- Thu Mar 26, 2015 3:52 pm
I too have run into this issue with both the ESP-03, and ESP-07/12 My findings are as follows:
My projects are all built on perfboard and soldered. I'm doing that before designing a PCB. I would not attempt to
prototype this on a solderless breadboard. That is just asking for trouble.
Here are my findings so far:
1 . A 5 volt relay coupled with an LDO regulator such as the ncp1117 series all driven from a common supply will experience resets. Looking at the 3.3V rail with my scope, I see spikes of several volts going in both directions on the 3.3V line when the board resets. These spikes have ~50nSec period and ring for a few cycles before they die off.
The LDO regulator has a worst case dropout voltage of 1.2V so the margin is 0.5V with a 5V input (which isn't much)
2. Using a 12 volt relay and a TO-220 LM1117 3.3V LDO regulator (to handle the extra heat dissipated). I did not see this reset problem when switching CFL's (known for thier horrible inrush current), or a space heater (high current) when using the 12volt source and a 12 volt relay. Here, the margin is much better at 7.5V.
3. Sometimes when I connect a clip lead or a ground probe, the ESP8266 will reset. This happens on both the 5V and 12V versions.
4. Adding 100nf from the reset 'rest' pin to 3.3V marginally helped on the 5V board.
5. Touching clip leads to unused I/O's will sometimes cause the ESP8266 to reset on the 5V board, but I have not tried this on the 12V board.
6. Choice of relay seems to be the driver of this issue. The 12Volt relay I'm using has a coil resistance of 780 ohms and is rated for 8A at 250V. It has UL and CSA agency ratings. The dollar 5V 'Qaingi' relay jas a coil resistance of 70 ohms, is rated for 10A at 250V and the 'CCC' safery agency rating.
7. Power supply selection may also play a part. I saw this more often when using the chinese AC-DC modules versus using my lab power supply. I have not yet tried a 12V Chinese power module, but it is on my to-do list.
If I were to guess what is going on: There is a noise immunity issue when working with these modules on prototype breadboards. I'm not sure if it carries over to a PCB design where there could be an opportunity to have better grounding, but I'm not going to risk a PCB design till I fully understand what the issue is.