Does Balance Wire Length Matter?
Introduction
This article is just a brief look into something that was brought up by a colleague, and it felt worth taking a look at in a more specific sense. There are a lot of areas of general electrical and mechanical knowledge that apply when working with PEVs (personal electric vehicles), but as niches grow, the knowledge in that niche becomes a bit more specialized, and the targets are more narrow. This is one of those cases.
T.L.D.R.
Not really, no. At least, not in this kind of application, with these kinds of lengths on these kinds of wires and conductors. And not to a degree that would actually have an impact on the charge lifespan of these kinds of batteries, or the total energy that they deliver to the ESC/VESC during the use of the battery.
Academically, and in line with the laws of physics, yes. But in a way that actually matters in practice? Not really, no.
Edit: Most of the answers to explain this, are probably best summarized by the Afterword by Battery Mooch at the end. He also mentions some things I missed.
Prior Knowledge
I’ve been building, repairing, and modifying electric skateboards at the Garage for almost 5 years now, and most of that time has been spent learning about, working with, and building lithium ion battery packs for these types of devices. Jobs would vary from the obvious Future Motion Onewheel to the more bespoke Kaly.NYC or LaCroix Boards eskates. Each device has its own vibe and build patterns, and each device usually has a way it’s made that suits the purpose more than it would something like a scooter. There are also trends in how certain things are done in the DIY/enthusiast space, but that’s more of a topic for another time.
This is all to say that electrically speaking, battery pack building and assembly is largely similar between eskates and onewheels, and so the approach has a lot of crossover.
The majority of batteries that power Onewheels and/or DIY/custom VESC based onewheels are all a variation of what I call a brick pack. A brick pack is essentially, a battery pack that’s shaped like a brick. Usually some kind of rectangular shape, or maybe a parallelogram, where the cells are placed close together (either in a frame or not), and the terminals usually end up close to one another.
Below is a Meepo electric skateboard battery, and it mostly illustrates what I mean.
That being said, this is the KIND of pack that’s generally in a VESC one wheel, and it’s the majority of the KIND of pack that’s built in the shop each week. And so, this article will take a look at one of those. The way balance wires are routed, chosen, soldered, managed, etc. are essentially the same here as they would be from any other similar pack.
Desired Outcomes
Desired outcomes sounds more fancy than it really is. Basically, when dealing with balance wires, there are a few things that take priority over others. Most things in DIY/custom building are like this, since nothing is ever perfect. Since nothing is perfect, one has to just prioritize things over others to make sure that those are taken care of before other things.
For me, the main outcome is to have balance wires go to the cell group they need to go to, to be connected there properly, and to be installed/placed in a way that reduces the risk of them either being broken or being short circuited. Both of those things would be bad. And when batteries are compressed into a box, like they are in VESC one wheels, then that’s something to really pay attention to.
After that, an important outcome is that the balance wires end up providing a reliable connection to the BMS so that it can monitor the voltages of the cells.
The Sample
Since this is a brief look, it’s just one battery. Specifically, this is one of the “GT ME4T Packs” that I offer at the shop. It’s a 20s2p pack that’s meant to fit into a Onewheel GT sized battery box for DIY projects. It’s made with the Samsung 50S cell, and of all the VESC one wheel packs I do, it currently has the longest balance wires (not including the split pack, which will be discussed separately).
This pack was wired for an ENNOID XLITE V3 BMS. The longest balance wire was approximately 410mm, and the shortest was approximately 190mm ‘
These balance leads are relatively long mainly due to the larger size of this pack over an 18s2p, and also due to the balance connectors needing some slack for how the GT battery box divider is.
Just in case anyone is wondering, there is indeed insulation between parallel groups. It may be a bit hard to see, because it’s a bit transluscent. I’ve tested it extensively against other insulation types, and arrived at this one for this (and another) specific purpose. It is more than adequate for this build due to its material properties, and it’s one of the few things I would rather not just openly share. The temp sensors here haven’t been adhered yet. That’s usually the last thing I do before wrapping up a pack (after testing), since I use a 2 part thermal epoxy to reduce the thermal lag and spread between the thermistor and the cell can.
By the way, while these are the longest balance leads in a VESC one wheel pack, they are far from the longest balance leads that are encountered in electric skateboard packs. Aside from all of the massive LaCroix batteries I’ve seen, a recent custom build was what I’d consider a “medium large” battery (18s5p). Incidentally, it was also wired for an XLITE BMS (bottom potted in thermal epoxy for more thermal mass).
Checking Voltage Discrepancies With A Multimeter
So, there are essentially, 3 places that one can measure/sample cell voltages.
The closest, obviously, is nearest the cells. That was done with my multimeter (Fluke 17B+ w/ ProbeMaster probes) at the nickel conductor balance tabs. I set the probe tips under the tab so I was measuring from the nickel and not the solder pool on the tab where the wire goes. I usually leave these probes in the drawer, reserving them for special cases. The tips are very sharp, and all contact surfaces are gold plated. The probes that came with the meter are fine, but these are much better, get stable readings more quickly, and have very soft leads that don’t pull the meter around. I highly recommend Probe Master if you’re using your multimeter all the time. The sharp tips also make probing very small connector tabs much easier.
Next, would be measuring at the most common place it gets checked, which is at the balance connector. This is measured across the entire length of the balance wire (26AWG copper wire), and through the JST-PH crimp, since the probe touches the tinned copper crimp bit.
Last, would be the reading of the voltages by the BMS itself. Keep in mind, that this reading is done differently than it would be through a multimeter, which itself, is an important thing to note. It’s here that the actual discrepancies probably begin to add up.
