The D-LI90 battery is a fairly standard DSLR battery used by Pentax/Ricoh for most of their medium to high end DSLRs from the last 10 years or so, including the K-7, 5, 3, and 1.
Since I started using my 300 mm F/4 lens I've noticed that performance across my old batteries was quite variable, with random weird glitches in camera performance. I assume this is caused by the high current draw required to drive the SDM autofocus motors in such a large lens (as compared to the tiny screw drive primes I've previously used).
Buying new batteries helped a lot, but I now had a stack of 3 old original batteries from around 2012 that never quite made it to the recycling center.
Even with new batteries I've noticed that the battery life in low temperatures is not as good as in warmer weather, and e.g. after around 1 hour of light use in -10 degree (C) weather my old K-5 would often struggle to actuate the aperture lever in stop down mode, sometimes resetting the camera if I did this a few times in a row.
These battery packs are all made using cylindrical Li-Ion batteries, and I know from reading a lot of datasheets that there can be a large difference in battery performance across temperature. Comparing detailed specifications between two common 18650 cells I found that at in the 0 to -10 C temperature range you can get around 2-3x better performance from some batteries that are otherwise equivalent at higher temperatures.
In general it seems that Panasonic NCR18650Bs are a good choice for e.g. flashlight batteries.
Getting back to the D-LI90, these batteries all use a slightly smaller battery with a standard size of 18500.
Disassembly is slightly variable, but was mostly done by using a nylon recoil-less hammer to whack the seam of the battery. After a bit of this I could sometimes loosen most of the weld/glue, scoring the seam with a scalpel helped a bit. Eventually poking a screwdriver in also works but avoid doing this near the side where the battery contacts are since there is a PCB near it that might be damaged.
Interestingly the "Made in China" labelled battery had Panasonic NCR18500 branded cells, while the other two were unbranded though apparently made by a Japanese company. The Panasonic celled battery also had a different protection PCB and a different design for the battery-PCB interface.
Looking at my local supplier I found that the only real option here would be Panasonic NCR18500As. This is a decent choice assuming they are similar to the NCR18650Bs I've used for my flashlights.
I was not able to find a good datasheet for the A variant, but the base NCR18500 datasheet indicates a performance of ~2 Ah for the 20-60 C temperature range, and at least 1.7 Ah at -10 degrees:
While not a true 1:1 comparison, e.g. a Samsung ICR18650 has a rated capacity of only 50% at -10 degrees when discharged at 500 mA (which might reduce measured performance since the battery will not heat up as much during the test).
You can get an indication of this performance delta when looking at the start of the Panasonic curve, the X-axis is implicitly a 1 hour at the 1.9 Ah point, and the low temperature curves dip down a bit at the start before coming up. This could indicate battery self-heating, and if the batteries were simply placed in a thermal chamber with poor airflow they could have warmed up a fair bit.
Unfortunately the exact test method is not specified, and also this is not representative of typical DSLR camera use outside of extensive live-view or video recording. For my primary use case there is little continuous loading and therefore little chance for battery heating.
I ended up ordering 2 sets of these unprotected cells and using a DIY capacitive discharge spot-welder to assemble new packs for a winter test.
Battery reassembly was fairly simple, and for now I've pushed the shell back together, and held it in place with a layer of non-trademark infringing Koptan tape. Kapton tape is thin enough that one wrap of it will not cause issues with the camera.
Note that my reference here is how many SLR pictures I could take before the battery either was unusable, or the battery indicator started moving. The Pentax battery indicator stays full until around 50% capacity remains.
The capacity measurement is purely voltage based, so it will tend to move around if a large load is placed on cold batteries, such as using live view or video mode.
In general these tests are for bird photography with the DA* 300 f/4, and the main LCD turned off. I rarely exceed 1000 pictures per outing.
- Official Pentax battery from 2020
Capacity at around -2 to +5 degrees C (variable due to solar heating) was around 1000 pictures over 2 hours before the battery indicator moved into the 50% area.
- NCR18500 battery pack first test
Capacity at approximately -5 degrees C (fairly constant temperature, no sun) was also around 1000 pictures over 2 hours before hitting the 50% area.
- Third Party replacement batteries
I have several third party batteries which were likely around 2-3 years old, but have only been used for around 1 year.
These batteries frequently run down to 25% charge or less after less than 1000 pictures at moderate temperatures such as +10 degree weather.
The official Pentax (Ricoh) batteries are fairly expensive, but appear to use high quality battery cells.
Third party batteries are significantly cheaper, but appear to use worse cells that are particularly ill suited to cold weather.
Rebuilt batteries with new NCR18500 cells appear to match or exceed the performance of new first party batteries.
Updated february 2021: During southern Norwegian winter (down to -15 C usually) a NCR18500 celled battery still has no issue taking 800+ shots over several months in service. With these cells I have not had any reason to swap battery during a typical outing.