ESP32 on 3 x D alkaline batteries?

I am working on updating a client's product design. It currently uses a 3 x D battery pack to power a 8 bit PIC and a micro servo.

I would like to update it to a ESP32 for its BLE capabilities as well as other capabilities.

What options do I have to efficiently power the ESP32 on the current battery design?

Thanks!

botbuilder:
What options do I have to efficiently power the ESP32 on the current battery design?

3 off D Alkalines would be my suggestion, with a good low drop out regulator.

For a different answer, you will really need to provide a lot more information on this 'product', what components it currently uses, what it does etc.

What is your experience with ESP32 ?

Get a boost converter that will take 0.9V to 5V input and output 5V with 3A max rating, I wouldn't push 2A.

That little gadget will deliver 5V until all 3 cells in serial add up to 0.9V, it will drink those cells down from the usual device quits at 3V or perhaps 3.3V total. Die at 1V each or at 0.3V each, which way runs longer?

GoForSmoke:
Get a boost converter that will take 0.9V to 5V input and output 5V with 3A max rating, I wouldn't push 2A.

That little gadget will deliver 5V until all 3 cells in serial add up to 0.9V, it will drink those cells down from the usual device quits at 3V or perhaps 3.3V total. Die at 1V each or at 0.3V each, which way runs longer?

Wouldn't it be better to regulate to 3.3V so that it can be connected directly to the chip rather than the board's converter? Would have to believe that its less efficient to convert to 5v then back down to 3.3v?

botbuilder:
Wouldn't it be better to regulate to 3.3V so that it can be connected directly to the chip rather than the board's converter? Would have to believe that its less efficient to convert to 5v then back down to 3.3v?

It depends entirely on the useage pattern of the project, powered all the time or asleep some of the time or most of the time.

But if you keep the details of the project secret all anyone is able to do is guess.

srnet:
It depends entirely on the useage pattern of the project, powered all the time or asleep some of the time or most of the time.

But if you keep the details of the project secret all anyone is able to do is guess.

Not trying to keep it a secret, just don't know how to answer a relevant question that hasn't been asked.

Since you have asked...

The first phase of the project involves a periodic predefined motion of a mini servo. I plan on the ESP going to sleep for 30-60s between cycles. An optional switch closure will be used to trigger a LED and piezo buzzer.

The second phase of the project involves connecting to the ESP via bluetooth to define the servo motion patterns as well as other control of the unit and possibly notifications based on battery level and the optional switch closure.

The servo I am using is capable of being powered by either 3.3v or 4.5v source voltage. It also can handle the 3.3v signal from the ESP.

botbuilder:
Wouldn't it be better to regulate to 3.3V so that it can be connected directly to the chip rather than the board's converter? Would have to believe that its less efficient to convert to 5v then back down to 3.3v?

It'd be best to convert voltage than to use a linear regulator. The thing is that 3 alkaline cells start out > 4.5V and can be drained to less than 1V using boost to 5V. An adjustable boost/buck converter could take a range of voltages and output 3.3V, they cost a bit more but they do exist. I picked up 5V to 3.3V converters that are the same size and pins as regulators (replacements) but only good for 100mA or so for less than $1.50 ea. I got those to power the 3.3V side of voltage leveling using diodes or 74HC4050 hex buffer.

Converters are usually >95% efficient. Linear regulators drop all excess, a 9V supply feeding a 7805 has to provide 180mA for there to be 100mA of 5V output. The current dropping potential makes the heat, more current with same drop makes more heat, linear regulators waste current as do voltage dividers. Running on batteries, go for even small efficiencies. I'd rather have 2 x 5% losses than 1 x 33% loss, 3.3V = 2/3 x 5V.

botbuilder:
Not trying to keep it a secret, just don't know how to answer a relevant question that hasn't been asked.

Since you have asked...

The first phase of the project involves a periodic predefined motion of a mini servo. I plan on the ESP going to sleep for 30-60s between cycles. An optional switch closure will be used to trigger a LED and piezo buzzer.

The second phase of the project involves connecting to the ESP via bluetooth to define the servo motion patterns as well as other control of the unit and possibly notifications based on battery level and the optional switch closure.

The servo I am using is capable of being powered by either 3.3v or 4.5v source voltage. It also can handle the 3.3v signal from the ESP.

Ah some details at last, would it have been so difficult to put them in the original post ?

If by 'efficiently' you mean longest battery life, then you do need to work through the numbers, do not assume an obvious solution is the best.

For a project that runs continuously, with a moderate power drain, then a switched converter might indeed lead to longer battery life, you can then drain the batteries below 1V.

However if the project will be put periodically into sleep mode, and an ESP32 can go down to 8uA, then a switched converter may not be a good choice. The switched converter will have a quiescent current, so even with the ESP32 at 8uA, the switched converter might draw 2-5mA. This will obviously drain the battery. In these circumstances a good low drop out regulator could give a longer battery life, even if the batteries are then only drained down to 1.1V each.

Now if you could find a switched up\down converter with a quiescent current of a few uA, then it would be a good, almost universal solution, but otherwise its a case of doing the numbers, taking measuremtns of run current, run periods, sleep current and sleep periods, and only you can do that.