DC DC Isolated Opto-Relay Charging??? Why Not?

[deVries]

I've been charging high capacity packs with three different DC-DC converter chargers for well over a year now, with excellent results, Doctorbass has done the same long before I did it.

I just bought the celllog breakout module from Robin of RWaudio. they fit 4 celllog connected for up to 32s and are isolated with very small relay.

I know Jeremy & Doctorbass are experts at using the small DC DC converters to do tricks with charging high voltage bulk charging (Doctorbass) & also using for cell level charging too (Jeremy & Doc). You guys are great for that.

Why isn't there an easy way to charge at the cell level using a single, yes, just one, DC DC single cell isolated output converter at high amps (maybe 3.6v @ 50+A or 100A for A123) & have it "shared" on an Opto-Isolated Relay to send the charge in sequence through the opto-isolated relay to one cell at a time while also connected in series in an isolated switching mode sequence for x-string of cells in series using heavier balance tap wires to carry the higher amps??? The wires may not have to be too heavy, since the charge is switching between the cells one at a time in series with the opto-isolated relay.

I know this idea has been thought of, but why hasn't it been done already?

(Btw, I'm not at all knowledgeable in understanding the electronics design of charging, so I'm probably missing the obvious not knowing what is "obvious". )

This was pointed out to me by PM:

problem is that the relay failure is important risk and you also need a delay between each relay to avoid the condition where two or more relay are ON state at the same time due to magnetic hysterisis of their coil.

[amberwolf]

This has been discussed before, in context of a BMS MCU reading and balancing the cells one at a time, but I don't have a link to the thread.

You could do it, but you will need a separate "relay" or isolator for each side of each cell, to cut the charger off all the other cells except the one you want to connect. It can be simplified a *little*, in that since the positive of one cell is the negative of the next, you can use just a single relay on that interconnect point, but then you need a circuit that will correctly connect the positive of the charger to the positive of the cell, and neg to neg. So if you have 16s, you need 17 "relays", plus whatever else is in that other circuit.

It would be simpler to just use 32 "relays", so that every cell has it's own positive and negative connections to the charger, independently controllable.



Then you need a circuit that reads the output of the charger, and once it reaches the point you determine (based on volts, current-dropping-curve, coulomb-counting, Ah, Wh, or whatever method you choose), it would disconnect from all cells, then connect to the next cell.


If the relays are just regular relays, it's probably gonna be fairly bulky, but there's no worries about figuring out what parts have to be rated for what voltages and so on--just get relays that are rated for the maximum voltage you might see across the cell being charged, and the maximum current you would draw at completely discharged--and make sure they're rated for DC.

Then install fuses rated for the pack voltage and the max cell-charge current on EVERY charging connection to the cells on the pack, so that if any relay ever gets stuck, and the charger switches to the next cell, then the short that will create across the previous cell will blow the fuse and not the pack.


I'm sure this can be done with solid-state electronics, too, but that gets into driving circuits and all sorts of other complications I don't have the time to think out. (I figured 5 minutes was enough to get you started, though. )


Either way, this is gonna probably be bigger and more complicated than just doing it with the separate DC-DC units.

[deVries]

This has been discussed before, in context of a BMS MCU reading and balancing the cells one at a time, but I don't have a link to the thread.

I'm sure this can be done with solid-state electronics, too, but that gets into driving circuits and all sorts of other complications I don't have the time to think out. (I figured 5 minutes was enough to get you started, though. )

Either way, this is gonna probably be bigger and more complicated than just doing it with the separate DC-DC units.

I searched and couldn't find it. Can you? I found one that had all the key posts deleted thanks to the "mbknight sellout".

Hey, pretty good for 5 minutes of thinking!

Ok, would the parts for the relay/fuses cost more then $8-10/cell???

AmberWolf, thanks for your ideas & fast thinking.

[amberwolf]

I searched and couldn't find it. Can you? I found one that had all the key posts deleted thanks to the "mbknight sellout".
I dunno--that could be the thread. Do you have a link to it I could check and see if I remember it?

Ok, would the parts for the relay/fuses cost more then $8-10/cell???

Depends on your parts sources. I haven't bought any of that stuff new in a long time, as I keep salvaging stuff people throw out that has most of what I need in it. So I don't know what they'd cost new--depends on what ratings you need out of it.

Also if you are going to make PCBs or just handwire it all, what kind of enclosure(s) you use, etc.

[deVries]

This has been discussed before, in context of a BMS MCU reading and balancing the cells one at a time, but I don't have a link to the thread.

Is this the thread you were thinking of ???...

Distributed PIC BMS

That looks great so far!
You're getting pretty close to what I figure will be the ultimate topology.
Do you need a voltage regulator? I thought the PIC had an internal reference that would work. Perhaps I'm thinking of the ATTiny. If the regulator protects against unwanted resets, then it might be worthwhile.

Turning the shunt on and off in a hysteresis loop seems to work fine. If you can run the shunt transistor in the switch mode, it won't need to dissipate much heat. I was also thinking the PIC could generate a PWM signal to drive the shunt. Once the PWM gets up to 100%, the PIC can signal the master to back off on the charge current. I think that part is doable.

Getting the cell MCU's to sleep and wake up was one thing I had no clue about. It should be possible to get a signal from the charger and one from the controller to tell everybody to wake up. During standby, they should sleep indefinitely.

Actual cell votages don't change that fast, so the cycle time to measure all the cells does not need to be that fast.

[Jeremy Harris]

Funnily enough I looked at doing something like this two or three years ago. My plan was to use a motor driven rotary switch (one of the solenoid driven ones that click round one stop each pulse), fitted with two high current wafers to switch each individual cell in turn. I actually have a couple of the rotary switches (made by Ledex) but couldn't find any high current wafers for them. The ones I have look like this:

swr-a48336-001_lg.jpg (137.14 KiB) Viewed 122 times

[amberwolf]

You know you could probably *make* high-current wafers, though it would take a bit of time and pondering.

Is this the thread you were thinking of ???...

Distributed PIC BMS

That might be it. It sort of sounds familiar. I see what you mean about it's swisscheese status. I tried to find archives of it on the web but no luck.