heathyoung
100 kW
Analysis paralysis time again (oh dear). 
(for those who are unfamiliar with the term - http://en.wikipedia.org/wiki/Analysis_paralysis )
After coming from Konions that required very little in the way of mollycoddling, I'm moving to something far lighter and more powerful for a GNG build on a downhill frame.
The thing that annoys me the most about Lipo is the sheer number of wires and connectors required when you start parallelling packs - it all takes up a lot of room, and looks like a rats nest. It also needs HVC, LVC and in all probablility some balancing.
OK. So - looking around, there really isn't that much in the way of what I want - a PCB that fits over the top of all of the packs, that allows paralleling at the balance tap and pack level, and provides a LVC and HVC function, with balancing, and takes up bugger-all room. The packs also need to be removable/replaceble quickly - so it can be swapped from 10-15-20AH with no great effort. And not require another 500g of cable to do it!
So I designed one. Its called analysis paralysis BMS, or APBMS for short
I'm using PCB mounted right-angle anderson connectors to parallel at the pack level (along with the usual JST-XH connectors) spaced equidistantly across the PCB (so all the wires are the same length, no guitar strings) and a bog standard TL431/TC54 setup for HVC/LVC and balancing. Balancing current is pretty low, but adequate for Lipo thats not really out of balance. 95% of the components are surface mount, so its slim - only 5mm high on the top, and there is nothing to get squashed, shorted or snapped off - I really treat my gear like crap so it needs to be solid!
Its also open source, help yourself. Single sided, so no trickey vias or through-hole plating to get expensive. Parts come to a princely ~$32 (not including PCB) without anderson connectors (bloody expensive - $1.34 each X 16) For those who don't trust andersons (and want a permanent pack) you could just solder the battery leads to the board.
Its not stupidly small SMD either, I used 0805's for resistors and SOT-23's. PDF's are actual size, so you can print these onto transparency film, or toner transfer film to make your boards.
I havn't built this thing yet - so check the design carefully, and watch pinouts on the packages.
BOM (sort of)
ZTL431 SOT-23 X 12 - (0.40 ea) (Farnell 1225191) $2.40
40R 1206 X 12 (0.008 ea) = 0.10 - EDIT - Now 24 X 1210 120R (330mW)
22K 0805 X 12 (0.008 ea) = 0.10
33K 0805 X 12 (0.008 ea) = 0.10
1K 0805 X 24 (0.008 ea) = 0.20
TC54 2.9V SOT-23 X 12 (0.43 ea) (Farnell 1852223) = 2.58
Kingbright KB817-B X 24 (0.29 ea) (Farnell 2001651) = 3.48
Anderson 1336G1 - 90 degree contact $1.34, 16 of (farnell 1601683)
JST-XH 7 Pin Right-angle header X 8 $1.00 each (farnell 1516295) OR
JST-XH 7 Pin Vertical header X 8 $1.00 each (farnell 1516283)
In true AP style, this will probably get updated again... and again... and... hey I wonder if a cell-log should be incorporated into this?
(for those who are unfamiliar with the term - http://en.wikipedia.org/wiki/Analysis_paralysis )After coming from Konions that required very little in the way of mollycoddling, I'm moving to something far lighter and more powerful for a GNG build on a downhill frame.
The thing that annoys me the most about Lipo is the sheer number of wires and connectors required when you start parallelling packs - it all takes up a lot of room, and looks like a rats nest. It also needs HVC, LVC and in all probablility some balancing.
OK. So - looking around, there really isn't that much in the way of what I want - a PCB that fits over the top of all of the packs, that allows paralleling at the balance tap and pack level, and provides a LVC and HVC function, with balancing, and takes up bugger-all room. The packs also need to be removable/replaceble quickly - so it can be swapped from 10-15-20AH with no great effort. And not require another 500g of cable to do it!
So I designed one. Its called analysis paralysis BMS, or APBMS for short
I'm using PCB mounted right-angle anderson connectors to parallel at the pack level (along with the usual JST-XH connectors) spaced equidistantly across the PCB (so all the wires are the same length, no guitar strings) and a bog standard TL431/TC54 setup for HVC/LVC and balancing. Balancing current is pretty low, but adequate for Lipo thats not really out of balance. 95% of the components are surface mount, so its slim - only 5mm high on the top, and there is nothing to get squashed, shorted or snapped off - I really treat my gear like crap so it needs to be solid!
Its also open source, help yourself. Single sided, so no trickey vias or through-hole plating to get expensive. Parts come to a princely ~$32 (not including PCB) without anderson connectors (bloody expensive - $1.34 each X 16) For those who don't trust andersons (and want a permanent pack) you could just solder the battery leads to the board.
Its not stupidly small SMD either, I used 0805's for resistors and SOT-23's. PDF's are actual size, so you can print these onto transparency film, or toner transfer film to make your boards.
I havn't built this thing yet - so check the design carefully, and watch pinouts on the packages.
BOM (sort of)
ZTL431 SOT-23 X 12 - (0.40 ea) (Farnell 1225191) $2.40
40R 1206 X 12 (0.008 ea) = 0.10 - EDIT - Now 24 X 1210 120R (330mW)
22K 0805 X 12 (0.008 ea) = 0.10
33K 0805 X 12 (0.008 ea) = 0.10
1K 0805 X 24 (0.008 ea) = 0.20
TC54 2.9V SOT-23 X 12 (0.43 ea) (Farnell 1852223) = 2.58
Kingbright KB817-B X 24 (0.29 ea) (Farnell 2001651) = 3.48
Anderson 1336G1 - 90 degree contact $1.34, 16 of (farnell 1601683)
JST-XH 7 Pin Right-angle header X 8 $1.00 each (farnell 1516295) OR
JST-XH 7 Pin Vertical header X 8 $1.00 each (farnell 1516283)
In true AP style, this will probably get updated again... and again... and... hey I wonder if a cell-log should be incorporated into this?
