adc.readvdd33()

[lmercucci]

adc.readvd33() not working....
in my init.lua have:

if adc.force_init_mode(adc.INIT_VDD33) then
  node.restart()
  return -- don't bother continuing, the restart is scheduled
end
print("System voltage (mV):", adc.readvdd33(0))

but

=adc.readvdd33()
=adc.readvdd33()
65535

[ildar]

If the ESP8266 has been configured to use the ADC for sampling the external pin, this function will always return 65535. This is a hardware and/or SDK limitation. https://nodemcu.readthedocs.io/en/latest/modules/adc/#adcreadvdd33

[HHHartmann]

That's exactly the example in the documentation. So this is a software or documentation bug.

[marcelstoer]

According to our documentation (and the code) readvdd33() doesn't take any parameters. Yet, the code above and in the documentation calls it with 0.

[HHHartmann]

For

NodeMCU 3.0.0.0 
	branch: dev
	commit: 0b90495517c8e2b34dbb3a2a99455ac3c4f05f1c

calling it with no argument yields the same result.

But for

NodeMCU 2.2.0.0 built with Docker provided by frightanic.com
	branch: dev
	commit: 5514475ab0b152eaa994495e6d034123ae811786

it returns 3015 which is what I would expect

[TerryE]

Doing a quick git diff 5514475a 0b904955 -- ../modules/adc.c shows that adc_sample() hasn't changed (except removing a trailing whitespace on the return). It's essentially a wrapper for system_get_vdd33() which only works if byte 107 of esp_init_data_default.bin is set as 0xFF, and this in turn breaks the read() function.
xxd -s 107 -l 1 sdk/esp_iot_sdk_v3.0-e4434aa/bin/esp_init_data_default_v05.bin returns 0, so it shouldn't work by default. If we want it to work by default we need to override byte 107.

[marcelstoer]

If we want it to work by default we need to override byte 107.

Hhm, that's what adc.force_init_mode() i.e. adc_init107() does - or used to do. However, git blame shows it hasn't changed in years, either.

If the test script keeps rebooting the device it means that adc_init107() detects that byte 107 isn't 0xFF but it fails to overwrite it.

[HHHartmann]

adc.force_init_mode returns true for the first call only.

[TerryE]

Oops. Definitely bug. I missed adc.c:L132. The Phy Data sector is no longer at END-4. We need to read its address from the PT

  partition_item_t pt = {0,0,0};
  system_partition_get_item(SYSTEM_PARTITION_PHY_DATA, &pt);

[TerryE]

I have fixed this in my working directory and will add this fix to my #2912 next commit.

[Michal78S]

Please clarify this: adc.readvdd33(0) returns system voltage in milivolts, but adc.read(0) returns what?
Voltage on ADC pin in milivolts or direct convertion resulst in range of 0-1024?

[ildar]

adc.read(0) returns what? Voltage on ADC pin in milivolts or direct convertion resulst in range of 0-1024?

Both are the same. That's because adc measures only up to 1000 millivolts.

[Michal78S]

Are You siure?
I got as below:

mV	ADC
0	14
100	114
500	525
963	1000
986	1024
So 14 is ADC offset, and then when substracted offset I got exacly 977 for 1V. It is the same when divide 1000mV/1024.

[jmd13391]

@Michal78S The ESP8266 ADC has inherent stability issues as a result of an internal hardware design flaw. You may find some of 2864 and esp8266/Arduino:2070 relevant.

[rptnscz]

According to our documentation (and the code) readvdd33() doesn't take any parameters. Yet, the code above and in the documentation calls it with 0.

There are two examples in the doc https://nodemcu.readthedocs.io/en/master/modules/adc/
The example in adc.force_init_mode() chapter shows 'readvdd33(0)' with '0' as a parameter.
But in adc.readvdd33() the syntax is without any parameters.

Anyway both doesn't work in the branch master.

@TerryE: how can i checkout your fix of the bug? I guess there are no change in adc.c module.

[marcelstoer]

@rptnscz maybe this one 32e161b#diff-1925935f8d24083ba3e86383f92b7ea8 It's a bit unfortunate that many fixes went into that branch (rather than dev) but that's just how it is. Please be patient.

[TerryE]

Fixed in #3075