Battery paralleling circuit

[coulomb]

Since this paralleling circuit is not really about homegrown BMSs, I'm starting a new thread for this circuit. This is version 4, where the positive side of the charging bus is replaced with a pair of inspired multipin sockets (thanks, acmotor!).



The advantage of version 4 over version 3 is that you now never parallel chargers. Each charger has its own little bus, and the size of that bus is determined by the plugs. In the above, you can charge with just the onboard charger, or you can charge with the onboard and two external chargers. The onboard charger charges 4 modules; the external chargers 5 each. If you added another charger, just make up another lead with 2 more plugs. Should one of those chargers fail, use the old lead.

You can also check the voltage of each module separately without opening up the vehicle, charge or discharge just one module, etc.

The disadvantage is that there is no interlock to prevent only part of the pack getting charged. I'll have a think about how this could be done.


So people don't have to refer back to the other thread, here are the previous circuits.

Version 3:



Version 2 (two charge buses):



Version 1 (same as V2 but no module fuses and no module boundaries:



Woody's original circuit:

    48V Bus + ---O  O--- + 48v block - ---O  O--- - 48V Bus
                   /                        \
prev 48V block ---O  O-----(bypass)-------O  O---  - next 48 V block 

My original response to Woody's circuit:


Comments welcome as always.
Edit: [ FONT=Courier ] doesn't preserve spaces.

[acmotor]

Hey, version 4 looks just like my red suzi wiring !
( with the addition of some 1 ohm charge balance resistors before the diodes )

Well done coloumb.

... and I can assure you it all works.

Battery pack broken into modules, safe voltages, flexible charging options are all the go !

[acmotor]

I have been tempted to put say 100K across the diodes so I can read each module voltage at the multipin plugs, just as a diagnostic tool.

[coulomb]

Weber has suggested that the pack can be used for running the house in a blackout, or load shifting, especially if the module voltage is a fairly standard one, like 48v. All those "V2G" things. So that would mean taking the diodes out altogether; maybe put them on the charger itself, or leave them out altogether. You'd still need to fuse the connector pins, of course, perhaps with higher current fuses so you can take out say 6kW to run the house for a while. (Assuming you don't charge at 6kW).

So now the connectors are basically access to each module separately. If you had trouble sourcing changeover contactors, you could just use single throw contactors and use the connectors to common the negative sides (I think that's how acmotor does it). Then you have the maximum flexibility, with the hassle of having to fiddle with the connectors every time you want to charge, even for opportunity charging on the road.

So maybe this: changeover contactors, access to each module's positive side, and a separate "charge bus" with the diodes connected to a smallish onboard charger. The DC bus is available via a beefy but convenient connector (say at the front of the car, through the grille but protected from bug splats) for extra charging at home. Now you can't series any of the modules, so that's a little less flexibility, but that's the price for not having to access the now largish connectors for internal charging, and for home charging. So you have one two pin "charge only" connector, and the two (or so) larger multipin connectors for draining applications.

[Nevilleh]

I'd be interested to see any thoughts/opinions on paralleling LiFeP04 batteries. I was thinking about making a 120AHr cell from 3 x LFP40s. I read the stuff on the "Battery University" and it seems quite practicable, but I don't see how to monitor individual cells.
Has anyone tried this?

[Johny]

My understanding of this is since this is the way larger packs are built using smaller cells you don't actually have to do anything. Parallel the cells and rate your BMS for the total cell capacity. The only downside is that if a cell decides to go low impedance, it will take it's paralleled brothers with it to battery heaven. Of course the BMS should pick up on this before it happens.

[coulomb]

I'd be interested to see any thoughts/opinions on paralleling LiFeP04 batteries. I was thinking about making a 120AHr cell from 3 x LFP40s.

There are also Thunder Sky 60AH cells; they seem to be less common, so fewer companies stock them. Probably less work paralleling two than three, at the expense of slightly fewer packing options.

I read the stuff on the "Battery University" and it seems quite practicable, but I don't see how to monitor individual cells.

My understanding is that you just don't monitor the individual cells. Internally, a 90AH cell might be two pouch cells in parallel, each of which is like 45x1AH (wild guess) sheet cells. So they already paralleled internally, and any mismatch is slight enough that it doesn't cause any problem. You certainly can't reverse cells that are in parallel, for example, or have one at 2.2v when the other is at 2.7v (assuming "normal" currents and adequate paralleling wires/straps).

Edit: grammar; specifically TS 60AH cells exist

[Tritium_James]

The problem with paralleling cells is what Johny mentioned, if a cell decides to fail low-impedance. You've now got a couple of good cells bolted to it that can dump in very high currents.

