The Actual Mountain Wheelchair Motors are Working!

…plus lots of other updates

It’s been a little quiet on the blog over the last week or thereabouts so I wanted to I’d like to share a number of things that have happened during that time…

Spokes

After dropping off the motors and wheel rims at Halfords in Llandudno, the guys were able to take all of the measurements, but weren’t able to locate the correct spokes from any of their suppliers. Just like me, they did some digging around and the only suppliers they could find were in China. I’ve since ordered the spokes from China, but of course they’re all celebrating the Chinese new year at the moment so it’s going to be a number of weeks before they arrive.

Battery Pack

The first 48v battery pack is ready. After bottom balancing all of the cells, I was impressed as they were charging as most of them remained within 0.001v of each other, and there was just one other cells which was 0.002v behind the others. Fully charged, this is a 56.5v battery:

Mountain Wheelchair 48v Battery Pack

Interestingly, John Williamson (Aka Burgerman) of www.wheelchairdriver.com has been reading through my posts, and based on his background working with charger manufacturers in an advisory capacity for 25 years, he’s suggested that bottom balancing is the wrong way to go about this. Thanks for taking the time to read my posts John, I think more research is required on my part.

Grinding away the Dropouts

With much trepidation, I spent a large part of yesterday grinding away part of the BMX forks to get the motor axles to fit. Looking at them though, I’m really worried that the dropouts on the forks (the part where the axle will sit), aren’t going to handle the torque produced by these motors. I’ve had to file so much away, that there’s far less metal on the forks now than there used to be. Originally, I imagine an engineer would have calculated how much metal needs to be there:

Mountain Wheelchair Filed Dropouts

Snapped dropouts is a common problem with e-bikes:

To overcome this, you can buy and attach what’s called a universal torque arm. This arm takes the torque produced by the motor and transfers some of it higher up the forks where they are stronger:

a-pair-electric-bike-hub-motor-torque-arm-front-motor-torque-arm-for-every-electric-bike.jpg_640x640

My only problem with this is that I think they look untidy. I gave it some thought last night and have made some rough sketches of a purpose built “torque arm” that will bolt to the frame. If I can get those made up at a reasonable price that’s what I’ll do, if not then I guess I’ll have to go with these unsightly universal torque arms.

At least the motors look cool now that they’re sitting in place and Ada’s really pleased with the purple dust cups which she chose.

Mountain Wheelchair Purple Dust Cap

Motor Controllers

Six motor controllers arrived from China yesterday and at first glance I’m really happy with them. They’re far smaller than I expected them to be which means they’ll be easy to fit within the mountain wheelchair frame.

Mountain Wheelchair Motor Controllers

The main problem I have with these controllers though is that the Chinese user manuals have been badly translated into English. Trying to understand what the manual is trying to say is rather difficult.

Whilst I was waiting for the controllers to arrive, I was however able to produce a wiring diagram to provide power the 6 motors:

6 motor wiring diagram for mountain wheelchair

Once I had the wiring diagram and knew what I wanted to achieve, I then spent days searching through online components for the parts I needed. Trying to find a simple thing like a 48v switch (200a and waterproof) has been impossible in the UK so I’ve had to resort to getting parts shipped from China again.

Actual working Mountain Wheelchair Motors

Nonetheless, and this is quite a big moment, I’ve just managed to get the actual mountain wheelchair motors running for the first time!

You’ll have to excuse the temporary wiring, but, Woohoo! It’s working!

Battery Charging for the Mountain Wheelchair

The post man arrived today with more goodies for the mountain wheelchair; another 8 LiFePo4 batteries, a 48v LiFePo4 charger, a 3.65v LiFePo4 charger and a new multi-meter.

I’ve had my existing multi-meter since I was in high school, more than 20 years ago. It’s lasted all this time, never failed me, and so I’ve never thought to replace it until now. Following on from a previous post where I discuss bottom balancing LiFePo4 battery cells, I needed a multi-meter that would give readings that were accurate to +-0.001v, which my old meter wasn’t capable of. Now armed with two multimeters I realised I had a problem…

In the post mentioned above, I’d connected the wheelchair batteries to a set of motors to try and discharge them. It took days (literally) of leaving two 300w motors running to get the battery cells down to about 3v each. As there were 8 batteries in the pack, I needed to get the pack down to about 24v (8 * 3), and as was to be expected, the voltage remained stable for days then suddenly started to plummet. At this point I disassembled the battery pack and started to discharge them individually.

