ESP-07S reset circuitry improvement?

Development Other Hardware Development
1

I have built a few projects using the ESP-07S device with external antenna and they have worked OK.
But now I have a borderline case with an electricity meter reader (HAN protocol) where the board must be powered from the meter's external (isolated) 5V supply. And on one of the meters it is acting up such that the 5V is seemingly overloaded sometimes, especially on startup. Then the sketch does not run.
The 5V supply is dropped to 3.5V using 2 diodes.

I have a 1000 uF capacitor across the ESP supply pins in order to have a reserve for surges during WiFi transactions and now I think that the power output is limited especially during startup.
So I want the ESP to remain reset for say 2-5 seconds after power is applied before it starts working.
The standard circuit consisting of a 4K7 resistor to +3V and a capacitor 10nF to ground is not providing the needed delay and if I put a big capacitor there then the rise time becomes too slow...

So what is a good/recommended reset circuit which can provide a say 5s delay after power application until the ESP starts running?
It needs to provide a fast rise time level on the RST pin.

2

The easiest is an ATtiny13a.

And what happens to the voltage drop when these diodes get warm ? You should use a 3.3v regulator in a to-220 package and i am confident your issues will subside.

You are probably better of having that on the 5v input, and just have 1uF on the 3.3v pins.
I recon if you create a proper power-rail, you won't need the ATtiny.

3

Well I have measured the supply current of the ESP-07S and it comes out as 79-81 mA on a digital meter. Occational surges do not count here so I think the power loss in my diodes is 80 * 1.5 = 120 mW in two diodes, 60 mW in each...
That should be fine, really.
However, I have ordered a pair of 3.3 V regulator chips to use as regulators.
TLV1117-33IDCY in SOT223 package could be soldered in when I come back home in 4 days.

I thought that the WiFi surges should be supplied by the caps and the bigger the better. It looks like the boards are stable with the 1000 uF caps on 3.3V rather than the 5V side.
Except for the current installation which acts up, probably this meter's circuitry is more sensitive to power usage and/or surges.

If the meter detects overcurrent when the ESP sends WiFi data or starts up then it might be the reason for some of the issues...
I will re-make the connection cables as well before I re-install into the meter cabinet on the pole outside the lot...

4

I think you should make sure that you have a sufficient heatsink.

When using your unpredictable diode setup yes.

Anyway this is the recommendation, a regulator will regulate to a fixed output voltage, diodes will provide you with a relative voltage drop, which means that the output is an image of the input.
A 1000uF .. beyond a 3.3v regulator may actually damage the regulator (it shouldn't but it might)
During WiFi connection the current may peak up to 300mA, but there should be enough power there (or not i don't know where you are getting it from)
ESP's can be fragile to over voltage in particular.

5

Thanks a lot!"
I will be back on this again come Sunday/Monday when I will also reconnect to the actual meter.

6

RESOLVED
So now I have done this:

  • Replaced the 2-diode voltage drop connection in the power supply with a proper 3V3 regulator TLV1117-33IDCY to generate the ESP8266 supply.
  • Replaced the RJ-12 cable from the meter with a new shorter one (the existing had issues in seating the connector properly in the meter socket)
  • Added delay(1) at the top of loop(). This reduces the average current consumption from 80 mA to 30 mA
  • Added delay(2000) at the top of setup() so that the system voltage would stabilize before starting operations.

With this in place I no longer have any issues with the reporting or resets.

1 Like
7

This topic was automatically closed 180 days after the last reply. New replies are no longer allowed.