I guess it depends on what happens in that circumstance: smoke, fire, explosion, it's going to factor in on the chemistry and the size of the cells. But the safe solution (and this is apparently what Tesla have done in their pack) is to fuse *each individual cell* so that if it fails low impedance, the fuse blows and it drops out of the pack. Note they have ~8000 cells in the pack... lots of fuses.

[Johny]

You get the feeling that the interconnects somehow act as the fuses with the Tesla.

[acmotor]

The fuses only have maybe 2VDC across them so any suitable piece of wire or PCB track does the job. No special fuses are required.

I am making up a module of cheap 8C Lipo (4.4Ah) at present. 4 wide x 10 in series (37V 17.6Ah). I have used the BMS (at the start of the Home made BMS thread). One BMS across each 4.
I have used PCB track as fuse that lets go around 60A per cell in a slow fault without drama (240A total). This is about 4 x the working current of 60A in red suzi. The 8C LiPos are already in trouble at this current and a mostly discharged cell will not blow the 'fuse'. 3 cells ganged up on one s/c cells would manage though.

ESR of these LiPo is ~35mohm so not great compared with the 20C cells so I have gone for 4 in parallel.

Monitoring a blown 'fuse' is purely visual at this stage and surely needs to be 'alarmed'.
Not all cell faults would blow a fuse so it is more a worst case catastrophic (maybe) protection.

All experimental at this stage.

[Nevilleh]

Yes, fusing the individual cells would be an answer. That means I would need an interconnect that would melt at say 10C, or 400A in the case of LFP40s. Any idea what that might look like? And still have a low enough resistance to not cause much of an IR drop at "normal" motor currents.
And with 132 cells to connect, it shouldn't cost much.

Having said all that, I just saw acmotor's post and maybe a bit of pcb might be an easy solution.

[acmotor]

There are plenty of track calculators out there e.g. mine was loosely based on this track calc.   
http://www.circuitcalculator.com/wordpr ... alculator/

[acmotor]

Another calculation to look at...
(from http://www.epanorama.net/documents/wiri ... tance.html )

Wire sizes used in fuses
The Standard Handbook for Electrical Engineers lists the following formula:

33 * (I/A)^2 * S = log( (Tm - Ta) / (234 + Ta) + 1 )

I = current in Amperes
A = area of wire in circ. mils
S = time the current flows in seconds
Tm = melting point, C
Ta = ambient temp, C

The melting point of copper is 1083 C.
See pp. 4-74 .. 4-79 of the 13th edition of the Handbook for more info.

[Tritium_James]

If it's relying on blowing bits of track on a PCB then your 'fault' detection could be a smoke detector in the battery box. There will be a crater in the fiberglass and little chunks of melted copper everywhere. Burnt PCB stinks!     

[acmotor]

True, but <2V doesn't seem to 'blow' the track. It just melts and goes open. Not a violent operation as with higher voltages.

There is some smoke so yes, smoke detection ? You could coat the track with odour encoded paint !!

400A is a lot for PCB copper Nevilleh. i.e. very wide !
The point of using the PCB was that the fibreglass board held things together. A straight wire (strap/bar) could be used attached to the PCB for mechanical support.

[acmotor]

If my LiPo experiment ends up not too embarassing , I'll post some details.

[Johny]

Please do acmotor. A couple of weeks ago I was flirting with the idea of LiPo (after reading about Kokams). I found that the cheapest LiPo battery system I could put together would be composed of 810 PSP batteries. But I guess PSPs don't need 5C so that wouldn't work!
When you say "cheap" - how cheap?

[woody]

George Symons of evmotors.com.au and AEVA Sydney Prez has been trialling some LiPos in his Charade for a while, I'll email him and see if he can tell us where he got them and how they're going (assuming his charade hasn't burnt to the ground).

[Nevilleh]

Another calculation to look at...
(from http://www.epanorama.net/documents/wiri ... tance.html )

Wire sizes used in fuses
The Standard Handbook for Electrical Engineers lists the following formula:

33 * (I/A)^2 * S = log( (Tm - Ta) / (234 + Ta) + 1 )

I = current in Amperes
A = area of wire in circ. mils
S = time the current flows in seconds
Tm = melting point, C
Ta = ambient temp, C

The melting point of copper is 1083 C.
See pp. 4-74 .. 4-79 of the 13th edition of the Handbook for more info.

Here's another link that has some useful stuff on currents, fusing etc.
http://wiki.xtronics.com/index.php/Wire-Gauge_Ampacity
It tells me that a piece of copper wire of 2.9mm dia will fuse at 400A!
As someone said, might be a bit much for a bit of pcb track....

