A little nervous I've underengineered my pack, thoughts?

I am in the final welding of my 11P 16S pack of 18650 cells (LG MJ1), and I am starting to wonder if I should go over the whole thing and weld even more nickel to the series connections, or mayve solder some copper wire to it (OR whatever you experienced pack builders might suggest!)

- I bought pure nickel strips from banggood.com. They are 0.1x8mm, and "rated" at 5A continuous
- The pack will be 60V, as you may have calculated already
- My motor is a QS 4kW hub motor, and the controller is a very capacle Kelly KEB70601. People say these motors are beasts and capable of much more than their nominal rated power, which is a plus, but of course one would have to accommodate this if one were to take advantage of it...

This is how I calculated my series connections:
1. 4000W/60V= 66.7A
2. 66.7A/5A per strip = 13.3 strips per series connection.
3. To give an example: some places I have only 5 cells making up a series connection. So what I did was calculate 13.3/5 = 2.67 strips on top of each other. Of course I rounded up to 3, so a total of 15 strips is making up the series connection. calculating backwards, this should be able to hold a continuous current draw of 15x5=75A. At no given place on the pack are there less than 15 strips making up a series connection.

But I am starting to worry about what bursts I will be seeing with my motor, and what pain they can/will cause to my battery pack? I've spent ages welding this thing together using my homemade welder, and I do not want to do the whole thing again. Nor do I want to ride my scooter 1km only to find that my battery pack is a huge bottleneck.

So I'm reaching out to you to see if you think I've underengineered my battery pack, or if you think it should be capable of handling bursts from acceleration up to say 150A (the BMS will have a 150A discharge limit, so this is what I will have to limit the whole system to anyway).

Project thread for the battery pack and the whole scooter is located here, for those interested in the looks of it:
https://endless-sphere.com/forums/viewtopic.php?f=10&t=89582

I will be very thankful for any input.

I think it really depends on how long you think you might "burst" to 150A. If we are talking a few seconds here and there, no problem at all. If you think you might burst for a few minutes while going up a steep hill, then there might be a problem with overheating of the nickel strips.

Here is a test I did on 0.15 * 8mm nickel. You can assume your 0.1mm nickel would be around 2/3 the performance.



So with 0.15mm * 8mm nickel @ 15A, the nickel reached a temperature of around 42C above ambient temperature. So in your case this might be a similar situation to running 10A on the 0.1mm * 8mm nickel. Which could approach almost near the boiling point in the right environment, which could start to melt plastics, heatshrink etc. But it took around several minutes to reach this peak temperature.

Also, how do you plan to do your final + and - connections on the pack?

Hi
You'll need to determine current draw based on the controller not the motor.... So what is your controller max amps?
Then apply nickel to series connects based in this. Check out the spot weld repository first page for amps nickel can carry but 4-5 amps is reasonable for .1x8mm. .2mm would have been a better option or even .15mm but no matter it's done now!
If you need to beef up current carrying solder some copper over the series connections to carry current. Thin strips of copper sheet works well ( carries 4x nickel for equivalent cross section area)

I think 60amps can put you on your axx 150anp full throttle is much to much for you. Seat over the front end squeeze your ass and put your head over the handlebars. Plus start on. Number 1 on 3 speed switch then they 2 0n three speed switch. O.K. oh OK. With helmet. And knee and elbow pads. Or try one first. First.

150A on a Kelly will be more like 80A on the battery. You can program the current limit lower if you run into heating issues but I think it will be fine if your cells are up to the discharge rate.

The biggest mistake is actually using the MJ1 cells at around 2C. Have you seen how much sag those cells have when being ran at 2C?
Those cells are kinda saggy at 3A.. quite bad at 5A, which will probably be continuous on your build:
http://lygte-info.dk/review/batteries2012/Common18650comparator.php
^-- check this

Personally i would run those at no more than 1C if you want them to last and deliver close to full capacity.. that would mean either running more in parallel.. or just choosing a different cell.

Using the nickel strips at their maximum rating is not the best idea either. But the hottest part of this setup is probably going to be your battery cells.

Another thing.. 60v x 65A = 3900W. This motor can produce much more than that in bursts ( go ahead and throw 100A at it ), and you're probably going to want to bump the current up further.
But as you go from beyond 2C continuous, this battery really starts turning into a saggy and inefficient space heater requiring active cooling if you want it to last more than 50-100 cycles.

I would put way more headroom in your battery pack. At a minimum for 4kw continuous, i'd be designing for 60AH with a cell like that.

