Nuts and Bolts of ADC Conversions

[jecottrell]

Hello All,

I'm trying to revise some code and would like to know the ins and outs of ADC reading. I previously used the INT_TIMER1 and INT_AD to precisely time my readings. In my revision I'd like to make sure I understand what I'm doing and do things right.

The biggest question I have is about what goes on during ADC reads. Can I start and ADC read on the timer interrupt and read it on the following interrupt?

[rwyoung]

If the time between reads (your timer 1 tic I would assume) is greater than the minimum time it takes the PIC to complete an ADC conversion (consult datasheet) then the value you read back from the registers is a latched copy of the last completed conversion.

If the time between reads (timer tic is less than minimum time required) then you can either check the adc status bit or accept the value in the adc data registers, knowing that it may or may not be 100% correct.

You can use a static variable inside your ISR to help you keep track of the "freshness" of the ADC data. But what I would definately NOT do is sit-and-spin inside the ISR waiting on the ADC done bit.
_________________
Rob Young
The Screw-Up Fairy may just visit you but he has crashed on my couch for the last month!

[jecottrell]

OK. That answers my question.

Re-hash... to ensure correct understanding:

Once an ADC read is requested, the conversion is started. After it is complete the value that was read is held until another read is requested.

Can you assume that readings will take the same amount of time, each time? So, can you start a read on a timer and finish the reading when the int_ad fires?

Or.. what is the 'recommended' way of timed ADC reads?

I've seen bits and pieces here in the forum, but is there a complete solution or example?

Thanks for all the help,

John

[Ttelmah]

There are three 'parts' to an 'ADC reading'. The first is triggering the adc, The second, waiting for it to do the conversion, the third is reading the result. The CCS function 'read_adc', called without a valuie passed to it, _performs all three automatically_. However provided your compiler is reasonably modern, the parts can be requested separately. So:

read_adc(ADC_READ_ONLY);

returns the value of the _last_ conversion requested.

read_adc(ADC_START_ONLY);

triggers a conversion, without waiting for it to complete, and without returning a value.

So, if in your timer interrupt, you call:

[jecottrell]

RJ,

BTW, 18F6627 & 3.236 @ 40MHz

Thanks. That is a big help in understanding. I had read one of your previous posts with a a similar code snippet and that's what got me interested in cleaning up my sloppy approach.

Your suggested solution raises another question for me.

[Ttelmah]

You don't have to enable an interrupt from the CCP. One of the options on the CCP, is to have it trigger a 'special event'. This can be setup to be the
ADC conversion. You then just get the ADC interrupt when the conversion is complete. If the timing of the CCP can be set to a value to suit you, you can read the ADC value, then handle the 'time' events, all in the one handler.

Best Wishes

[jecottrell]

OK. That sounds exactly like what I'd like to do.

I just made a quick read over of the Special Event feature of the CCP module. I'll try reading some more, but I can't see a couple of things and they aren't necessarily obvious to me because my unfamiliarity with the PIC. Can I use the special event feature with it not being connected? In other words can I use that feature of the module without any manifestations on pins and other things?

My general approach would be:

Setup CCP to fire on my desired timing
Setup special event feature to start the ADC
Write an AD ISR to read the completed AD reading

BTW, thanks for all the help... this is not an attempt to get you to spoon feed me. I just need a few questions answered here and there and I think I'll be able to figure it out.

John

[Ttelmah]

Yes. Mode 11, gives a trigger, without driving the output pin. Even modes that do drive the pin, can be used, by programming the TRIS for the pin to '1', which makes it useable as an input.

Best Wishes

[jecottrell]

As I get deeper into this it gets more difficult.

It looks like I'll need to manually enable the special event mode. I can do that. I'll use:

[Ttelmah]

You don't have to set things up yourself. The CCS functions support this. It is 'CCP_COMPARE_RESET_TIMER'. 'OR' this with 'CCP_USE_TIMER3', if you want to use timer3, but otherwise it'll use timer1. Your timing is dependant on the clock selected for this timer, and the value used in the compare (remember it counts 0...compare value, then restarts at 0. So for a compare of '199', you get 200 counts). The timer will never interrupt, since it gets reset before reaching the terminal count. You will get a CCP interrupt,when the comparison ocurs (which you just leave disabled, and ignore), and a ADC interrupt when the conversion completes. You just set the CCP register (CCP_1, or CCP_2), to the required count.
The limit, is on the lowest rate possible with the standard clock. If you want a really slow count, then as you mentioned before, drive Timer1, or Timer3, off an external source (but this costs pins).

Best Wishes

[rnielsen]

I, personally, like to do things manually instead of having the built-in functions control the ADC procedure. I'll declare the individual bits, like GO_DONE(which is the ADC status bit), and have one of the timers set that bit when it's time. When the GO_DONE bit is set it will start the ADC on it's way.

Then, I do an if(!GO_DONE) evaluation(in my main() body) and when the GO_DONE bit clears that means the ADC is finished. My code then enters that if() statement and reads the registers. I don't need to worry about timing because the hardware will clear the GO_DONE bit when it's finished the conversion.

Ronald

[jecottrell]

RJ,

I've got the following and the ADC is returning all zeros. Could you take a look and see if you see anything obvious. It looks like everything else is working... The sample array is getting filled, etc.

[jecottrell]

I got it working. It wasn't a setup problem. It was a variable problem that got overlooked in the porting to the new method of conversion.

A big thanks to all for the assistance,

John

[Ttelmah]

Obvious comment. Check the ADC channel selection. Your 'comment' refers to AD0, but you are selecting AD1.

Best Wishes