Universal PI Micro

[bugwhiskers]

Hi BjBj,

Any of bbsailors dual coils should work very well. A coaxial coplanar is another possibility.

The AtoD reference of 2.5 volts, divided by 24 bits (16.8 million) is 0.149 microvolts, just as well the PGA is quiet eh ?


regards
bugwhiskers

[bugwhiskers]

Hmm, i really wonder if a internal CMOS OP amp will give you any
advantages over the bipolar 5534?

Which micro have you finally settled for?
ADUC845?

The resolution of 144db cant easily be realised, the 4051 will swamp
it out, Rds of the switches shift due to temp and voltage variations.

I recall sigma-delta AD quite, aduc 845, 1.37kSPS!!
Carl said in some papers that there whas this 3khz pulse rate but
one should do with far less i suppose?

Hi unregistered,

The RDS of the analogue switches is actually beneficial in conjunction with the caps in filtering out some noise and changes due to temperature will be miniscule.
The 1.37ksps is the maximum rate of the ADUC845. In an earlier post I did mention that the full 24bits may not be achievable.
I want to try just the internal OP-AMP firstly as it will make for less hardware and if need be I can always go back to an external OP-AMP.
The problem with "on the fly projects" is that there will always be changes of direction in persuit of excellence.

regards
bugwhiskers

[BJBJ]

The idea with DD coils are going to be interesting indeed!
Perhaps one could use sailors mini coil arrangement!

But on the other hand the buffer+PGA seams to be quite silent,
noise in uV region.

Hmm, i really wonder if a internal CMOS OP amp will give you any
advantages over the bipolar 5534?

Which micro have you finally settled for?
ADUC845?

The resolution of 144db cant easily be realised, the 4051 will swamp
it out, Rds of the switches shift due to temp and voltage variations.

I recall sigma-delta AD quite, aduc 845, 1.37kSPS!!
Carl said in some papers that there whas this 3khz pulse rate but
one should do with far less i suppose?

[bugwhiskers]

Hi BjBj,
Thanks for your interest in the project.
The AD8055 was initially chosen for it's slew rate and current drive in the previous bus arrangement. With the evaluation board as the micro platform a lot of things will change.
At the moment I am contemplating using no preamp at all, instead using the micro's on board programmable gain OP amp that is user S/W configurable for gains from 1-128. The advantage of this is the minimisation of noise and the fastest possible sample time with the advantage it brings for detecting small fast targets like small nuggets.
Amplification of any signal also involves amplification of noise. If instead of amplification the resolution of the AtoD (144dB) achieves the same objective without the noise and the extension of the earliest sampling time.

The TC424 was used because it allows a 5v micro output to switch the MOSFET at the full supply voltage ensuring full and fast switching.

The number of samples required is under review.
The coil charge time is dependant on the coil used and can be made automatically detectable (see previous post) and user adjustable.

Re the number of bits required, as stated above, the more bits the more "gain".
Another change is the use of a 2 coil arrangement, either double "D" or seperate RX TX so the RX signal is the correct polarity for the AtoD. This arrangement will also allow either "P" or "N" channel MOSFET's to be used.
Regarding the 4053, it isn't required in the new arrangement.

regards
bugwhiskers

[BJBJ]

Splendid project BW!

I have also done some designs around this type of detector, but since
i came across your thread i think i have try on this multi sample cell idea to.
I use a MSP430F1232 (low power).

I have some questions and ideas.

1: What is the purpose of AD8055?
If no additional gain are needed the 5534 can drive the sample cells
directly.1n over a range of 0 to 5V are not a heavy load for it!
You still have the slewrate of the 5534 anyway!

2:Why not remove the 4053 and simply switch the AD pin to three state
or just ignore read it during sample storage?

3:Instead of TC425 as charge pump driver (which are quite difficult to find)
an idea could be to use a HC14 with gates in parallel.

4: How many samples of the knee are actually needed?

5: What maximum coil charge rate are you going to use?

6: After your initial tests do you believe you need more then 10 bits (AD).

