...70&postcount=1

Tinkerer

[msysta]

Hello from white and cold Finland.

What I've learned after the last post (thanks bbsailor, mikebg and others) is that the flyback voltage nulling has a theoretical minimum time so that just simply short circuiting the coil doesn't give the ultimate result in this manner. There is a way called critical damping. That is actually term describing the optimum attenuation of any oscillating system, like spring and shock absorber for example.

See this post for details

http://www.geotech1.com/forums/showt...304#post101304

And what comes to interaction between target eddy current and flyback voltage...

Any coil having current flow generates a magnetic field around itself. Now to be able to induct current to target object using this magnetic field there are three options:

1. move the magnetic field
2. move the target
3. tamper the current

In metal detector use choice number two is usually eliminated. And when searching, the operator is moving the magnetic field but not straghtly to charge the target but to cover larger area . This leaves us with number three.
Also it can be proven that with faster current change, a larger induction effect is achieved.
Term self induction must be known. It is the phenomenon causing the flyback voltage. In short, when current in any coil decreases, also coil's own magnetic field decreases. At the same time coil's magnetic flux decreases, generating self induction voltage which according to Lenz law objects the current reduction.

If a PI detector coil has a current that is cut off it behaves like described above, generating flyback voltage.
We must observe that coil's voltage and current are in different phase; when voltage is in maximum, current is in minimum and vice versa.
Now when maximum flyback voltage is taken to zero as fast as possible, it generates the highest possible current in coil, inducting the highest possible energy to the target.
Also when doing this fast, we are faster to set in receiving mode, listening the echo from the soil.
So there are two flys in the swatter.

This again took the conversation out from the original topic, apologies for that.

With Best Regards,
Markus S.

Happy New Year to everybody.

[simonbaker]

Hello from white and cold Finland.

What I've learned after the last post (thanks bbsailor, mikebg and others) is that the flyback voltage nulling has a theoretical minimum time so that just simply short circuiting the coil doesn't give the ultimate result in this manner. There is a way called critical damping. That is actually term describing the optimum attenuation of any oscillating system, like spring and shock absorber for example.

See this post for details

http://www.geotech1.com/forums/showt...304#post101304

And what comes to interaction between target eddy current and flyback voltage...

Any coil having current flow generates a magnetic field around itself. Now to be able to induct current to target object using this magnetic field there are three options:

1. move the magnetic field
2. move the target
3. tamper the current

In metal detector use choice number two is usually eliminated. And when searching, the operator is moving the magnetic field but not straghtly to charge the target but to cover larger area . This leaves us with number three.
Also it can be proven that with faster current change, a larger induction effect is achieved.
Term self induction must be known. It is the phenomenon causing the flyback voltage. In short, when current in any coil decreases, also coil's own magnetic field decreases. At the same time coil's magnetic flux decreases, generating self induction voltage which according to Lenz law objects the current reduction.

If a PI detector coil has a current that is cut off it behaves like described above, generating flyback voltage.
We must observe that coil's voltage and current are in different phase; when voltage is in maximum, current is in minimum and vice versa.
Now when maximum flyback voltage is taken to zero as fast as possible, it generates the highest possible current in coil, inducting the highest possible energy to the target.
Also when doing this fast, we are faster to set in receiving mode, listening the echo from the soil.
So there are two flys in the swatter.

This again took the conversation out from the original topic, apologies for that.

With Best Regards,
Markus S.

Happy New Year to everybody.

Critical damping may be "fastest" way for linear systems (depending on how you define fast). There still may be faster methods using non-linear techniques.

-SB

[moodz]

I agree with Simon ... active damping should be possible. Active damping is used in areas like car suspensions ( upmarket models ). When you hit a pothole you want to damp the natural tendancy of the suspension to oscillate.

Moodz

[msysta]

I agree with Simon ... active damping should be possible. Active damping is used in areas like car suspensions ( upmarket models ). When you hit a pothole you want to damp the natural tendancy of the suspension to oscillate.

Moodz

I agree that there may be better results available with active damping. Maybe the flyback voltage curvature from peak to zero has different optimum values for damping resistance per derivative moment. Then one could use transistor with control from the pulse itself. Or what I've seen with oscope, the high voltage range decays very fast but range below 50 volts is slower in decay. So maybe the active period could be only in that part.

