Rechargeable 21700 Power Supply Project

I’ve heard some of you guys are good with batteries.
I’ve personally gotten interested in the lithium ion tech over the last year with the idea to eventually make a small off grid solar system with battery storage on my detached garage.
Until then… I have a friend who wants a small portable power supply that’s rechargeable with the ability to supply a constant discharge of about an amp for about 3 hours at a time. The output voltage will vary from 1-12 volts. He also wants the whole thing to be compact and wrist or belt mountable.
I got a good deal on some 21700 model 3 cells (4800mAh) from big battery and figured I could use one cell with a boost converter to output my max voltage, then I could use something like a Darlington transistor and microcontroller to set the output voltage (and use the micro to monitor output voltage for steady power).
With this setup I could add one of those mini seven segment voltage monitor things and a rotary encoder with tact switch to turn the output on or off and change the voltage. I would just use the boost converter with a pot to adjust voltage but I need an output below the input voltage so that wont work on its own.
I would also use a usb jack with a 4054 charging IC to handle charging.
With one cell, a small pcb for logic, a rotary encoder and small display it would make for a pretty small package.
Some questions I had for the battery experts-
Using a charging ic takes care of possible over charging issues but doesn’t do anything to protect from over discharge voltage wise. So I was going to add a bms.
In some raw testing I found that the standard DW01 bms used for single series cells cuts off output at 900MA which is around the ideal constant output I need so this is a pain.
Even ebay bms modules that claim to handle 3A discharge for single cells are still just the DW01 and cutout just under an amp.
Does anyone have suggestions for a different bms ic or any other suggestions to make a more efficient setup?
I’m wondering what kind of output I’ll get with the darlington but should have some lying around somewhere to test when I get some free time.
Should I just use the micro (bare bones arduino) to handle everything? bms, boost, and output?
Let me know your thoughts.

I know a thing or two. I hear they’re chemical devices for storing electrical energy and making YouTube videos! :smiley:

As long as you don’t expect it to make any financial sense whatsoever, go for it. Though the right answer, in terms of cost and capability, is to trench power over in almost all cases. I’ve spent more than enough on my office system to trench power… but it’s way more fun to have a standalone system. So, happy to offer advice there when you get to that point.

With regards to your weird 1-12V device, are you looking to piece it together out of existing parts, or to do some board design? In any case:

The more voltage translation stages you have, the more your losses - and a lot of the power conversion circuitry has some fixed losses on top of the conversion losses, so doing low power stuff like you’re talking about is often quite inefficient.

If you’re doing your own work, I’d just implement a buck/boost topology (you can get controller ICs for that fairly cheap) that will go from “whatever you’re at” to “where you want to be.” Toss a uC on there to run your variable output and you’re good.

If you’re looking to piece it together out of parts, you’re on the right track, but I wouldn’t go to 12V - I’d go to 5V. There are tons of dirt cheap 1S lithium USB circuits that will handle charging and discharging, including a low voltage cutoff, and will regulate the battery to 5V out. Toss a little boost/buck converter on the output there, and you’ve solved your problem - almost. Some of the USB battery pack gizmos will cut out under light load, and some will hold it just fine. You’ll need to find one that manages a light load properly, and I’ve not particular advice on that right now.

Alternately, something like this would handle the charging and 5V output:

“I have a 1S lithium battery that I want to charge and emit a good amount of 5V from” is a far more solved problem than going to 12V and bucking down, because almost every USB power bank on the planet handles it this way.

Hang a little buck/boost converter off that output, and you’ve got what you want for a lot less fiddling about.

I hate ‘mAh’ as a measure of anything. It’s meaningless without knowing the voltage. But some independent google says 3.65v nominal, so ~ 15 watt/hours per.

But I’ll go with your max case scenario.

1amp @ 12v * 3 hours = 38 watt hours.

I’m kinda pessimist for battery capacities, but to me you can’t do it with one cell. Maybe 2-3.

So either a 3v BMS/usb charger with them in parallel, or 12v with them in series.

