Knuckles
10 kW
Maybe something like this ...
Horizontal Version
Vertical Version
Horizontal Version
Vertical Version
Fechter's Capacitor Coupled Cell Balancer
- Thread starter fechter
- Start date
Knuckles
10 kW
I got the idea from this source ...
http://www.celnav.de/hv/hv9.htm
(scroll thru)
I expect a three phase 'generator' at 20-50 kHz would be good. At higher frequencies the system is more efficient and smaller caps will work for a given power input. Individual battery cells act as capacitors so the caps adjacent to the bats probably are not needed (but can't hurt). The bats must remain connected at all times while the phase generator is 'running'. Cheap (general purpose) 5-10 volt electrolytic capacitors should be fine.
A high kv outrunner (or inrunner) would be a great 'mechanical' three phase generator.
There might be a potential for LVC and HVC control here also.
And a 'chainless' drive system and an end to 'fake pedaling'.
http://www.celnav.de/hv/hv9.htm
(scroll thru)
I expect a three phase 'generator' at 20-50 kHz would be good. At higher frequencies the system is more efficient and smaller caps will work for a given power input. Individual battery cells act as capacitors so the caps adjacent to the bats probably are not needed (but can't hurt). The bats must remain connected at all times while the phase generator is 'running'. Cheap (general purpose) 5-10 volt electrolytic capacitors should be fine.
A high kv outrunner (or inrunner) would be a great 'mechanical' three phase generator.
There might be a potential for LVC and HVC control here also.
And a 'chainless' drive system and an end to 'fake pedaling'.
Look who's back...
The problem with that arrangement is the cell furthest from the AC source has to pass through all the series capacitors and will have lower current than the ones closer. If the capacitors in series were big enough, the total ESR might be low enough for it to work, but I think you'd still have uneven current distribution in a long string.
I'm not sure if using a 3 phase setup will be worth the extra components, but you could probably push quite a bit more current. You could use the same series arrangement single phase. If there was a 'smart' element between the rectifier and the cell to regulate the voltage, it could compensate for the uneven supply (probably needed anyway).
The problem with that arrangement is the cell furthest from the AC source has to pass through all the series capacitors and will have lower current than the ones closer. If the capacitors in series were big enough, the total ESR might be low enough for it to work, but I think you'd still have uneven current distribution in a long string.
I'm not sure if using a 3 phase setup will be worth the extra components, but you could probably push quite a bit more current. You could use the same series arrangement single phase. If there was a 'smart' element between the rectifier and the cell to regulate the voltage, it could compensate for the uneven supply (probably needed anyway).
docnjoj
1 GW
Welcome back, stranger!
otherDoc
otherDoc
I have been thinking about this. I need to re-read this thread. But I was thinking how one might use isolated dc-dc supplies to pass current from a Hi cell to a low one. I would like to think of a bms that can be turned on when ever you ride the bike and when one cell (or string of cells in parallel) get higher then the average it starts taking power from them to the others. If you can prevent one cell from dropping before the rest on the go it would help, well it would help me with my crazy experiments.
Arlo1 said:
I've seen some setups like that and it does have some merit. Parts count becomes sort of a killer again.
I worked on a simplified version of this a while back. Each cell has a little transformer that takes excess charge and boosts it to pack voltage and feeds back into the main pack terminals during charge. The transformer takes the place of the shunt resistor in a typical setup. This allows very high shunt current without generating a lot of heat. I found some nice transformers that were about $2ea in quantity. These could easily handle about 3A shunt current.
The TI PowerPump arrangement is similar but shuttles charge to the adjacent cells only. In more complex arrangements, there is a 'share bus' that high cells dump into and low cells draw off of. For sure this requires microprocessor control. This would allow for full time balancing and help out weaker cells during discharge. Very elegant, but way too complex for me.
I just thought of another SIMPLE B.M.S. I think you could use mosfets hooked to each Cell that when turned on would connect the cell to a cap bank.
So it would be this simple. You pulse the mosfets on at different times. All feeding or pulling from the same cap bank. If all fets turned on at once You would blow it all because it would connect all cells in parallel. But if pulsed at different times it would work.
So it would be this simple. You pulse the mosfets on at different times. All feeding or pulling from the same cap bank. If all fets turned on at once You would blow it all because it would connect all cells in parallel. But if pulsed at different times it would work.
You'd want fuses on each cell or cell group, though, because FETs tend to fail shorted, so if one did fail for any reason then you switch to another one, you'd then be shorting *at least* two groups together. 
The fuse would prevent this from being a serious problem, especially if your system has a monitor to detect that a group is no longer connected and stops the process, and alerts you.
The fuse would prevent this from being a serious problem, especially if your system has a monitor to detect that a group is no longer connected and stops the process, and alerts you.
Arlo1 said:
The FETs would need to be controlled by some kind of smart software to shuttle the charge around. I think the FETs would all need to be rated for full pack voltage since one side of them would always be connected to a cell and you would need two per cell, maybe 4 due to the body diodes. It gets complicated to drive the FETs from a common microprocessor. For pulse only operation, you might get away with coupling capacitors for the gates.
