ESP32 specifiaction

[frenkR]

is there any info on ADC I/O support? I mean channels, type of chanel, internal, external reference, speed?

[ESPGURU]

is there any info on ADC I/O support? I mean channels, type of chanel, internal, external reference, speed?

think we have 13 BIT, 8 channel, can measure 0..4 V in 0..4096 by setting atten 0,1,2,3

http://www.esp32.com/viewtopic.php?f=5&t=19&p=79&#p79

[kolban]

Isn't a range of 0-4095 12 bits?

[rudi ;-)]

Isn't a range of 0-4095 12 bits?

adc_tune.png (264.23 KiB) Viewed 19150 times

Links / Info:
===========
the ADC Test from 20 DEC 2015
Firmware & Pics
Posting theme


hope this helps

best wishes
rudi

[rudi ;-)]

btw..
"..GURU" information is ok, have the same info from code..
but the info "range" 0, 4096 is wrong and i started practical with firmware and test like you see in video.

adc_info_bug.png (24.25 KiB) Viewed 19149 times

btw, we have up to 16 channel



hope now is clear
..

[ESP_Me-no-dev]

I have found the ADC to be really bad.
Full range at max attenuation (0-4V) actually is 0-3V or a bit below VCC
In the middle of the scale (2048) voltage is actually above 2Volts which is way off of the expected 1.5V (half of full)
Same can be noticed throughout the range and attenuations.
Basically you can not do any real measurement using this ADC.
Observations done with two multimeters.

[thereza]

Actually, this isn't clear to me at all. When talking about the 'N-bits' of an ADC, it's commonly understood that you have a range of values from 0-(N^2-1) for a total of N^2 divisions (assuming non- 2s complement encoding). 13 bits would equal a range of 0 - 8191. If the range is truly 0-4096 then that doesn't make sense (from a hardware/implementation perspective), and at best it would be called '12-bit plus 1' or something.

Reza

[ESP_Me-no-dev]

Why are you guys going over those bits so much
It's really simple. 12 bits = 0 -> 4095 and those are actually 4096 steps, because 0 is a value as well.

[rudi ;-)]

It's really simple. 12 bits = 0 -> 4095 and those are actually 4096 steps, because 0 is a value as well.

( fun! )

we have then 4096 steps?

example: 1v in
atten= 0

1V / 4096 = 0,000244140625 V
we have then max value 4095 = ( ( 1V / 4096 ) * 4095 ) = 0,999755859375V

if we have 4095 steps
1V / 4095 = 2,442002442002442e-4 V
we have then max value 4095 = ( (1V / 4095 ) * 4095 ) = 1V

..

btw
if we have value : 0
then
( ( 1 V / 4096 ) * 0 ) = 0
( ( 1 V / 4095 ) * 0 ) = 0

have a look : same here:

steps_in_adc.png (17.02 KiB) Viewed 19017 times

(/fun)

i allways calc

Value ?
Value : 4095
then 1 step = 1V / 4095

best wishes
rudi


btw
value 1023 same ( blue circle )
we have no 1 / 1024
we have max 1 / 1023

[rudi ;-)]

.. 13 bits would equal a range of 0 - 8191. If the range is truly 0-4096 then that doesn't make sense (from a hardware/implementation perspective), and at best it would be called '12-bit plus 1' or something.

Reza

you are right.
the 13 bit is value range 0..8191

but we have range 0..4095.
so we have 12 bit adc.

'12-bit plus 1' there is no chance to get this in real

nice talk, this fun!

think there is a description mistake in the code / understand mistake.
we have 1 / 1024 but we have not really to calculate the step 0
because value 0 = 0 V
our calc base is max value.
this is 12 bit, 4095
so i think ( hope )
1 / 4095 is the right
or
1 / 1023
and so on.

best wishes
rudi