Below are the readings I got with the meter, both at the nickel and at the balance connector. Forgive the rudimentary chart. Squarespace doesn’t have a chart function and so I just put the values into Excel.
As can be seen, there are some differences in the reading at the millivolt level (thousandth of a volt). And at first, it appears that there may be SOME voltage drop. That is, until you see some voltage rise at end of the balance lead.
So, what I’m guessing I’m seeing here are the effects of the 0.5% accuracy of the Fluke 17B+ meter.
In actuality, a millivolt doesn’t impact anything. At least, not in these cases where there is variation not only in the equipment, but in the cells themselves. What I mean by that, is that cell capacity can vary by several milliamp hours, and the same batch of cells being “full”, would net an even larger spread in voltages as a result of that.
So in practice, the differences seen in this chart aren’t significant. That is, if they weren’t probably the result of component accuracy (or lack thereof).
Does The BMS Care?
No one is manually checking cell voltages once a device is made, closed up, and deployed into use. Most folks don’t actually care. Plug it in, charge it, use it, repeat.
So, what the BMS sees would probably be more important to the overall lifetime of a battery pack, and in the case of this BMS, we can get that information relayed to us.
The following image is a screenshot take of this exact battery, immediately after it was plugged into the BMS and VESC tool opened. No charging has happened at all.
So what gives? Well, a lot of things. BMS’s (battery management systems…s) are complex devices. And they’re made in various ways. However, what they need to do, is convert the analog signal of a voltage (3.45443435546 volts or whatever) into digital signal (binary, usually) that the IC (integrated circuit) can understand and do something with.
And so, there’s usually some sort of analog to digital converter that does this, and there is more than one kind of those that a BMS can use. Some are all included in an ADC chip, some are made from scratch from discrete components. It depends on the design, the designer, and what they want to do.
These converters themselves can introduce signal noise, or have readings that are slightly off from a true absolute reading (not that those actually matter here). The signal, however pure it may be from the wire, will get filtered through countless components, each with their own variance in accuracy and fidelity.
Here’s a good video on that sort of thing, if it interests you:
What mostly matters when it comes to cell voltage sensing and measuring, is having a configuration that keeps the cells in a safe range.
Generally speaking, it’s really just about the range.
Lithium ion cells are “full” at 4.2 volts. That’s not necessarily 4.200 volts. It could be 4.204, or 4.227, or 4.199, or 4.244 volts.
4.3v and above is generally not good. 4.4v is bad. 4.5v is a lot of heat you don’t want, and chemical damage that’s probably irreversible and legitimately dangerous. Very bad.
Those same cells are “empty” at 2.5 volts. Most devices don’t drain cells that low, they’ll drain them to 3v. However, it doesn’t actually matter (in this context at least) if the cell stops being used at 2.853 volts or 2.940 volts.
It should NOT be drained below 2.5 volts, and so 2.205 is probably bad. That’s another area where chemical damage will occur.
Also, 1v for a cell is bad.
What Does It Mean?
So, while the length of a conductor can impact the resistance, in this specific case, at these lengths, it doesn’t seem to be enough to impact voltage readings more than all of the other factors along the conductor path.
Solder, since it’s a mix of tin, lead, silver, etc., will have impacts on resistance across the balance connection. So will the crimps that terminate each lead into the connector, and that connection to the header.
Note below the photos of the common JST-PH female connector, and how little contact is actually made with the header pin. Those connections aren’t gold plated, they’re tinned copper pieces, and so those small contact areas themselves are likely to introduce more discrepancies than the wire itself.
And even then, the variances in readings that these things introduce don’t actually have an impact on the overall function of the pack, the BMS, or the life of the pack in these kinds of devices.
Especially when any given BMS has a large spectrum of ADC component selection and quality, and that voltage readings are often subject to noise from current flow of different amperages, as well as the internal resistance of the cells themselves, since their electrochemical makeup will resist the flow of current to some degree.
And, since the BMS units used in these applications use passive balacing, the current flowing through these wires are usually between 20-50 millamps (thousandths of an amp), and so even at 26awg, the voltage drop across these conductors isn’t significant enough to have an impact on the readings through the entire maze of components. Those components generate noise, as does the current flow happening all around them via the traces of the BMS itself. There are many strenghts of electrical fields present in a complex PCB, and those can (and do) affect the fidelity of the signals.
Here’s some observable noise:
Conclusion
Having a conclusion is probably a bit of a misnomer here. Usually things like this are pretty open ended, since new information can often change existing opinions.
In this case, as far as these sizes and implementation of packs go, it seems currently like there are probably more pressing issues to focus my attention on.
Afterword by Battery Mooch
I agree with your conclusions but for different reasons…
I don’t think any of the multi-channel BMS controller chips and MCU-only BMS designs measure voltage while balancing current is flowing so balance wire resistance doesn’t matter. These BMS’ read the voltage between pulses of balancing current. This covers most of the BMS’ being sold.
The one-chip-per-channel China overvoltage chips used for balancing in the ZBMS (and stolen designs based on it) and a few very cheap BMS’ do read the voltage while balancing. They will stop trying to drop the cell voltage early because of the balancing wire resistance causing a voltage drop. But as soon as the balancing current stops the voltage shoots back up and the BMS starts balancing again.
This could extend balancing time a little bit depending on what stop/start delays are built into the BMS. But in the end, wire length doesn’t matter here either.
What can affect balancing is, as you noted, the voltage reading accuracy and the amount of electrical noise on the balancing wires. Additionally, some BMS have a fairly complex balancing algorithm and that can affect how well a pack balances out in certain situations. For most setups this isn’t a concern though IMO.