Discharging Individual Cells

There are lots of ways to discharge an battery, some of which are far safer than others, but I used what to hand; about 350 Light Emitting Diodes:

LiFePo4 Single Cell Discharging

Incorrect Multi-Meter Readings

Whilst in the process of discharging the batteries, I noticed that my old multi-meter is giving different readings to my new one. To see which one was wrong, I got hold of a third one and lo-and-behold, the multi-meter that I’ve been using for the last 20+ years is reading higher voltages than it should do. Hooked up to a 3.65v charger, the old one is giving me a reading of 4.09v!

Different Multimeter Readings

I’ve no idea how long it’s been like this but it’s likely that I’ve been using the wrong measurements for years. Oops!

The good news is that at least now I know I have one that is accurate and I can go about getting all of these battery cells to exactly 2.750v.

Anyway, I’m digressing somewhat; the reason for writing this post is because I wanted to share a modification that I’d made to the 48v LiFePo4 charger.

Factory LiFePo4 Charger

In my previous post (see link above) I explained that after bottom balancing all of the cells to 2.750v, I will then wire all of the cells in series to give me a 48v battery. To this end I purchased a 48v LiFePo4 charger. From the factory, it charges the batteries up to 57.6v (which is about normal for a 48v charger).

LiFePo4 Charger with Factory Setting

If there are 16 cells in the pack, then that means 3.6v per cell (57.6 / 16). This presents a problem because as was shown in my previous post, the charge/discharge curve for LiFePo4 cells is very steep at both ends. As a cell starts to reach its maximum charge, it suddenly starts to shoot up. If one cell were to reach 3.6v before the rest of the cells in the pack then its voltage would increase faster than the others and likely result in permanent damage to that cell. The absolute maximum voltage these cells can safely handle is about 4.2v but I’ve decided to aim for 3.5v to give the whole system a bit of “room to breath”. This meant that I needed to modify the battery charger.

Modified LiFePo4 Charger

Rather than have a charger that tops out at 57.6v, really what I wanted is a charger that tops out at 56v (3.5 * 16), so I went ahead and opened up the charger. My luck was in as I found that it had a very small adjustable VREF relay. By making a small adjustment to this, I now have a charger which tops out at exactly 56v and is perfect for charging the batteries safely.

LiFePo4 Charger with 56v Setting

For anybody else wanting a charger which tops out at 56v, I purchased this one from Eclipse Bikes. The adjustable vref can be found on top of a small blue relay next to the green/red lights and is labelled VR2 on the circuit board. Obviously this will void the warranty and more importantly; the charger is running off 240v mains supply so there is inherent risk involved. For me it’s been a pleasant surprise and I now have a means to charge the mountain wheelchair without damaging the batteries.

In Other News

The 6 hub motors previously ordered have successfully passed UK customs, and I’ve also gone ahead and ordered 6 motor controllers to go with them. When they arrive I’ll have everything I need (apart from a few small pieces) to make an actual moving vehicle. Yeah, I’m making progress with the mountain wheelchair!!!

LiFePo4 – Power for the Mountain Wheelchair

It became apparent quite early in the project that batteries for the mountain wheelchair would need considerable research. In a previous post I estimated that the wheelchair would need a 3 tonne battery to get up the mountain. Luckily, the project has come a long way since then.

When I built the first test platform for the mountain wheelchair I decided to use lead acid batteries (think car battery). This was partly because I’m already familiar this technology, and partly because they’re inexpensive; the four batteries on this prototype were £20 each.

Lead Acid Batteries

The problem with lead-acid batteries is that (1) they’re heavy, and (2) the way their voltage drops over time isn’t particularly useful.

Voltage Drop

Take a look at the blue line in the graphic below, this shows how a lead acid battery’s voltage drops over time.

Battery Discharge Curves

If you imagine for a moment that this battery is powering a wheelchair motor, then you would expect that over time the battery’s voltage would drop. In fact, looking at the blue line for the Lead Acid battery, you can see that the voltage drops in quite a linear fashion, so there’s nothing really surprising here. Batteries discharge as you use them right?

