Help to design a PCB for my PI design!

[barry]

Like Carl I do PCB layout to learn PCB layout better.

[Carl-NC]

Was not my intention to start an argument over the code. 1843 seemed to be making little progress getting any help on this project so I thought he might get people more excited by opening it up. But it is his project and he is free to handle it how he wants.

Let's move on...

- Carl

[1843]

porkluvr,
Thanks a lot for the schematic.
There are some errors in the schmatic:


Best regards,
1843

[1843]

There is a difference. They design their own boards or pay someone to do it for them. They don't expect someone else to do that part of the work for them at no charge.

If your code is going to be closed and your ego is so big that you believe you are the only one who could improve it, design your own board. Surely you are the only one who could do it right anyway.

HH,
Rip

Hardware of my PI is so simple and it's almost nothing in compare with its source code.
Believe that if I had enough time, I'd design the PCB by myself.

If anyone are needing to change in firmware, I'm doing it friendly. But do not expect to make a lot of changes so that it becomes a deep seeking PI. The frequencies below 650Hz and the pulsewidths more than 100us will be ignored.

Thanks to Ivconic and Barry for their nice layouts.

Best regards,
1843

[okantex]

Hi ,

Thanks to İvconic an Barry for pcb layouts ,they show their good wills again even argues going on her.

1843 ,
why not to do deep seeking PI like delta , can not you do necessary changes , and to get close samples can you add S&H units ,I mean outer ADC unit for micro.

Regards to All
Okantex

[Götz von Berlichingen]

Touche. Posting a URL would have been very illuminating. I will do some searching.
But still, does spinning a PCB for the PI-111 equate to writing its source code? Sorry, but I don't think so.


Not necessarily a logical extension. But what you were after is not a fair trade.


Good job, Ivconic

The Stuart II Project is right here, on Geotech : http://www.geotech1.com/cgi-bin/pages/common/index.pl?page=metdet&file=/projects/stuart2/index.dat
Forum for Stuart I and II : http://www.geotech1.com/forums/forumdisplay.php?f=17
The PIC is only used to generate 2 Rectangle Signals, so i think there is Room for a few Extras.

Back on Topic : of course 1843's Firmware must be something special -
but the low-noise PCB that he wants is really hard to do, even more on a single Layer.
All i know is that smart routing of Ground and Power Supply Lines is one of the Keys. We have a mix of sensitive Analog Stuff and a MCU here...
I did a few Experiments with the PICs 10Bit A/D, and i had some trouble to
get the Noise out to make good use of these 10Bits.

[Aziz]

Hello,

to get rid of of the "noise", you have to "correlate" it. Everything, that is deterministic and predictable, that can be eliminated. This counts also to the phase of the disturbing source. "True uncorrelated" noise can not be. Only the band limitting can help in this case (filtering).

One of the biggest noise sources of the design is the 7660. It is not synchronized to the processing element. The ATmega8's PWM output should synchronize the 7660.

Next is the ATmega8 itself, but as long as the code is deterministic on every cycle, this can be correlated and eliminated therefore. Some caution to code branches must be made. So that the branch and the not taken branch should take same time period.

Other noise sources can be the audio part (on target presence) and 50/60 Hz mains hum. As 1843 low passes the amplified signal before digitizing, he eliminates lots of noises. A typical averaging of the signal which achieves obviously good results even the ADC resolution isn't big enough. The other fact is filtering the power supply of the pre-amps (op-amps).

1843 gave you all a possible way of using the micros internal ADC again. But one sample on every cycle wouldn't be enough and you have to take a moving average of the ADC samples (like boxcar integrators).


You may ask yourself and think of a bit, why I am investing so much effort in synchronisation of all the parts. The answer is simple: "correlating possible noise sources".


Aziz

[Eugene52]

Help to design for my P.I. design !

Hello,

to get rid of of the "noise", you have to "correlate" it. Everything, that is deterministic and predictable, that can be eliminated. This counts also to the phase of the disturbing source. "True uncorrelated" noise can not be. Only the band limitting can help in this case (filtering).

