I converted a 1975 Triumph Spitfire to a fully electric vehicle in 2014. The project is described here, including the specifications of the motor, inverter and other components. And a word of warning: this project changed my life infinitely for the better. Be careful what you get up to in your spare time!
https://www.linkedin.com/pulse/e-fire-triumph-spitfire-ev-paul-martin
After the E-Fire was destroyed, I dropped its electric drivetrain into a 1973 Triumph TR-6, converting it to a fully electric car, as my COVID project in 2020. The project is described in this article:
https://www.linkedin.com/pulse/er-6-electrifying-conversion-paul-martin
This article gives considerable detail about my recent battery pack upgrade on the ER-6 project.
After 11 years, the lithium iron phosphate (LFP) battery pack from the E-Fire was starting to show its age. Many cells were below 80% of original capacity, and several failed outright. The range of the car was therefore reduced by quite a lot- from a low starting number. The pack also had considerable voltage sag in particular cells during hard (650A) acceleration, resulting in nuisance battery management system (BMS) alarms while driving. The older Sinopoly cells I used in that build had a design flaw: they used a plastic case which was fitted with a relief valve rather than a rupture disk. Over time, the electrolyte- a solution of LiPF6 in ethylene carbonate, dimethyl carbonate and other proprietary additives, does suffer from some solvent loss to evaporation through these relief valves, which are quite flimsy things which don’t seal tightly. The trunk (boot) of the car always had a vague smell of these solvents. Modern LFP cells have metal cases and are fitted with rupture disks, not relief valves, to protect against venting caused by a short circuit or damage to the cell.
For some time I’d been buying the new generation of LFP cells directly from China via Alibaba. The first order took some effort and research to ensure that we were getting adequate value for money and weren’t being taken advantage of by the seller, but after buying cells for the solar installation at my farm and for my friend’s huge home battery system, I eventually became comfortable that I knew what was going on. An Austrian living in Australia and his YouTube Channel “Off Grid Garage” was also of particular help.
Sadly, the original cells we were buying, though inexpensive, had specifications adequate for solar installations of limited power, but had too low a “C rate” for use in the car project. The C rate is the maximum current draw (or current supply for charging) that the battery can handle, expressed in amps per amp-hour, i.e. in hours^-1. A C rate of 1 means that the cell can be safely discharged from 100% to 0% state of charge (SOC) in 1 hour. For the 280 Ah cells we were buying originally, the C rate for discharge was 0.5 C (2 hrs or 140 A) continuous and 1 C (1 hour or 280 A) for short periods. Since the inverter was capable of drawing 650 A- and that extra current makes the car fun to drive, there was no point in using these initial cells. Cells with higher C rate were available, but tended to be quite expensive.
That all changed however in 2024. LFP cells popular with the solar community, made by manufacturer EVE, were finally made available with an English language specification sheet showing the following C rate data:
EVE LF304 LFP Prismatic Cell Data
Standard charging current: 152 A (0.5C)
Maximum continuous charging current: 304 A (1C)
Maximum instantaneous charging current: 2C, 30 s, <80% SOC, 25+/- 2 degrees C
Standard discharge current: 152 A (0.5 C)
Maximum continuous discharge current: 304 A (1 C)
Maximum instantaneous discharge current: 3C (912 A), 30 sec., > 20% SOC, 25+/- 2 degrees C
Initial internal resistance: 0.16 milliohms at 1 kHz, 15-40% SOC
DC internal resistance: 1.2 milliohms, 50% SOC, 1C, 10 seconds
Cycle life: 4000 cycles at 25 C 2000 cycles at 45 C
The car’s inverter is capable of drawing only 650A for short periods- a 30 S discharge at 650A would have the car up to its top speed, at a motor RPM where current draw falls off due to back EMF. So these cells can definitely handle, in theory, “fun” currents for fast acceleration. In fact a more detailed table in the spec sheet shows that the peak discharge is 2.12C or greater even down to -5 C temperature as long as the SOC is greater than 30%. That’s the coldest weather this 2 seater convertible will ever be driven in- in fact as soon as road temperatures threaten to be frosty, the local road authorities dump salt on them- and this car goes away for the winter, because British cars are soluble in saltwater!
