Article Next leap in battery performance for ebikes?

[knut7]

In 2013, Bosch released the 400Wh battery. It was a 40 cell battery, and it was their biggest battery for a few years. Already in 2015, the lithium batteries were drastically improved. So Bosch launched the 500Wh battery for 2016. It was still a 40 cell battery. And pretty much nothing has happened since.



When I talk about battery improvement in this context I’m talking about “energy density”, how many Watt-hours we get per kilo battery. Even the new 625 and 700Wh batteries have pretty much the same energy density. So when will we see the next significant increase in energy density? I share my thoughts in this video.

Continue reading...

 

[steve_sordy]

Thanks for the work done Knut, well worth it and a very interesting summary.

 

[RazorBlade]

Any increase in capacity should also be coupled with increased motor efficiency, thus will we need bigger heavy battery's, or just make the the batter smaller and lighter, and increase bike efficiency?

 

[Mabman]

Awhile ago I got pronged on a post here for saying that the 700wh Levo battery was due to 21700 cells.....it's really the only thing that made sense and I always thought it odd that Spec and the others that are using the same cells to increase their wh's are not advertising the fact? Not a huge gain but significant enough especially in regard to C rate.

 

[Beekeeper]

Great article Knut. Thanks

 

[Loamranger]

Thanks for posting

 

[wildsau2]

Bike manufacturers should care about standard formfactors for their batteries. Customers should get batteries for 10 years for their bikes. Does anyone know of laws or regulations (EU, US,...)? thx

 

[knut7]

Any increase in capacity should also be coupled with increased motor efficiency, thus will we need bigger heavy battery's, or just make the the batter smaller and lighter, and increase bike efficiency?

I guess there is some gain to be made by improving motor efficiency, but the biggest gain is in battery tech. When we get new battery tech, the ebike-motor/battery manufacturers can decide if they want to increase capacity or reduce weight. Up until now, they have mostly picked range over weight, but that might change in the future. With an energy density of 310Wh/kg, the Bosch 625 battery would be about half a kilo lighter.

 

[knut7]

Awhile ago I got pronged on a post here for saying that the 700wh Levo battery was due to 21700 cells.....it's really the only thing that made sense and I always thought it odd that Spec and the others that are using the same cells to increase their wh's are not advertising the fact? Not a huge gain but significant enough especially in regard to C rate.

The 21700 cells have higher charge and discharge rate as you say. But they didn't necessarily have to use the 21700 cells. Greyp has a 700Wh battery too, it's made by 60 pieces of 18650 cells! That should indicate a weight of over 4kg, but I weighed it to just 3,58kg, which is lighter than the Spesh 700Wh. And it is really small. I'm not sure what they have done, but I have a theory. To get 700Wh from 60x 18650 cells, they must have cells of about 3300mAh cells. A less energy dense cell will typically have better charge+discharge rate. Even though the 18650 cells has lower charge+discharge rate than the 21700, Greyp makes up for it by using the lower density cells in a 10s6p config, meaning 6 cells in parallell. This increases the charge and discharge rate. The power draw will be spread over several cells, so the cells will be nowhere near their limit. This means less heat developing, so Greyp could pack the battery thighter with a lighter case. No need for internal vents/ducts/etc.

Hopefully I didn't misunderstand your post completely!?

 

[wildsau2]

what about 48V systems belonging efficency? thx

 

[knut7]

what about 48V systems belonging efficency? thx

48V is more efficient because of lower current draw compared to a 36V - at the same power output. But this is usually only an issue when the batteries are being pushed close to their maximum discharge capacity. I think the biggest advantage of the 48V is the ability to put out a decent amount of power in the mild (low power) emtbs, shuch as the Levo SL. The small range extender would most likely not work in a 36V system, it would have just 10 cells in series and wouldn't be able to deliver enough current. But with a 48V system it does work, even if it's on the limit.

 

[Flatslide]

48V is more efficient because of lower current draw compared to a 36V - at the same power output. But this is usually only an issue when the batteries are being pushed close to their maximum discharge capacity. I think the biggest advantage of the 48V is the ability to put out a decent amount of power in the mild (low power) emtbs, shuch as the Levo SL. The small range extender would most likely not work in a 36V system, it would have just 10 cells in series and wouldn't be able to deliver enough current. But with a 48V system it does work, even if it's on the limit.

