PI Induction

[Gilmar Ribeiro de Oliveira]

I have worked in Pulse induction project, and I
use a FET FQP3P50, FET P-channel, because i'm using microcontroller and
P-channel is easer to interface with microcontroller.
The circuit is working well, but I need more distance to get metal, more
sensibility, I got 30cm steel that have in the standard of metal detector. I tested different delay, pulse width, frequency and I don't have success.
If I increase the gain of the op-amp the output signal is unstable and it isn't good read with the AD from microcontroller.
I use a coil with 22 turn and the size is 9.8" x 19.6" (25 cm x 50 cm), but
I tried many other number of coil turn without good results.
The breakdown voltage of the PFET is 500V.
I tried to use post-gain filtering but I think that the my circuit is on
the maximum gain and after that there is a lot of noisy.
I use the 70uS on the pulse on the coil and 20uS how delay and 100uS window
of the sample. I tried to change all this parameter without good results.
I'll try the coil concentric, I think I'll get more sensitivity.
How many turn does RX coil ? How can I make this kind of coil ?

If you had some tips, or some good idea, please help me.

Best Regads,
Gilmar

[FJ]

Re: PI Induction

Hi,
You offer lots of info but really not enough. You really shouldn't need post gain filtering. My detector is getting incredible depth without it as most all I've seen do. What is the gain of your initial op amp? You mention "the op amp", is there more than one? Are you using a differential integrator after the initial opamp? What is the gain of that stage? Do you have a final amp after that stage? What is the gain of that stage? Also, very important, have you looked at the op amp rails with a scope to see how clean the voltage is? If you are using a charge pump or some type of boost circuit, that can be a source of nasty noise. How is the PIC supply, are you filtering it well? Here is a real key with using a PIC, how about board layout. I had been through all of the above and found problems and cleaned them up and still had problems. The detector was unstable and drifted like crazy. After a lot of time spent on analyzing and burned up brain cells, I decided I had done everything right however I was a novice at PCB layout and that had to be the problem. I searched the internet and found good articles on proper board layout with combination analog and digital (PIC) circuitry. I followed these rules with completely laying out the board again...no circuit changes. The detector cleaned up and came alive. I was able to crank the gain up on the final amp more and the detector has unbelievable depth. I have done some other improvements over the published designs I've seen on the net since then and improved depth but the board layout to isolate the digital from analog was the key when employing a processor on high gain amplifier circuits like PI detectors use.
Good luck,
FJ

[Kev]

Re: PI Induction

Hi,
Something FJ didn't mention is your sample window at 100us? This is very long, although you gain a large slice of signal, it is also accumulates heaps of noise which may not equalise over such a long time.

Cheers Kev.

[Gilmar Ribeiro de Oliveira]

Re: PI Induction

Hi Cheers.
Thanks for you reply.
I tried change the sample window and didn't have good results.
What is a good sample window ?

Gilmar

[Gilmar Ribeiro de Oliveira]

Re: PI Induction

Hi,
Thanks for reply.
I use 3 op-amp, the first has 2000 of gain(NE5534) and the next (LT072)is a integrator with 100K resistor and 100nF and 1k to ground, after that i have a filter 100K and 100nF to ground. In this point the signal is very stable and I have the last op-amp low gain, because if I increse the gain the signal to be instable.
It get a metal of 10x2 cm with a deep of 35 cm, with about 1V in the output of the last op-amp.
As for the power of the op-amp I have different +5V and -5V for the op-amp.
send me you E-mail that I send to you see my schematic.

Thanks.
Gilmar

[FJ]

Re: PI Induction

Hi Gilmar,
I answered you in a private email. Hopefully I can help eliminate the many hours I spent getting through the very same problems you describe.
Good luck to you,
FJ

[Kev]

Re: PI Induction

Hi Gilmar,
From what I've gleaned, a receive window from 10 to 20 us should suffice with a pulse width of 70us.
Once you try what FJ recommends you may be able to successfully reduce this, and get a better s/n ratio.

Regards
Kev

[Eric Foster]

Re: PI Induction

Hi Gilmar,

I find that I get good results if the sample window is the same as the delay. i.e. 20uS delay and 20uS sample window.

Eric.

[Eric Foster]

Re: PI Induction

Hi Gilmar,

Gain of 5534 may be a bit high. I run this stage at about x500 and put more gain later on, after the integrator and futher filtering. At higher gains for the front end, the bandwidth drops off, affecting the recovery time.

Not sure if I understand your integrator values right? Usually between 1k and 10k for the input time constant resistors, and 470k or 1M for the discharge resistors, one of which goes to ground in the case of a differential integrator. 100nF is OK for the capacitors. Is it a differential or single ended integrator? If differential, then the capacitors need to be matched.

I use 10k and 100nF for the post integrator filter.

To check if the noise is internal or external, wind a figure of 8 coil. This cancels pickup noise, particularly at low frequencies.

Eric.

[Gilmar Ribeiro de Oliveira]

Re: PI Induction

Hi Eric,
I'll try this 20uS delay and 20uS sample window.

Thank you,
Gilmar

[Gilmar Ribeiro de Oliveira]

Re: PI Induction

Hi Eric,
I use a single integrator with 1k to ground and 100k for discharge resistor and 100nF. On the output of this opamp I use a filter, 100K and 100nF to ground, It reduce the noise.
My coil is 26 turn with 550uH. Is It correct ?

Thanks.
Gilmar