PICKINI V4 - an easy to build, self adjusting PI detector

[SaltyDog]

Thanks guys ..

I have installed mikroC Pro for PIC under WINE, and that works well, I was able to compile the code just fine (once ticking the ADC & PWM module boxes)
I will have to play around with fonts, as the defaults are horrible under WINE ..

Just a question: What is the story with licensing for mikroC? It seems to work ok without registering the product, is it time limited?

For programming I have found PICPgm for linux, that seems to be well supported .. I will try it out when I get my hardware built ..

[Altra]

Just a question: What is the story with licensing for mikroC? It seems to work ok without registering the product, is it time limited?

There are no time limitations, I believe you are limited to a 2kb size program.

[Qiaozhi]

Thanks guys ..

I have installed mikroC Pro for PIC under WINE, and that works well, I was able to compile the code just fine (once ticking the ADC & PWM module boxes)
I will have to play around with fonts, as the defaults are horrible under WINE ..

Just a question: What is the story with licensing for mikroC? It seems to work ok without registering the product, is it time limited?

For programming I have found PICPgm for linux, that seems to be well supported .. I will try it out when I get my hardware built ..

As Altra said, you cannot generate hex output over 2K of program words with the free version. I don't know how that relates to source code, as I have a dongle-licensed copy.

To get the fonts corrected, you will probably need to install the Microsoft fonts. You can do that easily as follows:
sudo apt-get install ttf-mscorefonts-installer

I know that fixes the font problems with Firefox, so it will probably do the same for other programs.

[SaltyDog]

As Altra said, you cannot generate hex output over 2K of program words with the free version. I don't know how that relates to source code, as I have a dongle-licensed copy.

To get the fonts corrected, you will probably need to install the Microsoft fonts. You can do that easily as follows:
sudo apt-get install ttf-mscorefonts-installer

I know that fixes the font problems with Firefox, so it will probably do the same for other programs.

Thanks. I am on ArchLinux so doing "winetricks corefonts" did the trick .. (it installs the ms core fonts within wine )

[SaltyDog]

Another couple of questions regarding the schematic.

1. Q2 is shown as a transistor, where as my understanding a BS170 is a MOSFET, just a bit puzzled ..
2. Bernard, did you think about adding a ground-plane to the top of the PCB, or is noise not an issue ..
3. Another point of confusion. In the software it looks like you are doing a calibrate_offset() every 2ms, yet in the project specs you state that the calibration is only done at startup, what am I missing?

[waltr]

1. Q2 is shown as a transistor, where as my understanding a BS170 is a MOSFET, just a bit puzzled ..

A MOSFET is a type of transistor which come in N-ch & P-ch. JFETs, NPN BiPolar & PNP Bipolar are also Transistors.
https://en.wikipedia.org/wiki/Transistor

In the v3 schematic I see Q1 shown as an NPN BiPolar transistor but marked BS170.
Since this drive the Gate of Q2, the Coil driver, either an N-ch MOSFET (BS170) or an GP NPN BiPolar (2n3904) would work.

[SaltyDog]

A MOSFET is a type of transistor which come in N-ch & P-ch. JFETs, NPN BiPolar & PNP Bipolar are also Transistors.
https://en.wikipedia.org/wiki/Transistor

In the v3 schematic I see Q1 shown as an NPN BiPolar transistor but marked BS170.
Since this drive the Gate of Q2, the Coil driver, either an N-ch MOSFET (BS170) or an GP NPN BiPolar (2n3904) would work.

I am an electronics engineer of some 50 years, I do know what the difference is between a transistor and a mosfet is.

I am referring to the V4 schematic, it shows Q1 correctly as a MOSFET, but Q2 is shown as an NPN transistor (BS170) it is not, the BS170 is a MOSFET and should have that symbol, you normally don't replace one with the other ...
The fact that it shows BS170 and a NPN symbol, is what is the confusing part ..

[SaltyDog]

Hi Bernard,

In the software code listing, I am not seeing any code for the error tones (i.e missing coil etc ) Is the published code on http://users.telenet.be/willaert/MD/.../software.html the latest?
(the header shows it as v4.1)

Thanks

[F117]

@ SaltyDog:

Q2 is shown as a transistor, where as my understanding a BS170 is a MOSFET, just a bit puzzled ..

The BS170 is a MOSFET. The symbol in the schematic is not correct. Something to do with the eagle library.

Bernard, did you think about adding a ground-plane to the top of the PCB, or is noise not an issue ..

Noise is always an issue - especially noise picked up by the coil when testing indoors. I never tested with a dual plane PCB. It seems to work fine the way it is ( outdoors ).

Another point of confusion. In the software it looks like you are doing a calibrate_offset() every 2ms, yet in the project specs you state that the calibration is only done at startup, what am I missing?

The function calibrate_offset() is used to do a successive approximation setup of the DC offset of the opamp. The goal is to get the steady state DC of the output signal at 8V = 4V at the input of the microcontroller comparator. After startup, this function is called a number of times - instead of the normal tx_pulse_processing(), until the DC level is OK. Then, the normal pulse processing takes over every 2 ms. Check the flag if (!calibration_busy )...

In the software code listing, I am not seeing any code for the error tones (i.e missing coil etc )

There are 2 different tones: multi frequency tones generated wit the sound() function, like void start_sound() and ready_sound().
There are also monotonous sounds, that just fill in a beepdivider value and generate a continuous tone : beepdivider = alternate_beepdivider;
That is the case for the error sound - or low battery sound: a low frequency tone ( in the case of coil error, a continuous tone - in the case of low battery: a low frequency tone : alternate_beepdivider_time.

[SaltyDog]

That explains it all many thanks..

[waltr]

I am an electronics engineer of some 50 years, I do know what the difference is between a transistor and a mosfet is.

A MOSFET IS a Transistor!!!!

[SaltyDog]

A MOSFET IS a Transistor!!!!

True, but as Bernard picked up, I was referring to the symbol .. seems I was not making myself clear enough for you ...

[SaltyDog]

Hi Bernard,

I am thinking, without too many modifications, it is possible to do simple two tone discrimination, by just some simple code changes (as per the White's PI)

If you have a look at the attached image, it shows the pulse decay for different metals (ferrous, non-ferrous). If you judiciously choose the slice level (i.e maybe some manual control), then you would get the pulse width "decreasing" for non-ferrous.
(i.e if you put a line at 0.3v on the graph, then non-ferrous would give you 5us, no target 7us, and ferrous 12us .. as an example)

If you map that to a different tone (and/or light) we could get some form of discrimination ..

Let me know what your thoughts are ... I could have a go at the code changes to try it out ...

You would of course need a fast coil ..

[F117]

I tried it once with 2 comparators at different slicing levels: one low and another one closer to steady state.
You will notice with a scope that the decay curve angle is different for ferrous and other metals.
Taking into account these 2 pulse widths, you can derive the decay angle.

Tried this and eventually gave up on it. I had a max of 500 Hz for ferrous and 1kHz for other materials.
But I abandoned it after some field tests. Don't remember what was wrong exactly, but it was never a guarantee not to dig any iron...
Feel free to adapt the code any way you like.
You could also consider adding 2 ADC samples on the flyback and combine differential voltage samping with pulse width measurement...

[waltr]

I have two tone working on my HammerHead 2. To do this I use the Three Sample GEB method with the first two samples going in the differential integrator. Low conductors swing the integrator output low and high conductors swing the integrator output high. This then goes to a VCO circuit (actually a small pic that does a Voltage in to tone out function).

Works great to determine a high verse a low conductor but Iron can go either way depending on its size, shape and exact composition (alloy). So never worked to disc out iron.