My take on the HH2

[waltr]

hello my masters in this forum I'm new this is the shared part of the circuit you have 16F88 pcb drawing and you can send all documents from my gmail address.
Thank you

No need since the files are posted here.
Just click on then and SAVE to your local drive.

[waltr]

Iron Discrimination

PI Detectors typically can only resolve differences in the Target's TC (Time Constant). The low the targets conductivity and/or size the shorter the TC. The GEB sampling method I use (see above posts on GEB) can give a different output and audio tone for low verse high TC targets. Also medium to small Iron can be put into the GEB "hole" (the GEB cancels the TC response). This is with the GEB sample at about 60-70 in this detector but does not cancel all iron objects. The TC response Hole is another issue to be solved and it seems that MineLab's SD & GPX detectors have a method to eliminate this 'hole' but that is for another time.

Indication of Iron (Ferrous) targets is the goal here. Tinker hinted in his threads on the possibility of detecting iron. He showed scope captures of the per-amp output During TX pulse on with a Balanced coil (concentric or DD). I have been using balance coils on my PI detector (see posts above). I did not adjust these for a perfect, zero output balance but slightly more over so that during the TX on the pre-amp output is at the positive rail then drops to zero Volts at about half the TX pulse period (see post #6 for wave form of coil balancing during TX on and post #14 of waveforms without and with various targets.).

What to look for is the Time the waveform is are the pre-amp positive rail. Iron target (screwdriver) causing the Voltage to drop sooner and both non-ferrous targets (gold ring & silver coin) cause Voltage to drop later. This is a basis for one method of iron discrimination.
A second thing to note in the scope pictures is the shape of the decaying Voltage during TX on. Without a target the Voltage decays to just below Zero then raises a little. With an iron target the Voltage drop well below zero and raises but not to zero whereas with a non-ferrous target the Voltage decay curve is nearly unchanged.
These are the two characteristic I am trying to use to discriminate iron.

First method - Timing the TX ON high pulse.
To keep programming simpler I am using a second PIC microcontroller (PIC16F8 to measure the time of the signal when high. This uses the PIC's Voltage Comparator and an external Vref (10k trim pot) to adjust comparator thresholds. On a high going signal a PIC timer is started (actually just zeros TRM0) and then reads the timer value when signal goes low. This gives the length (time) of the high part of the TX waveform. After power up and a delay to allow the analog circuits to stabilize, multiple time measurements a taken then averaged and saved as the "no target" time.
Now the time of the high TX on is measured and compared to the 'no target' time. If the time is shorted then target probably ferrous. This 'difference' is used to set the duty cycle of a PWM output (PC PWM module running at 19kHz and 50% with no target). An RC filter (2.2k & 1nF) turns the PWM output into a Voltage which can indicate Ferrous or non-ferrous target.
That is the basics on this method however due to the time change is small, I needed to increase the difference. Simply multiplying the difference did not work well since and noise and/or jitter caused hugh fluctuations . Instead I do a digital integration of the timing difference and take and average 8 timing samples (this sets the 'zero' time difference and allows changing coils that have different balance). This did two things: 1- it amplified the timing difference and 2- increased/decreased the difference slower. This works just like the op-amp integration stage seen in most PI detectors.
I did the simulation in an Excel spreadsheet to work out the math. If anyone is interested I can write up how this is done and post the spreadsheet.

Second method - sampling the decay curve during TX On.
This uses a differential input integration circuit. I build a second board with just the sampling switches, integrator and SAT circuits. Then coded the same PIC used for timing to do two samples during the decaying part of the TX ON waveform. The time of the samples is: After TX ON delay 55us and sample for 10us. Then delay 7usec and take second sample for 10usec. These are just like the main sampling of most PI detectors except they are taken during TX on. The integrator goes to a SAT circuit which DC blocks (motion detection) and gains the level.
If the target is ferrous then there is a change of the SAT output level whereas non-ferrous targets have no to very little level change.
I have not yet checked how this changes with different coil having difference balance. Maybe it will just work or I may need to have a calibration done after power up.

Attached is the PIC code used.

