This is not only my process, motivations and design goals but a living document for the length of the project.
I'll keep publishing more steps as I move towards the end design goals. Feel free to comment and send suggestions for the steps that are yet to be done.
In a nutshell, all I wanted was to start with a standard, monolithic electric motorcycle and replace the battery pack for a removable, smaller but sufficient one. The definition of sufficient will depend on your daily routine. Mine is ~1 kWh. More about this in the design step.
Design goals are :
- Light enough to carry it from parking spot to charging spot (office, home, etc)
- Removable -- Park, battery removal/insertion procedure < 2 min.
- Small enough to charge anywhere
- Not an artisanal battery pack - instead, composed of a few safer and smaller RC Hobby batteries (Turnigy, Zippy, etc)
- Serviceable battery pack - when one cell goes away, you shouldn't need to pay for a whole new pack, instead, just a fraction of it.
The only aspect of this instructable is the mechanical and electrical aspects of the Battery Pack design. I purposedly started with a working electric motorcycle. There's plenty of good documentation out there about conversions, chassis, etc.
I'm in the school of thought that advocates for "don't reinvent the wheel". On the other hand, most wheels are one-size-fits-all and when it comes to Electric Vehicles, I'm yet to see an option that is not a huge commitment, financial and in terms of adapting your environment(s) to charge it.
In principle, a removable battery pack should enable you to use it in the same way as a gas vehicle. Park it anywhere. You'll still have to charge the pack, but that's assumed to be ok.
Note: I need to highlight the ZElectric guys, during my state of the art research I found that they do this business professionally in the best way I've ever seen - keeping it simple, straight to the point and still providing great specs in their vehicles. Not adding overly complex technological features (smart everything) and with that keeping the price down. For ~ $5k you can already get a high end electric bike from them, good for the highway.
Step 1: The Donor Bike
Specs first, then you can read the story if you're more interested :
- Purchase price : $700
- Power : ~ 5 kW
- Controller : 60v, 80A
- Brushless hub motor
- 60v ; 40 Ah Lead Acid battery pack - 3 yo and a lot of abuse :
- 8 months, no maintenance charge storage
- When driving it you could feel that the battery pack was in the end of its life
So I could feel that it was a good machine, it just needed some attention for it's battery pack. As usual with batteries, once they're damaged or old, there's little to do about it but getting a new pack. Perfect case for our project.
Description for humans is - the batteries would actually last long, but any high current demand would make them kneel (voltage would drop from 62 to under 44). I could drive it for about 5 miles but whenever I'd try to climb a gentle hill, it wouldn't respond well to throttle.
At night, lights would turn off when accelerating.
One factor to motivate people out of Lead Acid for EVs is : look up Peukert's law for capacity reduction over high discharge rates. And that's what we do with EVs - average power is pretty ok, but "acceleration punches" are pretty demanding to the batteries. Lead Acid really doesn't like it.
I know, decent LiPo packs are not cheap. Which is why it's great to go small. I can do my commute with 1 kWh and that costed me only $200in hobby LiPos. If I were to replace the original 40 Ah 60v lead acid pack I'd pay more than that.
Step 2: Dimension the LiPo Equivalent of Your Lead Acid Pack
No matter what you had before, we need to figure out how to power the controller. Here's how :
This is not exact business as batteries have a significant higher voltage when charged than when they're discharged. This is more subtle (and way less linear) in LiPos than in Lead Acid, so we need to be careful here.
In my case, I had 5 x 12v Lead Acids. That gave me 66v charged and 57v discharged. Look at the Lead Acid discharge curve at ~2C for reference.
While trying to provide a similar range, I went for a no matter how, 16S pack. That'd give me 67.2 charged and 59.2 discharged.
I know, it's not exactly the same, but we can only work at increments of one cell. So that's the closest possible. Also, most circuits don't have a mandatory input voltage, is more of a range, as they generally have internal regulators and protections.
If it was real electronics, we'd have all the specs from the manufacturer and be sure. As we're talking about cheap, brandless circuits we have to hope for the best and buy a 50 bucks replacement if something goes wrong.
FWIW In the beginning of my LiPo journey, I thought that voltmeter readings would be close to useless or misleading. In practice they're pretty helpful due to the LiPo discharge curve. You just need to be very careful, as after it reaches 3.7v per cell, the discharge to damage will be pretty fast. Which is why most BMSs have a cutoff close to that point.
Step 3: Design Considerations for the New Pack
Here we'll discuss :
- Calculate energy required
- Finding the appropriate batteries to assemble the pack
- Charging time
I'll share my rationale and process. You could apply a similar approach by changing a few parameters to adapt to your case.