On the other hand, a Samsung 25R pack of the size you're thinking will handle the same load pretty well. 38.5AH of samsung 25R would have a maximum output of 308A, so it would barely sneeze at being ran at 100A all day.

I checked them out compared to 30qs. Sure they're not as good for high discharge, but at 5amps they were fairing pretty well. Over 7amps- they start to look average and by 10 they're struggling. But this is continuous draw..and I don't think he'll be hammering at 3+kW all the time. Well on an uphill- yes, but if it's just cruising round town, on residential streets etc I reckon they'll be fine.
Guess it depends on the terrain/speed.

That motor will burst to 20,000 watts easily since my 205 can do 14,000 without breaking a sweat. As previously mentioned you need a 30 to 40 amp pack of some 15 to 20 amp per cell....cells.

Have fun!

Tom

Thanks for all your answers guys.

The controller is rated at 110A with 280A peaks (Kelly KEB72601). It is programmable, and I never intended to run my batteries at the limit of the controller. So I have decided that I will limit the battery current to a maximum of 150A within the controller software. Am I misunderstanding something here, is this not possible?

I didn't think my cells would be an issue, but who am I to know. This is what I thought when I decided what cells to use:
At nominal power (4kW), the current draw will be 66.7A. The cells are rated at 10A continuous discharge, and with 11P they will only see 6A each at nominal power. Almost half of what they're rated for.
Now, are you guys saying they can't really handle 10A continuous and still keep close to their rated capacity? Are you even saying they will get hot and/or damaged?
I did check them on the graph on lygteinfo.dk, but what am I looking for? All I can see from the chart is that (for example) for a 7A continuous draw their capacity will drop to 3.1Ah instead of 3.5Ah. I can live with that...

kdog said:

If you need to beef up current carrying solder some copper over the series connections to carry current. Thin strips of copper sheet works well ( carries 4x nickel for equivalent cross section area)

Click to expand...

I like this idea. Now that I have 3 to 4 layers of nickel on each series connection, maybe it's OK to solder a little on top of this? Plus, maybe I don't have to solder right over the cells, depending on what kind of strips I find.

redilast said:

Also, how do you plan to do your final + and - connections on the pack?

Click to expand...

I was planning on doing it like ebikeschool.com shows in one of his youtube videos; soldering the cables directly to the nickel on the + and - side of the battery, but only solder between the cells. Ideally I would attach them mechanically to the nickel strips, but I would have to have 14 layers of nickel welded on top of each other; not an option! Any ideas on this subject is also greatly appreciated...

On one of my packs, I used some heavy (8ga) copper wire along the ends and had the nickel strips extended so they could wrap around the wire, then I soldered it. Each cell has a tab going to the wire.

Fechter what kind of battery porn is that ? Can we at least see her shoulders or her knees the only thing I see is feet and ankles.
How about a little true strip tease you can keep the feathers up if you must. A little more then a tease.

999zip999 said:

Fechter what kind of battery porn is that ? Can we at least see her shoulders or her knees the only thing I see is feet and ankles.
How about a little true strip tease you can keep the feathers up if you must. A little more then a tease.

Click to expand...

Here's the whole thing:
https://endless-sphere.com/forums/viewtopic.php?f=14&t=79324

fechter said:

On one of my packs, I used some heavy (8ga) copper wire along the ends and had the nickel strips extended so they could wrap around the wire, then I soldered it. Each cell has a tab going to the wire.

Click to expand...

That's really slick! I could do something similar, but I will have a maximum of 5 cells I can "extend" out of on each side/pole, so it would still be 3 strips of nickel on top of each other.

EddySPalm said:

Now, are you guys saying they can't really handle 10A continuous and still keep close to their rated capacity? Are you even saying they will get hot and/or damaged?
I did check them on the graph on lygteinfo.dk, but what am I looking for? All I can see from the chart is that (for example) for a 7A continuous draw their capacity will drop to 3.1Ah instead of 3.5Ah. I can live with that...

Click to expand...

The problem is with how much of the pack's energy is turning into heat. Lithium does not like heat and will degrade at 160F and beyond.
Even at 1/2 the continuous maximum.. you are going to get a hell of a lot of heat.

Here's the thing. These cells are rated for their maximum continuous discharge based on the result of a single cell in a laboratory at room temperature. When you finish a pack, you have a bunch of cells heating each other. Enclose the pack in a box or with shrinkwrap and you retain even more heat.

