felezjoo PI(the best pulse induction metal detector that I made until now)

[jladre]

--------------------------------------------------------------------------------
Hi Jladre,

The detector audio on this is not 'loud and clear' and is difficult to hear. How much does your 6X6mm soda can side wall target weigh in tenths of grains.

Dan

i don't know why you can't hear the audio, it is pretty loud even the mic is far from the machine..im sorry i don't have weighing tools to measure in grains.

[jladre]

Thanks for the information.
A couple more questions if you don't mind:

• How do you set the delay to auto? Is this option available in V3.30, as I seem unable to find it?
• What is your Iron Reject setting?

P.S. I've just noticed ... if you press the Down button, it displays 19 (+0) when the delay is set to zero. So I presume this means the minimum delay setting is 19us, and the delay setting in the menu represents a positive offset to this value. Is this correct?

the delay 0 is the auto mode. its not actually always 19 because it vary on your coil(coil&cable) my old coil got 18 (+0) sold coil 10(+0)and the new got 16(+0). iron reject setting also vary in the coil. one of my coil set on 13 the other is 16 and another of my coil set to 23.. screen shot below

[ivconic]

the delay 0 is the auto mode. its not actually always 19 because it vary on your coil(coil&cable) my old coil got 18 (+0) and the new got 16(+0). iron reject setting also vary in the coil. one of my coil set on 13 the other is 16 and another of my coil set to 23..

Yes, in previous versions it was "AUT" and now is "0".
Iron rejecting starts accurately at "8" with my coil.
I haven't enough time to experiment with all the possible adjustments, but as i assumed; it can detect small Al can cut decently.
...
Qiaozhi have you made it already?
What are your impressions?
Any photo?

[reptooyep]

I am getting constant email&pm demands for sharing hex code with patched counter.
Sorry folks but can't do that. I already explained my reasons.
Simply it's not fair to do something like that against colleague.
But i can help you just a bit; by giving you small hint.
If you put an effort you will gain the result on your own.
And that's about all what i can do on this, now and never again.
Please understand.
Here is the hint:

I cannot find any result on google for Owoshuma soft ...
It looks pretty cool. I love HEX editing but HXD is too hard to work with.

[ivconic]

I cannot find any result on google for Owoshuma soft ...
It looks pretty cool. I love HEX editing but HXD is too hard to work with.

Oh now you are provoking me to continue with revealing the method!
It's not about editing hex. It will lead you to nowhere.
It's about executing it step by step and tracing instructions.
To see what instruction doing what in code.
How it allocates the memory. At what moment what register is holding what data... etc.
Nice thing is because you may freeze it and analyze it in peace.
And that's how...
AVR Studio is just fine for that job.
I am using obsolete software by habit.
Reason why i am running it in vbox is because it will not work under win7.
In "compatibility mode" under win7 it is working... but producing lot of bugs and can freeze for no reason.
So it's not about plain hex editing.

[ivconic]

I cannot find any result on google for Owoshuma soft ...
It looks pretty cool. I love HEX editing but HXD is too hard to work with.

Strange?
I guess it's up to the firewall of your provider.
Some countries are having really strange policies upon net surfing, kinda "Orwellian" situation.
Same thing also on torrent sites.
Luckily some of us are still living in a free world:




Not really a problem, like i said; Atmel Studio is having all the "tools" too.

[jladre]

just found out that 16 us delay displayed on my lcd is actualy 8us on oscop. not accurate.. anyway it's time for vacation.. hunting in the south..

[reptooyep]

Strange?
I guess it's up to the firewall of your provider.
Some countries are having really strange policies upon net surfing, kinda "Orwellian" situation.
Same thing also on torrent sites.
Luckily some of us are still living in a free world:

[ATTACH]36030[/ATTACH]
[ATTACH]36031[/ATTACH]

Not really a problem, like i said; Atmel Studio is having all the "tools" too.

Hi Ivconic,
I'm realy puzzled with ATMEL studio 6 ..
What i did is dissassemble the file using CMD and the command "avr-objdump -j .sec1 -m avr5 FelezJoo_pi_v3_30.hex > Felez.txt"
Now i have a file containing all adresses and the asm code for each adress.
I copy/past in a atmel studio asm project, i keep just the asm code (which is very long to do). Some of the statements are not recognize by ATMEL (although they appears in blue) :



I did that to have the opportunity to use the debugging cause when i open the .hex file directly by "file-open" in ATMEL, i cannot do anything (the file i see in not disassemble). I don't have acces to the debugging functions (need the ASM file ??).

Is ther a simplest way to do that in ATMEL STUDIO 6 ?

I'm realy interrested into your owashima soft, could you share a domnload link please, i tryed everything but my reasearch does not give any result, even with THOR BROWSER !!

