Help to design a PCB for my PI design!

[Pennyweight]

your new PI design

Hi 1843,
one of my technicians tells me that your design seems rather good but he likes Carl's additions too. Are you going to implement all of his suggestions?
If so, an update of your schematic should be posted. I may arrange a PCB layout for you in a few weeks if no one else does. We have 2 PI designs currently being etched by a commercial PCB company here in Australia. I could pass your updated schematic to them also. Look forward to any updates. Cheers, Dwt

[1843]

A 2-stage preamp is a good idea... I gonna use a TL072.

Best regards,
1843

[Carl-NC]

I suggest using a 2-stage preamp to get faster settling. Also a diode on the FB resistor can help. Finally, use a more generic offset adjustment (like Hammerhead) so you can try out other opamps.

Might want to sync the 7660 off the micro.

- Carl

[1843]

Qiaozhi,

I read your PM... I think that the circuit is simple enough that a single-side PCB is fine.
About the size of PCB, it doesn't matter... The smaller, the better!
And finally about the output files (Layout and component-side), a gif, TIF or pdf is OK as I'm not going to reproduce...
In my opinion SprintLayout-5 is a very good software... it can make Gerber files.

Best regards,
1843

[1843]

How are you planning to make the PCB ... by hand (using nasty chemicals), or sending the layout (or Gerber files) to a PCB manufacturer?

If someone design it and put it here, I make one by hand to test and final adjust the firmware.

Best regards,
1843

[Qiaozhi]

Hi Aziz,

Yes, The performance drops because of the low ADC resolution.

I'm putting two versions of firmware on the forum; One optimized for coin (and ring) hunting and another for nugget.(They both include ground balance.)

But I still looking for someone design a nice PCB.


Best regards,
1843

How are you planning to make the PCB ... by hand (using nasty chemicals), or sending the layout (or Gerber files) to a PCB manufacturer?

Also, I'm guessing that you are not located in the U.S.A., which would mean that a simple layout tool, such as Pad2Pad, could not be used (no Gerber-out).

[1843]

Hi 1843,

it is quite surprizing, that you can achieve such a sensitivity with the internal 10-bit ADC resolution of the ATmega8.
Reducing the amplification and sampling earlier would give some more signal-to-noise ratio. Did you try this?

Aziz

Hi Aziz,

Yes, The performance drops because of the low ADC resolution.

I'm putting two versions of firmware on the forum; One optimized for coin (and ring) hunting and another for nugget.(They both include ground balance.)

But I still looking for someone design a nice PCB.


Best regards,
1843

[Aziz]

Hi 1843,

it is quite surprizing, that you can achieve such a sensitivity with the internal 10-bit ADC resolution of the ATmega8.
Reducing the amplification and sampling earlier would give some more signal-to-noise ratio. Did you try this?

Aziz

[1843]

Hi 1843,

In the photo of your prototype PCB, there are 4 control knobs on the front panel, but there are 3 variable resistors shown in the schematic. I can only see the volume and contrast controls, and the offset-balance (which is probably the preset). What are the others, and where are they on the schematic?

Also, there are 5 switches (not counting the on/off switch) but I cannot see these on the PCB. Or are they jumpers?

Hi Qiaozhi,

You're right. The panel and the prototype PCB are from a German PI (Minediver) I had made before.(and due to bad performance it was hammered!!!)

In my PI, settings are done by the 5 switches (push button) that they wrere not connected when I taked the photo.


Best regards,
1843

[Qiaozhi]

Hi 1843,

In the photo of your prototype PCB, there are 4 control knobs on the front panel, but there are 3 variable resistors shown in the schematic. I can only see the volume and contrast controls, and the offset-balance (which is probably the preset). What are the others, and where are they on the schematic?

Also, there are 5 switches (not counting the on/off switch) but I cannot see these on the PCB. Or are they jumpers?

[1843]

If you like to have 12% more depth, then try the following:
doubling the coil voltage from 400 to 800V. Take the STx11NM80 (x=P, W, F, B) MOSFET (800V, 0.4 ohm, 11A). The diode 1N4937 could probably work further. If not, take the STTH812D (1200V) or 1N4007 if you can't get the ultra-fast diode.

Aziz

Thanks for your ideas;
One of my purposes to design this PI was to use ordinary and easy to get components so that everyone would be able to build it at everywhere. So further developements are possible...
Now I really need a good PCB to continue developing the firmware.

Hi 1843.
I've been designing cct boards for 13 years and have experience in analog, digital, high-speed and RF layout (2.4 GHz is the highest freq layout I've done).
Perhaps I might look at your schem and study it a bit.
I'm getting married this year and don't have as much time as I did have in the past
but I'll see if I can estimate a time and get back to you.

Yes, I still layout circuit boards for a living...doing one now at work @ 2GHz.

Beav

That's very good if you design the PCB. I hope you get enough time and design it in a few weeks.


Best regards,
1843

[Beav]

Hi 1843.
I've been designing cct boards for 13 years and have experience in analog, digital, high-speed and RF layout (2.4 GHz is the highest freq layout I've done).
Perhaps I might look at your schem and study it a bit.
I'm getting married this year and don't have as much time as I did have in the past
but I'll see if I can estimate a time and get back to you.

Yes, I still layout circuit boards for a living...doing one now at work @ 2GHz.

Beav

[Aziz]

Hi 1843,

another improvement is to increase the 1k resistor going to the anti-parallel clamping 1N4148 diodes. Increase this to 5.6k (or 6.8k) and adapt your damping resistor value to this change. The series clamping resistor has less effects during the high voltage clamping period (until ~0.7 V) and you can sample up to 2 µs earlier with higher target signal levels.

Aziz

[Aziz]

Hi 1843,

Hi Aziz,
See the 1N4937 in series with the MOSFET. It works great! 600V...15pF...fast enough...

1843.

glad to see, that you have found a better diode than the 1N4007. Even it is rated as a 1A diode (RMS value), it can withstand more current flow in pulsed operation. The forward drop voltage of 1.2 V is low enough.

If you like to have 12% more depth, then try the following:
doubling the coil voltage from 400 to 800V. Take the STx11NM80 (x=P, W, F, B) MOSFET (800V, 0.4 ohm, 11A). The diode 1N4937 could probably work further. If not, take the STTH812D (1200V) or 1N4007 if you can't get the ultra-fast diode.

The damping will be finished earlier due to high voltage. The target stimulation will be increased but signals only available, if you sample earlier.

Aziz

[1843]

1843,

The 1N4937 with 15pf of capacitance helps to isolate the coil circuit from being affected by the MOSFET full capacitance and should allow a higher value of Rd than the same circuit without the diode. The higher Rd value will increase the slope (more vertical) of the current discharge Time Constant (TC) to better stimulate small, low conductive targets.
Yes! I observed the results after using the diode.

What is the actual inductance of your 19 cm, 27 turn coil and the adjusted value of Rd for minimum delay?
The inductance is about 380 uH and the Rd should be around 600 ohms.

What is your minimum delay?
The delay is set to 15 us for the small coil.

Have you tried a larger coil yet?
No, I can't test larger coils indoor.

Thanks

bbsailor

Best regards,
1843