Look at the 10A graph. The cell goes from 3.7v nominal to 3.3v nominal. ( get nominal voltage by looking at the voltage in the center of the graph )
If we take the nominal voltage x the capacity, we get: 8.58 watt hours.
But at 1 amp, we get 3.6v nominal x 3.2ah = 11.52 watt hours.

The difference between 11.52 watt hours and 8.85 watt hours? you're losing 25.6% of your cell's battery energy to internal resistance. Therefore, 25.6% of the battery output will be heat.
So let's say you've built a pack and are running the full 10A at 4000W. The cells will be producing 1000W of heat in an enclosed space. That will go kaboom in short order without liquid cooling.. and it'll be extremely inefficient. suddenly your wonderful 250whrs/kg pack is delivering an effective 187.5whrs/kg to your controller.

Ah, here's the other problem with that. You built a 17S pack thinking that 3.6v nominal x 17 = 61.2v nominal, right? but when the cells sag to 3.3v nominal, you get 56.1v nominal and take a hit in performance and speed. You have a major resistance problem here. It is like running 40 amps through a 12 gauge wire.

Now let me make this simpler for you so you don't have to sit around with a calculator.
The thing you want to look for when figuring out how a cell with perform at a certain amp level is the voltage sag. The voltage sag is an indicator of resistance going on ( just like in a wire ), which indicates a loss in efficiency and an increase in heat.

Compare the 0.2A and 5A discharge graphs to see what the voltage drop is relative to baseline performance at a super wimpy discharge..
Notice that we've got a difference of 3.45v to 3.7v, right in the middle of the discharge? well, a drop more than 0.2v is quite big. We've got a 0.25v loss though.

Let's do the math on the 5A scenario, using the voltages in the middle of the graph as the nominal voltages.

3.2AH x 3.7v = 11.84watt hours at a 0.2A discharge.
3.125 ah x 3.44v = 10.75 watt hours at 5A discharge.
We've got a 9.3% loss of energy to heat, so at 4000W, we're making 372W of heat, which is still pretty bad.

You might say.. 'yeah, i can figure out how to disperse 372W of heat in this pack'.. and you could be right, when the pack is brand new.. but as it ages, it's discharge rating is going to go down as you put more cycles and calendar life on it.

I have 7 year old 20C turnigy packs here and their internal resistance has gone up 5X. They're effectively 4C rated packs by now. But i've always used them at 2C maximum, so i've always had lots of headroom. The amount of headroom you have is zero here. But your pack is going to need some form of cooling right out the gate, whereas my massive headroom has allowed me to just put my packs in a bag and forget about the thermal considerations.

I would design your pack with more cells in parallel or start thinking about a cell with lower internal resistance. Samsung 25Rs are the champion in this regard.. but there are a few cells that are closer to a middle ground. Maybe the Samsung 30Q is better? look for no more than 0.2v drop when you are comparing discharges at various C / amp in order to find the sweet cell for your needs.

Oh, one last thing.. with a 4kw rated motor on hand, you are going to want to push it beyond what you're thinking.. electric power is addictive

In this scenario, your controller runs super cool because you're at half it's limit.. the motor runs barely warm because your peak power is right at it's rated power.. and your battery after being fully discharged makes a thermal cam cry

neptronix said:

lots of clever words

Click to expand...

First of all, I need to thank you for taking the time to educate me right here in my own thread, instead of just pointing me to the great wide web. Things become much clearer now, and I realize that however I look at this, I would have been better off with another cell.

There are a few things I am wondering about, still... Where's the time factor in all these calculations?
The tests where the cells have been ran at 0.2V and 5V lasted for 16hrs and 1hr, and never will I run the motor at close to it's nominal power for a full hour. It's a moped, and limited by law to go no faster than 50km/h. Yes, I should have mentioned that.

And will I actually produce 1000W of heat the second I start drawing 10A? I'm not so sure.. The 25.6% difference between the 0.2A test and the 5A test is given in capacity, and so the 1000Watts is the average heat loss between the two. The 10A test lasted for what looks like 15 minutes, so it had significantly more than I would call the duration of a burst to heat up. Also surely, heat builds exponentially, so not only will the heat loss start very low, I believe it won't have to build up much at all during the 5-10seconds I imagine a burst will last.

