In the beginning…
It’s been in the back of my mind for a while now to upgrade my scooter to lithium, and I’ve been watching the prices of lithium cells gradually fall for a year or two now. But having noticed a slight deterioration in the performance of my lead-acid batteries over the winter I realised that after 2 years and about 2000 miles of riding, it would not be long before I would be looking for some form of battery replacement.
Replacing like-for-like is the cheapest solution in the short term, but is a tad short-sighted as it’s generally accepted that lithium batteries can last up to 6 times as long as their lead-acid counterparts, but cost nowhere near 6 times as much to switch to (probably in the region of 2-3 times the cost of lead-acid). There are many advantages to lithium over lead-acid, not least their tendency to retain a large portion of their voltage until they are nearly exhausted, for the voltage not to “sag” when drawing large amounts of current, and their lighter weight. The drawback is that lithium cells are very sensitive to being overcharged or overdrained, meaning that a Battery Management System (BMS) has to be installed to ensure that any cells that reach their maximum voltage are not charged any further whilst the other cells continue to charge (balancing), and that if any cell reaches a preset minimum you can be warned, and it can even cut power to the controller to ensure that no further discharge occurs. This obviously adds to the cost but can still bring the whole system in at about 3 times the cost of replacement lead-acids, depending on which BMS you choose.
With this in mind I gradually put together a plan to perform the upgrade; I can fit approximately 70 or so volts’ worth of cells in the original battery box (thus regaining the underseat space I lost with the 72V upgrade), but I wanted to maybe gain a little extra performance at the same time. I realised from a previous attempt to remove the original 4 batteries from the battery box that this was not going to be possible without some serious “modification” of the box as it appears when it was welded to the frame it distorted, trapping the batteries in there for good. With this in mind I decided that a slightly larger box could be accommodated within the frame, and that this could be dimensioned to suit 50 lithium cells, giving me 80 volts! Add to this the weight saving (about 20Kg of lithium vs. 47Kg of lead-acid) and you can see that I may gain more than a little performance 🙂
As for a BMS, I looked at some of the “off-the-shelf” options and was not really taken with any of them. Then one evening I decided to register on the Battery Vehicle Society forum, and discovered that a few of the members on there had been developing a “home-grown” solution based on microcontrollers. This idea immediately appealed to me as I am a Software Engineer by trade and always enjoyed doing what we call “embedded work”. So I set about ordering some bare PCBs from the guy that originated the project, and as it happened he had a 25-cell “slave” board which is perfect for my needs (I’m going to arrange the 50 lithium cells as 25 “supercells” of 2 cells in parallel, which will give me a capacity of 24AH at 80V (not a lot less than the 28AH of lead-acid I currently run). Needless to say this is going to increase the cost of the project as it not only performs the function of a BMS, but also monitors and displays many other parameters (current, temperature etc.). You can even connect the pulse output from the controller to show an accurate speed reading, and attach a radio-controlled remote display so that you can monitor the charging process indoors!
Sourcing the parts to populate the boards proved to be a bit of a headache as this project has been a “work-in-progress” for about 3 years now and has undergone a few revisions, and no-one has produced a definitive up-to-date parts list. However with a bit of head-scratching and perseverence all the correct bits were identified and ordered, and within a few evenings of receiving them I had the “master” board (the one that does all the clever stuff) and 2 of the 25 circuits on the slave board populated and running. One thing I did save some money on is the display. The system is designed to drive it in a TV-like way (composite video), and most people use the LCD monitors designed for reversing cameras in cars. One of these would be way too large for me to fit anywhere on the scooter, so I came up with the idea of adapting an old Casio pocket TV that has been no use since they switched off the analogue TV transmissions. It already has a composite video input, and I reckon the display will fit nicely into the dash of the scooter where the (soon to be redundant) battery meter currently sits. For more information on the BMS, see here.
One of the problems with this “home-grown” approach is that in order to customise it to your own needs you need to be able to edit the software and reprogram the chip. It’s written in PIC Basic Pro, but having looked at the price of this compiler (and not being a great fan of the Basic programming language anyway), I decided that I would attempt to rewrite the Master code in C.
Fast-forward a few weeks and the C rewrite is almost complete, the slave board is completely populated and working, the lithium cells have arrived and been assembled into a pack, and work has started in earnest on the actual conversion.