Another advantage of higher voltage battery packs is the ability to be charged faster without stressing the cells, as I understand it. Certainly my 48V pack charges quickly in comparison to 36V times.

 

[knut7]

Another advantage of higher voltage battery packs is the ability to be charged faster without stressing the cells, as I understand it. Certainly my 48V pack charges quickly in comparison to 36V times.

Yeah, that's right. It works both ways. Charging a 48V battery with 4A will create about as much heat loss as charging a 36V battery with 4A. But the 48V battery is charging at higher power.

 

[boBE]

48 V systems are also more efficient at each step after the battery: wiring; connectors; motor windings; due to lower I*squared*R losses. At some point the regulations require greater protection, I do not know at what voltage this comes in.

 

[Mabman]

The 21700 cells have higher charge and discharge rate as you say. But they didn't necessarily have to use the 21700 cells. Greyp has a 700Wh battery too, it's made by 60 pieces of 18650 cells! That should indicate a weight of over 4kg, but I weighed it to just 3,58kg, which is lighter than the Spesh 700Wh. And it is really small. I'm not sure what they have done, but I have a theory. To get 700Wh from 60x 18650 cells, they must have cells of about 3300mAh cells. A less energy dense cell will typically have better charge+discharge rate. Even though the 18650 cells has lower charge+discharge rate than the 21700, Greyp makes up for it by using the lower density cells in a 10s6p config, meaning 6 cells in parallell. This increases the charge and discharge rate. The power draw will be spread over several cells, so the cells will be nowhere near their limit. This means less heat developing, so Greyp could pack the battery thighter with a lighter case. No need for internal vents/ducts/etc.

Hopefully I didn't misunderstand your post completely!?

Certainly you can add more cells to achieve what you say. My point was that Specialized and others than GreyP are making cell packs that fit into the same integrated frames with increased wh's and the reason they can do so is by using 21700 cells. It's a smart move and I think more will come in the future. As well as upping voltage which is starting at appear from the major manufacturers for reasons as noted above.

I have 4 bikes w/ 21700 cell batteries that reside in external cases that total about 2900wh's between them that cost me $2000. Aside from one 52v the rest are 48v. So using that as a metric I can't see that 21700 battery packs should cost significantly more to produce either.

 

[knut7]

48 V systems are also more efficient at each step after the battery: wiring; connectors; motor windings; due to lower I*squared*R losses. At some point the regulations require greater protection, I do not know at what voltage this comes in.

Yeah, it's the Amps that is the limiting factor. Increased current means increased heat loss due to increased resistance. That's why a battery works fine at high state of charge, and gets really warm at low charge. Voltage is higher in the beginning and current draw is lower. And, as you say, this happens in the entire circuit. But it it shouldn't be much weight or cost added having power wires big enough so there is no significant increase in resistance at max current draw.

As you say there are stricter regulations at some point. The IEC (International Electrotechnical Commission) has a classification called ELV (Extra Low Voltage). A device cannot exceed 50V to be classified as ELV. But not all agree. The IEC defines ELV as sub 50V AC and sub 120V DC!. And the EU commission don't talk about ELV, they have a Low Voltage classification that starts at 50V AC and 120V DC.

So it seems there could be complications when servicing a 48V system, as this will exceed 50V at full charge. But I'm not sure how the various regulations apply in this case.

 

[knut7]

Certainly you can add more cells to achieve what you say. My point was that Specialized and others than GreyP are making cell packs that fit into the same integrated frames with increased wh's and the reason they can do so is by using 21700 cells. It's a smart move and I think more will come in the future. As well as upping voltage which is starting at appear from the major manufacturers for reasons as noted above.

I have 4 bikes w/ 21700 cell batteries that reside in external cases that total about 2900wh's between them that cost me $2000. Aside from one 52v the rest are 48v. So using that as a metric I can't see that 21700 battery packs should cost significantly more to produce either.

