Power Supplies, multiple output, negative included

[skimask]

Just a quick thought for you ski... properly charging that Battery-Pack is going to be more tricky than you think because the level of discharge across the entire pack isn't going to be constant... because of those taps you're going to be discharging some cells at a greater rate than others. So, if you're going to be charging the entire pack connected in situ, a lot of cells will reach full charge perhaps a very long time before others.

That's what I'm talking about. But as you implied back in post #2, ground is ground, wherever it is, it's ground, or at least it can be. So, I whip up a 36v battery pack, 40 cell to get 36v at end-of-charge .9v/cell, 36v total, with 4 regulators:
-28v -> 0v (i.e. ground), -6v -> +22v, 0v -> +28v, +5v -> 33v.

The unit, during testing, isn't connected to anything else besides the tester while be tested, so it shouldn't matter where ground is, as long as the right potentials are there to drive things, even the chassis is completely electrically isolated from the circuitry. Overall current draw is going to be far less than 3 amps, once in a great while peaking near 3 amps.
So using linear regulators to keep parts count and cost down, I regulate out the 33v direct from the batt pack, regulate 28v off the 33v, regulate 22v off the 28v...instead of putting all of the regulators in parallel (40v regulated to 22v using a linear is a lot of wasted heat), use pass transistors on all of them and I should still have enough overhead V+ to run them all, although the 33v might be pushing it at full load with a near end-of-charge batt pack.

Charging however, as you eluded to, will be another story. 40 cells would max out at around 58v, probably need something like 64v to push any current. Some of the cells will undoubtedly charge before others. So, either I go with a very slow charge (less than 1/10C) to keep out-gassing down, or I split the pack into smaller segments during charge. I've charged a 16 cell pack successfully before using a simple 'linear' (not sure what you'd call it) charger that charged the whole pack at once, started off with higher current, as the pack charged up, the current dropped to something like 1/50C. Probably not healthy for the battery pack, but it's still alive and kicking after 3 years.

Or I just skip the battery pack, and go with a unit that runs off the mains.

More thinking required...


PS.: Mel, and BTW, that fix involving the 2 swapped wires, was done while I was at Fairford, practically in your 'backyard'. And I couldn't have done it without a book from Mike M. at nikamelectronics, and PIC guy.

[T.Jackson]

Here I was thinking that Melanie was merely referring to a balance type charging arrangement -- whereas each cell is individually charged to ensure optimal life expectancy, by eliminating the possibility of overcharging. But don't take my word for it, I'm just a pathological liar.

[skimask]

Here I was thinking that Melanie was merely referring to a balance type charging arrangement

Yes, I know, charge balancing, always the best way to go, with an unlimited budget, loads of time, a nice looking PCB, and so on and so on...(i.e. educational environment)
Then there's the practical way of doing things (i.e. welcome to the real world)...

I'm not planning on using rechargeable li-ion batt's (which would be the obvious way to go unless I want to start semi-contained chemical reactions, i.e. fires). I've also got to think about the (and I use this term loosely) morons (read that as uneducated persons who don't know the ins and outs of various battery types) who are going to be using this tester long after I'm gone. Therefore, Ni-Cad's are out due to the memory effect (these guys won't know how to do the charge/discharge thing for max life, even if it's specifically stated in the manual).
So, I'm basically left with NiMH if I want a rechargeable setup. Memory effect is practically nonexistent, low temp performance isn't the best, negative delta-V during charge is small, offset that with a low charge rate and a timer.

I took a good hard look at the box I'm testing. There's only one component in there that's polarity sensitive...and that's dump diode across a single relay coil. That one thing ruins everything for me. Without that one diode, I could easily use N-ch MOSFETs under PIC control to self-test the whole thing for me and tell me where the fault is, if there is one. And it turns out I only need 28v. The extra 5v section is completely separate from the rest of the box. Figure I only need 33v (28v + 4v of regulator overhead with pass transistors plus some fudge factor), brings me down to a nominal 32 cell pack, 4 packs of 8 cells. The circuit can tolerate a couple volt drop at end-of-charge on the pack.

So, it's time to figure out how to get rid of all those rotary switches and replace them with banks of P- & N- channel MOSFETs and automate something...

P.S.

I'm just a pathological liar.

Don't be putting words in my mouth there sparkles.