The problem with this is that the power available to the motor steadily gets weaker and weaker over time. On the mountain wheelchair, it would mean you’ve got lots of power to begin with, but as soon as you start driving, the available power drops in a linear fashion. Half way up the mountain you can no longer get over the same obstacles which you could at the start.

LiFePo4 Batteries

Now take a look at the red line on the graphic above, this represents LiFePo4 batteries. As you can see, the LiFePo4 batteries lose power in a far less linear fashion; they maintain a constant, high voltage for a long period of time and then quickly drop off at the end. This is great for the the mountain wheelchair because it means that you’d have fairly constant power for the duration of the trip. If you managed to drive over a boulder near the foot of the mountain, chances are you’d be able to drive over a similar boulder on the summit.

Not only this, but LiFePo4 batteries are also typically half the weight of lead acid batteries, so not only do you have more usable power, but you also won’t need as much power because the wheelchair will be lighter.

The other main advantage of LiFePo4 is the amount of power (current) they can provide in a single burst. If the motors needed a sudden boost of power to get the wheelchair over a particularly large step, LiFePo4 batteries would be more capable of supplying that boost than lead-acid.

The two main disadvantages of LiFePo4 batteries though are their price and the additional care needed to look after them.

LiFePo4 Price

In the prototype shown above I had four 12v 7ah batteries. These batteries were wired together to give me a total of 24v 14ah. In total this cost £80.

LiFePo4 batteries only provide 3.2v. So to make up 24v, you need 8 batteries. The cheapest I’ve been able to find suitable LiFePo4 batteries in the UK is £23.45 each. Multiplied by 8 this is £187.60 for 24v (More than twice as much as the lead acid batteries).

That being said, LiFePo4 are good for 2,000 cycles (fully drained and recharged again), whereas lead acid has a maximum of 300 cycles. In the long run then, LiFePo4 might prove to be more cost effective as they won’t need replacing as often.

As you can see, these batteries are worth the premium and not too long ago these 15ah cells arrived in the post. Shiny!

24v 15ah LiFePo4 Battery Pack

Caring for LiFePo4 batteries

As I said above, the other problem with LiFePo4 batteries is the additional care required.

Batteries can catch fire. I imagine it would be a terrifying thing to see somebody in a wheelchair engulfed in flame part way up a mountain.

One of the reasons they catch fire is because there are lots of individual battery cells all working together. If one battery fails, then all the other batteries will keep putting demands on the failed battery until it expands, pops, and eventually has the potential to ignite.

The way to avoid this is to make sure that all of the batteries are well matched. This doesn’t mean just buying batteries with the same specification though, as over time each one will deteriorate at a different rate to the others.

There is conflicting information on the internet regarding how to make sure all of your batteries are well matched – LiFePo4 is a relatively new technology and many opinions are based on previous experience with lead-acid batteries.

Battery Management System

One method is to use a BMS (Battery Management System). A BMS monitors the state of every individual battery. If one battery runs flat, the BMS will stop the vehicle.

The problem with LiFePo4 batteries, as can be seen in the graph above, is that when the voltage of your weakest battery does drop, it will drop quickly. If your BMS isn’t fast enough, kaboom!

The other problem with this is that you have to prematurely stop the wheelchair just because of one weak battery cell.

A better solution is battery balancing.

Battery Balancing

Top balancing means charging each individual cell as much as possible until they all have the same voltage.

Bottom balancing seems to be the best option for LiFePo4 cells though. Instead of trying to balance the cells by forcing them to a certain voltage, in bottom balancing you drain the LiFePo4 cells until they’re in the steep downwards part of the discharge curve (Usually about 2.75 volts).

As I write this post, I have my LiFePo4 cells wired up to two motors which have been left to run.

Discharging the Battery Cells

There are 8 cells in total and I want to get them down to about 3.1v each to begin with. 8 x 3.1 = 24.8. Once the pack reaches 24.8v, I will begin to discharge the cells individually until the each have 2.75v.

The cells then need to be left for 24 hours for their voltage to stabilise. At which time, it might mean charging or discharging them a little more, until they are all within a +-0.001v tolerance.

Once all cells have stabilised at 2.75v, I will wire all of the individual battery cells together, and charge them as one unit, as if it were a single battery. Before I can do this though, I need to buy another 8 cells (to give me 48v in total), and a 48v battery charger.

