Universal PI Micro

[bugwhiskers]

Hi Sean,

I did a search on "CCD delay" and it just returned the 2 references in this thread.

regards
bugwhiskers

[bugwhiskers]

Hi Tinkerer,

The following link may help you make your mind up.

http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail?name=ATXMEGA128A1-AU-ND

Event system to handle and prioritise interrupts
DMA
12 bit DAC's
12 bit ADC's
31nS per instruction @ 32 MHz
8 * 16 bit counters
2 * 2 analog comparators etc etc etc

All for less than 10 bucks.

regards
bugwhiskers

[Tinkerer]

AVR or PIC?

Hi Bugwhiskers, Tinkerer, all...

You COULD use C for the signal evaluation part, then have the timing loops written in assembler. A countdown/ reload sequence for the critical timing using the AVR's (or PIC's) counter timer registers works well.

Tinkerer have you tried PROTON BASIC? VERY easy to learn and the interrupt routines are pretty fast enough for a 1K pps PI with a 10uS sampling delay .

Bugwhiskers, did you read my post about the CCD delay idea? An thoughts on whether that might be useful of not?

Nice to see you are still around here somewhere . Like I said, When you have something that is usable, I'll buy one, just let me know when it's reay please.

Good luck.

Sean,
thanks for the help.
I looked at the PROTON BASIC site and I don't like their sales tactics.
I got the timing under control now.
It is time to move on to the signal evaluation part now.

All the best

Tinkerer

[Sean_Goddard]

Hi Bugwhiskers, Tinkerer, all...

You COULD use C for the signal evaluation part, then have the timing loops written in assembler. A countdown/ reload sequence for the critical timing using the AVR's (or PIC's) counter timer registers works well.

Tinkerer have you tried PROTON BASIC? VERY easy to learn and the interrupt routines are pretty fast enough for a 1K pps PI with a 10uS sampling delay .

Bugwhiskers, did you read my post about the CCD delay idea? An thoughts on whether that might be useful of not?

Nice to see you are still around here somewhere . Like I said, When you have something that is usable, I'll buy one, just let me know when it's reay please.

Good luck.

[Tinkerer]

ATMEL and assembly

Hi Tinkerer and Sean,

I remember those days Sean, my first was the Sinclair ZX80 and I spent hours pouring over the Z80 opcodes, even memorised a lot of them.

Tinkerer, if you look through the SD2000 full schematics thread you will see a downloadable file that I posted there that performs all the timings. With minor modification the code can be changed to suit any ATMEL micro and is a good starting point requiring only minor changes to the code to suit your application.

One or the benefits of going down the ATMEL path is

http://www.avrfreaks.net/

Once you join the forum you can have any question answered by a host of experts. You'll be spewing out code in no time.

Assembly Language produces the fastest and most compact code. It also allows you access to every function in the micro.
If you were writing programs that worked with floating point math then "C" would be much easier but just toggling pins can be done in assembler with ease.

Just to get you started.

http://www.avr-asm-tutorial.net/avr_en/

regards
bugwhiskers

Thanks for the encouragement, Bugwhiskers, and thanks for posting the timing schedule.

These links really look enticing. The day may come when I will know enough to be able to really compare PIC and ATMEL.

In the meantime I would love to see an educated discussion by more knowledgeable people than myself, of the advantages and disadvantages of each kind.

All the best

Tinkerer

[bugwhiskers]

Hi Tinkerer and Sean,

I remember those days Sean, my first was the Sinclair ZX80 and I spent hours pouring over the Z80 opcodes, even memorised a lot of them.

Tinkerer, if you look through the SD2000 full schematics thread you will see a downloadable file that I posted there that performs all the timings. With minor modification the code can be changed to suit any ATMEL micro and is a good starting point requiring only minor changes to the code to suit your application.

One or the benefits of going down the ATMEL path is

http://www.avrfreaks.net/

Once you join the forum you can have any question answered by a host of experts. You'll be spewing out code in no time.

Assembly Language produces the fastest and most compact code. It also allows you access to every function in the micro.
If you were writing programs that worked with floating point math then "C" would be much easier but just toggling pins can be done in assembler with ease.

Just to get you started.

http://www.avr-asm-tutorial.net/avr_en/

regards
bugwhiskers

[Tinkerer]

writing assemby

OOER....I hope you don't think I was having a "go" at you Tinkerer . I know your skillset outclasses mine in so many areas, no really I MEAN that. I have the greatest respect for you. Having read many of your previous posts.