Getting my hands dirty – the build
The first thing to do was to strip the bodywork off the scooter and attempt to get the old batteries out. This was not so easy because of the aforementioned “Chinese welding” job which had distorted the top of the battery box inwards, trapping the batteries in the box. This was soon remedied with judicious use of the hacksaw, halfway across the bracket. This meant that I could double the bracket up and drill and bolt it back together, with the side of the box in its intended place.
Batteries removed, I then realised that the box had a “front” which would need to be bent down to accommodate the longer battery pack that I was installing, which would also mean the pack could slide in from the front as the seat-support crossmember restricts access to the battery box from the top. A couple more hacksaw cuts of the spot-welds at the top of the front panel, and a bit of brute-force bending later, and I had a compartment that looked something like this:
Next thing was to see if the pack was actually going to fit in there – it was a tight squeeze under that crossmember, but it went in!
I’d realised for a while that in order to accommodate the extra length of the pack I’d need to cut some of the plastic and have it poking through, and make up a “power bulge” to cover it up. What I didn’t know was how far it would protrude, so I was pleasantly surprised with this:
Of course now it had to come back out again, as I still needed to connect all the wires from the BMS slave board, a job I was not looking forward to.
Delaying the inevitable, I moved on to charger wiring; I had purchased a couple of the highly-recommended Meanwell 48V power supplies which have adjustable voltage, and can be modified for “constant current”, a requirement for a lithium charger. I had done the modifications a couple of weeks ago and attempted to adjust the “series” voltage down to 92V which I reckoned would be about right for the number of cells involved. Unfortunately I could only go down to 93V but I’m hoping the extra 0.04 volts per cell won’t make that much difference! I’d already tested the current limit modification by attaching electric fires, power tools etc. to produce a load, and then turned the current down until it had just started limiting – I’ll possibly adjust this up a bit when I actually have the supplies connected to the battery.
I need a 12V supply to power the master board during charging as I obviously won’t have the ignition on which will normally power it, so I’ve made up a little board which allows me to have 2 separate 12V supplies plus a line that will be shorted to ground when the charger is plugged in so that the processor knows it’s in “charge” mode. This board also has a 6V regulator on it to provide a supply for my pocket TV for when it’s installed in the dash 🙂 The charger 12V supply will come from my scooter’s original 12V DC converter which is connected across one of the 48V Meanwells.
One of my dilemmas was how to connect the current sensor – this needs to be “in line” in the main 80V wiring. Luckily Greg Fordyce has just produced a few PCBs just for this purpose, one of which he kindly sold me. So the sensor is now firmly soldered to the PCB, along with 2 stonking great 170A cables 🙂
I’ve decided that the slave board will actually be taking up some of my underseat space after all, just in case I need to disconnect it in a hurry. This will only occupy about an inch in the bottom of the space so I’m going to make up a “false bottom” that will sit over the top of the board so that I can still stow my essential wet-weather gear and towel to wipe off the seat 🙂
I’m going to mount the master board and the little 12V/6V daughter board mentioned above in the compartment under the footplate which was presumably originally for the battery on a petrol scooter. I think there’s just enough headrooom to mount it on the 2 little pillars that are already there and drilled to accept self-tapping screws. If not there may be some more plastic-cutting going on!
I finally plucked up the courage to connect up the BMS wires to the battery pack and all was going extremely smoothly until I was 3 wires from finished when I just touched the connector with my meter probe, and there was a blinding flash and a puff of smoke, and this was the result:
I’m still not sure how it happened – I did have the negative battery terminal connected to the shell of the connector (but not any more – it’s out on a separate connection now!), so whether I accidentally shorted one of the pins (possibly about 74V) to the shell with my meter probe (which you can see is now a little worse for wear). Luckily, that connector was only on there for testing purposes so it was expendable, and miraculously a peek inside the important connector housing showed no apparent trauma at all (also since borne out by actually plugging the slave board into that connector, which worked perfectly! 🙂 ).
The next job is to install the battery pack back in the scooter and find a suitable mounting point for the current sensor, and then I’ll be pretty much ready to switch it on and test if it runs!
That’s pretty much the state of play at the moment and although the job is nowhere near finished yet, I’m going to post this now as I’ve been writing it over a period of a couple of months and it could end up like War & Peace!