At highway speeds, current draw is around 150 A, so right about at the “normal” discharge rate for these cells, much less the maximum continuous discharge rating of 304 A.
The inverter is capable of generating 200 A of regenerative braking current, so that too is well within the capabilities of the cell. And no charger I can afford will ever charge the car at 152 A, much less 304 A!
So the specs looked good- time to negotiate!
I ordered 36 cells- 33 for the pack replacement and 3 spares. Wise move as it turns out!
New (EVE LF304, left) versus old (Sinopoly 180 Ah, right) cells, for comparison purposes.
Each cell holds about 1 kWh (972 Wh), and costs for the cells were on the order of $60 USD/kWh. Shipping door to door, taxes (including Canada’s 13% HST), and the small (~3%) duty on energy storage batteries from China, added another $21/kWh. A whole pack replacement for $2675 USD. 33 kWh this time. My previous pack cost around $4000 USD for 18.5 kWh…that’s a pretty dramatic reduction, from $216 to $81/kWh, between 2013 and 2024. And the prices haven’t stopped falling yet.
Better still, the new cells have a higher energy density per unit mass and volume:
Volumetric energy density: old cells 164 Wh/L vs 395 Wh/L for the new cells- a 2.4x improvement
Mass energy density: old cells 110 Wh/kg vs 183 Wh/kg for the new cells- a 1.66x improvement
The mass of the old and new cells were nearly identical- 5.3 kg each, so the new larger pack would not change the weight or weight distribution of the car in any respect except one- it would allow the centre of gravity to be lowered slightly, because the new cells are shorter than the old ones.
For comparison, Tesla 2170 NCA cylindrical cells are ~ 711 Wh/L and 247 Wh/kg- but remember that cylindrical cells don’t pack to the same volumetric density as prismatics do- and are way sketchier from a safety perspective for a DIYer to handle.
Turns out I had to wait a long time for delivery of my cells- much longer than the normal 5 weeks or so. They were ordered in mid December, 2024, and weren’t received until March 24th. A combination of freight forwarder incompetence and congestion in the Canadian ports led to a very long wait, and with Alibaba you pay 100% up front, so there were some nerve-wracking days while I wondered whether or not the order had disappeared. Although they did show up undamaged, sadly the late delivery ate much of my schedule window to install the pack before other tasks ate my life. So the project slipped a bit onto the back burner.
All the cells were capacity tested using a very practical method: charge at 40A to 3.65 V, then wait 10 minutes, then discharge to 2.5V at 40A, then recharge a bit for storage. Why 40A? Because that’s all my friend’s tester can do. It took weeks to test all those cells, old and new! All cells had at least 304 Ah capacity when tested that way. The usual soak at 0.05C (15A) to get the cells with certainty to 100% SOC was ignored. Capacities varied more with cell temperature in my partially temperature controlled shop than they did from cell to cell.
A typical charge/discharge curve is given below:
The charge voltage is very flat, varying little with state of charge between about 5% and 95% SOC for LFP cells- unlike with high nickel cells, voltage is nearly useless to infer state of charge with these cells. Coulomb counting (integrating current versus time) is about the only reliable way to assess SOC, and even that must be re-set occasionally with a full charge to 100% SOC. That’ s my normal practice with the car anyway- charge until one cell trips the BMS, then reset the coulomb counter in the car to 100% SOC/ 0 Ah drawn. It’s worked well for years.
A graph for an old cell (purchased in 2013) is given below for comparison:
You can see that at 40 A, the energy efficiency (discharge energy divided by charge energy) has dropped only a little, from 95% to 89%. As the cells age, internal resistance increases, and that will show up as more cell heating and lower energy efficiency at high discharge rates. This was observed in the pack’s performance in the car in its last year, with several cells alarming the BMS by dropping below 2.5 V during heavy acceleration. But these cells are more than good enough to put into a solar storage application, where currents will be much lower. So that’s where the entire old pack is going. I expect this pack will still be doing its thing, storing and returning energy, for another decade.
Once all the cells passed their capacity tests, the “fun work could begin.
A preliminary measurement of the new cells made me quite confident that they would be a drop-in replacement, with only minor modifications required to my battery boxes.
