2021-12-12

The DIYMore 18650 Battery Shield V3 su... isn't very good.

Earlier this week I was working on a project where I wanted to power some electronics with an 18650 cell. Because I needed 5V and 3.3V and an 18650 doesn't provide either I dove into my grab stock and came out with one of the "DIYMORE" battery shields I had in stock, ordered on the cheap from AliExpress when I stumbled upon them one day. So I put an 18650 in, soldered the electronics to the 3.3V and 5V rails, checked if everything worked and went on my merry way.

It wasn't until a bit later when the idea came up to have the project measure the battery status and, since I was using a Wemos D1 mini with WiFi capabilities, have it report the battery voltage by "phoning home". I added a little voltage divider over the battery's + and - and hooked it up to the analog port of the Wemos D1 mini. A few lines of code later and V1.1 of the project was done.

I hadn't run the project for longer than a few hours total so until that moment I didn't have a clue what the actual power consumption was. But since the project mostly consisted of the Wemos D1 mini in deep sleep for most of the time I assumed it would be able to run at least several days, probably weeks or months. Looking at the reported data I had graphed using Grafana I came to a different, shocking, conclusion: this was barely going to last the weekend!

Battery voltage dropped from ~4V to ~3.5V over the course of 16 hours

This is when you're glad you decide to plot data instead of looking at the raw numbers - it would've taken me a while longer to realize what was going on had I been staring at:

Raw data of voltage readings over the course of an hour

What was going on? Why was the battery depleting SO fast? Were my calculations wrong? Was the Wemos D1 mini not deep sleeping correctly? Was there a short?

The investigation

I wouldn't call it an 'investigation' per-sé but, hey, I needed a caption. Anyway, I took out my multimeter and started poking around in my project. Everything was close to even perfectly at the expected values. So that left me with only one conclusion: It must be the 18650 shield. And that's when it hit me: it had 5(!) LEDs on the PCB to indicate the battery level so I immediately jumped to the conclusion that those must be the culprits! I took my soldering iron and whipped the LEDs of the PCB. I turned the power to my project back on and then it was a waiting game for the data to come in every few minutes.

After about an hour I came to the conclusion that, nope, removing the LEDs didn't do squat! The slope of the graph was still very much like a ski slope, and a steep one at that! So the next thing was to take out the 18650 and use my bench power supply to 'emulate' the 18650 for a while. I set it to 4.1V, hooked up the leads to the + and - terminals and switched it on. To my surprise, and dismay, I saw a draw of about 100mA! What the...

100mA draw - with NOTHING being powered or attached to the board

Since my project was soldered to the shield and, by now, I started to suspect the shield itself, I went and got another of the shields from my stock and hooked it up to the bench power supply. And there it was, staring me straight in the face. Just the shield, nothing else, hooked up to the bench power supply. Around 100mA current draw. Since I had a few more of these shields I tested all of them. And, yep, all of them had the same problem. I even had 2 "double" shields that take two 18650's - same problem again!

"Dual" shield. Same problem

So after googling for a few minutes I stumbled upon this Reddit post and from there it was downhill all the way. I found a ton of complaints all over the web about these shields.

Desoldering the battery status indicator LEDs did next to nothing

Since my project was due in a few days I had no real viable option other than to find a similar solution. So I took to the web once more and found 'the same' (but different - going from the photos on their website) board and ordered a few of those. Luckily, they came in the next day.

What a difference

I immediately hooked up the new board to my bench power supply and switched it on: < 1mA! That's what we're talking about! I took out my soldering iron and soldered my project to the new board. Measured again: < 1mA and < 100mA during the (short, few seconds long) operation the project had. The difference is night and day; but you can see for yourself:

Yes, 0.000A. Yes, it's turned on
Battery voltage over a period of 16 hours, again

The difference in battery usage is night and day. The difference between the "good" boards and "bad" boards, however, is much more subtle.

Battery side. Left is the "bad" board, right the "good" board. Note the short(er) USB A connector at the top, the spacing between the 5V and 3V connections at the sides of the boards and, funnily enough, the switched polarity; the "bad" board has the positive at the bottom, the "good" board at the top.
Again, "bad" board on the left, "good" board on the right. Note the USB-C connector on the "bad" board and the 5 battery status indicator LEDs. The "good" board has less components and no battery status indicator LEDs.

Conclusion

Well, if it wasn't already obvious: Don't buy or use the "DIYMore" boards; or at least not the version I have. Below I have provided clear photos of the "bad" and "good" boards so you'll be able to compare them to a board you're considering buying and I hope you won't get stuck with these battery draining things like I did. Since I am a bit of a cheapskate because, after all, I'm Dutch, I did order a bunch of new boards off of AliExpress again - they're about 60% cheaper than the ones I can order locally so that's worth a gamble. I will update this post once these arrive and have been tested.

I'm not sure what causes the huge power consumption on the DIYMore board. Maybe someone more knowledgeable can say something about this. I have added some close-ups below of the main IC's on the boards.

DIYMore ("bad") board. SW6115
DIYMore ("bad") board. IP3005A
"Good" board. 4056E
"Good" board. FP6298
"Good" board. 8205A

Update 2021 - 12 - 23

Well, the shields I ordered from AliExpress came in and I tested all 10 of them: all of them get the green light. They appear to be the same exact PCB (besides the color) and components as the (good) shield I already had from the Dutch supplier.

The "good" and "bad" shields side by side.
Close-up of back side of both "good" shields.
Close-up of front side of both "good" shields.
Only the used 18650 battery holder is slightly different.

For completeness' sake the microscope shots:

"Good" board. 4056E
"Good" board. FP6298
"Good" board. 8205A

Also... I just noticed the AliExpress page states:

5V 2A

But the silkscreen states:

5V 4A

Hmmmm🤔 Maybe I'll test that too, one day. Who knows.

1 comment:

  1. Turn-on current is mainly due to 68 ohm chip resistor. The 68 ohm chip resistor is directly connected between 5V out and GND. The two pads next to the 68 ohm resistor are actually connected. Remove the 68 ohm resistor or cut the line between the two pads.

    ReplyDelete