No I did link the correct one.
As the guy in the tread stated later it depends uf wifi is on.
Same for me.
Wifi off works fine.
And no, im not trying to land a spaceship but when I attach a common widely used temperature sensor that has 10mv/degree Temperature it’s jumping for more than 3 degree and you can’t use the output reasonably.
Tried disabling and reenabling wifi for reading but that took 5. 2 seconds which is was to much for reacting in real-time at the App.
But at least there are many people by now discussing about this so it will probably be fixed soon and I can continue.
I also wrote codr for interpolation of the readings but to get a stable 0.1 degree reading I need 0.5 seconds which is way better than disconnecting each time but no real solution either.
Yes the results I gave you are with your sketch and AO grounded.
Ok so if your output also isn’t 0 you’ve seen the problem yourself and I didn’t got defective parts which I suspected already 
Now disable WiFi (remove commenting slashes in my code) and it will work. You will get 0.
That’s probably interference of wifi into the Analog pin so the solution would probably be to just pull the register when wifi isn’t transmitting. I hope they will integrate that soon.
An ADC that “guessing” isn’t useable but I got no pins left for an external one.
If you took the time to read my posts you will see that I wrote the results are the same, very accurate, with or without WiFi.
Good luck with your project.
I took the time BUT
Variation 5-13 isn’t good even for any cheap ADC for 1V when grounded ANY ADC needs to be zero since there’s nothing to measure at all. For example compare that with any cheap OpAmp that can apmlify microvolts without problems.
You didn’t tes with wifi off!
You did test with connecting to wifi and without the routine for connecting to Wifi.
That’s not the same!
Even if you don’t ask the ESP he automatically connects to the last wifi it was connected to.
You do need to WiFi.disconnect();
Try that!
I am fortunate enough to have stock of the WeMos that have never been connected to WiFi.
AO to ground gives a fluctuating reading from 1 to 2, so 0.1% to 0.2% accuracy. NOT the zero reading and zero margin of error you appear to be dreaming of.
For MOST systems a margin of error of 1.5% has no impact on the practical use of a device.
So a temperature reading of 21.15% might actually be 21.47%. For $4, based on all the things a WeMos or ESP can do, this is EXCELLENT value.
If you can’t code around the issues perhaps spend a few weeks of your project time looking for an alternative piece of kit!!!
Fluctuations of 1 would be acceptable but not 10 or 50 as on my other esp.
But anyway even that device should at least fluctuate between 0 and 1 not 1 to 2.
And when grounded there can’t be any reading at all no matter what the tolerance for the specific part is.
Price… Compare that to the price an OpAmp which is around 10 cent and can amlify microvolts correctly. That’s absolutely not dependent on price.
That’s electronic elements not Gucci bags
But anyway, we don’t need to argue about such things. I just stumbled from one problem to another.
I initially plannes a pretty simple project which got more and more complex by now which didn’t make it easier…
I initially planned to be done in 2 max. 3 weeks but since I got that much ideas now I’m thinking of selling my thing commercially when done so it makes sense to invest more time.
Will continue writing my code and deal with that problem later, since that’s a problem I would have a solution for even if it doesn’t get fixed soon.
Way better than the disconnect problem I had before where I didn’t had a clue on how to solve it or work around it.
so you have been logging @Costas hours and will be reimbursing him i suppose?
all i see is “tilting at windmills”…
which is my you might need to re-think from square one?
Dave thanks for your very nice comments as always…
Firstly I didn’t plan to make something to sell.
Secondly I don’t know if I can but anyhow at least I want to make the one for me as good as I can.
Thirdly I don’t know if Costas is from Blynk but in case I really could sell something they need to agree with Blynk anyway for custom Server/App software.
So I assume that’s not Bad for Blynk.
And I did rethink and plan my Idea with the ESP since you guys told me it’s stable.
How can I or anybody know beforehand of other problems that will appear?
someones gotta say it 
i’d just be re-thinking the “home-made sensors” if it were me…
this is what i mean by going back to “square one”:
set out:
but yeah, i kinda did this for my project, and it is still going 10 months later… but i can see an end in sight, and it is actually well beyond my expectations…
so keep it up, but you need to stop going around in what seem like circles…
That’s nothing about my sensors since they work fine with Arduino but I couldn’t use the Arduino together with Esp and AT software.
Also just some are ones I did myself the temperature sensor is an integrated circuit for example.
Anyway as said I know a way to get around this analog in problem but it bugs me that its not working as it should and you need to program around that to get correct readings.
When reading A0 the correct reading is always the lowest one. Which also is a clear indication of radio interference when the module is sending since the value is raised always.
so what RF shielding do you have?
Either RF from WiFi is coming in inside the module or some kind of voltage drop because of the module sending and stopping to send must be the cause I assume.
But anyway have put that issue on hold since I can work around.
But now that I have changed to ESP because of the instability of Blynk and EPS with AT firmware I probably found the reason (at least I assume).
I connected Arduino GND/RX/TX to the ESP as in the Example I got in this Forum.
Always the connection got lost because of Buffer overflow that was probably caused by the “Busy p…” message of the ESP when polling to quickly.
When Uploading to the ESP you need to put it in Download Mode by pulling GPIO0 down at poweron.
But the Wemos D1 does that without your need to press the button and pull the Pin down and I always did wonder how they do it.
Now I stumbled over a circiut diagram that shows the connection to the ESP that probably is also used by Wemos (didn’t check).
The difference it uses DTR for what you normally do manually. So probably you need to use DTR as well as TX/RX, then you would see the ESP isn’t ready and wouldn’t send too quickly and therefore wouldn’t get “busy p…” and the buffer wouldn’t be messed up by that.
At least thats my assumtion.
Can’t test that since I don’t have that Arduino-ESP AT Firmware circiut built anymore but that would make sense!
Maybe that helps others…
That is correct. It’s described here also: http://reflowster.com/blog/2015/05/11/esp8266.html
i’ve been using that since I bought my 2nd ESP to flash them. It works very well, if you have access to your DTR and RTS pin on the USB-to-Serial 