Random Power Banks for low-capacity cells

thanar

Member
Joined
Feb 12, 2018
Messages
155
I have been building Power Banks using the lowest-quality cells I scavenge that are still above 1000mAh at a discharge rate of 0.5A. These cells obviously will not end up being used on a Powerwall project, they did however pass all testing.I will share my thoughts on the projects and my questions, since I was expecting this venture to be more successful that it actually is turning out to be.

First Power Bank uses 6 Samsung 28A cells from a laptop battery pack. They sum up to 6600mAh(3.0V to 4.2V),discharge at 0.5A, which adds up to 24.42Wh (nominal 3.7V). The emptypower bank was bought through e-bay for a couple of bucks. These power banks need some grinding on the inside, since the cells sit too tightly and sometimes a couple of them tend to disconnect from the positive bar once you forcefully close the lid. That's why all testing is being down with the cover open, for the moment.

image_jewchn.jpg


The power bank is supposed to be able to provide both 1A and 2A of current on two different USB ports. I found out that both ports provide up to 1A maximum, which calculates to 10W max power when both ports are being used. That calculates down to 450mAh (nominal) per cell. And here's where things get interesting...

I first completely discharged the power bank until it shut off. It shuts off at around 2.85V, which is not too bad, the cells will shoot up to over 3V as soon as current draw disconnects anyway. I then completely charged the power bank and it took 7766mAh at 5V nominal, which calculates to 38.83Wh (that's 60% MORE than the cells' discharge capacity).

image_hqwxej.jpg

I then completely discharged the power bank and it gave out just 2887mAh, which calculates to 14.44Wh, which is 40% LESS than the cells' discharge capacity.

image_tzaacy.jpg



All of the above, end up at a 63% efficiency while charging, and 59% efficiency while discharging, with an overall efficiency of 37%, at least by my calculations and the hardware at hand. This is kind of awful. Can you see any obvious mistakes?


image_bpritk.jpgimage_azksah.jpg


There's a critical bug with these power bank cases...[size=large]Because the cells are tightly squeezed, looks like that if using the build-in springs/strips, a couple of cells get disconnected as soon as the case is closed, which could even be dangerous in some cases. That's why I've opted to scrap the strip/springs altogether and solder the cells with wire instead.[/size]

 
A few thoughts during lunch pause :

-According to what I find on Aliexpress the board is obviously the 1A model not the 2A one
- I can't find any real data about the IC that'son the board but from whati can seei would bet the charge circuit is a linear one wich mean (for 5.15V input to 3.6V mean at cells) 3.6/5.15 = about 70% efficiency
- Accounting a 90% efficiency for the cells themselves (i have no precise data about that but it doesn't sound too bad for old cells that usually have a higher internal resistance *) we should have about 3.6*7.766*0.9= 25Wh available when discharging the pack
- You got 5.21*2.887 = 15Whout ofthe bank
- 15/25 = 60% efficiency wich is pretty bad for asmall boost but nothing to be really surprised of from a cheap gadget

... Andwe get back your overall 37% = 0.7*0.9*0.6 !

Even witha betterboost part at 80% the overall efficiency would be only 0.7*0.9*0.8 = 50%

* If anyone has measured datas aboutcells efficiency with age please tell us !

Edit : typos, poor english
 
That powerbank uses a linear regulator.

I don't have it anymore, since it has a major fault in which sometimes, the output of the IC can cause a micro short, which causes the capacity to be drained in about a day.

Also, it has a tiny inductor for boosting 1.5A 3.6V at 5V 1A. The boosting IC is an unmarked one, but it looks decent.

If you want a good one, get one with a TPower IC, which is the company which manufactured the original high quality TP4056, TP5000, TP5100, and others.
 
Assuming that you tested that each cell could hold about 4.1V for a few days without losing more than 0.03 V ( ie they did not self discharge )

Then there can be no explanation for this poor efficiency , I think , except that the electronics if not allowing effective use of the cells potential....

Even with a terrible internal resistance of 400mOhm per cell , the losses on one charge discharge @0.5 A is only 11% ...