The unregistered post I saw seemed to be telling BW that he didn't know what he was doing it, and it kind of hacks me off that some are all too ready to point out problems without providing suggestions as to answers. I used to do it until someone pointed out that is was just NOT the "done" thing. Now at least I'm aware when I'm being an *** (mostly).

I have been coding MC (machine code) since I was 13 years old (I used to write SW in HEX for the ZX Spectrum) although I gave it up and now use compilers myself.

Took me ages to work out why my critical timing stuff didn't work when written using compiled C and the like. Turns out I'm not as smart as I thought I was . You learn something everyday eh?

Sean,

writing assembly code is a painfully slow process for me. But I have found nobody to do it for me so I have to bite the bullet.
People tell me I should learn C that is easier. Or C## or what not, well there just are things that one is too old to do.
Today I managed to have my PI with PIC timer running just exactly the way I wanted.
Exactly 10,000 PPS
5 different samples of 5us each at exactly the right time.
Blocking the flyback at exactly the right time within 1\2us in spite of the analog delays caused by the transistors.
First sample taken at 5us with a gain of 2000 on the preamp. 320mA coil current.
So I felt reasonably good, just that nagging thought:"got to start learning C"
So when I saw your post, it really made me feel better. For somebody with the vast experience like you to say that assembly has its merit and that a lot can be done with assembly has made my day. Thank you.

[Sean_Goddard]

OOER....I hope you don't think I was having a "go" at you Tinkerer . I know your skillset outclasses mine in so many areas, no really I MEAN that. I have the greatest respect for you. Having read many of your previous posts.

The unregistered post I saw seemed to be telling BW that he didn't know what he was doing it, and it kind of hacks me off that some are all too ready to point out problems without providing suggestions as to answers. I used to do it until someone pointed out that is was just NOT the "done" thing. Now at least I'm aware when I'm being an *** (mostly).

I have been coding MC (machine code) since I was 13 years old (I used to write SW in HEX for the ZX Spectrum) although I gave it up and now use compilers myself.

Took me ages to work out why my critical timing stuff didn't work when written using compiled C and the like. Turns out I'm not as smart as I thought I was . You learn something everyday eh?

[Tinkerer]

assembler

Yes and you're no programmer either!!!

Use C?????????? WHOOP DE DO....OMG!!!! In what way can you even PRETEND to accurately control the timings of loops down to uS with a C compiler that produces code bloat and thus timings in days rather than uS?? Sorry dude, but it's ASSEMBLER every time for an app. like this. If you have ANY embedded experience you would know this simple fact.

Q: What are C's LOW LEVEL (direct hardware) drivers written in??

A: ASSEMBLER..10 points if you got that one (which you didn't).

And as for implementing interrupt service routines in C, the same applies. they need servicing NOW, not when the compiler decides to put in the code with associated delays to say "anytime this week will do".

Amazing just how all these "know it alls" with nothing to say except criticise and pass snide comments hide behind the "unregistered" tag.

No wonder poor BugWhiskers left this forum, all the trolls asking him stupid questions when they should read the whole thread. "The answer is out there" to quote the X-Files

Thanks for your post. It makes me feel a lot better with my feeble attempts at getting the timing right.

Tinkerer

[Atul Asthana]

Assembly is the way to go

I agree with Sean for use of assembly.
Though a little tough, assembly gives you the control and flexibility. And, a large amount of libraries are available for AVRs, on the web to reduce your burden.

[Sean_Goddard]

Whats assembly going to do ..use c program, its standard and easy, and run the micro at max spped ie 40-80Mhz, c avoids complex prog patterns and helps in seeing interrupt latency and delays.

Yes and you're no programmer either!!!

Use C?????????? WHOOP DE DO....OMG!!!! In what way can you even PRETEND to accurately control the timings of loops down to uS with a C compiler that produces code bloat and thus timings in days rather than uS?? Sorry dude, but it's ASSEMBLER every time for an app. like this. If you have ANY embedded experience you would know this simple fact.

Q: What are C's LOW LEVEL (direct hardware) drivers written in??

A: ASSEMBLER..10 points if you got that one (which you didn't).

And as for implementing interrupt service routines in C, the same applies. they need servicing NOW, not when the compiler decides to put in the code with associated delays to say "anytime this week will do".

