The "opposite" of a diode OR (sort of)

[ZachR]

I have an interesting design challenge.  I am working on a system that has multiple battery packs.  The existing design uses a simple diode OR to draw power off of the pack with the highest voltage.   However, the end user for this project has come back and said "we want the system to draw power off of the lowest pack, as long as the lowest pack is above some voltage threshold".  I could build some FET switching array controlled by a micro that samples each pack, but figured there might be some other elegant solution using analog components, maybe not as simple as a diode OR, but similar.  Any ideas?

Thanks,

Zach

[Zero999]

There's no simple way to do it using diodes. It's possible with comparators switching P-channel MOSFETs but not so simple.

[sleemanj]

I could build some FET switching array controlled by a micro that samples each pack

I think this is probably the best (read simplest and least likely to behave in unexpected ways) idea.  I'd guess use a  diode-or to power just the mcu (and gate drives if necessary), done right it's not going to be drawing enough to care about, then use back-to-back P-fets (because don't forget about the body-diode) to switch the main supplies based on sampling.

[Lightages]

If these batteries are rechargeable, and are anything other than NiCads, this is crazy. No rechargeable battery wants to be discharged anymore than necessary except NiCads which prefer to be fully discharged. Discharging the other types more than necessary kills their useful life. If these are primary batteries and the intent is to get the longest use from them and have the biggest reserve left in the other batteries, then this is not a simple diode problem as already mentioned. It is a ADC for current, voltage, and microcontroller solution.

[wraper]

If these batteries are rechargeable, and are anything other than NiCads, this is crazy.

I'd guess the user wants to FIFO them, so the empty batteries get swapped while the device is still powered.

[T3sl4co1l]

Connect them in series, with antiparallel diodes for each cell.  Use a constant current load (tolerates wide voltage swing), rather than a constant voltage load (exhibits wide current swing; which is characteristic of the conventional diode-OR case).

Best done with NiCd or NiMH (that's not utterly destroyed by discharge to zero), with a wide voltage range SMPS on the input.  Or better, with some logic to handle the switching.

Tim

[German_EE]

Objective 1
Draw power from the lowest pack providing that the pack is above a certain threshold.
Each pack has a relay or contactor on the output that drops out when the pack voltage drops below the threshold voltage. Some industrial contactors have adjustable voltage thresholds.

Objective 2
Draw power from the pack with the lowest working voltage.
Suppose you have four packs, A,B,C and D. Connect all the negative points together then connect a relay coil and a diode between the + outputs of A and B. Depending on which is higher the relay will be on or off, powered by the voltage difference between the two packs. Use the contacts on this relay to select pack A or B. Do the same with packs C and D. Now wire a diode and relay coil between the AB and CD output and use the same system to select which pair of outputs has the lowest voltage. If you use 3V relays you can detect a difference between the packs of 3,7V.

[con-f-use]

For Obj. 2, a microcontroller + fet's is the cheaper solution and draws negligible hold/switch current compared to the relays.

[German_EE]

True, a microcontroller with an 8-Input ADC could do the job in a dozen lines of code but a box full of relays and contactors is more fun.