It seems the only solution is to bypass the electronics and then test cycle efficiency.
 
ozz93666 said:
Assuming that you tested that each cell could hold about 4.1V for a few days without losing more than 0.03V

Of course I did.
I actually charge them to 4.2V and let them restfor around two weeks.
Anything that stays above 4.1V gets discharged and thencharged to 3.7V and stays for another couple of weeks.
Anything that is still above 3.65V at the end passes.
 
Did a couple more of the same power banks, this time soldered the cells with copper wire. Total efficiency jumped to a little over 60%, which I was not expecting... I suspect some cells were disconnected from their contacts when I closed the unit prior to doing the initial test, which is a bummer and also a word of notice for said power bank shells. Will do all my powerbanks by soldering copper wire to the cells from now on.
 
thanar said:
Did a couple more of the same power banks, this time soldered the cells with copper wire. Total efficiency jumped to a little over 60%, which I was not expecting... I suspect some cells were disconnected from their contacts when I closed the unit prior to doing the initial test, which is a bummer and also a word of notice for said power bank shells. Will do all my powerbanks by soldering copper wire to the cells from now on.

That's interesting - but not surprising. I shall try soldering mine from now on.

How did you test the cells? If you used the Opus, it often overstates capacity by about 10%. Another factor is what voltage does the bank cut off at?
 
I test all my cells on an iMax charger. These banks cut off at around 2.85V (voltage drops to under 3V for a very small amount of time under load on the bank), and charge the cells up to spot on 4.2V. Discharge tests done at 500mA for these cells, since they will never see more than that in these power banks.
 
BlueSwordM said:
That powerbank uses a linear regulator.

I don't have it anymore, since it has a major fault in which sometimes, the output of the IC can cause a micro short, which causes the capacity to be drained in about a day.

Also, it has a tiny inductor for boosting 1.5A 3.6V at 5V 1A. The boosting IC is an unmarked one, but it looks decent.

If you want a good one, get one with a TPower IC, which is the company which manufactured the original high quality TP4056, TP5000, TP5100, and others.

I am waiting for a DIY powerbank.. hope it isn't using linear regulator..
 
hermitdave said:
BlueSwordM said:
That powerbank uses a linear regulator.

I am waiting for a DIY powerbank.. hope it isn't using linear regulator..

To be realistic, I believe most powerbanks have a linear regulator for charging, since I don't think lots of people will care if the wall brick works twice as hard to get their power bank to full. It's what goes out of a power bank that matters, actually.
 
Note: the Samsung-28 cells can handle fine 1A discharge, so since you have 6 in parallel and only a maximum drain of 2A there can be no problems there. From the anomalies you describe you seem to have a problem with the actual circuit.
 
I'm having a bit of a trouble with tracking the various boards for this power bank case. So far, I've built power banks using the following three boards (presumably belong on the same line): HKS-806D-2, HKS-806D-2F and HKS-806D-3.
The first one seems to be unable to provide over 1.1A at the 5V output. The second one seems to be working fine, but it's been having trouble detecting when a low-draw device (for example a BT headset) is connected into the USB. The last one seems to be working best, looks like it's probing the output USB every now and then, and even if there's a very low draw device connected, it eventually starts getting power.

So I guess the latest and greatest one is HKS-806D-3, latest one I used had a production date of around May 2018, which however I am unable to find online any more. Anyone has been having any luck? I have enough not-good-enough cells to build around 20 giveaway power banks for friends...
 
In theory all should handle 2A...but D-2 is designed for low current applications (<1A), D-2F for higher current ones (1-2.1A) and D-3 is the universal/improved version that should work well in most scenarios.
 
This is all useful information! Do you have links of where you bought the D3 version from?

I've been using 2 cells and 4 cells as giveaways as well, but one person did say his died. I suspected it drained to 0V and killed the cell. I'll post up some photos of my 2 cell versions with some tests. One for maximum output, the other for idle drain.
 
not2bme said:
This is all useful information! Do you have links of where you bought the D3 version from?

Latest bunch from a random e-bay seller arrived with version 3 board inside, although all photos on the page showed a ver. 2F board. I guess they all now ship the updated version, which does have a drawback, it doesn't seem to be working properly with a couple of USB lights I have around. It switched off after a few seconds for no apparent reason, no matter the intensity of the light.
 
Back
Top