The reason brands like Specialized can fit bigger batteries into existing frames is that the old 500Wh battery was made bigger than necessary. The 500Wh Specialized battery is way bigger than the Shimano BT-E8035 released last year. Shimano is sort of doing the same thing for 2021. Their new 630Wh battery is longer than last year's 504Wh battery. So for 2021 Shimano introduces the 504Wh BT-E8035L. This is the same size as the 630Wh battery, so they will both fit the same frames.

That being said, I agree with you. Specialized might not be able to stack 60 cells in a battery and fit it into their frame, certainly not with the layout used by Greyp. And yeah, the 21700 based batteries doesn't have to be more expensive. It's still the same chemistry.

 

[wepn]

the 21700 based batteries doesn't have to be more expensive

Very true in fact theoretically they should & may already be less expensive per kWh.

The extra 5 mm of length and 3 mm of diameter means almost 50% more volume for anode, cathode, and electrolyte in every unit of production. The benefits cascade - from cost of production and distribution as well as energy densities, runtimes & mass. Also less complexity and failure modes in higher capacity battery packs.

 

[Paul Mac]

This is all just variables on the technology we have today.
I want to see what the future will look like.
Already Tesla have introduced a new way of making the batteries for their cars using dry anodes.
This process gives a significant improvement on charging speed, battery life, power and weight.
It will be the car manufacturers that will push the science that we will all reap the benefits of later.

 

[MrSimmo]

Graphene based Batteries will be a massive advancement, or Silicon-Graphene Anode as a discreet application.

 

[wepn]

This is all just variables on the technology we have today.
I want to see what the future will look like.
Already Tesla have introduced a new way of making the batteries for their cars using dry anodes.
This process gives a significant improvement on charging speed, battery life, power and weight.
It will be the car manufacturers that will push the science that we will all reap the benefits of later.

Well it's variation in format that will have most impact on the next bikes we buy. Variation in chemistries will follow.

Tesla acquired Maxwell & along with it their dry electrode tech. The most obvious advantage is the manufacture of electrodes without hazardous solvents and complex expensive equipment. Apparently allows for higher electrode lithium content which extends expected lifespan & improves capacity. Of course new chemistries & formats usually take quite a while to appear in bikes.

 

[Flatslide]

Here in NZ, as I understand it, up to 80Vdc is okay to work with but over 80Vdc and any AC system requires a certified electrician. I'm not sure if any of you watch Robert Murray-Smith's channel on Youtube, he is a chemist who works with graphene batteries amongst other graphene products. I've learned much from watching his videos.

 

[knut7]

This is all just variables on the technology we have today.
I want to see what the future will look like.
Already Tesla have introduced a new way of making the batteries for their cars using dry anodes.
This process gives a significant improvement on charging speed, battery life, power and weight.
It will be the car manufacturers that will push the science that we will all reap the benefits of later.

Yeah, as I said in the video, several companies are developing the current battery tech. I set out to find what would give us ebikers the next leap in, well, energy density. And as far as I can tell, that has to be an improved lithium-ion battery. I also mentioned the Maxwell dry electrode technology. And the tabless patent. It is really interesting, I'm excited to see if they exceed 300Wh/kg. We might get the answer in a couple of months.. But the problem is Tesla is the worlds largest consumer of batteries. So I assume they will use all they can produce for their cars and energy storage products. I believe the NCM811 will be the first improved battery tech to reach ebikes.

 

[knut7]

Graphene based Batteries will be a massive advancement, or Silicon-Graphene Anode as a discreet application.

Graphene batteries could be in mass production in a couple of years. It's expected to be an anode using "few layers" graphene. Or "nano platelets" , which is close to graphite. So they won't be using the true, single layer graphene. This is expected to improve charge rate and lifespan, but not energy density. And the complicated process of producing graphene means the cells will be more expensive.

I'm excited about graphene, and it seems it's finally becoming a reality. But I think it will have bigger impact in other areas than batteries. I might of course be wrong.

