ADC Voltage <> integer values(compiler)

[FJMSoft]

Hello.
This may be a very basic question, my english is poor so I will try my best to explain what I need.

I want to work with voltages using the ADC.
Several times I just want to check for a voltage using ADC then I need to do the math by hand, lets say 8bit ADC, want 4.2v:
4.2v ÷ (5v ÷ 256) = ~215
Then put in the code something like: if ADC > 215...
No problem with the math at all, but it consumes time and code gets less straightforward to understood.

My question is, there is a way to directly put directly the voltage value in the code (instead of the hand calculated integer) and without making the uC spend time calculating this?
I mean, make the compiler do this math.
Kinda: if ADC > 4.2v... or maybe
if ADC > (4.2/(5/256))...

Thank you.

[jgschmidt]

We often do this with a MACRO. Rather than burying constants deep in the code we do:

[Ttelmah]

You can always find out 'what' the expansion codes to, by compiling and looking at the result.

If you try FJMSoft's suggestion, it won't work. Reason is that maths defaults to using integer, and 5/256 in integer, gives 0. So the compiler will complain about a divide by zero error.

However:

[FJMSoft]

Wow!
VERY NICE!

Ttelmah code looks more interesting for me, more flexible and voltage value keeps exposed in the code.

So, I can securely make any math like this "#define int_adc(x) x/0.01953" and it will be placed as integer in the code?

Also, 5/0.01953 = 256... being 256 will not make a problem here?

Thank you very much! :D

[Ttelmah]

It'll mean that if you did use 5, it'd switch to doing an int16 comparison.
However since you can never test for the ADC reading being above 5v, it shouldn't apply....

[FJMSoft]

I dont want to jump into int16.

Above 5v for sure not, but for exact 5v is possible.
There is a way to compare ADC to 5v without using int16?

Using 0.0196 instead of 0.01953 will add a small error but maximum comes back to 255.

[Ttelmah]

You can't compare for 5v, with the ADC at all....

It'll 'read' it's maximum, 1.5bits below 5v.

Historically, if you had an 8bit ADC, you'd have to do division by 255, not 256. This makes it slow in a processor. So some years ago, Texas produced an ADC, that was scaled _as if_ it counted to 256, but then stopped at 255. Just about every ADC since has done the same. The PIC ADC behaves as if it counted to 256 (1024), but stops at 255 (1023). It gives 255, nearly 0.03v _below_ 5v, and then doesn't count any higher.

Just do a #define for (say) MAX_ADC, and MIN_ADC, and use these for the maximum and minimum values.

[temtronic]

There is a 'cheat' I did years ago on an 877 project....which should work for most 5V PICs.
Supply the PIC with a VDD of 5.25 volts.Since the VDD max is 5.5 , this is just below the max and the PIC is happy to function fine.
Another possible 'cheat' is to use the Vref option and ,again, feed that pin with 5.25 volts. This is within the spec of VDD+.3 as noted in the datasheets.
If you really,really need to read 5.00 volts this should work for you. Obviously the volts per bit will have to be recalculated, that exercise is left to you.

Jay

[FJMSoft]

I got it.
Thank you very much!

[FJMSoft]

Hello again.

Friends, I was trying to use this technique in my code, test code is like this:

[PCM programmer]

Get more accuracy with floating point math:

[FJMSoft]

[Ttelmah]

Also, step back. Remember that comment that the ADC _behaves_ as if it has 256 'risers', in terms of it's actual transfer function.

Have a look at AN536 from Microchip.

[FJMSoft]

I was giving a little check in the ASM code:

[PCM programmer]

Well, what is one way to convert a float to an integer in C ?
Casting. Do the math in floating point, then cast the result back to an int.