ESP8266 Community Forum

- Thu Jun 16, 2016 7:47 am #49267 Hi all, I've been racking my brains for a couple of days now to try and get my head round this, please help to put me out of my misery.
I want to run an ESP project from three 1.5V batteries and I’m not happy with the full 4.5V on the ESP when the new are new. But as soon as we move down the discharge curve the voltage will stabilize around 3.3V. I’ve got it in my head that I should be able to put an N channel FET in the supply but with drain and source in reverse to the normal convention (source to battery and drain to ESP) so that the internal diode is conducting. This with no connection to the gate will drop me 0.7V so when the batteries are new the voltage is only 3.8V (still a bit high but better).
Now the fun bit, as the voltage drops I want to switch on the FET to short out the diode giving me the full battery voltage to the ESP. Normally to switch on an N channel FET the gate voltage is increased with respect to the source. Is that still the case if the FET is connected in reverse (in which case I need to use a P channel) or should the gate voltage be reduced below the source?
If I can get this to work it will make a nice little self-switching volt dropper for battery applications that increaces efficientcy (it's noted that the ESP draws more current at higher voltages) with no dropout so the full battery reserves can be used.
Thanks in advance.

- Thu Jun 16, 2016 8:27 am #49269 Hi!
Basically yes, you can use a fet s body diode quite easy. Please see the specs oft the fet for the absolute value oft the voltage drop for the specific fet you wanna use. Technically you can still fire the fet to bypass the diode, that is the way it is done in syncronous rectifier. But Herr comes the funny part... To completely switch a fet you usually beed a voltage from gate to source of 4.5v, skme fets even need 10v. And now you also need some clever circuit to detect and controll the voltage level...
And you would also burn the power over this body diode without any use oft it... If you want to burn energy and stabilize the voltage use a zener diode and a resistor..
Maybe it would work with one oft those fest that is per default with an open Chanel and you can jncrease its resistance by applying a voltage in the gate, but with those types i am not that familiär...
Hope that helped a little
Tom

- Thu Jun 16, 2016 8:44 am #49272 Pulling the gate low will switch the p-FET on, the potential between gate & source is the only significant factor, in which direction you mount it is irrelevant.

The configuration you are suggesting is actually often used for reverse polarity protection as a more efficient replacement for a diode:
https://circuitsalad.com/2012/09/04/usi ... rotection/
In your case you could use a resistor and a zener diode (or a TL431) to control the gate.

For production use I would simply use a modern, low quiescence LDO however. They to pretty much the same (except the use the resistance of the mosfet, not its body diode), only use a couple of uA and everything is contained in a single part.

On the other hand, this circuit is a pretty good learning example how mosfets & zener diodes work. So if you want to tinker, you should definitely have a go at it ;)

- Fri Jun 17, 2016 5:13 am #49322 Hi, thanks Tom & lethe. It does sound abit like I'm thrying to re-invent the wheel. If only I'd thought of this 10 years ago. Anyway as I'm running a matrix of 35 WS2812b LEDs (text is never more than 50% on at a time about 110mA + ESP) and just tring to get as much out of the batteries as possable. I think i'll check the battery voltage with the ADC, as the voltage drops I'll use a spare GPIO to switch the logic level FET on. The project is a set of 10 wearables sync'ed up and once powered up they run untill the 3 AAA alkaline batteries are flat (Lipo's not an option), changing the batteries or powering down will reset the FET anyway. A bit rough but it should work and keep the cost down.

A nice write on how LDOs work here:
http://focus.ti.com/download/trng/docs/ ... ropout.pdf