With all 16 cells at 2.75v, then wired together in series, I will use the 48v charger to bring them all back up to almost maximum capacity. Maximum capacity for these cells is 3.6v, however, I’ll only bring them up to 3.55v in order to give them a little room to breathe. For all 16 cells, this will be 56.8v in total, and should be plenty enough to get the wheelchair into the mountains.

*Update* See the modification I made to the battery charger.

Once they’ve been charged, they need to remain in their pack. All cells should be discharged and recharged together. If for some reason you disconnect one of the cells, then you will need to go through the whole bottom balancing process again.

How much power will the mountain wheelchair have?

The current design for the wheelchair will accommodate a maximum of 112 x 15ah LiFePo4 cells.

112 LiFePo4 Batteries

To make 48v, I need 16 cells. If I then divide 112 by 16, this gives me 7 x 48v battery packs. If each of these packs has 15ah, then this gives me a total battery capacity of 105ah (7 x 15).

I don’t know how many amp/hours the mountain wheelchair will need. Some people have said that 105ah is far too much, others have said it won’t be enough. As I don’t have the mathematical ability to calculate this figure for myself (although I have attempted it here), and I seem to get conflicting opinions depending on who I ask, my plan is start off with one 16cell 48v pack which gives me 15ah. I’ll see how I get on with this and then add more packs as required.

It makes sense to start off with as little as possible, not just because it makes for a lighter wheelchair, but also because each 48v battery pack costs nearly £400! If I use up the available space on the mountain wheelchair and have 112 cells to give me 105ah, then I will have spent £2,626.40 just on batteries.

What’s next?

The other 8 batteries have been ordered, along with a 48v battery charger and a single cell 3.65v charger. When they arrive I’ll get the whole pack balanced. The hub motors have been posted and are currently en-route from Cina. One of my next tasks will be to invest in some motor controllers so that when the motors arrive, I can begin testing the electronics.

Still Confused by Power Requirements

I’ve posted several times now about batteries and power consumption but they’re still a cause of substantial concern.

Ultimately, I think it’s the batteries, or rather the power requirements, that are going to make or break this project.

My first calculations suggested that the wheelchair would need a 3 tonne battery, or in other words; this project is an impossible dream.

Employing a large dose of optimism I made some very arbitrary amendments to tailor the results of the calculations to make the project look possible but ever since it has been niggling away in the back of my mind.

Looking at it again, the Llanberis path has a distance of 7.23km and a total height gain of 0.966km (measurements taken from Ordnance Survey 1:25,000 map). Using trigonometry, we can work out that the average slope of the Llanberis path is 7.7 degrees:

Llanberis Path Gradient

To calculate this as a gradient, we simply divide the height by the length and then multiply by 100. (0.966 / 7.23) x 100 = 13.361%

Obviously the gradient of the path will be steeper in parts than others, but having this average should give us more accurate calculations.

To pull a 100kg weight up this slope at 5kph needs a 250w motor. That’s a far cry from the 3,000w motor I’d quoted previously.

As you double the weight though, the power requirement doubles. So a 200kg weight needs a 500w motor.

Obviously, as the power requirement for the motor increases, so does the power consumption and the weight of the required battery.

It’s clear then that power requirements, and therefore weight reduction, are going to become an important part of this project.

A 500w motor seems awfully small though when you consider that a child’s quad bike might have a 1,000w motor.

Looking at existing mobility scooters though, 500w seems very common. The only one I’ve seen which uses a larger 800w motor is this “All Terrain” Mobility Scooter:

“All Terrain” Wheelchair

Perhaps then a smaller motor would be sufficient? I guess there are currently too many unknowns to be able to find the right answer. With that in mind, I’m going to make a list of questions which need answering in order for me to get to the bottom of this:

Ultimately I want to know how powerful the motors need to be and which battery will provide that power?

In order to answer this question I need to know:

Weight of wheelchair – impossible to say without knowing the other values
Weight of passenger – 40kg approx, will need to get Ada on the scales
Weight of Motor – difficult to say without answering other questions first
Weight of Batteries – as above
Gradient of Path – 13.361% average
Steepest gradient likely to be incurred – need to take some measurements
Length of path – 7.23km to the top, 14.46km return.
How is power distributed between the motors – If six motors produce 500w each, does that mean you have 3,000w power?
If a 500w motor needs 30 amps to get up the hill, does that mean that the six motors need 180 amps (6 x 30) or does it mean that because there are six motors all doing the work they don’t need to draw as much power and therefore they have a combined power requirement of 30 amps?
On top of this I’m also going to need to know the power requirements of other devices, such as lights, battery indicators, linear actuators/hydraulics, any software controllers that might be on board, motor controllers etc.

