-->
Page 1 of 7

Data pins are NOT 6.5v tolerant

PostPosted: Fri Jul 10, 2015 11:36 pm
by Stoney
Changed title to better reflect what was going on..

I posted the text below .. then recently found that the micro I was using at the time was operating at around 6.5V due to a client loading the incorrect regulator .. so it is likely entirely incorrect and I apologise for misleading anyone.
I am not aware of any issues with using 5V.


So THIS is all wrong
I had seen quotes from the data sheets saying the data pins are 5v tolerant, even quoted the passage.

I was wrong, do not do this. Maybe it would have been ok with a series resistor, maybe not, I will always be using at least a voltage divider in future.
I now have an ESP12 with no functional Rx pin that was connected to a 5v PIC tx line. It was posting data to a thingspeak channel and stopped doing so after about 10 hours or so.
The device still functions, I can see my standard output data on both serial and serial1, complaining it is not getting packets but it is no longer reprogrammable and is going in the bin, after I have removed the flash chip for experiments.

Re: Data pins are NOT 5v tolerant

PostPosted: Sat Jul 11, 2015 4:05 pm
by tytower
I think what you refer to is this datasheet extract
All digital IO pins are protected from over-voltage with a snap-back circuit connected between the pad and ground. The snap back voltage is typically about 6V, and the holding voltage is 5.8V. This provides protection from over-voltages and ESD. The output devices are also protected from reversed voltages with diodes.


So did those protections not work for you ? I don't understand at this point what a snapback circuit is so I can't comment on your post really except to say that a 5 V input should be safe if the protection cuts in at 6 volts.
I think you mentioned at one time the pins can sink or source 12 mA so could your current have burnt it out?

I found also wrapping it in alloy foil for a day brought back one of my chips I thought was done.

Re: Data pins are NOT 5v tolerant

PostPosted: Sat Jul 11, 2015 8:15 pm
by martinayotte
tytower wrote:I found also wrapping it in alloy foil for a day brought back one of my chips I thought was done.


Is that Grand Mother recipe ? :D

Re: Data pins are NOT 5v tolerant

PostPosted: Sun Jul 12, 2015 4:58 pm
by Stoney
tytower wrote:I think what you refer to is this datasheet extract
All digital IO pins are protected from over-voltage with a snap-back circuit connected between the pad and ground. The snap back voltage is typically about 6V, and the holding voltage is 5.8V. This provides protection from over-voltages and ESD. The output devices are also protected from reversed voltages with diodes.


So did those protections not work for you ? I don't understand at this point what a snapback circuit is so I can't comment on your post really except to say that a 5 V input should be safe if the protection cuts in at 6 volts.
I think you mentioned at one time the pins can sink or source 12 mA so could your current have burnt it out?

I found also wrapping it in alloy foil for a day brought back one of my chips I thought was done.



well heres an article on snapback but a fresh degree would help http://www.fujitsu.com/downloads/MAG/vol39-1/paper14.pdf

all I know is the device worked, then after installation and during use, it stopped working, only on the Rx pin which was the only pin subjected to 5v, 12mA should not be a concern as input current should be almost zero (several uA).
It simply cannot be reprogrammed though, so any further testing is limited, but I can see my original sketch still running and complaining about no serial input, resistance and diode test measurements are still identical with the Tx pin so it looks to me that the ESD protection is still fine but the input buffer stage has died.
I managed to hotair desolder the esp12 from the electrodragon carrier and fit a socket for esp01's .. added a divider for the serial (1k and 470) and stuck it back in the cupboard on an undersink boiler I designed for a client, its for long term testing. This one has already lasted about 12 hours so far beating the previous record by 2 hours..

http://thingspeak.com/channels/45627