First, knowing how much is sufficient is the core business. Let's start with distance. Ask your preferred map provider the raw distance of your commute. Keep that in mind and use this geographically less featured calculator :
Which also considers elevation gain. As this if for light bikes, make sure to specify a higher weight for your motorcycle. On the calculator, I got 6 Ah needed to go one way on my 60v battery pack. So the full commute will need 12 Ah.
As usual, don't consider that this is all you'll need as there are always losses and safety reserves that are good to exist.
Once you get fluent with these measurements of capacity, I found pretty helpful to translate the battery pack capacity to Wh. It's a much more comparable unit. Just multiply your Ah capacity and the pack voltage. I got ~960 Wh.
It's much easier to compare to other electric vehicles and get a real world feeling on energy spent per mile. For example, I found that electric scooters should consume about 60 Wh per mile, given that electric bikes do much less and performance motorcycles do about 80 Wh per mile. This is a great ballpark value. use it to compare to your numbers from the calculator.
Going to the physical world, I found a pretty good deal in hobbyking.com for a 16 Ah 4S battery :
Minimum Capacity: 16000mAh
Configuration: 4S2P / 14.8V / 4
CellConstant Discharge: 10C
Peak Discharge (10sec): 20C
Pack Weight: 1290gPack Size: 173 x 74 x 45mm
Charge Plug: JST-XH
Discharge Plug: XT90
The link will eventually break, but here it is :
I got 4 of these for $50 each. Pretty good deal, total of $200 for the entire LiPo pack. That's rare and that's one of the reasons I wanted to go for a Hobby pack based design. It sells a lot and you can find spares and deals much more easily.
That gives me the 16S I need, once I connect the 4 packs in series. It also allows me to disassemble the 16S pack into the smaller 4S ones and use a standard model plane 300W charger, which is what I did during all the testing period, before I bought the 16S no balancing charger.
I don't think we even need to discuss current capacity here. Lead acid would need a lot of care. But using LiPo is just cheating. Peukert's law doesn't apply and if we're going for 16 Ah, 10C is 160 A , double than our maximum current. So you just need to be aware that your acceleration peaks are under the 10C (or higher) battery spec.
For charging, you will usually will do pretty well charging at 1C in 1h. I preferred going the safer route and got a 5A charger for my 16Ah pack, which will give me roughly 3h charging time.
For chargers, Aliexpress and the next step's BMS suppliers are pretty ok. I'm sure there are better ones, let me know in the comments!
I'll fast forward to the current draw tests and let you have a bit of fun on how that turned out in the garage.
Of course that while driving in the streets currents were much higher. A 2 seconds period on 80A to get to cruise speed and averaging in 10 to 20 Amps depending on elevation gain to maintain cruise speed.
I recommend everyone to do this test, so you know the characteristics of your circuit and get familiar with the values of current depending on what you do in the road (climb hills, accelerate, cruise). This will hint you how not to increase heat in the circuit by operating it at max current for too long.
I went for a 7 bucks analog ammeter + 75 mV shunt, which was the cheapest high sample rate ammeter I could find. The digital ones (like you see in my dashboard) have a so low sample rate that you have no idea of what's going on. EV load is full of peaks worth understanding.
It's also useful to have a voltmeter, digital is fine. That sample rate is satisfactory for some reason. One crucial information I got from it, is that when I go to maximum current (80A), I get a voltage sag that goes down to ~45v.
With that in hand I was able to calculate the internal resistance of my battery pack, it was not small but still within the expected range. Which explains both the voltage sag and the low price of the pack. If you aim for higher C (discharge) specs, I bet resistance is much lower, and voltage sag is much smaller.
Step 4: Using a BMS
One thing I assumed and learnt that not so easy - beyond 6S it's very hard to find a fast enough charger that also balances.
Most EV setups will require a BMS and it's a very bad idea to think you don't need one. First months will be fine, but once the first cell fails, you're in for a world of flames. You better have a way to detect that in time to retire the defective cell/pack.
So it might be ok to prototype without a BMS, but put in your plans to get one. It's not that hard to install even if you barely know what you're doing : http://www.batterysupports.com/60v-672v-16s-100a-...
If you're still thinking you can get away without it, consider the following scenario :
Most stealth LiPo failures come from cells that due to age have a << capacity than the rest of the pack. You're charging and that one will charge "just fine", until it reaches 100% of its reduced capacity and the charger doesn't see that, keeps charging the pack. Voila - our cell goes over 4.2v and becomes a fire starter.
A similar one but in the discharge scenario - you're riding your bike and the reduced capacity cell gets to 3.7v much faster than its neighbors. Your monitoring (whatever it is, unless it's per cell) won't catch it, you'll keep riding and guess what? Depleting LiPos under 3v also ends up in flames.