Boy, that was wrong!
The combination of a very small (~ 1 mm) nominal width increase, plus a considerable stack-up variation in width due to small amounts of bulging in the cell sidewalls, plus the thickness of the epoxy fibreglass plates that are recommended to separate the metal-cased cells from one another in a pack (which I’ve never bothered with for my solar installations, but seemed necessary with the vibration of a car installation), meant that I had to totally re-do my front and rear battery packs. While the new rear pack fit with ease, and the shorter cells gave enough space to finally work on the rear pack while it was installed- the front battery pack was within a few millimetres of being too wide to fit between the former engine mounting points. Fortunately, I had a few millimetres to spare, and there was sufficient room for the polycarbonate sheets I used to armour my battery boxes- never liked the idea of using steel for that purpose.
Removing the old battery box allowed me to pull the motor, mounting plate and clutch/pressure plate assembly. This allowed me to replace and properly re-install the clutch throw-out bearing. The clutch is barely ever used in the EV, but the old bearing’s squeaking was driving me nuts because other than that, the loudest sound in the car is the buzz of the little 12V vacuum pump I use to provide vacuum to assist the mechanical brake master cylinder (because there is no intake manifold vacuum to do that job any more!).
A considerable amount of grinding, welding and cursing later, the battery boxes were finished and painted and ready to drop into the car.
Note that there is no thermal management on the batteries, because none is needed. It’s a 3 season car, which goes away before temperatures drop to zero degrees C because that means salt will be put on the road- and British cars are known to be quite soluble in saltwater. The pack also doesn’t need cooling, because it will never be rapid charged. Its heat capacity is large, and the most I’ve seen the battery pack rise in temperature during a drive is about 5 degrees C. Thermal management is simply unnecessary with LFP cells under my driving and charging conditions.
Now, the really fun stuff: the new battery management system.
Back in 2014, there seemed to be two options for BMS: an open-source project called “miniBMS”, which I used because it was simple, brainless and affordable, though not cheap at about $15/cell, and the Ewert Orion BMS- an expensive Cadillac of a thing which gave cell by cell data which you could display on a Bluetooth device. I balked at the price which was over $1000 USD.
Sadly the miniBMS developer sold his project to a company which abandoned it. Some of the boards failed in nasty ways, drawing 12 mA instead of their usual 2 mA and causing cell imbalance. Replacement boards were not to be found. Fortunately a donor came forward. Rasmus Banke of Banke ApS (https://banke.pro) , a maker of electric drivetrains and electric power take-off units for heavy duty trucks, had a professional quality BMS lying around that couldn’t go into a client project for a variety of reasons- and was grateful to me for dissuading him from further work in relation to hydrogen trucks…so I was happy to accept his donation.
Sadly, the BMS requires an expensive software tool to program it, so I’d need to waste the time of the Banke team to use that option. And in conversations with others here on LinkedIn, a couple Chinese BMS options came to my attention that I hadn’t heard of before.
I settled on the AntBMS: https://antbms.vip/products/ Things have come a long way in 11 years- this thing was about $100 CDN including shipping on AliExpress, and is full featured with Bluetooth output. No more watching blinking LEDs to tell the status of my pack, or using my Back to the Future group of six voltmeters to compare voltages on groups of cells- I can l look at every cell voltage on my phone, at a glance.
I bought the lowest current version they sold (40A) which was capable of supporting 32 cells. That would leave me with one cell to manage manually, which I was comfortable with.
The BMS strategy I used in the E-Fire was a tried and true one recommended by Randy Holmquist of Canadian Electric vehicles back in 2013 when I was shopping around for conversion parts. Randy has since retired, but he was very much a no nonsense “do what works and no more than that” kind of guy. That strategy was to interlock the BMS to the charger, such that charging was terminated when any cell rose above 3.65 V. He simply allowed the charger to “bounce” on the BMS signal, turning the charger back on again if any of the celltop BMS boards reached its reset voltage of about 3.5 V. During discharge, i.e. while you were driving, a BMS alarm would only trigger an audible alarm- it was not interlocked to shut down the inverter or even to trigger a “limp mode”. This would startle the driver and might be unsafe. Risking a cell voltage reversal from over-discharge- in practice requiring deliberate ignoring of the BMS warning alarms- was worth it from a safety perspective while driving.