[ivconic]

Hi Ivconic,
I'm realy puzzled with ATMEL studio 6 ..
What i did is dissassemble the file using CMD and the command "avr-objdump -j .sec1 -m avr5 FelezJoo_pi_v3_30.hex > Felez.txt"
Now i have a file containing all adresses and the asm code for each adress.
I copy/past in a atmel studio asm project, i keep just the asm code (which is very long to do). Some of the statements are not recognize by ATMEL (although they appears in blue) :

[ATTACH]36049[/ATTACH]

I did that to have the opportunity to use the debugging cause when i open the .hex file directly by "file-open" in ATMEL, i cannot do anything (the file i see in not disassemble). I don't have acces to the debugging functions (need the ASM file ??).

Is ther a simplest way to do that in ATMEL STUDIO 6 ?

I'm realy interrested into your owashima soft, could you share a domnload link please, i tryed everything but my reasearch does not give any result, even with THOR BROWSER !!

I can't share software, it's internal company property, sorry.
But nothing special with it. Any other will do the same.
...
If you want to see textual strings within hex file; you have to convert it to bin format.
But plain editing those textual strings will lead you to nowhere.
You have hex2bin, bin2hex, avr-objdump, disavr... and plenty more tools available.
Ok here are two files, you may try with them just to exercise:

[reptooyep]

Wat programmer are you using?
Presumably some Usbasp or similar?
Also, are you typing avrdude parameters in cmd or you are using some of the gui's ?
Why don't you try something odd; just flash hex file without worrying too much on fuses?
I am suggesting this because i notice that E fuse is always turning different from what i type in Avrdudess.
Now.. i am too lazy to inspect why it is happening, but i flashed hex once without setting up the fuses and it works the same good as when i adjust fuses precisely!?
I assume it's because Avrdudess's defaults are alright.
Here is the reminder:
http://www.engbedded.com/fusecalc

...
I recently added cheap Chinese Usbasp dongle to my "arsenal". And on first try i also got "dead" LCD!!
Flashing went "alright" but later device was dead with blank LCD!
So stubborn as i am; i spent whole afternoon to inspect that.
And the whole thing was about obsolete f/w in dongle (slow sck provoked constant alerts and errors).
So, what i did; i opened it and located "self-programming" jumper somehow (not marked, hidden, so i had to trace chip pins and traces) and than using
Arduino UNO as ISP i flashed new (and very good) f/w in it. From here: http://www.fischl.de/usbasp/
On next try it started to work like a charm (no more sck alerts), again flashed Atmega and detector works 100% correct.
I did whole thing just to avoid bootloader presence on Atmega, because newest version 3.30 has conflicts with it.
The results you already seen on my last youtube video in one of the previous posts.
So... sometimes even when flashing went "alright" - actually it is not alright. (Goof old times of parallel programming! Now i am stuck with USB because my laptop is not having parallel.)
Especially if you mess too much with typing in cmd. Huge chance to miss or mismatch some parameter.
Good old DOS times has passed long time ago!
Maybe this story will give you a hint and help you ... and maybe wont.
But there is no way to help you like this remotely, you must do it on your own.
...
BTW Extreme Burner sucks! Avrdudess and eePe RULEZ!

Hi,
I still have trouble with my avr, i think i tryed everything :
1 : I tryed several atmega but they all work on other projects
2 : I tryed several crystal but it's the same as another project that works.
3 : I checked the wiring ten times without any bad connection
4 : As you suggested, i put the firmware usbasp.atmega8.2011-05-28.hex in the two usbasp unit i have.
5 : i wired only pins 1 (dumping resistor), pin 7 and 20 to vcc, 8 and 22 to ground, pins 9 and 10 to crystal and pin 17 to audio circuit : No vital sign
6 : I checked the pins on my oscilloscope and on pin 10, i can't see the crystal oscillations, only a 5v DC

I put this chip in another project and it work great, i can see the crystal oscillations.
The fuses are OK, the power supplys of the board are 12v and 5v, all the ground are grounded.

I think it drives me crazy

[ivconic]

Hi,
I still have trouble with my avr, i think i tryed everything :
1 : I tryed several atmega but they all work on other projects
2 : I tryed several crystal but it's the same as another project that works.
3 : I checked the wiring ten times without any bad connection
4 : As you suggested, i put the firmware usbasp.atmega8.2011-05-28.hex in the two usbasp unit i have.
5 : i wired only pins 1 (dumping resistor), pin 7 and 20 to vcc, 8 and 22 to ground, pins 9 and 10 to crystal and pin 17 to audio circuit : No vital sign
6 : I checked the pins on my oscilloscope and on pin 10, i can't see the crystal oscillations, only a 5v DC

I put this chip in another project and it work great, i can see the crystal oscillations.
The fuses are OK, the power supplys of the board are 12v and 5v, all the ground are grounded.