Step 5: Prototype Pack
More show than tell here. I did what I like the most. Got the job done as fast as I can, which we can afford here as it's the prototyping phase.
A wood board to hold everything together, AWG 8 wire (spec'ed for the nominal average current) and the XT-90 connectors to match the batteries.
I assume that you already know what needs to be done. If you don't it's probably not a good idea to be messing with this big LiPos without any protection circuit.
Yes, it's a rig. I'm not proud of it. But it works, it got me 1~2 months ahead with actually testing the thing on the road. Don't worry. We'll get back to it to do the proper design, which is the whole focus of this instructable.
Basic features will be taken into account in the final design these are :
- Each of the 4 packs could be connected/disconnected/removed individually for balancing/maintenance
- Fits the battery storage bay of the bike. Has anchors so that it can be strapped in place.
- Can be carried by hand - not in the pictures but I have a small rope that works as a handle
- Should be able to slide in and out from a Rail.
Step 6: Next Steps
So we're at a point that the whole upgrade was done, tested and prototyped. It works. I'll start commuting with that and work on the next steps :
- Find the right BMS+Charger approach
- Provisory 5A/3h charger is already here
- Having something that takes pure DC and manages the battery charges will eliminate the requirement of a charger, so we could plug it to a standard DC power supply that supports higher current.
- That's where the fast charging fun starts. < 1h
- This will be a digital design I'll share here.
- This should have a rail system counterpart to live inside of the bike. Should reduce battery removal/insertion from time from 2~3 min to < 30s.
- Battery should fit the front port, protected from a key, very easy access. Picture to follow.
Step 7: Feedback From Community
Thanks to the community, some very interesting ideas showed up in the comments. I'll respond to some in a longer way than just a comment would allow me.
It was well suggested that I use supercapacitors to provide high current at the motor start from a full stop. That's the most current demanding action of the vehicle.
It actually makes sense as I experience a voltage sag from 60v down to 45v at acceleration.
This is more due to the internal resistance of the pack (calculated to be ~200 milliohms for the whole pack, 12.5 milliohms per cell) doing what it does at high currents (V = RI) more than the battery not being able to supply this current for other reasons. Max current spec for the pack is 160 A.
But indeed a good application of a supercapacitor is to compensate for voltage sags, like in a rectifier. I would need the precise I and V curves from a data logger before find a good fit capacitor.
2 other fun facts :
- To utilize the full capacity stored in the cap, we'd need to discharge it to 0 volts. This is not going to happen because when we get to 45v the battery will sustain that voltage and we won't get current from the capacitor anymore.
- I calculated a 300 F capacitor to store 150 Joules. Considering a full acceleration being 80 A for 5 seconds, it would cost 2400 joules. So a big capacitor bank.
If I was to pursue a solution like that I'd go for a small pack of Graphene/LiPo batteries, they can do 65C sustained, 130C bursts, which means I could likely use a 600~1000 mAh pack as a buffer, which would be relatively cheap and small.
It'd also contain about 100 kJ, which is enough for 4 full accelerations. A management circuit could then recharge them at 15C while we're not accelerating.
But considering all the trouble to get it in place, I actually have a pretty good performance with the current pack. I was even able to get out of a traffic light faster than a gas bike, just for the gas biker to get angry and try to prove that he's faster :-)
That's a good point. I've seen most EVs to use LiFePo cells. I'm still happy that I've chosen LiPo though, I'm optimizing for cost. I haven't looked at LiFePo a lot, but the few times I did, packs of similar size were costing about 800 to 3k depending on where you look. I wouldn't trust my life to the $800 packs as they're in the low end of the spectrum.
Out of curiosity I'm quoting a LiFePo pack from a chinese company to see how much they charge. It's looking like in the $600 range. Nice features is that has the BMS and case included.
I still think I made the right choice as I'm trying to come up with a framework to use replaceable RC Hobby packs as the building blocks for the EV batteries. It'll be a bit more work but I really like the fact that one can pretty much open the case, throw the 4S pack away and find a deal on a new one for really cheap.
LiPo buffer for Lead Acid pack
It's a pretty good idea and I thought of it too, at the early stages when I tried the Lead Acid pack and it was not up to standard. It's perfectly doable, but if I have to mess with LiPos at all, I preferred the short route that got cheap enough today -- spend $200 in the LiPo pack and walk away. For reference, my previous Lead Acid pack, 9 Ah costed me $150.
But it would be beautiful to see Lead Acids not being affected by Peukert's and actually delivering the capacity they have in the label.
Still, I'm a fan of the lighter stuff these days. It's much more manageable and much less time in the metal shop making precise cases, enclosures etc.