Monthly Science: Sony NP-BX1 Battery Status

Having had the weaker of the two surviving STK batteries die 36 minutes into a ride, I tested them all:

Sony NP-BX1 - 1 A test - 2016-08-17

Sony NP-BX1 – 1 A test – 2016-08-17

The X axis shows W·h, rather than the usual A·h, because that seems more useful in a world of constant-power supplies.

The test current is now 1 A, rather than the previous 500 mA, to more closely match the camera’s actual load. The CBA tester doesn’t have a constant-power mode; I think that doesn’t make much practical difference.

The orange curve (STK D) is the failed battery, ending after 1.4 W·h. At an average 3.2-ish V, that’s 26 minutes, which is close enough to the actual run time, given the different current.

The red curve (STK C) is the sole STK battery of the original four from last November that actually worked.

The upper two curves come from the mostly unused Wasabi batteries (F and G), also from November. They have lost a bit of their capacity, but show the highest voltage out toward the end, so that’s good.

The black curve is the lightly used Sony OEM battery that came with the camera. Although it has about the same ultimate capacity as the other three “good” batteries, the voltage depression suggests it’ll trip out early.

The others are pretty much debris by now. I suppose they might be good for LED blinkies or some other low-voltage and low-current application, but …

So I’ll start using all four of the better batteries and see how the run times work out in actual use.

  1. #1 by david on 2016-09-29 - 01:36

    BTW, quick question for you: I am about to have a stack of big (100AH @ 3V) LiFePO4 batteries on hand to make a 2p4s pack. I’d like to match the cells in each pair as closely as possible, so I need some way to accurately (1%? better?) graph SoC/Voc over a wide range. The West Mountain CBA would be best, of course, but I don’t have one. Any thoughts on whether wiring up my Tek scopemeter to a datalogger and bolting on a constsant-resistance load would generate useful data, or if I really need to come up with a CBA? Any idea where I might find someone willing to loan me a CBA or let me use their for a few runs?

    • #2 by Ed on 2016-09-29 - 09:47

      I like the CBA, but it’s a lot of coin for what it is, particularly with locked-down Windows-only software. I use it just often enough to not want to mail it off. [grin]

      You could gimmick up a good constant-current load (*) from a hulking MOSFET + heatsink, an op amp, and a handful of parts. That, plus a datalogger, gives you most of the functionality of a CBA, minus fancy graphs and Windows driver issues. You probably want an over-discharge cutoff, too, although you could use a stock battery protection circuit if you weren’t fussy about the cutoff voltage.

      I’d be mildly surprised if the RC / quadcopter crowd didn’t have a decent high-capacity battery tester.

      (*) Which is what you really need for mA·h vs. time plots. Throw an Arduino + good DAC into the mix to drive the MOSFET and you could do constant power vs. time tests, which would be more useful these days.

  2. #3 by david on 2016-09-29 - 17:49

    Thanks! Actually I just found one that’s pretty cheap, so maybe I’ll just do that after all. Do these multi-pack traces you always post come from sequential runs, or does it have multiple channels to run them all at once?

    • #4 by Ed on 2016-09-30 - 08:54

      The CBA does one test at a time, but you can add each new test to an existing file: every one of those multi-trace plots represents a long day of battery shuffling.