Feedback from the v1
As mentionned in my last article, the first version of the power bank could only power up my DIY box google, and not my Topsky F7X. So the main goal of this revision is of course to increase the output power. I also though it could be convenient if I can use it to power my TS100 solder iron.
I also noticed that my phone don’t detect the power bank as a high power adapter, and therefore only charge at 500mA, half of the available current.This will issue will be adressed as well.
I had more idea and improvement in mind, but I decided to keep it simple and cheap for now. However, in anticipation of a potential upgrade, I ditched the Attiny9 microcontroller for a more capable and future-proof IC.
One last thing that encouraged me to keep working on this project is the third place it got at the FR1-Challenge of CircuitMaker. I’m very proud of this result, and it push me to do more open source projects.
The low current charging in OTG mode is quite simple to explain. Phones expect to have the USB data line shorted together on high power wall adapter, which was not the case in my design. However, data line must only be shorted in OTG mode. Otherwise, the power bank own detection will be disturded in charging mode.
The solution was then to add an analog switch between the USB data lines, activated by the OTG_EN signal.
Here is the list of the criteria I used to find the new microcontroller:
- Wide range of input voltage, 5V to 3V. Because I will power it directly from Vbatt
- ADC with internal voltage reference. Will allow a more precise battery level monitoring
- USB device capabilitie with bootloader. This is more a future-proof requirement
- At least 10 I/O pins
- Smallest package
I went deeper this time on the boost circuit design. I first re-measured more precisely the power I needed to power my devices:
- Topsky F7X: 13W at startup, and then 7.5W idle. Voltage doesn’t impact the power consumption.
- TS100: 18W at 12V (which is the minimum working voltage).
I used the Texas Instrument Webench tool to find a suitable boost IC, and I picked the TPS61089. It can generate 12V from a single cell, and up to 18W. So on paper, it can power the TS100 solder iron. It’s package is also very small, which allowed me to use a bigger 5×5mm inductor. But not big enough to reach the full potential of this chip, at least according to the simulation.
Routing and Manufacturing
Not much to say here, I followed the same workflow as for the v1. Here is the PCB fully assembled.
The boost circuit was heavily tested to find out it’s maximum capabilities in the real world. And there was a lot to do. As expected, the boost output power didn’t reach 18W but not very fat at 16W. However, it took me quite some work and testing to get a solid and stable design. During my discharge run, I soon realised that the IC was overheating, causing the output to brown out. So I went through an optimization process to improve the boost efficiency.
On the list of parameters that affect the circuit efficiency, I could tweak two of them:
- Output voltage. The more difference there is between the input and the output, the less efficient the converter will be.
- Switching frequency. Lower frequency will improve the efficiency by reducing the switching loss of the MOSFET.
Because of the totally different microprocessor used from the v1, I re-wrote the software from scratch. The capabilitie to measure precisly the battery voltage allowed me to add one more low battery warning level. Now the status LED will blink when the battery is low, and blink quickly for critical battery level. I optimized the code to minimize the standby current consumption of the board, which ended up around 500µA. For a 10000mAh pack, it will take more than a year to drain 50% of the battery.
I tried a new enclosure design, as the two joints on the previous version turned out to be very visible.
This time the main shell goes all the way along the cells, and two flat cover close the power bank.
This power bank has finally became my daily driver for my FPV gear. With the recycled cells I used, the capacity is around 8000mAh, which gives me roughly 4h of run time on my google. With brand new cell, the capacity can easily reach 14000mAh.
I’m not entirely satisfied by the new enclosure, I miss the chamfered edge. Also, because the TPU is so flexible, the cover don’t fit very well, even when glued. Probably a mix from both design will do it.
If you’re interested to get one, send me a message, I may still have some board around.