picbits,
Interesting idea, but for industrial use, I tihnk they are after a tried and tested method wit 'off the shelf' parts. If I hadmore time I would look into it, as it would be a novelty for them. Of course parts with wear over time, but thats what 'spares' contracts are for
Hero, as I mentioned, charge time is not overly critial, but preferably under 8 hours. Therefore, I should probably use more than just a current source, but we're not talking a 2 hour charge time here.
Apparently, Ni-MH indeed have a 'knee' but according to several datasheets form chip manufacturers (I'll get back to you with a link) they must switch to trickle current (or indeed turn off the charging current altogether) when the cells voltage is at its peak, where-as nicads, should be roughly 45mV down from the peak. This makes thinks trickier for me, as I designed a peak-hold circuit to detect this.
In order to detect the drop in voltage for 3*AA NiMH, it looks like I will have to detect something silly perhaps a maximum of 15mv (5mv per cell).
Of course some analogue and a uC's ADC will see this no trouble, but primarily it looks like I'm going with temperature change. Looking for a peak of 10C above ambient (30-35C). Secondary to that, the voltage knee, or rather, the 'peak', which I will have to calibrate for these particular chemistry (comparator + precision reference). Thats two comparators to turn it off.
Although it is very similar to NiCD's, I have read in several places subtle differences which dictate the order of switch off.
Thankfully, I have just been informed they want this project to have a face-lift....ergo...the charging circuit can go on a new PCB, not seperate. Which sort of makes this post redundant as I can now just use a maxim chip, or something specifically designed for the job (including a uC, OR the existing uC with a bit more code!). That said, its been a great learning experience, always good to know you guys really know your stuff. I will do some tinkering over the next week, and post my solution. As I said, now I have a hell of a lot more options since they want the entire thing redesigned for some extra's, which can include the charging circuit ON the main PCB.
Finally, Nigel, fantastic link! Very detailed. Usually I just like to get a schem, but he describes its opertation very well, a great basis for a true Ni-MH charger. I'll go with 400mA charge current, for 2500mA, thats roughly 6 1/2 hours. Not too quick to make timing overly critical, but not too slow for them.
Thank you all,
Blueteeth