And those what I've posted earlier are wrong because I am somehow saying that flyback voltage causes the eddy current. That is not the case of course.

[w5cdt]

You're going to need an input offset trim on that CA3130. Your DC gain is so high that you will also amplify the input offset voltage significantly.

[moodz]

You're going to need an input offset trim on that CA3130. Your DC gain is so high that you will also amplify the input offset voltage significantly.

... maybe ... however if you look at the scope pic a few posts up ... this shows the actual output of the amp ( overlayed with output of the first sample switch ) and it works fine without any offset trim. moodz.

[w5cdt]

CA3130 input offset voltage

You got lucky. Check the CA3130/CA3130A data sheet. Max input offset voltage is published as 15mV and 5mV respectively. So for a gain of 1000 and a Vos= +/-5mV the output would be sitting at +/-5V. Clearly the specific 3130 that you used was better than that. But the manufacturer considers a Vos of +/-4.99mV as "good" for a CA3130A factory test screening.

[moodz]

balanced PI backend ....

.... after a few months of busy times in the coalmine ... I now have time to return to this project. Below is a picture of the backend of the balanced PI. The foundation is a generic FPGA development board. The system supports a 50 Mhz 32bit soft RISC system with 16 Mb of RAM 4 Mb of Flash, ether port, USB, 2 x serial, 1 x VGA, rotary switch, LCD etc. ( ie everything you need for a detector box ... just add code )
Code is well developed to implement an inbuilt Digital Oscilloscope that will be able to display TX and RX waveforms in the field without lugging a real one around. The display, timing and triggering will be specially adapted for displaying the interesting parts of a PI detector waveform on a small vga display. The input is a full differential 24 bit ADC sampling at 4 Million samples per second ... this greatly simplifies the frontend electronics ( Goodbye to analogue sample and holds ).
All the housekeeping code is programmed in C ( future plans basic interpreter ) whilst all the time critical stuff is programmed in HDL ( hardware descriptor language .... goodbye nasty assembly code )
All rx processing is done in the digital domain and tx is also generated using programmable state machine to produce arbitrary waveforms including a "listen" function where the detector can receive the TX waveform of another detector and reverse resolve the timing to clone the pulse waveform no matter how complex. All timing clocks are programmable and the test version can implement a 5ns resolution TX PWM modulation wave form with upto 32000 pulse transitions in a single TX cycle. The board utilises high efficiency low noise switch mode supplies to support battery operation.

The total cost of the project to purchase new modular components is pegged at $US 300. ( make your own coil / wand and no cost for software image ).

moodz

[WM6]

.... after a few months of busy times in the coalmine ... I now have time to return to this project. Below is a picture of the backend of the balanced PI. The foundation is a generic FPGA development board. The system supports a 50 Mhz 32bit soft RISC system with 16 Mb of RAM 4 Mb of Flash, ether port, USB, 2 x serial, 1 x VGA, rotary switch, LCD etc. ( ie everything you need for a detector box ... just add code )
Code is well developed to implement an inbuilt Digital Oscilloscope that will be able to display TX and RX waveforms in the field without lugging a real one around. The display, timing and triggering will be specially adapted for displaying the interesting parts of a PI detector waveform on a small vga display. The input is a full differential 24 bit ADC sampling at 4 Million samples per second ... this greatly simplifies the frontend electronics ( Goodbye to analogue sample and holds ).
All the housekeeping code is programmed in C ( future plans basic interpreter ) whilst all the time critical stuff is programmed in HDL ( hardware descriptor language .... goodbye nasty assembly code )
All rx processing is done in the digital domain and tx is also generated using programmable state machine to produce arbitrary waveforms including a "listen" function where the detector can receive the TX waveform of another detector and reverse resolve the timing to clone the pulse waveform no matter how complex. All timing clocks are programmable and the test version can implement a 5ns resolution TX PWM modulation wave form with upto 32000 pulse transitions in a single TX cycle. The board utilises high efficiency low noise switch mode supplies to support battery operation.

The total cost of the project to purchase new modular components is pegged at $US 300. ( make your own coil / wand and no cost for software image ).

moodz

THE FUTURE OF MD IS HERE!!!!!!!!

Fantastic moodz. Congratulation!

What is proper ordering code of your TI Spartan -3E board?

How to get your software image?