As you can see, most of this can’t yet be answered so for the moment I’m going to concentrate on question 8 and 9. They seem like they shouldn’t bee too difficult to find the answers to and I expect they will influence the outcomes of the other questions too.

*Update* here are the answers to questions 8 and 9.

Just a dream?

So the other day I estimated that the wheelchair would need to be driven by six 3,000 watt motors. At a constant gradient of 40 degrees, to do the 18 mile round trip, the 4QD calculator suggested the motor would have a constant draw of approximately 65 Ampheres.

I have in my shed a very large 12v 120AH leisure battery which I use to power an electric outboard motor on a canoe. The 120AH rating means that with a 6 Amp draw, the battery provided 12v for 20 hours.

With the 65 amp draw of the 3000w motor, this battery would last 1.8 hours (120 ÷ 65). Let’s say it took 5 hours to reach the top, then you would need 3 of these batteries.

Of course, this is only for 12 volts. For 48 volts we’d need to multiply this number by 4. 4 x 3 = 12.

That’s 12 very heavy caravan batteries. At a guess, I’d say my 120AH battery weighs about 30kg. So overall, that’s a weight of 360kg just for the batteries.

But… Because I want to use 6 motors, does that mean I have to multiply this number by 6? If it does, then it would mean that the batteries weighed over 2 tonnes!!!

And of course, if you’re dragging a two tonne battery up the mountain then the power requirements of the motors increase and then so do the battery requirements. Perhaps then it isn’t possible, and this is why it appears something like this doesn’t already exist.

How about looking at other battery solutions?

I came across a 1.2v 500ah battery which weighs 15.9kg dry. To make a 48v battery, you would need 40 of these. 40*15.9. That’s 636kg for one motor and that’s without the battery fluid. You’d then need to multiply this by 6 which is nearly 4 tonnes and that’s before adding battery fluid.

It was starting to look like a bit of a pipe dream so I got out my map and actually measured the distance to the top of Snowdon rather than relying on secondary information. The distance to the top of Snowdon (On the Llanberis Path) is just over 7km, so a 14km round trip, which is far less than the 18 miles that I’d been using up until now. The internet lied!!!

I’ve also made a very basic estimation of the Llanberis path’s gradient. I had originally been using a value of 30 but keeping in mind that I’m sat at my desk and not out in the field, I’ve estimated it to be more like 10 degrees.

So if I took these new values, slowed down both the the acceleration and top speed, and gave the calculator what could be an unrealistic weight of 200kg, the current draw comes down to 20 Amperes. This would mean that 4 of the caravan batteries (120kg in total) in my shed would be able to get us to the top.

The problem now is that I don’t really know if I then need to multiply this by 6, so that each motor has the same battery pack.

I think the answer isn’t going to be as simple as this.

With 6 motors, individually they won’t need to draw as much power from the motors because they’ll be sharing the load. Imagine trying to push a car by yourself. Then imagine how much easier it would be if you had five more people to help.

Using this analogy, I suppose the same amount of force is required to push a car no matter how many people are pushing. Perhaps it’s the same for the motors? It doesn’t matter how many motors you use, the amount of current drawn will be the same.

I don’t know enough about electricity to be able to say if this is the case, and I imagine it’s not as linear as this, but it does at least sound logical. If this is case then it would mean that 120kg battery (or thereabouts) might get the wheelchair to the top of Snowdon. 120kg still sounds like a lot, but really it’s just four caravan batteries and this to me sounds doable. It is at least far more doable than a 3 tonne battery.

Moving forward, it’s clear that I don’t have the required underpinning knowledge to make these kinds of calculations and I think it’s therefore time to ask for some expert advice.

I think I might also have to stop feeding Ada! Or as mum suggested, ask Elon Musk for help.

Power, Speed and Motors

One of the biggest unknowns at the moment is which motors are going to be used in the final wheelchair.

The problem is that my background is in computing not engineering and I don’t know much about motors. As an experienced walker though I can make some calculations about speed.

