Is ADC inaccuracy 0.1V normal for STR71X?

[hg-chen]

Posted on September 04, 2008 at 11:59

Is ADC inaccuracy 0.1V normal for STR71X?

[hg-chen]

Posted on July 18, 2008 at 09:16

STR711FR2H6 is used.

the ADC inaccuracy is about 0.1V,is this normal,

any hardware and software information for improving STR71X ADC accuracy?

[avelez]

Posted on July 29, 2008 at 05:01

I don't know what have you tried before, but did you try to calibrate the ADC before taking samples of the voltage? There's a section in the ADC section of the reference manual that explains how to do it.

[bruno239955_stm1_st]

Posted on September 04, 2008 at 07:13

I'm personnaly using a STR710FZ2, and I have quite the same question.

In the Reference Manual (Rev1, July 2007), there is a figure (figure 74) which gives the ADC output value versus the input voltage. There are 2 constant mentionned, ADC_DATA(0v) and ADC_DATA(2.5v), which are the results for 0v and 2.5v inputs respectively.

On this figure we can see that thoses values are not 0x800 and 0x000, as I first expected. They don't even seem to be close to it (but the figure isn't probably to scale).

For now, I assume those values are 0x800 and 0x000, which leads me to a 0.1v measurement when I ground the pin, and a 2.1v measurement when I put 2.5v on it.

On the contrary, when I left the pin open, I read 1.25v almost exactly.

Does all this mean we have to apply external voltages on the ADC pin, read the values, and use them for calibration ?

I don''t have needs for a very high precision on this ADC. Are there maybe some ''average'' values for ADC_Data(0v) and ADC_Data(2.5v) available ?

If there isn't a way to go without calibration, could the input/output buffer structure be used to apply a 0v (throught a weak pull-down maybe) ?

Thanks !

[kleshov]

Posted on September 04, 2008 at 11:59

Well, as the manual says, you have to calibrate the ADC. The manual also explains that as many as two different bandgap reference voltages within the ADC are inaccurate (+-5% variation). Assuming the two bandgap references are independent, you'll have to make at least two measurements to calibrate the ADC.

There is another cause for concern. The forum topic here

http://www.st.com/mcu/forums-cat-4734-17.html

suggests that you have to re-calibrate the ADC after each power-up.

Well, I'm not putting up with this nonsense. I'm using an external ADC.