Implementing that same strategy, I used the BMS’s main 40A current path to drive a very simple circuit: a 2 watt resistor was connected to cell 32, and then to the input side of an optoisolated solid state relay on the AC side of my charger. The input is an LED which operates between 3 and 32VDC and draws about 10 mA at 12 V. I chose the resistance to drop the voltage at the input to 12V at peak pack voltage. I’ll use a single miniBMS board to monitor cell 33, once I get around to it. For now, cell 33 has been discharged by 5 Ah relative to all the other cells, ensuring that it will never be the cell that trips the BMS during charging.
In hindsight, I could have used two of the somewhat better supported JK BMS units that are talked about in gruesome detail on the Off Grid Garage YouTube channel. One would handle 16 and the other 17 cells, with each acting on its own solid state relay. But then display management for my planned Bluetooth BMS display would have been more challenging.
I don’t yet have an audible alarm from the BMS, but have some ideas for how to implement one. It will take a bit of logic to tell when the car is charging (and I don’t want the noise) versus discharging (when I want a LOUD beep any time there’s a BMS alarm).
Sadly, when tightening up the bus bar connections, one of the studs welded to the battery terminals, broke off. The studs are welded to the batteries by the supplier, not the manufacturer, and they obviously had some quality problems on their end. I’m negotiating with them to get some spares- hopefully they live up to their responsibilities.
I wired the connections of the BMS and dutifully checked every one with my multimeter- and did find that one of my solder joints had come loose on the way to the rear pack. Once that was fixed, I connected the BMS and pressed the on button – and was greeted by a cheery “beep” and some blinkenlights – LED status indicators that to this day I still can’t find any description of.
Off to install the ANT BMS phone app- and that was an experience of full contact Chinese software-wrangling. No troubleshooting information available- the unit didn’t even come with a wiring diagram. Installing the app required me to defeat every safety feature on my Android phone, and requires full-time location access (GPS) even when the app isn’t running. And the download page is in Chinese without an English translation. User-carnivorous software.
Downloaded the app, finally, and turned it on. No joy. No Bluetooth connection- and without a Bluetooth connection, the BMS is a brick. There’s no way to configure it.
Uninstalled and re-installed the app, this time with location permissions on by default. Powered down the device completely, and then re-started it. Finaly the app connected – and it has never been a problem since. But still far from a satisfactory experience.
The BMS is quite full featured, but because neither my charge nor discharge current is flowing through its main metal oxide semiconductor (MOS) switch, many of its features are useless to me. One thing I did notice during initial pack charging was that the primary overvoltage settings don’t actually open the main MOS circuit. The device still thinks it has separate MOS switches for charging and discharging, even though it comes with two wires (B- and “everything else”, ie. a common point for both charge AND discharge) rather than with three (separate wires for charger and discharge device connection). So after taking one cell briefly to 3.8 V (not good!), the problem was fixed, and I set all overvoltage settings to 3.65 V and reconnection to 3.55V. It worked flawlessly throughout the arduous process of charging and balancing the pack, because my cell SOCs were all over the map before I started.
Although the BMS has a very low balancing current, that doesn’t bother me. I don’t mind manually balancing my pack once per year- something that’s easy to do when every cell is accessible to you directly, unlike with an OEM pack.
My charger might also be of interest. It is the cheapest, dodgiest, simplest charger imaginable, as I had no desire to re-purchase an ElCon unit after my first one let out its $1000 USD worth of smoke soon after I’d installed it into the ER-6 (after four years dutifully charging the E-Fire).