The average healthy person walks in the mountains at a speed of 4km per hour (2.5 mph) so there isn’t much need for the wheelchair to go any faster than this. To work out what speed the motors need to turn, I’m going to use a 12″ (30cm) wheel as an example. A 12″ wheel has a circumference of nearly 1m. To travel 4km then it would require 4,000 rotations. This tells us that this wheel would need to rotate at a speed of 4,000 rotations per hour, or 66.66 rotations per minute. If we rounded this up to 67RPM we have an idea of what speed a 12″ wheel needs to rotate in order to travel at a walking pace.

This example though presumes that the wheel turns at the same speed of the motor and doesn’t take gear ratios into account. However, even with this simplified example it becomes clear that the motor doesn’t need to operate at high speeds. What’s going to be more important is torque, but more on this in a moment.

Providing power to the motors is also a current unknown. To complete the 18 mile journey to the top of Snowdon and back at 4kmph would take about 7.5 hours (without stops). This would mean that the batteries would need to be able to supply constant power to the motors for an absolute minimum of 7.5 hours, although this doesn’t take into consideration things like terrain, angle of inclination, or wheel spinning/slipping.

Another consideration is the battery voltage. It’s a lot more complicated than this but in very general terms, the higher the voltage rating of a motor, the higher the torque. Also, as a battery’s ability to supply a higher voltage over a longer period of time increases, its size and weight also increase.

I don’t yet know what calculations I need to perform, but, I don’t think a 12v battery (similar to what you’d find in a car) will be big enough. To get enough torque out of the motors, I’d guess that we’d need at least 36 volts (3 car batteries).

If the final design employs a rocker bogie mechanism then it will likely need 6 motors (one for each wheel). 3 car batteries for each motor (18 car batteries!!!) is clearly out of the question. I think it will most likely either use one battery for the whole system, or two batteries so that there is one for each side. This could be a 48v battery overall, or 2 x 24v batteries.

At the moment I’m thinking of 12v/24v leisure batteries used in caravans and boats but there are alternatives such as the small 12v batteries you get on motorbikes (although I doubt they’d up to the job), the 48v batteries that you get on electric bicycles, and of course I should look at the batteries used on a typical wheelchair.

I don’t think it’s possible to decide on a battery until we know the power requirements of the motor though, which brings me back tot he topic of this post.

Motors

Motors are powered by magnetic fields. These magnetic fields are created by passing current through a wire. The stronger the current, the stronger the magnetic field. So as I said above, in very general terms, the higher the voltage rating of a motor, the more torque it has.

There are different types of motors such as brushed, brushless and servo motors amongst others.

According to sparkfun.com, brushed motors have the advantage of being simple to control, have excellent torque at low RPM and are inexpensive to manufacture. They sound perfect for our needs. The disadvantages of brushed motors are that the brushes can wear out over time, they generate electromagnetic noise and are usually limited in speed due to brush heating. As mentioned above though, speed isn’t an issue for us and brushes can be replaced. Noise could be an issue though as it’s going to be important not to upset other hill walkers.

Brushless motors are becoming more and more popular as they are reliable, efficient, produce high speeds, are mass produced and easy to find. They are however difficult to control without a specialised controller, and importantly they require a low starting load.

In comparison then, brushed motors are better for low speed torque, whereas brushless motors are better for high speed where torque isn’t a concern. Of the two it’s clear that a brushed motor is the more suitable.

The motor manufacturer MAHLE sell 24v DC motors with a power rating up to 3,5 kW. These motors are used in winches and the like. If they can pull a Landrover then they shouldn’t have difficulty pulling a wheelchair. However MAHLE also sell an AC induction motor which is specially designed for low-voltage applications and for use in electric vehicles. Although they use alternating current, they will still run off a direct current battery with the use of an ECU which converts DC to AC. MAHLE’s 24v AMT has a power rating of 5kw. Although I still don’t know what power rating is needed, just to give some context a child’s electric quad bike typically uses a 1kw motor.

MAHLE AC Induction Motor

1kw Quad Bike

In summary, this feels like progress towards choosing a motor and it seems the first step is to calculate torque requirements and then research AC induction motors.

Mountain Wheelchair

The Off-Road, All-Terrain, Mountain Wheelchair

Tag Archives: batteries