Running the ESP8266 - best practices

So, I spend the weekend getting to run the ESP8266*. Now I got to like it and want to have some PCBs made for quickly building WIFI enabled devices with the module and Arduino Nanos. Though it works, I have to questions, what the best practice would be:

  • 5V->3.3V, power supply: Currently, I am using an LM1117T33 LDO. Is there any other supply that could be more feasible and is still cheap and low footprint? I could look for switching ICs, but they also have a voltage drop, might need additional parts or introduce noise ... And the ESP8266 consumes so much power that all battery application would be short-lived anyway ... Any good reason to use something else?
  • 5V->3.3V, logic: ATM, I am using a 1k/1.5k voltage divider for the RX input of the ESP8266 (the other way around, the Arduino works fine with the 3.3v logic level). As far as I understand, that would be fine as long as the input of the ESP8266 as very high impedance. IN any case, the voltage might only get too low, but not too high. I cannot find any data for the input impedance of the ESP8266. It works, however. Alternatively, I would have MC74HC4050N logic level shifter available. They would, however, cost a little more, have a larger footprint and could introduce an additional point of failure, especially if they are cheaply bought, like mine :wink: SO, what are the opinions on this question?
  • For completeness:
  • baud rates totally differ from most tutorials (mine has 9600)
  • some responses overfill the serial buffer of the Arduino, so the initial OK gets lost
  • responses that differ from the ones in example codes ("OK" or "Ready" instead of "ready")
  • my Arduino Pro Micro can't provide enough power even on the 5V output (via an LM1117 and even though the usb charger provides enough current - there might be a polyfuse somewhere)

ElCaron:
As far as I understand, that would be fine as long as the input of the ESP8266 as very high impedance.

I think you can take that as a "given". While it clearly has a RF section (which is probably responsible for most of the power usage), the logic section would be entirely CMOS with a very high input impedance. Mostly capacitive in fact, though an input-only port - as distinct from the current-driving and multi-purpose ports of the ATmega - would have a relatively low capacitance in any case.

I know you said your using a nano but could you replace it for a 3.3V Pro Mini?
That would solve some problems.
As for a switching power supply I use these with the ESP8266 modules without problems.
Their size is 1" x 3/4". Very low quiescent current draw in standby.

Pro Minis are relatively expensive compared to the Nano clones. Also, they don't have a USB port (so I have to cut a USB cable if I want to use one of the cheap and ubiquitous USB chargers).
What happens while flashing with a USB2Serial adapter? Will they run on 5V then? That would be bad for the ESP if it was still connected.

The LM2596 you propose has about 67% efficiency according to the datasheet at 5V->3.3V. That would be quite exactly the same as a linear regulator. On the other hand, the "packaging" is far less PCB friendly then a TO-220. When You feed with higher voltages, you are better off, I think, but for me, this doesn't seem to help ,or am I overlooking a problem?

Actually the chip is an MP1584 not the LM2596.
A linear regulator consumes around 3mA in standby, the one I linked is around .0025mA
Using a USB/TTL like this works fine with the 3.3V Pro Mini and ESP8266 it has a 3.3V pin.
At $2.50 for a 3.3V Pro Mini clone it's around the same price of a Nano clone.

Ok, you are right, they are really cheap. Then that is an option. I could put the module on the backside of the pcb ...