Bipolar PI MD with GB and DD Coil

[dbanner]

Looking at the factory timing diagram, when the transmitter shuts off:

• The TGT-STROBE (Target Channel) fires with a 5µs to 20µs variable delay (controlled by the user/trimmer) and stays open for a duration of 5µs.
• The GB-STROBE (Ground Balance Channel) is hard-locked to fire exactly 15µs after the variable delay begins, staying open for 20µs.
• The DISC-STROBE (Discrimination Channel) is fixed to fire almost instantly at 2.5µs after the TX pulse ends, staying open for 8µs.

[dbanner]

(We know the problem, looking at the timing sheet, that 15µs to 39µs window is hard-scaled relative to the target pulse. When you push the TGT-STROBE down to a 5µs delay, a 15µs–39µs ground delay forces the GB channel to sample between 20µs and 44µs down the timeline, the sheer delta in ground voltage between 5µs and 20µs+ creates an impossible amplitude gap for the differential summation op-amps to bridge cleanly, causing the circuit to saturate or chatter.)


( the ground mineralization signal has decayed exponentially compared to what it was at 5µs. Because the disparity between the ground signal at 5µs (Target) and 20µs (GB) is too vast, the differential amplifier runs out of headroom, making a clean manual ground balance impossible.)

There are other potential obstacles as well, having to do with search coil and front end design.\ optimization.

[dbanner]

A very nice project that is complete with nice PCB ready to populate. I like it very much, good work guys! Great job.

[Detectorist#1]

Hi Carl,
My problem isn't overloading of the front-end. The problem occurs with one type of red bricks - the target signal and the GB signal for delays under 10us become with the same polarity and it is not possible to make ground balance.

[pechkata]

Hi,
After many tests in real different grounds last weeks, I understand why the serious manufactures not use in PI MD delays under 10us ( the reason isn't the level of the skills of the engineering staff). All tests in air with lower delays are OK (possibility to catch more smaller nuggets including the sizes of 1grain) but the tests in real grounds shows many problems. This have as result some changes in the schematic of "Dual Tone Pulse Detector Rev.B" The first change is increasing of the min "Target Delay" from 5us to 10us. For this change, the value of C41 have to be increased from 220pF to 470pF. For compensate of the "base sensitivity" level, the value of RT1 have to be changed from 20K to 100K and set from average 10K to average 50K. For avoiding of opposite second tone after the target detecting, the C6 capacitor have to be shorted. Good news is that for these changes no need to make any changes is the PCB project.

[Carl-NC]

Hi Carl,
My problem isn't overloading of the front-end. The problem occurs with one type of red bricks - the target signal and the GB signal for delays under 10us become with the same polarity and it is not possible to make ground balance.

Single-sample target responses don't change polarity when changing the timing. Something else is going on.

[pechkata]

( the ground mineralization signal has decayed exponentially compared to what it was at 5µs. Because the disparity between the ground signal at 5µs (Target) and 20µs (GB) is too vast, the differential amplifier runs out of headroom, making a clean manual ground balance impossible.)

I had the same problem too...

[dbanner]

This is a good guess?:

This brick is maybe superparamagnetic. with remanence,VRM. Worst enemy of metal detectors.

the signal amplitude is extremely huge! in the first few uS after the transmit cutoff. op-amp integrators are blown away! The integrator capacitors drank from a firehouse instead of sipping slowly.

[dbanner]

Possible enclosures for PCB's : Hammond 1598RGY or 1598R BK (Side-by-Side Mounting)

Hammond 1554V (Watertight Polycarbonate / ABS) (Rugged Field Setup (Waterproof Shielding)

Hammond 1599TGY (Flame Retardant ABS) or 1599TABK (Stacking the PCBs (Compact Setup)

[dbanner]

I am looking over some enclosures to house the PCB's:
Possible Enclosures : Hammond 1455Q1601 (Extrusion Aluminum), stack boards
Possible Enclosures: Hammond 1598SGY or 1598SBK (Plastic Instrument Case)both boards flat on the same horizontal plane


the Hammond 1455Q1601 case matches your front and back panel layouts.

[Detectorist#1]

Hi Carl,
I will make additional measurements for target delays from 5 to 10us but in the next week I will be at SPA vacancy.

[Detectorist#1]

Hi dbanner,
Some clarifications regarding the issue with the "special" brick.
This issue is observed in "HYBRID" mode:
- TARGET DELAY is set from 5 to 20 us after the end of the TX pulse. The duration of the TARGET STROBE is 5us.
- GROUND BALANCE DELAY is set to 39us after the start of the TX pulse ( just after the peek of the truncated half-sine wave TX current).The duration of the GB STROBE is 10us.
- DISCRIMINATION DELAY is 2.5us after the start of the TX pulse. The duration of the DISC STROBE is 8us.
After the impossibility to eliminate the signal from the brick with GB POT, the picture on the oscilloscope shown that "target" and "GB" signals after integrators have the same polarity. For Al and Fe targets, these signals at 5us delay have opposite polarity with different amplitudes. I will make additional measurements soon.

[boilcoil]

HI Detectorist#1,
Congratulations again for the wonderful project.

I read that you have problems with the detector working with short target samples. As far as I understand, the problem is that when working with early samples, the signal is reversed after demodulation.
Here is what I think about it:
To graund the device using the subtracting method, the condition must be met > DeltaTG=k*DeltaGB, where "Delta" is the difference in the signal received on the ground and in the air. When DeltaTG is larger, it is necessary to increase the gain in the GB channel.
Unfortunately, demod works not with DeltaTG and DeltaGB, but with the difference of the actual (TG) and (k*GB) samples. In case DeltaTG is much larger than DeltaGB, it is possible that (k*GB) to become larger than (TG). In this case, the signal after demodulation is "reversed" - if it was in quadrant I, it goes to quadrant IV. This can happen to you, especially in half-sine mode, when you take the GB sample in a region where the field change rate is low and the ground response is weak (from the timing you showed, the GB sample is taken just before the end of the on-time, where the current, respectively the magnetic field is almost flat).

This can also happen in pure PI mode, if the GB samples are pulled very far from the TG samples.

To solve this problem (if I understand it correctly), the GB samples must be moved to a region where the field change rate is higher and the ground response is also higher. Of course - you will lose sensitivity, but you will be able to graund MD in the same quadrant where the target sample is.

That's what I think, but I might be wrong - you know I'm not very good at metal detecting, and I'm just thinking in general.

[Detectorist#1]

Hi boilcoil,
First, thank you for your efforts and comments directed to solving of the problem with GB feature for very short "TARGET DELAYS". This week I'm far from my oscilloscope (at SPA vacancy). Next week I will make additional measuring in attempt to find where is the problem. I hope, we will find the solution with common efforts.