Amazing just how all these "know it alls" with nothing to say except criticise and pass snide comments hide behind the "unregistered" tag.

No wonder poor BugWhiskers left this forum, all the trolls asking him stupid questions when they should read the whole thread. "The answer is out there" to quote the X-Files

[Sean_Goddard]

Every one here seems to be a nut case mumbling to themselves, just like a cpu talking to ram without any proper output, with their clone schematics, try to get to the front end circuitry you fellows. Just changing the display with graphical one and cloning the hardware wil not help you fellows, since year 2000 i am seeing this site, and no improvement, many 2nd grade manufacturers in India has a PI detector which can detect a cannon ball 5 feet away in air, you fellows dont even come to this limit, and you cant come to the speck of 20 feet detectors....

SURE I CAN!!

I could pulse 200+ Amps through a coil and detect a cannon ball at 40feet plus, but why would I want to do a lame thing like that? Moreover Why would I want to dig 20ft for a cannon ball?

You're talking bovine faecal matter mate. Have some constructive input or shut up and shove off..troll!

[Sean_Goddard]

Apparently not everyone is driven by greed. I've followed BW's threads with great interest as it appears to be one of the most superior projects here. What disapoints me is that status is not merited by intelligence but by favouritism. It's blatantly obvious that this guy knows more about ee than most that post on this site. I have no doubt he's capable of a design that matches or exceeds any commercial design made.

I will agree there are freeloaders lurking about but I also see lots of "wannabee's" with little to no knowledge. If the commercial project can come up with anything even close to this design I'll be happy for them but at the same time extremely surprised.

IMO

And you are WHO now? Could YOU do better?

I think you serisously underestimate some of the people here. Yes a lot of ideas are thrown around, so why not sign in with a REAL name and join the debates constructively instead of just passing semi snide remarks and hiding behind anonimity? that smacks of a UKDN member (= BIGTIME, know nothing LAMER ) to me.

Yes BW's project is VERY promising, but as you can see a few others have thought about and recognised what he is trying to achieve, and suggested more technical and effective ways to get the same results and as for favouritism, that's just plain BS. A few post regularly anad ALL are welcome, there is no discrimination of favouritism here that I am aware of, Carl ensures this. Please cite examples.

Not ALL of us have the time to do as much as we would like. If I didn't have to "work" to pay the bills, then MAYBE I could be churning out projects like this or, more likely, I'd just be drunk in the corner of a bar somewhere, still we all have our crosses to bear.

[Sean_Goddard]

Hi All, just "re-found this thread".

BW I'll take one of those when it's finished. I suggest a PAYPAL account and website so that we can buy directly from you .

I assume the reason for the 64 caps is that the Atmel A/D is not fast enough to directly sample the decay, so you are "retiming" the waveform and using the blanking period between Tx pulses to do the processing? Or am I just being an idiot (again) and missing something else?

In the mean time, Can I suggest these;

http://www.cirrus.com/en/products/pro/detail/P1024.html

You can interleave left and right channels and sample at a whopping 384KHz with inherent 120dB s/n. Talk to J.L. King about this IC I sent him one to evaluate.

Instead of the 64 stages why not use 2048 in a BL3208A (or similar) Bucket Brigade Delayline??

If you don't need all 2048 stages I have an idea. Say for example that you only need 128 samples, then what you could do is take your 128 at say 11Ms/s then ground the BBD input (to reduce noise) and "up" the clock frequency to, say 5MHz, "race" the data through the remaining unused stages until the wanted data was then near the output pin (near in terms of number of internal caps to traverse). Then simply drop the clock frequency to match your required A/D sample rate and hey presto, you can accurately digitise the decay curve.

You could also use a television CCD device (stores one line of analogue video) as these things are meant to work REALLY fast (around 5MHz)

If you did the timing circuit in an FPGA ,then you could experiment with different sample rates and number of samples taken very easily.

Of course the above may not be what you want to do, but the BBD is just a small DIP8 instead of a bank of 64 caps and takes up a LOT less space in board real estate.

Hope this gives you some ideas.

GREAT project BTW

[amtech2005]

Universal PI Micro

Where is bugwhiskers ?? Did he ever complete his PI MICRO Metal Detector ?? I will go through all the posts here again and try to refresh my memory , if he posted all the complete final schematics and software , or if his P.I. Micro metal detector is only for sale , and not for us to build ?? This post went dead Quick !!!..........Eugene