[Hemonster]

Hi Coulomb and ACmotor

With regards to the diodes, aren't these going to prevent the chargers from sensing the cell voltage? Or is that that you are just using a constant voltage current limited PSU (ie. not smart chargers), and bump up the output voltage a bit to take into account the diode drop? (0.5V for a Schottky?)

As for safety, I've read in other forums that you shouldn't tie the centre pack to the chasis as touching any part of the pack to the chasis will short out - and the breaker won't do its job?? Are there safety devices that will cause a trip by just monitoring currents going in and out of +300/-300V? without requiring a ground to chasis?

I'm thinking I would use a fuse to the main traction line, will this not suffice? Would it be better for me to have a overcurrent breaker as well?
Do you guys have fuses for the main traction line, or in each battery module?

Thanks for your input, I really appreciate it.

Hemonster

ps:
I do like the idea of the connectors, however won't this all just be easier if each battery had its own charger? Greensaver told me that the warranty on their battery is 1 year, ONLY using their "official" charger - so looks like somebody in their marketing department saw a "value adding" oppurtunity. This charger is a smart type which has various charging steps and protection built in (including a beep warning if the battery has issues) - though the good thing is these are 48V 5A chargers at USD$41.18 (15-Apr-09) so about USD$500 for 11x48V modules (I'm still debating if I should have 50 or 52 20AHr cells) - ie. that's USD$500 for a 2.6kw charger - might cause an overload at the AC plug but for these cells I'm looking at, that's a 4-5 hour charge period using "their" smart charger.

[acmotor]

Hemonster, I use CV supply adjusted to allow for diode.
I also use a series resistance (1 ohm) to give a tapered off charge to 13.8V and allow at least 12 hours for the last of the charge to occur and battery eq. to work. IMHO some smart chargers kill SLA batteries by being impatient. Lead acid takes time to charge - simple reality and some chargers just try to charge in minimum time. Using the manufacturer's own charger is good idea for warranty though.

Diode is for safety from feedback in the case of fault (as is fuse) and means the multipole connector is not live when in charge mode.
Diode would reduce the voltage at battery from a smart charger although this would only be 0.1-0.15V per battery on 48V charge so may not worry
the final charge (it is typically warm in Oz and unless the charger reads the actual battery temperature then the voltage will be out by -10mV/°C anyway)
This is just one slant on the connection method.

If you use multiple chargers then just leave them connected (via fuses in both lines at the batteries). Assuming the chargers have floating outputs and you have space onbard vehicle for them.

Fuses....
Both on traction lines (-,+) to controller and on all charger connections. Fuse at at least each location of the battery pack sections i.e. bonnet and boot.
For an overcurrent breaker to trip means controller has failed, so you won't just be resetting the breaker ! The time for breaker to trip is likely to be a lot longer than a fuse to blow ? and a 600V 100A DC fuse is likely to be a lot cheaper than a 600V 100A DC circuit breaker. (remember AC CB is not suitable). You need more than one fuse anyway.


Re the greensaver chargers....
15A socket may be required for 2.6kW although if power factor is high >0.9 then you may well get away with 10A socket as the 2.6kW is probably not for the whole charge time. Anyway, Greensaver, or you could wind the charge current back a bit to fit into 10A (2.4kVA).

Grounded chassis ? mixed bag. AS3000 says SLV (separeted low voltage) is not to be used above 500V. What ever you do, some form of chassis (ground,earth) leakage detection should be used. If you've complied with other wiring standards (and also broken pack up with contactors to safe voltages)then you won't be able to 'touch' anything live anyway.
There are other threads on this forum on this topic.


edit: fuses

[Hemonster]

Hi acmotor,

Thanks for that reply.

Fuses....
Both on traction lines (-,+) to controller and on all charger connections. Fuse at at least each location of the battery pack sections i.e. bonnet and boot.
For an overcurrent breaker to trip means controller has failed, so you won't just be resetting the breaker ! The time for breaker to trip is likely to be a lot longer than a fuse to blow ? and a 600V 100A DC fuse is likely to be a lot cheaper than a 600V 100A DC circuit breaker. (remember AC CB is not suitable). You need more than one fuse anyway.

A fuse for each 48V module is probably excessive and may become more of a nightmare for fault finding, however might be safer when considering inadvertant shorts to the chasis from within the larger battery pack. Should I really be concerned about this? or just ensure that this can't really happen within the pack?

