Status:
works, but poor performance: excessive (maybe - ?) ripple, especially in boost mode and at higher power levels. Mac laptop charges sucessfully off of it, sometimes. Diodes and inductor heats too much (80c at full load). This might be a straightforward fix, or a more siginficant design error (wrong component choice, values, or layout...)
Why:
The goal is that I can charge my laptop off an ebike battery directly, or any other DC source (solar system). No other solution that accepts this high input voltage in existence when I started this, now there is YZ826S from china which does the same thing (which I'll try before continueing to work on this project).
Prototype:
Features:
- Done in Kicad 6.
- Single-sided board for easy assembly.
- The single-sidedness means that it may be possible in a future revision to replace the remaining THT parts with SMD, and then attach a heatsink or enclosure to the back, for through-the-board cooling.
- As compact as possible.
- Input protection:
- Reverse-polarity protection provided by two high-side n-channel mosfets back-to-back, driven by HT0440.
- Inrush current limiting provided by same mosfets, by slowing the switch-on speed significantly.
- Fuse
- both of these work well, I will reuse this in other designs.
- Low standby / idle power consumption: the buck-boost converter is disabled when no device (power consumer) connected.
- Adjustable low-voltage cutout.
- Datasheet implementations for TPS25740B and LM5118 are mostly folowed, only the input protection, 5v supply, and power-saving stuff has some clever stuff.
- XT60 power connectors.
- Status indicator LED.
Warning:
TPS25740B seems to be discontinued with no obvious replacement. I had to buy them on Aliexpress. Has some minor bugs (I forgot what exactly. Notably needs an additional / different mosfet somewhere. I'll update this when I find my prototype board where I bodged it on. Also requires one cap around LM5118 replaced with different value).
Play with this at your own risk..