 

[csj]

48V is more efficient because of lower current draw compared to a 36V - at the same power output. But this is usually only an issue when the batteries are being pushed close to their maximum discharge capacity. I think the biggest advantage of the 48V is the ability to put out a decent amount of power in the mild (low power) emtbs, shuch as the Levo SL. The small range extender would most likely not work in a 36V system, it would have just 10 cells in series and wouldn't be able to deliver enough current. But with a 48V system it does work, even if it's on the limit.

Yeah, that's right. It works both ways. Charging a 48V battery with 4A will create about as much heat loss as charging a 36V battery with 4A. But the 48V battery is charging at higher power.

Neither these claims is correct. Pack voltage has negligible impact to efficiency, power delivery, or charging.

The only way to compare packs of different voltages is to consider different configurations of the same number and type of cells. This is where the mistake is being made here. A common example being used is the 40 cell 36V high energy pack already mentioned, so start with that. This pack has a 10s4p configuration, cell blocks of 4 parallel cells are wired in a series of 10 blocks.

Now consider an 18V variation, 5 blocks in series but 8 cells per block. Still 40 cells, still 500 Wh, still the same cell type. Twice the Ah rating, half the voltage.

What the maximum power draw? The same, because the cells and cell count are the same. What's the efficiency? No matter how you define that, it's the same, because for the same draw, the same power is demanded of each cell and same internal losses occur. How much heat is generated when charging? The same, assuming the same charging power.

The internal resistance of one of these cells is roughly 30 - 40 mOhms. That means the 36V pack has an internal resistance of 80 - 100 mOhms. The 18V pack though, has an internal resistance of only 20 - 25 mOhms because it has half as much resistance per block and half as many blocks. Therefore, at the same charge current (the claim above) the 36V generates 4 times the heat of the 18V, not "about as much".

Now, of course, the 18V is being charged slower, and when you double the current to fix that, the heat generated becomes equal (because it scales with the square of current). Why does this work out to be equal? Because you have the same number and type of cells. The charging claim above is simply wrong.

Regarding the first claim that "48V is more efficient", that is also wrong for similar reasons. Yes, a higher voltage pack provides less current for a given power, but that does not mean it is more efficient. A higher voltage battery, all else equal, will have higher internal resistance that will exactly offset the reduced current and produce identical internal losses, by definition.

There are other claims here that higher voltage improves efficiency outside the battery. Broadly speaking that is not always true, but at lower voltages it is but only insignificantly.

Voltage choices in low power eBikes are tactical, one is not better than another.

 

[csj]

Another example that may be more interesting is to compare the same 36V pack to a 48V. A 48V pack has 13 series groups which will not divide evenly into 40 cells but will with 39 cells. Comparing a 40 cell 36V pack to a 39 cell 48V is pretty close to the same (2.5% difference).

Assuming a cell resistance of 36 mOhms, the 36V pack will have a resistance of 90 mOhms (10 * 36/4) and the 48V pack will be 156 mOhms (13 * 36/3). The 48V pack will have 73% more resistance and generate 73% more heat at the same charge current. Now, it will charge faster because of the higher charging power, but it will get a lot hotter doing so and may suffer reduced cycle life.

Also, because of the higher voltage the 48V battery will deliver less current for the same power and heat losses due to the reduced current will only be 59% of the 36V pack, but the 36V pack has only 58% of the resistance so the I2R losses are essentially the same (difference is due to 39 vs. 40 cells).

This is all because the only thing that matters is the type and number of cells you have and not how they are wired within the pack.

 

[csj]

The small range extender would most likely not work in a 36V system, it would have just 10 cells in series and wouldn't be able to deliver enough current. But with a 48V system it does work, even if it's on the limit.

This is also not true. The range extender in the SL is a 13s1p configuration that uses a 3.4 Ah 18650 cell. There are a several possible cells that could be used that have the required 7 - 10 A rating which that battery is capable of.

Now, if you delete 3 cells to make a 36V battery, you'd have to up the current draw to 9A for equivalent power. 9A is still doable but worse because of the smaller battery, but if Spec had simultaneously switched to a 2170 cell (5 Ah Samsung 50E) they would have added 10% more capacity and had arguably more current headroom than the existing product.