So far this has only been bench tested. It will take me a while longer to put into my detector case so I can field test. Questions that need answering are:
Can this reliably indicate a ferrous target?
If not what types (sizes, shapes) does it not indicate as ferrous.
Do any non-ferrous target indicate as ferrous?
What if there is a non-ferrous target close to a ferrous target. Will the ferrous totally mask the non-ferrous? If so then I need to dig all anyway.
How should that Voltage levels indicating ferrous be presented to the operator?
One idea is to have an low frequency superimposed of the audio tones. This can be done in the Audio tone PIC processor (see post #25 on the PIC Dual tone VCO). This would then have a wobbling sound.
Another meter could be added for a visual ferrous indicator (see post #81 on a center zero meter).
Ideas on how to present ferrous info are welcome.

I will probably go with the second PIC to do the ferrous detection since it will get very tricky to do more timing within the current TX & sampling timing sequence. I do have the possibility to change to a larger, faster PIC processor (a reason the PIC is a separate board and not on the main PCB).

To put all into one I would use the PIC18F2610. This runs twice as fast (40MHz instead of 20MHz for the 16F8, has a hardware multiplier, more ADC inputs, more IO pins, a 4th hardware timer and of course more program and ram space.

[green]

I can indicate ferrous-non ferrous on the bench with my PI(looking at signal when Tx on). Problem comes when trying to do it with a varying ground signal. You might want to see what ground signal does while testing your circuit.

[waltr]

I can indicate ferrous-non ferrous on the bench with my PI(looking at signal when Tx on). Problem comes when trying to do it with a varying ground signal. You might want to see what ground signal does while testing your circuit.

Good idea and I will try with the bag of hot soil I have.

Still lots of testing to do and appreciate the suggestions.

[waltr]

Warning: there is a few major bugs in the PIC code is Post # 92.
I have them fixed, it was the timer integration, and will post new code soon.

[waltr]

Ok, new PIC code is done and tested with more comments added.

Currently this has just basic sampling pulses for the analog integrator but full code for the TX IB decay timing, integration and PWM output.

[waltr]

Here is the hex file for the PIC12F1501 VCO.
Look at the source code (.asm) comment at the beginning for the PIC pin definitions in post #25.

VCO-12F1501-HEX.zip

[6666]

Here is the hex file for the PIC12F1501 VCO.
Look at the source code (.asm) comment at the beginning for the PIC pin definitions in post #25.

[ATTACH]45388[/ATTACH]

Thanks thats very good of you, will try it .

T

[waltr]

Been snowing like crazy here so off work doing some coding and learning the new Microchip IDE X and XC compilers. Porting this VCO code to C to learn compiler but unsure yet if it actually works. I'l post the C source once verified.

Decided the PIC VCO circuit needed better documentation so drew a full schematic with all the options.

Examine while reading the assembler source code (attached here in zip).
VCO-12F1501 -Working1.zip
PIC1501-VCO.pdf

[6666]

Hi Waltr thanks for the update on the vco
back in post #25 you talk about the circuit posted by mickstv (I did prototype this circuit and it works very well):
the schematic in Micks post has now gone due to being posted on external site and expired
if you still have it would it be possible for you to post it here for reference and to last longer thanks.

[waltr]

Hi Waltr thanks for the update on the vco
back in post #25 you talk about the circuit posted by mickstv (I did prototype this circuit and it works very well):
the schematic in Micks post has now gone due to being posted on external site and expired
if you still have it would it be possible for you to post it here for reference and to last longer thanks.

Yes, I do have a copy and Posting here:


Mod'ed

[waltr]

Have you tried the VCO code?

[6666]

Yes, I do have a copy and Posting here:
[ATTACH]46430[/ATTACH]

Mod'ed
[ATTACH]46429[/ATTACH]

Thanks, this should last a long time here, hosting pictures off site can disappear quickly

[6666]

Have you tried the VCO code?

I have been trying, but my usual supplier does not carry that chip, so been trying to source one at decent price, postage really kills business here in Oz, would that hex go into another chip ?

[Polymer]

Wow, this looks great waltr. What sound "profile" does it have?
I am far far away from getting a pic programmer ... still trying to understand simple analog PI stuff.
Working on better / more informative sound is a field that is very much worth pursuing I find.

Have you come across these specialist IC's yet? I find them very interesting, quite out of the box and relatively new to the market.

http://www.alfarzpp.lv/eng/sc/application.php