I get about 17-18Wh/cell, but… yeah, same difference. If the whole regulated discharge chain ends up past 80% efficiency, I’d be stunned, so it is a multi-cell project.

As long as you don’t expect it to make any financial sense whatsoever, go for it.

Yeah this garage off grid would be a very small system mostly for fun.
All of my big tools like lawnmower, snow thrower and several other landscaping tools are electric and I would use the solar to charge them/operate the garage doors and various landscape lights. Maybe a mini fridge.
The battery backup (small, maybe 2kwh) would also be nice to have incase I lose power which in my area happens a few times a year for a couple hours at a time on average and running an extension cord from the garage to power the fridge and internet would be convenient when that happens.
We’ll see though, my house is gonna need a new roof in the next 5 years so maybe I’ll do tesla solar roof. Maybe just new roof and regular solar. What are your thoughts on solar roof?

I get about 17-18Wh/cell, but… yeah, same difference. If the whole regulated discharge chain ends up past 80% efficiency, I’d be stunned, so it is a multi-cell project.

Yeah I agree with you guys that I’m not gonna get much out of one cell at that 1A continuous.
I should clarify that the load for this supply will vary quite a bit and the 1A would in the end be more of a spike for a few second. I was just calculating for worst case scenario.
Real world I’m guessing he would probably be using 200-400MA at around 6V continuous. ~3W.
I think what I’ll do is use the setup you suggested - 5V charging with boost to 5V, and buckboost output from there - with a pcb for one cell and a pcb for three cells in parallel and have my friend try out both to see some actual use data.

From what I’ve seen, mostly unmaintainable crap for an obscene price compared to regular roofing/solar. It’s unclear how (or even if) you can replace individual tiles that have gone bad once the roof is installed, and I’m not clear on how they meet certain NEC regulations in 2017 and beyond. I’ve heard some assertions that they’ve written themselves an exemption that might or might not hold up to local scrutiny (I doubt my local pain in the rear would approve them, for example).

If you really value how the roof looks over just about anything else, then, I suppose, they’re fine. I know someone with them, and it’s a fresh enough install that they haven’t had issues that require repair yet, so… the service reputation of Tesla hasn’t been an issue yet (their service reputation is “Uh, yeah, we’ll get to that, sometime… please stop calling…”). But outside that, there’s just no reason to go that route.

Put a fresh roof on, put panels up on rails, with either per-module rapid shutdown gizmos and a string inverter on the side of the house, or microinverters if you must (I really, really dislike Enphase, related to the nice little handout/near-monopoly they bought themselves in NEC 2017, so I will try to avoid using them), and call it good. You shade the roof, which helps longevity and reduces cooling demand in the deal.

Or go ground mount if you’ve got the space, skip the rapid shutdown stuff (I think NEC 2020 may change that, because… well, Enphase probably has their hand in that too), and enjoy a maintainable setup.

No reason to have a separate PCB, just gang the cells in parallel. It’ll take 3x as long to charge, but a parallel group of cells is functionally one cell.

Yeah this is most likely the route I would go. I’m not a fan of micro inverters but also didn’t know enphase did that with NEC. How exactly did they do it?
Good to know you know someone with the solar roof to see how it ages. I have my own concerns about cold weather snow/ice build up causing issues. I feel like it’s another “california weather designed” product. Maybe in the future they’ll refine the design a bit. We’ll see.

Yeah I see what you’re saying. I was thinking a little too far ahead by designing a smaller prototype using one cell and a larger prototype using three parallel cells. I could just do the three cell design and give it to him with one cell populated for testing, to see how long it lasts.

Good question. I don’t know the details, but the NEC 2017 requirements are oddly specific to what Enphase offers - per-panel electronics with sub-80V shutdown voltages between any two points in the array. Tigo has some modules that do it too, but I don’t get the impression they’re in terribly good shape as a company.

Multiple solar installers I’ve talked to have mentioned Enphase in the context of NEC 2017, so I assume there’s something there.

Yeah, or just build the board separately from the pack, and let him wire the pack as needed. Being able to put the pack somewhere slightly different from the PCB would probably improve either balance or flexibility of mounting.