0
votes

This question is entirely unrelated to my code, but to satisfy the obligatory show your code directive:

file.format()

Before the call above returns, on this one SoC I always get a wdt reset. Sometimes but not always the flash does appear to be formatted when the chip is started again. And sometimes if freezes after the wdt reset message, and has to be powered off (looks like wrong comm parameters after pressing hardware reset, but none of the terminal app options seemed to match.)

(Note: since starting this draft I built another copy of my device, using another new, recently received ESP8266-12E, and it behaves identically. Previously built copies still work normally, with the identical firmware.)

So this must be a bad chip, right? Or maybe bad on-board flash? It is a brand new one I just bought. I've also seen file.write issues, with buffer size always 255 bytes or less, though no read issues at all.

One other quirk, after burning a cloud-built nodemcu image to this ESP8266-12E device, adc.read returned 65535 and adc.readvdd33 returned an apparently valid value. (I corrected that by burning esp_init_data_default.bin to 0x3FC000.) This was the first (out of 15, maybe 20) I have seen that was like that. I did not check to see if an older version of nodemcu was already on it.

This wouldn't be the first chip with which I've had issues on arrival; it's at least the 2nd, likely the 3rd or 4th.

So maybe the larger question, what percentage of the ESP8266's that you buy, are either DOA or suffer infant mortality? (Not counting the ones that you have reason to believe were inadvertently killed.)

2
Haven't had bad chips so far: ESP-201, NodeMCU 1.0, WeMos D1 mini (2 of each). Never used extensive file operations so far, sorry. Side note: ADC mode can now be set in Lua.Marcel Stör
I wouldn't call it extensive, just formatting and self-loading the app from a web server (after uploading 4 files with ESPlorer.) As of this morning I finally have a way to test the chips without soldering them in, that makes getting bad chips in the mail much less painful, unsoldering SMD with 16 connections is a PITA. Hey I got your Docker image installed/building, piece of cake, very cool -- thanks!Mark McGinty

2 Answers

1
votes

The problem can be something other than the ESPs, like a inappropriate power supply. I know from my own experience that the Arduino Uno and most USB-TTL converters cannot safely deliver enough current to ESPs. If you're not already, consider using a dedicated power supply circuit that are connected to a USB power source.

0
votes

It does indeed seem to be a hardware issue, 2 bad out of 6, not good! I think it might be a certain vendor but don't want to name names without being sure... Whatever is wrong with the chip hangs it up long enough to make the watchdog bark.

Much more than the cost of the part, the time consumed figuring out whether it's lua code, firmware, supporting connections, peripherals or the chip itself, is the costly thing (not to mention frustration, and wasted storage on SO.)