A high impedance leakage detection system can detect one short, but can't really tell if there is a 2nd short or not (or can it?). I was thinking if there was a method for using several high impedance leakage detectors to identify more than one short and also pin point better where the short is occuring - now that would be a smart craker

I was thinking that each chasis connection would be via (say) 100K resistor (rated at 2x full pack voltage) which feeds into the diode of an opto (with a C across it to filter out some noise). The output of the opto on the other side is amplified and fed into a micro. This is replicated elsewhere in the pack, except the opto diode is reverse so you can tell which of the optos are conducting and in what direction and what magnitude. Perhaps from this information some clever code can work out what region has been affected and if it is severe enough to warn the user or stop or limp the car. Sorry ... this is probably a bit off topic, and is just a wild idea at the moment. Need to think more.

[coulomb]

A high impedance leakage detection system can detect one short, but can't really tell if there is a 2nd short or not (or can it?).

The idea is to fix the first short before the second one happens.

Perhaps you mean you want an easy way to pinpoint the leakage/short. An interesting problem. If you have breakup relays, you could use those to isolate the problem. Though of course they are normally in parallel, so perhaps a push button per group (not accessible without tools) could be used to force a breakout relay on. You'd have to start with the relays nearest the detector, but to test the last group of a half string, you'd have to turn on all the relays up to that one. So I guess they need to be toggle switches. When you turn on a group and the detector fires, you know it's that group.

[Hemonster]

The idea is to fix the first short before the second one happens.

However the problem as I see it is that when you get the warning, you don't know if it was one short or two shorts - ie. do you stay in the car and drive home and put "check electrics" on your todo list, or get to the kerb ASAP NOW! stop the car and break the contactors and call the tow truck?

Perhaps you mean you want an easy way to pinpoint the leakage/short. An interesting problem. If you have breakup relays, you could use those to isolate the problem. Though of course they are normally in parallel, so perhaps a push button per group (not accessible without tools) could be used to force a breakout relay on. You'd have to start with the relays nearest the detector, but to test the last group of a half string, you'd have to turn on all the relays up to that one. So I guess they need to be toggle switches. When you turn on a group and the detector fires, you know it's that group.

This sounds manual and could even be a bit dangerous. What I was thinking was an "online" method for determining where the pack might be faulty without touching the pack. Say for instance you had a ground leakage resistor on each of the 48V modules, say at the +ve terminal. Then if a short were to happen say somewhere in the pack, all the resistor currents would register something based on how far they each are from the fault, the ones closest will register the least current, the ones furthest would register a higher current (higher potential)(ignoring my opto idea for now). If the short happenned at exactly one of the 48V +ve terminal, that sensor would register no current where as its brothers/sisters would register a current - bingo you know exactly where the fault it. If it is within the 48V pack, then you can tell from the average currents which 48V pack is the offending one. This system can also tell you if more than one short has occurred by looking at the profile of currents.

The main contactor isolates the battery pack from the inverter, so you can also tell if the problem is in the battery or in the traction circuit.

That's the theory anyway ... practically it might be expensive to put a ground leakage sensing circuit on each 48V module. However if I was desining a 48V BMS, I might choose to incorporate this functionality, and have it as one of the faults that drive the opto isolated interrupt/warning line (that is used for OV and UV over temperature measurement per cell).

Still, it could do with a bit more thinking

[weber]

However the problem as I see it is that when you get the warning, you don't know if it was one short or two shorts - ie. do you stay in the car and drive home and put "check electrics" on your todo list, or get to the kerb ASAP NOW! stop the car and break the contactors and call the tow truck?

Neither.

After the first fault happens the idea is to have a bloody annoying alarm beeping at you the whole time the contactors are on, so you don't need to write it on your to do list. You should pull over at your leisure to check for smoke when it is safe to do so, but the chance of a second fault at a different voltage point at the same time is _very_ remote, unless something catastrophic happened, like a collision, in which case you should certainly turn off your contactors and you probably won't be driving home anyway.

There are two bad things that can happen with a second fault:
1. Lethal electric shock
2. Fire

1. The only way to get a shock is if the second fault is you! You would need to touch the vehicle chassis and a conductor that is at a different voltage. These should all be insulated and inaccessible without using tools, and not in the cabin. Obviously you shouldn't go poking around your electrics with the contactors on and the alarm going.

2. Most second faults involving metal to metal contact (as opposed to metal to person to metal) should result in blowing a fuse. But yes, it's possible that one of the two faults is of medium resistance and is causing local heating but is not enough to blow the fuse, so do pull over some time, no hurry, and lift the bonnet and open the boot and walk around looking and smelling for smoke (but not touching).

Then make up your own mind whether to drive home.