Measuring Vdd with PIC12F1822

[40inD]

I can't understand, how to configure PIC12F1822 for measuring own Vdd with FVR? It is possible only with pin RA1? Where this pin must be connected?

[Ttelmah]

You do it 'backwards'.
You don't use a pin at all.

You set the ADC to use Vss and Vdd as it's references, and then read the FVR, with the ADC.

So, setup the ADC with VSS_VDD, Then set the adc_channel to 'FVR_CHANNEL'. Then do an ADC reading.

Now if you program the FVR to 2.048v, then if you get a reading of (say) 635, then you know Vdd is 1024/635 * 2.048 = 3.3v.

Generally, if you are using this to check battery status, then don't bother with actually calculating a value (uses a lot of space), instead set just 'limit numbers' for (say) 20%, 50% etc.. battery levels. So if perhaps you were taking 3v as 10%, you can calculate in advance that the ADC reading for this would be 699. With some knowledge of your battery, you can have just a table of 10%, 20% etc., values, and have almost no maths involved.

[40inD]

Thanks.
So, RA1 is free for use?

[Ttelmah]

Yes.

RA1, is not involved at all. What it is used for on the ADC, is if you want to use an external voltage reference to feed the ADC, this becomes the input pin for this reference.

Look at Fig 16-1 in the data sheet. You will see that the ADC can have as it's Vref, Vdd, The FVR, or AN1. Using Vdd, this pin is not used.

The same diagram also shows how you have the four 'normal' ADC channel inputs, or the options of reading from the FVR, DAC, or temp indicator.

The other way you could read the voltage would be to use the FVR to feed the ADC, and read the voltage using any ADC pin _off a resistive divider_. However the problem with this is that the divider would then draw power.
The whole point of the 'reverse' approach, is it removes the need for such a divider, and removes the need for a pin.

[PCM programmer]

Here is an example of using the FVR to determine the Vdd voltage.
This was used with a 16LF1847 running from a lithium battery at 3.6v initially.

[asmboy]

About a year ago i finished up a design for a two channel precision biologic incubator control with 6 LM34DH sensors selected by an 8 channel DG408 analog mux, with low offset DC amplifiers and a PIC 18f2423 / 12bit a/d. I ended up using a stiff - low dropout regulator for the pic at 5.0v +/- .2v - and use it as the A/D reference too. Calibration was done by reading an LM4040CIM3-3.0 on an uncommitted mux input- and then correcting the LM34 data - based on what the pic SAY's the true 3.0V reads. works perfectly. Method does a constant check of tracking adjustment - based on the LM4040 -and sure beats using cal pots ;-))

[temtronic]

One thing to remember is that battery voltage is NOT a true, good indicator of battery capacity or 'life'. You should always do a 'load test', validate results, consult battery datasheets, etc. based upon true loading of the battery. It's a mix of science and 'balck magic' to 'dial in the numbers'.


Jay

[asmallri]

[asmboy]

i started the project just as you say. 3v ref textbook input. amplifying the sensors dc output. But noise performance needed improvement, A LOT-- for an incubator that had to meet a long term stable spec of 37degC +/- .1deg
Not a school kid project.

i ditched the client supplied lm35 ( centigrade out deg =10mv ) sensors for lm34 which has 10 mv per deg F - and set the system display to present in centigrade. More millivolts per degree C ........

i needed the max resolution on the reading with my 12bits - but not allocated as you think - in the delivery version: the same 3v ref was used as an offset to a gain amp with neg value clipper such that only 200mv of the temp sensor delta - based at .85v - were offset at an amplified value across a span of 0 to 5 v on the pic input.

i made the controller pure click-bang for any temp below 85F or above 115F
( solid ON or SOLID OFF) the window that was left was perfect for my PID tuning in the servo mode.

the low offset low noise op-amp circuit converts a span of .85v -1.15 volts into a 0-5v signal where count0=85deg F and count4095=115deg F

using more analog gain and switching to the extant 5V PSU and the method i described led to the lowest system noise and best accuracy when compared to best practices absolute calibration validation. The reference is read along with each data point for both display and servo input -- and continuous correction is done - essentially on every reading set. in the end - one unexpected result was that the LM117 powering the pic was damned stable under load!!
Based on experiment - this combination provided better noise performance than trying to resolve 200 mv on top of 850 mv over a 3 volt span.
thus the most stable ( low jitter in lock ) PID servo -

noise can be as important as RMS accuracy - this was project that needed both.

[40inD]