My charger consists of a 120VAC input variable autotransformer (variac) of Chinese origin, purchased on Amazon for $100 CDN. Its switch and fuse were garbage and failed in no time- note that these things are not CSA/UL certified, should be illegal for sale as a result, and are a fire hazard until you remove their garbage switch and fuse components), but the variable autotransformer unit itself is absolutely robust and is rated for 20A. Its input is turned on and off by the BMS via an optoisolated solid state relay as already mentioned. The output side of the variac is fed into a 20A bridge rectifier mounted on a heatsink. Both the rectifier and the variac are cooled continuously by an AC muffin fan. The bridge output (full wave rectified DC) is fed to the pack via an analog ammeter and a fuse. I have an Anderson connector on the charger’s output, which allows me to unplug it from the pack and plug the charger into a pair of clip leads which I can use to charge any group of batteries I want- an extremely handy feature when top-balancing the pack. You simply start with the dial at zero, plug it in, and then turn the dial until the ammeter shows the current that you want. The unit can charge anywhere from one cell to the entire 33 cell pack at currents up to 15 A, though you’d need to feed it with a 20A AC circuit to keep the breaker from tripping. But let’s just not talk about the power factor…this thing draws current only at the peak of each pulse. It’s cheap, flexible and very cheerful though, with a “mad scientist” look, and will charge my new pack in, um, 20 hours. Longer than overnight…Oh well, fortunately I won’t be running the pack anywhere nearly completely flat on the regular. And of course forget about charging at even a Level 2 charging station. Maybe one day I’ll break down and spend another $1000 USD on a “real” charger- the ones from Thunderstruck look pretty cool. But for now I’m happy with my evil scientist Variac.
Initial test drives revealed some problems with the build which still need fixing. In particular the Hall effect device I’m using for accelerator pedal position sensing, doesn’t work well with my Curtis controller. I will be putting a zero/span adjustment device on its output to feed the Curtis a signal it’s more comfortable with, and that should avoid some seasickness from the thing jerking from 50 A drive current to 50A regenerative braking every time you go over a bump and your foot moves the slightest bit. I’ve adjusted the acceleration rate (throttle position response time constant) to compensate for now, but it makes the car sluggish off the line which is decidedly un-fun.
Note that the car is in what my friend Tom calls “50/50 condition”, meaning that it looks great at 50 miles per hour from a distance of about 50 feet. It badly needs bodywork and paint, and I’ve reached my lifetime limit of painting one car, so I’ll need to spring for somebody else to do this work for me-maybe next year.
The first long-ish drive I did, revealed another problem: out of laziness more than anything else, I’d used three of the vendor-supplied solid copper bus bars on the small removable 12V pack I have up front, rather than the heavier multi-layer bars from the E-Fire. Sadly, they weren’t up to the heavy currents drawn by my inverter. One got hot enough to melt the heat shrink. The hot bus bars also melted the insulation on a couple BMS wires, causing a short circuit which fried them. In the process of replacing the inadequate bus bars, another cell post snapped off- so now I’m down to one spare cell.
The pack is now all neatened up, with the inadequate parts replaced and all the BMS wires carefully routed away from the bus bars. A long drive today was very satisfactory and incident free. I’m going to enjoy driving the car this fall. After some range calculations I’ll hopefully find out if I have enough to make it both to and from the annual British Car Day show in Burlington, Ontario on a single charge. If so, the ER-6 will be there for the first time. It never made it there as a gasoline car- it was too unreliable to risk the trip!
Disclaimer
This article has been written by a human. A human whose original electric conversion project, totally changed his life for the better. A human who likes cars that drive rather than as art objects to polish or museum pieces to restore. A human who is 100% an amateur, i.e. someone who loves both British roadsters and EVs. If you’re pursuing a conversion yourself, do yourself a big favour and read all the threads you can at www.diyelectriccar.com, and do a bunch of additional research and reading. Rules and regulations vary greatly from country to country- what works for me could be 100% illegal where you live- or it might be legal, but you may not be able to insure it to drive it legally on the road even if you dot all the Is and cross all the ts. As they say, your mileage may vary!
A note to the “haters”: if you’re a classic car fanatic, and think that electric conversions are sacrilege, then please just f*ck off. An EV conversion is no different than an engine swap, and people have been doing that with classics since cars were invented. The TR-6 is not rare- there are tens of thousands of them still on the roads of the world, and this one was in rough shape anyway. And it’s my car not yours, so it’s none of your damned business either way. And not everybody is in love with toxic exhaust emissions, noise, grease, and constant tinkering with 50 year old junk made at the lowest ebb of car reliability in history. I’m in fact very grateful that most of the problematic components of this car are now being used in somebody else’s restoration project.