Reg
BJBJ

[bugwhiskers]

I dont think my vote will be counting then .. I would be interested in both approaches... a discriminating model .. and the one to work with as am really interested in prospecting model.. would probably work around it and see which one out performs the other ..
Excellent work BW
cheers
gef

Sorry for the late reply to your post gef12.
The evaluation board lends itself perfectly to the original UPIM concept and with the addition a few more components it becomes the discriminating PI so you will get your wish after all.
The 2 boards that have been ordered should be delivered around the end of this month so I can get back to work.

regards
bugwhiskers

[bugwhiskers]

The following link is I believe the book that comes with the ADUC845 evaluation board. There are some useful tools supplied to evaluate the noise performance of the board.

http://www.analog.com/Analog_Root/static/technology/dataConverters/microConverter/8XXGetStartedv1.0.pdf

regards
bugwhiskers

[bugwhiskers]

Hi rWAVE,

Thanks for your interest in the project.
If a 0.1 ohm resistor is inserted in the bottom leg of the "P" channel MOSFET and the top of it is fed to one of the AtoD inputs then a reading could be taken and the pusle width set for that coil based on the time it takes for the voltage to get to 90% of the supply voltage. The micro could be set to do this once on switch on so that virtually any coil could be adapted. There would of course be an over-ride to allow the user to shorten or lengthen the TX pulse to optimise for desired targets and/or conservation of battery power.
The article by bbsailor is an excellent guide to coil making. The coils he sent me work very well indeed and he has been a great help to me with this project supplying a lot of ancillary information.

regards

bugwhiskers

[rWAVE]

Greetings bugwhiskers,

I have just stumbled upon this forum and your VERY impressive project. CONGRATULATIONS! I am fairly new at metal detecting (Garrett Ace 250), spending most of my time at local, Southern California parks. I now want to head to the beach and search in the wet sand and surf, hence PI. Your project seems ideal.

After reading previous posts, it seems like your design will allow for lots of experimenting with different coil designs. I read with great interest (as I have no experience) "Making a Fast Pulse Induction Mono Coil" by bbsailor; I understand he provided some test coils to assist in your development.

Since your project is uC-based, will it be possible as various coils are built and tested to allow each coil to have a unique "ID" and profile, such that when a particular coil is attached and the PI detector is powered, it can sense which coil is connected and then load the appropriate profile, thus limiting user variables (delays, pulse widths, etc) to only those appropriate for that particular coil?

This would allow those of us outside the project to design and test coils unique to our type of ground and to targets sought. We could then provide coil design parameters (type of coil, diameter, wire size, number of wraps, inductance, resistance, etc) and profile information to the group at large, thereby sharing coils designs? This should allow those of us interested in your project to easily build a family of coils, aimed at specific ground types and targets.

Regards.
Richard

[bugwhiskers]

Analogue uC overview

The link has a good description of Precision Analogue Microcontrollers.

http://www.elecdesign.com/Files/29/12345/12345.pdf

24 bits equates to 16.8 million or 144 dB......however, many factors influence just how many of those 24 bits can be realised. Because the AtoD will be reading the voltage stored on a cap which is essentially DC I am hoping the resolution will be very high which equates to lots of gain/sensitivity.

One would assume the the evaluation board is very well laid out with noise minimization being the highest priority so the micro looks good. It will certainly be better than any board I or any other amateur could lay out.

regards

bugwhiskers

[solmer2]

I wannn one too



Hi bugwhiskers GREAT WORK !!!

I'll buy one so add me to the list.

Good luck.

Solmer2

[bugwhiskers]

MAX5165

I did a search and contacted the local distributor. Farnell NIL, Digikey NIL, local agent NIL plus quoting 19 week lead times and $20US each. Hmffffff !

regards

bugwhiskers

[bugwhiskers]

Hi Unregistered,

Thanks for the info. The chip doesn't work exactly the same way as the discrete approach but nevetheless may still be usable and save board space.

In the discrete approach the "IN" pin is an input during sampling and an "OUT" pin when digitising by the AtoD. With the single chip arrangement the buffered and clamped outputs couldn't be used unless de-multiplexed which would require more chips.
Do you know if they also make a matching de-multiplexer ?

regards

bugwhiskers

[felix56]

The following link shows a readily available platform with enough I/O ports and other usefull features.
The chip is an ADUC845 from Analog Devices, among other things it has 24 bit AtoD capabilities.

http://www.analog.com/UploadedFiles/Associated_Docs/671728330ADuC_845_PCB_0.pdf

It comes pre-asssembled with power plug pack cables and programming software for around $75US.

I have 2 on order to test drive but alas they won't be here for 3-4 weeks, Australia is an electronic backwater.

regards
bugwhiskers

Hi, bugwhiskers.
"Potential platform " Is the answer of my help request about hex or asm file for universal PI metal detector.
regards

felix56