PICKINI V3 - an easy to build PI detector

[N0QBH]

I'll have to add the caps and see if that rectifies the problem. Thanks.

My 2cts so far:
I have never had any trouble with my good old PICKit3 programmer in combination with any of the Microchip controllers supported.
I always provide power from the programmer and never had to lower the Vcc.
I plug in a homemade breadboard with an 8 pin socket and pin header to match the PICKit connector.
I have also placed a 10 microFarad elco in parallel with a 100nF ceramic capacitor between pin 1 and 8 of the socket ( Vcc and Gnd). Maybe this can make the difference when current peaks occur during the programming phase ?



Best regards,
- Bernard

[F117]

Next to the decoupling capacitors, I also have a 4K7 pullup resistore on the RST input pin, so between pin 4 and pin 1.

[N0QBH]

Experiences so far...
Did a few hours of tweaking and experimenting with the OSH Pickini board yesterday using a metered bench supply and a home made 8" mono coil at 344 uH.
Operating current is about 130 mA at 12v.
Pretty happy with it's performance in air so far - a penny at 6" with a nickel and quarter about the same.

Built up the 12v low dropout regulator circuit and powered the Pickini with a charged 4 cell LiPo 2200 mA/hr battery.
The circuit seemed unstable and I noticed the audio tone seemed to stutter on weak signals. Adjusting sensitivity and/or R23 didn't seem to help much.
A quick check of the power supply with my scope and didn't show anything.
Today I removed the regulator circuit and put a 3 cell 2200 mA/hr LiPo connected directly. Much better.
A couple questions.
Does the PIC ADC start measuring the positive going flyback at 3.25 volts?
What is the resolution and how long is the measurement taken?

A picture.


Mike

[F117]

Does the PIC ADC start measuring the positive going flyback at 3.25 volts?
What is the resolution and how long is the measurement taken?

The flyback pulse is going to one input of the internal comparator. The other input of the comparator is at a fixed slicing level. The output pulse of the comparator is the gating signal for an internal 16 bit timer running at 32 MHz. The sensitivity potmeter is a voltage between 0-5V. This is measured with the internal ADC and controls the number of timer samples that are accumulated. The resulting number from the timer is fed into running average low pass filter.
This is in a nutshell a description of the PICKINI archtecture .

As I already mentioned a couple of times: V3 does not have an onboard power supply regulator and when the power supply changes, the point where the flyback is sliced may shift. When this point gets too close to the DC end value, the reading will become unstable. When the slicing level is too low, the sensitivity will drop. It is important to adjust R23 accurately as described here http://users.telenet.be/willaert/MD/...ectronics.html:
Adjust the DC offset of the opamp output with multiturn R23:
Measure the DC voltage at pin 3 of the PIC12F1840 and adjust from low voltage until 5.3V is reached.
Stop adjusting when 5.3V is barely reached.
The voltage won't go any highter since it is clipped internally in the microcontroller with a diode to 5V.


When you compare this temporal measurement (pulse width) with a voltage measurement, you encounter similar problems in the voltage domain: delay too small = measurement unstable ( slope of the flyback pulse) and when the delay is too large: steady state and very low sensitivity.

I am currently field testing PICKINI V4, which has no more adjustments, but calibrates itself each time you switch it on. I will publish this here later on this year, when testing results are satisfactory.

Best regards,
- Bernard

[N0QBH]

Thanks for the info, Bernard. What gates the counter off?
Here's a photo of my pulse. I noted the shape changes (gets rounded off) at lower R23 settings.

Will Pickini V4 work with the present hardware? Your continued work is appreciated
Mike

[F117]

Thanks for the info, Bernard. What gates the counter off?
Here's a photo of my pulse. I noted the shape changes (gets rounded off) at lower R23 settings.
pickiniV3_01.jpg
Will Pickini V4 work with the present hardware?

Mike,
The timer that measures the puls width is gated by the output of the internal comparator.
The pulse on your scope looks OK.
PickiniV4 is different hardware - no adjustment potmeters.

- Bernard

[N0QBH]

I assume you start the timer on the positive going swing at the slice point of 3.25 volts.
Do you reconfigure the comparator to trigger at near Vdd in order to determine when to stop counting?
Just trying to get my head around the process.
Will V4 require a new pcb layout or will I be able to use the V3 pcb and omit some parts?

Mike

Mike,
The timer that measures the puls width is gated by the output of the internal comparator.
The pulse on your scope looks OK.
PickiniV4 is different hardware - no adjustment potmeters.

- Bernard

[F117]

Mike,
In your post #170, you see the flyback pulse on the scope.
When you slice it with a comparator at 3.25V, you get a positive pulse ( assuming the flyback pulse is at the negative input and the 3.25V at the positive input). This pulse enables counting of the timer.
If the pulse width is 100 microseconds and the clock is 32 MHz, you get a number of 3200. When a target comes near the coil, the number will increase. ( pulse width increases).
Not many secrets here...

- Bernard

[N0QBH]

OK. In the metal detector book it said the sampling was being done at the "knee" of the positive going part of the negative pulse.
While experimenting, I noticed the pulse width shift while putting a coin near the coil, so I guess you're using another way to "skin the cat".
Thanks for the info.
Mike

Mike,
In your post #170, you see the flyback pulse on the scope.
When you slice it with a comparator at 3.25V, you get a positive pulse ( assuming the flyback pulse is at the negative input and the 3.25V at the positive input). This pulse enables counting of the timer.
If the pulse width is 100 microseconds and the clock is 32 MHz, you get a number of 3200. When a target comes near the coil, the number will increase. ( pulse width increases).
Not many secrets here...

- Bernard

[6666]

Some progress on pcb's
Your boards are being made as you read this!

We've sent the panel containing your boards to the fabricator. We expect to get them back around June 10th.

[F117]

In the metal detector book it said the sampling was being done at the "knee" of the positive going part of the negative pulse.

In the book, it is "PI design 5" - p170 in the first edition.
What you describe is voltage sampling. PI design 5 and Pickini measure the pulse width of the flyback.

[N0QBH]

Thanks, I somehow missed that page.
Is there any clear advantage using either voltage or time change detection with the same coil?

In the book, it is "PI design 5" - p170 in the first edition.
What you describe is voltage sampling. PI design 5 and Pickini measure the pulse width of the flyback.

[F117]

Is there any clear advantage using either voltage or time change detection with the same coil?

The short answer is: with classic voltage sampling, you can determine a fixed delay for the sampling point, which may give you some kind of discrimination added value, eg for gold you want a very short sampling delay. With pulse width detection, all metals will increase the pulse width , I don't know of any way to have some form of discrimination. I already tried sampling the decay at different levels to determine the angle which reflects the material conductivity, but so far I didn't succeed in adding any value because of this.
There is also a longer answer, like this one: post #317 here: http://www.geotech1.com/forums/showt...etector/page13

- Bernard

[N0QBH]

What is the preferred cable between the coil and circuitry?
If twisted pair is used, what is the twist spacing?

[N0QBH]

I went ahead and used some coaxial type video cable between the coil and circuit. Performs about the same as the twisted wires it replaced but is much easier to manage.