One of the biggest noise sources of the design is the 7660. It is not synchronized to the processing element. The ATmega8's PWM output should synchronize the 7660.

Next is the ATmega8 itself, but as long as the code is deterministic on every cycle, this can be correlated and eliminated therefore. Some caution to code branches must be made. So that the branch and the not taken branch should take same time period.

Other noise sources can be the audio part (on target presence) and 50/60 Hz mains hum. As 1843 low passes the amplified signal before digitizing, he eliminates lots of noises. A typical averaging of the signal which achieves obviously good results even the ADC resolution isn't big enough. The other fact is filtering the power supply of the pre-amps (op-amps).

1843 gave you all a possible way of using the micros internal ADC again. But one sample on every cycle wouldn't be enough and you have to take a moving average of the ADC samples (like boxcar integrators).


You may ask yourself and think of a bit, why I am investing so much effort in synchronisation of all the parts. The answer is simple: "correlating possible noise sources".


Aziz

Hello Aziz , You are 100% correct . The Hammerhead has a 7660 also . It would be a better idea to use something different , or designing a high-Speed switching power supply would be better to get a more clean and stable higher voltage .

I am glad you posted this today !!! The programing is very easy . All P.I. programs are very similar , I just found out that the "Gold-Pic3 P.I." programming is Identical to the "Stuart P.I." Progamming ....Also , I have not seen the internal schematic of the atmel , but it must have a lot of internal filtering and also because of the High-Clock frequency . This will be a Great little P.I. when it gets done..................Regards.........Eugene52

[Götz von Berlichingen]

...
Also , I have not seen the internal schematic of the atmel , but it must have a lot of internal filtering and also because of the High-Clock frequency . This will be a Great little P.I. when it gets done..................Regards.........Eugene52

ATmega8 Data Sheet : http://www.atmel.com/dyn/resources/p...ts/doc2486.pdf
I see no filtering in there, its almost the same as the ADC in the PICs.
The only Measure against Noise would be running the ADC during sleep Mode.

If you want Filters inside, a Cypress PSOC would be worth a Try.
Configurable Analog Blocks could reduce the Part count (and cost) of a Metal Detecor a lot.
As far as i recall, the Hardware can even be reconfigured while the PSOC is running...

[barry]

ATmega8 Data Sheet : http://www.atmel.com/dyn/resources/p...ts/doc2486.pdf
I see no filtering in there, its almost the same as the ADC in the PICs.
The only Measure against Noise would be running the ADC during sleep Mode.

If you want Filters inside, a Cypress PSOC would be worth a Try.
Configurable Analog Blocks could reduce the Part count (and cost) of a Metal Detecor a lot.
As far as i recall, the Hardware can even be reconfigured while the PSOC is running...

The Cypress PSO is a interesting piece of hardware. Someone has already made a metal detector with it.

[1843]

You don't want help designing a board, you want someone to design the board for you. It would probably help if you gave a list of preferred footprints or lead spacings you desire.

This would be simple to do in Cadsoft Eagle, and would probably even fit the freeware version's 3-1/2" X 4-1/4" allowance.

Do you want to design a PCB?

What would someone get in return for turning out a PCB pattern to your satisfaction?

Nothing! As I get nothing in return for the software.


Best regards,
1843

[Qiaozhi]

Try this ->

Hi 1843,

Let me know if this is what you're looking for?
There may be errors, so please check. The usual caveats apply.

----------------------------------------------------------------

[1843]

Hi Qiaozhi,
Your PCB layout really skyrockets my motivation for development the software!

There are some minor errors... please see your inbox.

Thank you very much!
1843

[okantex]

Hi ,

is this design's TX-off fast enought to get early samples for discrimination tests.
what is teorical value?

[Muntari]

Hi Qiaozhi,
Nice work, I'm thinking the pot R26 and volatge into PC1 is going to be a problem as voltage top/bottom is +5/-5....

What PCB application do you use just as a matter of interest?

Cheers

Muntari