PI metal detector for really small nuggets

[green]

Just double the capacitance and keep resistors the same. Response speed should be the same as you had at the lower sample rate, but better noise averaging.

Eric.

(I use an integrator with the switches out side the feedback loop. Gain=Rfdbk/Rin*sample rate*sample time and TC=Rfdbk*Cfdbk. Doubling sample rate doubles gain and TC remains the same.)

Are my formulas wrong? Doubling the capacitance should double the TC lowering the response speed.

[green]

Picture of integrator differentiator I'm using. Max7409 does most of the filtering. Any suggestions for making it better or different?

[Ferric Toes]

Doubling the capacitance should double the TC lowering the response speed.

Yes, but if you then double the sample rate the response speed is raised back to where it was.

Eric.

[green]

Yes, but if you then double the sample rate the response speed is raised back to where it was.

Eric.

Still not seeing why changing sample rate effects response speed. TC=Rfdbk*Cfdbk. Sample rate isn't in the TC formula for a normal integrator with switches outside the feedback loop.

[eclipse]

I have preamp & preamp_inverted that pass through SPDT, they are rotated each cycle and feeded to single ended integrator onto 2 channels (target + ground).
Very interesting chip Max7409 you're using for filtering I may check it out.

[green]

I have preamp & preamp_inverted that pass through SPDT, they are rotated each cycle and feeded to single ended integrator onto 2 channels (target + ground).
Very interesting chip Max7409 you're using for filtering I may check it out.

I use amplifier normal and invert out with unipolar Tx. Was thinking invert out would be delayed causing a longer first delay. Hadn't looked at scope picture of N and I out until this morning. Might have to increase first delay 200ns. Might be worth trying.

Can detect #9 lead shot with GB off. Maybe not with GB on because of noise increase due to longer sample time. Need to solve noise problem.

[green]

Yes, but if you then double the sample rate the response speed is raised back to where it was.

Eric.

Wondering if we are thinking the same thing. Doubling sample rate didn't change response time but did double response speed(rate of change)?

[green]

I have preamp & preamp_inverted that pass through SPDT, they are rotated each cycle and feeded to single ended integrator onto 2 channels (target + ground).
Very interesting chip Max7409 you're using for filtering I may check it out.

Couple scope pictures of MAX7409 in and out. 66n on pin8(4.5Hz filter). Does some good, lot of the higher frequency noise is filtered by the integrator and differentiator.

[green]

Have read where others have gotten an increase in detection distance by adding a resistor in series with the coil. Have wondered why, tried adding a resistor little higher than coil resistance(5R resistor, 4.2R coil). Signal amplitude with resistor about 60% of signal with no added resistor. Decay slope doesn't appear to change.

Checked EF cancelling of the integrator. Adjusted amplifier offset to zero and read offset at integrator out with a DVM. Adjusted amplifier offset for 100mV and saw a change at integrator out. Replaced jumper at integrator input with a 100 ohm pot, wiper to amplifier out. Adjusted amplifier offset to zero and read offset at integrator out with a DVM. Adjusted amplifier offset for 100mV and adjusted 100 ohm pot for same offset reading. No change in integrator offset when adjusting amplifier offset back to zero.

Still wonder if the noise trace I see at integrator or differentiator out is typical of what I should see or am I missing something?

See I made another error. 5 ohms was added to Tx(4.2 ohms)not Rx. Need to do a better job of proof reading before I post.

[Qiaozhi]

Have read where others have gotten an increase in detection distance by adding a resistor in series with the coil. Have wondered why, tried adding a resistor little higher than coil resistance(5R resistor, 4.2R coil).

It can in certain circumstances produce an increase in detection distance because the coil current is allowed to "flat-top". The results also depend on the target time constant.
This is investigated in detail in Appendix C of "The Voodoo Project".

[bbsailor]

Qiaozhi,

Have you ever looked up the theory of how lock-in amplifiers work to extract weak signals from noise? Noise occurs randomly within a duty cycle but sampling many duty cycles in the same location seems to improve the signal to noise ratio, much like integrating many RX samples in a PI metal detector. The key variables that would affect a metal detector would be:
1. The sample frequency
2. The coil diameter
3. The sweep speed
4. The coil style

The above four things determine how many potential target signals could potentially be integrated and the improvement in the signal to noise ratio.

Could you look over the theory of the lock-in amplifier and verify that it is related to improving the performance of integrating many RX signals in a PI metal detector?

Thanks

Joseph J. Rogowski

[Qiaozhi]

Qiaozhi,

Have you ever looked up the theory of how lock-in amplifiers work to extract weak signals from noise? Noise occurs randomly within a duty cycle but sampling many duty cycles in the same location seems to improve the signal to noise ratio, much like integrating many RX samples in a PI metal detector. The key variables that would affect a metal detector would be:
1. The sample frequency
2. The coil diameter
3. The sweep speed
4. The coil style

The above four things determine how many potential target signals could potentially be integrated and the improvement in the signal to noise ratio.

Could you look over the theory of the lock-in amplifier and verify that it is related to improving the performance of integrating many RX signals in a PI metal detector?

Thanks

Joseph J. Rogowski

At the moment I'm working a different project, but I've added this to my long list of things to investigate. It's an interesting question.

[Carl-NC]

Both PI and VLF detectors work in the same way as a lock-in amp. The biggest difference is a lock-in amp doesn't necessarily know what the target frequency is until it manages to pick it out of the noise and "lock in" on it. With PI & VLF, you start out knowing the frequency exactly so you are already locked in.

[green]

Differentiator seems to be giving me some problems. I have been using the upper(a output). Didn't have a 5u film capacitor so I used two 10u X7R in series. Works but dielectric absorption causes output to be slow returning to zero after a large input signal. Need to order some film capacitors. Tried lower(b output). Works but output seems to drift over time. Don't know why. It does multiply input off set by 20. Wired it back to upper circuit to see if I still get drift.

[green]

The upper circuit worked better but still not perfect. Disconnected Tx coil, seemed to solve the problem. Probing the differentiator and Max7409 was causing some offsets with Tx coil connected, lot worse with lower circuit. If I bypassed the Max7409 things looked good with Tx connected so I removed the Max7409 with the upper circuit connected. https://www.geotech1.com/forums/atta...1&d=1611942093

bipolar Tx19 continues changing, maybe someday I'll get it right.