I've been procrastinating on writing this, but my "economy" conversion of a e-max sport to LiFePO4 power, using Thundersky LFP40AHA cells is now complete.
In the US/Canada, the cells are available from Electric Motorsport (EM) of Oakland, California (http://www.electricmotorsport.com/store/index.php and in limited quantities from Patrick Rentsch residing in the same area. Please PM me for info on purchasing from Patrick. EM sells them for $100 each and presumably has a warranty. Patrick sells them for a flat $87 each, including shipping, but all sales are final. He does, however, charge and discharge the cells before shipping.
The main issues for the conversion are physically fitting the cells in the scooter, and finding a suitable battery management system. You will want to upgrade the battery monitoring instrumentation a bit to protect your investment - as over-discharge is the most effective way to damage a LiFePO4 cell.
1. PHYSICAL INSTALLATION
As far as fitting sixteen cells in the scooter. I'll let this picture do most of the explaining:
The battery box did need some minor widening, (using a small jack and wood blocks) to accommodate this cell arrangement. Also, you may notice that the DC-DC converter was relocated to the frontmost battery box clamped on to the front edge using a small aluminum plate. This was needed because at it's original location, there were two protruding mounting nuts which interfered with this battery arrangement.
Here's the installation with all the connections and battery management leads installed:
The positive terminal of the pack is the cell at the left end of the front row of five; the negative terminal is the front-most cell.
The Thundersky cells are taller than the stock batteries, but I was pleasantly surprised to find that no raising of the seat well was required - but I did have to cut the protrusions off where the front mounting studs/nuts were used. Also, holes had to be cut where the seat and rider weight could potentially press down on the right and left front-most battery terminal bolts.
The battery hold-down bar is not used. Attaching the seat with the two rear bolts plus the front screw near the seat hinge works fine.
It is important to secure the batteries - they must not bounce or rattle in the battery box in use. The fit in the box is already pretty snug for all cells, but for additional assurance, I injected expanding builders-type expanding foam filler between the battery box and the foam ribber padding on the sides of the box. This foam secures the cells while still (hopefully) allowing extraction and replacement of individual cells. Note that Thundersky recommends clamping the cells together to resist swelling in use. This largely applies to the LCP (LiCoO2) cells, not the LFP cells as LiCoO2 cathodes expand and contract with charging/discharge, but LiFePO4 cathodes exhibit much less volume change.
The Thundersky cells are sealed and non-spillable, but use a liquid electrolyte with some air-filled head space. For this reason, I believe they should be installed as close to vertical as possible.
The older members of this board will recall that I'm running my e-max;s in switchable 60 volt mode using an extra 20AH SLA under the seat as a "booster" battery when a top speed of 45 mph/75kph. So, the real battery box that holds the single SLA in the stock setup now holds one of the now-two separately charges booster batteries. They are still SLA's for now, as this is an "economy" setup, and no 60 volt (20 cell) battery management systems are available. But one may be available soon.
If a 20 cell/60 volt system is desired, there is room for possibly two cells laying on their side in the rear-single battery box, and one more cell in the rearmost battery box. But, I'd recommend instead sacrificing some storage space and putting all four additional cells in the seat well. They can be secured with large-diameter type hose clamps.
2. BATTERY MANAGEMENT AND CHARGING
One of biggest problems was finding a suitable battery management system (BMS) that would be affordable, and compatible with my stock e-max chargers, which I was wanted to continue to use as an economy measure. For the e-max, we only need two thing from a BMS - a means of preventing any single cell being discharged below 2.5 volts, and a means of balancing the cells at about 2.65 to 2.75volts per cell. Many battery management system have too many features for our purposes. Particularly, we don't need a current limit cutoff that the BMS's have on these packs being sold for e-bicycles - which have a limit too low anyway.
I had originally hinged my whole project upgrade on the Bob Mcree designed charge-management and low-voltage cutoff board - only to see the whole project fall through. (go to the endless sphere forum - "new 16-cell BMS" thread) So, I ended up spending more money than I'd like on the the low-voltage cutoff 9LVC) board and less than optimal dual TP210 aeromodeler pack balancers sold by Gary Goodrum here:
The LVC board was mounted to a hardboard cover over the hole in the bottom of the seat well, and the balancers plugged into the connector, as shown here:
The LVC is primarily designed for e-bikes, not larger scooters, so the LVC's signal wires are connected a bit differently. They are connected to the throttle leads with a pull-up resistor added to the throttle signal wire so if the LVC detects a low cell, the throttle signal get grounded out, shutting the motor down. In practice, this will take the form of a studdering or stumble - just like the LVC built into the controller.
The stock e-max charger has a switch-to-CV charging voltage of 59.5 volts. The cutoff current for end-of CV charging is adjustable via a 10-turn pot on the charger control board. I turned it down from about 500 mv to about 250 mv. At 59.5 volts, this charges each cell to 3.72 volts if perfectly balanced. The dual TP210's don't do a perfect job - but generally keep them between 3.70 and 3.76 volts - good enough. The TP210's can be left kept connected full time, but since that are a dissipative type of balancer - bringing the high cell down to the low cell. I prefer to use them only during charging. This means that charging is no longer quite the plug-and-forget process that it was, since you need to unplug everything fairly soon after charging is finished.
The LiFePO4 cells hold a higher end-of-charge voltage than lead acid, so the float stage of the charger never gets actuated.
Thundersky recommends 4.25 volts per cell, but the consensus is that for maximum life, a LiFePO4 cell should be charged to no more than 3.65 to 3.75 volts. During charging, once the voltage reaches 3.65 volts, it shoots up exponentially, so the amount of extra charge capacity between 3.65 and 4,25 volts is only a few percent.
I am using the 2.1 volt version of Gary's low-voltage cutoff board - the only other readily available version being a 2.7 volt version. The recommended minimum voltage for Thundersky cells us 2.5 volts, but this goes down to 1.5 volts at -35C. So interpolating, that's 2.2 volts at 10C.
While a pack monitor, like a Pak-Trakr or Cycle Analyist could be used, I opted to just use a self-powered digital voltmeter for economy. If the voltage is sagging to 42 volts under load, I better be close to home.
The performance gain from the cells is impressive. Even with my two SLA booster cells, the scooter weighs about 40 pounds/20kg less than the old setup. Range is increased from about 20 mi/32 km with new SLA's to at least 33 mi/53 km. The pack provided much more consistent voltage during discharge, only sagging below 48 volts when 85% discharged. All the riding so far has been done in summer weather, however.
There is no requirement that LiFePO4's be kept fully charged, so the pack can be treated almost like a fuel tank. The general thinking is that liFePO4's will have the longest calender life if kept partially discharged.