Again, this conclusion is wrong because you compare different sizes of packs and attribute the result only to voltage. Also, there are a whole host of power tools that deliver more power out of smaller lower voltage packs than the range extender. That should tell you whether a 36V pack could be made to work.

 

[RobertR]

Are there any graphs/data showing what the max amp draw is for each brand motor?

Once you have that then you build a battery to those specs that will be the most efficient etc.

 

[knut7]

@csj I'm talking specifically ebike batteries using 18650 and 21700 cells, so we're limited to these cells and 10s or 13s configs. Not sure why they're not using other formats, like the 20650 etc. Therefore there are situations where the 48V system has an advantage. I still believe the Specialized range extender battery wouldn't be possible in a 10s1p config, based on my experience with it. The battery seems to be at its limit at low SOC, and the motor power ramps down. Sure, they could make it 10s1p, but it would have to ramp down way sooner. And at lower cell temperatures, the current output would be further reduced. Sure, it's unfair comparing a 13 cell battery with a 10 cell. But the range extender would be completely different in a 10s2p config. Chances are it would be too big/heavy for what they want to build, and it couldn't be brought on planes. I do of course agree that lower current output can be made up for by using more cells, like I wrote regarding 21700 Spesh (10s4p) vs 18650 Greyp (10s6p) batteries.

Here's my understanding of the advantages of using higher voltage. One advantage of a higher volt system is the ability to put out the same amount of power while drawing less current. This allows for a weight/cost saving using thinner cables/wires. But the distance between battery and motor is so short in an ebike, so this is a non issue. The higher voltage becomes more interesting when there's a limit to how much current you can draw. Like for EVS. A lot of 300A fast chargers are being put up now. They are rated at 150kW at 500V. But these support up to 1000V. The Porsche Taycan is an 800V system that should be able to hit 300kW. Now cable dimensions is a real concern too.

But I have been thinking about the effects of the rest of the system, and completely forgot about what happens when we connect cells with internal resistance in series. The internal resistance of the battery does of course increase.

 

[csj]

I'm talking specifically ebike batteries using 18650 and 21700 cells, so we're limited to these cells and 10s or 13s configs. Not sure why they're not using other formats, like the 20650 etc. Therefore there are situations where the 48V system has an advantage.

What situations? Why are we limited to 10s or 13s configs? Are you ignoring my 10s/13s example?

I still believe the Specialized range extender battery wouldn't be possible in a 10s1p config, based on my experience with it. The battery seems to be at its limit at low SOC, and the motor power ramps down.

What is your experience with that pack that leads you to make these claims? Are you an engineer? You make claims that require technical knowledge of the system that I doubt you have.

Sure, they could make it 10s1p, but it would have to ramp down way sooner. And at lower cell temperatures, the current output would be further reduced. Sure, it's unfair comparing a 13 cell battery with a 10 cell. But the range extender would be completely different in a 10s2p config.

Never suggested a 10s2p config but a 10s1p config with a 2170 cell. That would be superior to the current pack in power and capacity and only about 70g heavier. As for flight regulation differences, that's off-topic and unrelated to voltage. It's also trivial to address.

Here's my understanding of the advantages of using higher voltage. One advantage of a higher volt system is the ability to put out the same amount of power while drawing less current.

I already addressed this. A higher voltage system has higher battery IR that negates the advantages. If you disagree, then where is my explanation wrong?

As for the "thinner cables", you've already admitted that is irrelevant to ebikes. Yes, it matters for electric cars but that's not the topic.

But I have been thinking about the effects of the rest of the system, and completely forgot about what happens when we connect cells with internal resistance in series. The internal resistance of the battery does of course increase.

At the voltages and currents involved here, the external effects of different configurations are negligible. All that matters is what is happening within the battery, and that is literally nothing.

Different configs have different voltages and currents but also different IR and this different IR completely, mathematically negates any apparent benefit of different voltages and currents external to the pack. That's the entire point. You've recognized now that pack IR varies but you are still asserting benefits without considering IR. Why?

Finally, I believe no electrical engineer would make your arguments. Also, I am an electrical engineer.