[Solved] Ni-MH to Li-Ion conversion. Anyone managed this?

Not sure if this is the right section or not. But power supplies seems the nearest. This is in part a dump of what I have learned/know and in part a request for others to fill in the gaps and correct any bad assumptions.

I have a music streamer that I’ve just re-capped and is working nicely. It is my eldest daughter’s Logitech Squeezebox Radio. The device is simple has a nice sound and works well enough either as a line-in portable speaker, or integrated with my squeezebox server. While I was recapping it I took a look at the rechargeable battery pack that it comes with (but which has been non-fucntional for a few years). It is a 10S pack of 1.2V 2000mAh AA Ni-Mh cells. As is often the case with NiMh packs there is no dedicated BMS on the pack, instead we have a 10 pin connector of which 5 are used. Vcc (nominal 12V), Gnd, two voltage taps connected by resistors on the pack between cells 3&4 and 7&8, and finally one end of an NTC which is then tied to ground. The pack also has some kind of thermal fuse wired in series in the pack. I’ve never seen one of those before on a Ni-Mh pack but it adds an extra level of protection I guess, though the rating seems rather high.

I can buy a new NiMh pack for about 40 quid (50-55ish USD?), I can rebuild a new pack from brand new cells (these cost about £2.50 each when buying 10). Those are the easy options. However I thought I’d at least explore the possibility of using Li-Ion cells instead. 

Ni-Mh and Li-Ion have quite different charging needs.

Firstly Li-Ion are 3.7 nominal voltage (typically 3.2-4.2 operationally), versus the 1.2V of the AA cells. A Li-Ion pack needs to have a balanced charge to avoid under or overvolting individual cells in the pack. For this reason a BMS circuit is normally used in the packs. The BMS in this case would likely be a 3S (3 in series) giving two voltage taps like the NiMh and Vcc/Gnd. It may also provide the NTC depending on which BMS you buy. 

Li-Ion charging follows two cycles, a constant current phase for fast charging, followed by a constant voltage phase to trickle charge, where the current gradually declines until it crosses a threshold and chging is terminated of prevent overcharging.

Ni-Mh meanwhile, is far blunter, the charger will I believe apply the full voltage (12V in my case) to the pack, it tracks the temperature (via the NTC) and the voltage via the taps and B+. Charger termination looks at Delta V, a small drop in the charge voltage that can be detected at capacity, this is a small drop I believe and as such is often combined with deltaT temperature checks to make sure we don’t get a runaway charger. 

Another clear difference is that while the NiMH charging circuitry has the charge voltage (Vin) supplied on the Vcc (or Battery+) line, Li-Ion tends to separate those.

My conclusion so far is that soldering up a new pack of NiMH batteries seems a lot less painful and error-prone than the Li-Ion path, and that is almost certainly the route I will take. However, the concept intrigues me still. 

A “proper solution” requires some circuitry, a small microcontroller perhaps, monitoring the BMS and converting the signals such that it triggers a DeltaV and/or DeltaT detection. This seems like a nightmare and a lot more effort than it is worth that is for sure.

A simpler and perhaps more attractive option is tohave a separately supplied BMS (there is a convenient 3.5V supply that ought to have enough current) but I would need to prevent the main board charging circuit from providing 12V directly to the pack. A diode on the Vcc would do that I think, allowing the voltage to be drained but no charge to be applied, and perhaps that would be enough? But could that lead to issues on the main board charging circuitry? Ideally I’d like to see the onscreen charge indicator working.

Any thoughts and experiences to share?

Neil