What I will do now:
I will complete the welding on the pack and take it to my brothers' work, where he has a 150A power supply. I will run a duration test at 1C and 2C (I imagine) while monitoring the temperature of the pack. Then I'll simulate bursts of up to 150A. I will do this primarily to look for rising temperatures, I won't be doimg my own capacity calculations (unless my brother does them for me ) I'll use probes inside the pack and a handheld FLIR camera to get an overall condition of both cells and nickel strips.

I will post the results and photos here, but be patient, I don' t even have a charger or the correct BMS yet.

Kelley controllers are generally rated in motor current. Battery current is lower by about half. That's probably what folks were hinting at above.

EddySPalm said:

neptronix said:

lots of clever words

Click to expand...

First of all, I need to thank you for taking the time to educate me right here in my own thread, instead of just pointing me to the great wide web. Things become much clearer now, and I realize that however I look at this, I would have been better off with another cell.

Click to expand...

You are welcome. I've been repeating myself on this for many years, but many vendors have been outyelling me these days.. ES has a smaller reach than product websites though.

There are a few things I am wondering about, still... Where's the time factor in all these calculations?
The tests where the cells have been ran at 0.2V and 5V lasted for 16hrs and 1hr, and never will I run the motor at close to it's nominal power for a full hour. It's a moped, and limited by law to go no faster than 50km/h. Yes, I should have mentioned that.

Click to expand...

Woah, that flips things on it's head. I expected that you'd have selected such a huge controller and motor because you needed one that size.
Are you just looking for mega insane torque or something? because a 1.5kw rated motor and 80A rated controller would do exactly what you're looking to do. For example, i run a 1.5kw rated leaf ( more like 2000w rated actual ) at a peak of 4000W, but the 4000W only happens when accelerating or climbing a hill, otherwise i'm cruising at 1000-2500W or less. A pack like yours would handle that.

Okay, so a moped is going to have thick tires, yes? I'd imagine your constant current draw at 50km/hr would be around 1300-1500W or so ( using the ebikes.ca simulator with a 330lb bike to try to emulate the tire friction ). If that's the case, you're only loading this battery at 1C or under which would be totally fine. It will heat up and cool down at various periods and probably be okay.
The most annoying thing will be the loss of brute acceleration under throttle at high load from voltage sag.
But with what you're talking about, your setup is actually okay.

And will I actually produce 1000W of heat the second I start drawing 10A? I'm not so sure.. The 25.6% difference between the 0.2A test and the 5A test is given in capacity, and so the 1000Watts is the average heat loss between the two. The 10A test lasted for what looks like 15 minutes, so it had significantly more than I would call the duration of a burst to heat up. Also surely, heat builds exponentially, so not only will the heat loss start very low, I believe it won't have to build up much at all during the 5-10seconds I imagine a burst will last.

Click to expand...

That's 10A per cell i was talking about, multiplied over how many cells you have, to give an example that 10A capability of a single cell is pretty laughable when thinking of running that way in an assembled pack. Yeah, a single cell can discharge it's heat in a lab at full tilt while barely avoiding going pyrotechnic.. and resellers market the cell accordingly.. it's just that once you have a big pack assembled, it's another story.
Don't worry about the time the tests went on.. basically all these tests you see on lgyte-dk and the battery cell graphs provided by manufacturers etc take the battery from fully charged to fully discharged. At lower rates, it's of course gonna take eons. The most important thing to note is how much voltage sag is going on.

What I will do now:
I will complete the welding on the pack and take it to my brothers' work, where he has a 150A power supply. I will run a duration test at 1C and 2C (I imagine) while monitoring the temperature of the pack. Then I'll simulate bursts of up to 150A. I will do this primarily to look for rising temperatures, I won't be doimg my own capacity calculations (unless my brother does them for me ) I'll use probes inside the pack and a handheld FLIR camera to get an overall condition of both cells and nickel strips.

I will post the results and photos here, but be patient, I don' t even have a charger or the correct BMS yet.

Click to expand...

Good thinking. Hey, i hope you stick around.. you have the right scientific mind for this kinda place.

Me? i don't do any of that. I just wildly overengineer things and see how long i can run them :lol: generally what i aim for in a battery pack is to use 1/4th of the maximum rating continuously. Almost every pack i've built under that thinking runs ultra cool and lasts a long time.
Your use should run just a bit warm, i imagine.

neptronix said:

Woah, that flips things on it's head. I expected that you'd have selected such a huge controller and motor because you needed one that size.
Are you just looking for mega insane torque or something? because a 1.5kw rated motor and 80A rated controller would do exactly what you're looking to do. For example, i run a 1.5kw rated leaf ( more like 2000w rated actual ) at a peak of 4000W, but the 4000W only happens when accelerating or climbing a hill, otherwise i'm cruising at 1000-2500W or less. A pack like yours would handle that.