I think it drives me crazy

Maybe there is physical error on your pcb? Which one have you made?
And finally maybe you are using corrupted hex file?
Here are all the working and tested versions that i have, try each one of those and see:

[Qiaozhi]

the delay 0 is the auto mode. its not actually always 19 because it vary on your coil(coil&cable) my old coil got 18 (+0) sold coil 10(+0)and the new got 16(+0). iron reject setting also vary in the coil. one of my coil set on 13 the other is 16 and another of my coil set to 23.. screen shot below

Yes, I've been experimenting further, and my 300uH coil mostly shows 19us delay and sometimes 18us.

Yes, in previous versions it was "AUT" and now is "0".
Iron rejecting starts accurately at "8" with my coil.
I haven't enough time to experiment with all the possible adjustments, but as i assumed; it can detect small Al can cut decently.
...
Qiaozhi have you made it already?
What are your impressions?
Any photo?

Here's a photo of my Felezjoo ->
I designed a new layout that has the tactile switches on the PCB and/or off the board. Plus the LCD was meant to mount on top of the PCB, as shown in this 3D representation ->
However, the available capacitors were taller then expected, and the LCD pinout was different. Hence it was mounted on the end. It was only an experimental PCB anyway, so no problems.

With my 300uH coil I managed to get some settings that will reject an iron bolt and accept several coins, but some silver coins were then a bit iffy. I also probed the TX oscillator signal with the scope. I was expecting to see sets of pulses, maybe varying in width (after seeing your scope shots) but the pulses are actually all the same width (as set up in the menu). It seems that your PC scope is having a problem sync'ing because there are gaps in the transmission, presumably because the processor is busy doing something else, such as updating the display. As far as I can see so far, there's nothing magic going on here. It's simply some sort of conductivity discrimination based on sampling around the decay curve. The Earth field rejection works very well though. Also need to try the ferrous discrimination with different sized targets.

The interesting thing that has come out of this is that the LF357 is quite a lot faster than the NE5534 (or NE5532). However, comparing a single-stage preamp with an LF357 against a NE5532 as a 2-stage preamp, there is little difference if any.

just found out that 16 us delay displayed on my lcd is actually 8us on oscop. not accurate.. anyway it's time for vacation.. hunting in the south..

I'm not sure where you could possibly be measuring the actual sample delay for this detector, since it uses direct sampling, and hence the actual sample delay is hidden in the software. I suspect you are looking at the preamp output, and making a presumption from there. In which case the value displayed on the LCD is most likely the correct one. With a 16us sample delay, any target with a TC of less than 3.2us will have completely died away before sampling takes place. As far as I can ascertain, the detector continuously changes the sample delay, not just at switch-on, by checking when the preamp comes out of saturation. You can see this yourself by placing the coil near some metal, and then displaying the delay setting by pressing the Down button.

I also tried detecting a 5mm square of aluminium can (sent to me by Nupi, along with some hypodermic needles after an earlier discussion) and it was undetectable, even when using the settings you supplied. Then I cut my own small aluminium square, this time 10mm x 10mm, from a Schweppes Tonic can. This was cut from the side wall, and had a thickness of 0.1mm. Even trying many different menu settings, the best result was about 5mm detection distance with the target close to the actual coil windings.

At the moment I don't know what the relationship is between integw and sensitivity, but they appear to affect each other. It seems that integw is maybe the amount of averaging (integration) applied in software, and lower values give a sharper response.

I also tried increasing the frequency to its maximum value of 999pps, and reducing the TX pulse width to its minimum of 100us, in the hope of getting a shorter sample delay. However, the results was much the same.

In order to get consistent results, I suggest that everyone check the thickness of their 5mm aluminium square target. As noted in another thread, the material in the top and bottom of a soda can is substantially thicker than the side wall. I remember from some calculations I did previously that target thickness makes a huge difference to the TC.

[6666]

The interesting thing that has come out of this is that the LF357 is quite a lot faster than the NE5534 (or NE5532). However, comparing a single-stage preamp with an LF357 against a NE5532 as a 2-stage preamp, there is little difference if any.

I wonder how fast two LF357 would be in a 2-stage preamp ?, and thats handy to know if you cannot get a LF357, thanks.

My 5mm aluminium square target is .1mm thick.

[jladre]

You can get advantage by using low Rds mosfet.. You will found out that the usual minimum delay 18(+0) will reduce depending on mosfet.

[Chet]

Hi Qiaozhi
A few thoughts come to mind from your post #867;
If the discrimination is based on the target decay or time constant curve then a minimum delay of around 16us would delay the first curve sample past undesired signal excursions that could greatly affect the beginning of the curve. As you suggested for strong target returns the software could monitor for the flat opamp saturation voltage to end; then delay a bit more to take the first curve sample. After that further samples would provide the overall curve shape to be evaluated for a match or cutoff with software stored curves.
Have a good day,
Chet