Okay, so a moped is going to have thick tires, yes? I'd imagine your constant current draw at 50km/hr would be around 1300-1500W or so ( using the ebikes.ca simulator with a 330lb bike to try to emulate the tire friction ). If that's the case, you're only loading this battery at 1C or under which would be totally fine. It will heat up and cool down at various periods and probably be okay.
The most annoying thing will be the loss of brute acceleration under throttle at high load from voltage sag.
But with what you're talking about, your setup is actually okay.

Click to expand...

Yeah, ahem, things kinda just escalated partswise. The thing is: I'm converting a 1970 Vespa 50cc. In EU, for some strange reason, electric mopeds can have motors rated for 4kW's nominal (while gas powered mopeds are usually restricted to around 1.8, and that's peak!). So I "maxed out" on the nominal power because why not from there, I asked QS motors if they could match me up with a controller and they sent me the 72601.

That's 10A per cell i was talking about, multiplied over how many cells you have, to give an example that 10A capability of a single cell is pretty laughable when thinking of running that way in an assembled pack. Yeah, a single cell can discharge it's heat in a lab at full tilt while barely avoiding going pyrotechnic.. and resellers market the cell accordingly.. it's just that once you have a big pack assembled, it's another story.
Don't worry about the time the tests went on.. basically all these tests you see on lgyte-dk and the battery cell graphs provided by manufacturers etc take the battery from fully charged to fully discharged. At lower rates, it's of course gonna take eons. The most important thing to note is how much voltage sag is going on.

Click to expand...

Again, I'm so glad you directed my attention towards more than just the capacity&ampacity markings on cells. I'll humbly admit I've never even heard of voltage sag, and I'll put a lot more research into cell choice and pack configuration on next round of pack production! Let's hope the pack lasts a few years at least, so maybe even some new tech has reached the market by the time I need to replace it!

Good thinking. Hey, i hope you stick around.. you have the right scientific mind for this kinda place.

Me? i don't do any of that. I just wildly overengineer things and see how long i can run them :lol: generally what i aim for in a battery pack is to use 1/4th of the maximum rating continuously. Almost every pack i've built under that thinking runs ultra cool and lasts a long time.
Your use should run just a bit warm, i imagine.

Click to expand...

I'm not going anywhere 8) Though I am really impressed with people like you, with 12 000 posts and still you're writing text books' worth on info in a single post to a single newbie. Kudos to you!

First test done. We had to quit the test when the BMS died on us at 110A.
Unfortunately I have not made nice graphs like Henrik at lygte-info.dk does, but I have notes, pictures and videos.

What I intended to test:

1. Low load over enough time to prove the functionality of the battery pack, and to find any super weak spots:
- We took the battery thorugh 10 minutes of 20A current draw. The temperature rose 1 or 2 degrees after 5 minutes, and stayed constant for the rest of the test.

2. Nominal load for as long as we feel the battery pack is safe and sound:
- We ran the pack at 70A for 15 minutes.
- The voltage sag was 8.12V (started with battery charged to 66V).
- Temperature rose for what seemed linear for the first 10 minutes, to about 40 degreesC , then started rising quicker, or exponentially for the next few minutes, until we measured a maximum of 60 degrees C (pointing the thermocamera well into the core of the pack) and that's where we ended the test.

3. Peak load110A.
- Lasted for about 30 seconds, and that's when my (150A!!) BMS broke a transistor.
- No significant temperature rise during this time, though we did start at about a 40 degrees C core temp as we didn't take the time to wait for it to cool to room temp.
- Voltage sag was 10A

Conclusion:
- I have definitely chosen batteries that are not a good match for my motor and controller, given their capacities.
- I will definitely have to program the Controller to hold back a lot on current draw, both peak and continuous.
- I feel safe that this pack will do fine in a moped like it is supposed to do.
- A thermistor inside the core of the pack will be implemented and carefully monitored.
- Some cooling might have to be implemented, but I don't know what or how yet.

The end...?

I will attach some pictures of the pack/test equipment and eventually some pictures from the thermocamera tomorrow.

Video of transistor "blowing up (1m 25s, roughly):
https://www.dropbox.com/s/nd3348g3sw4wick/2018-02-23%2020.05.21.mp4?dl=0

Broken transistor:


Stats at 70A:


Test equipment:



Pack and BMS: