Automatic loading setup for an autonomous vehicle - Will this work?

jb84

10 µW
Joined
Jun 5, 2023
Messages
5
Location
Bonn
Hello,

I'm currently trying to build an automatic charging module for an autonomous RC vehicle. Before you notice it through my questions, let me say that I have very little experience on the field of charging and that I am aware of the risks.

The battery to be charged is a 4s2p Li-Ion battery with a capacity of 12.4 Ah. The vehicle carrying the battery will steer into a loading dock and power itself down. Now two contacts in the loading dock will connect a Li-Ion charger to the positive and negative pole of the battery and start charging it. The batteries' balancing wires are constantly connected to an active balancing circuit to equalize the cells.
The components to be used:
  • 4s Balance charger
  • A Li-ion Charger with 5-6 A, e.g. this or this (sorry, last one is german only).
  • The batteries are custom made and come without bms.

My questions are the following:
  1. Does this pass a basic sanity check? Did I correctly understand the functions of all parts involved? Is there something that won't work.
  2. Can the balancer be constantly connected to the balancing wires(during charging and draining). The vehicle can drain the battery with up to 120 A.
  3. I want to charge the batteries with 5-6 A max. Does the balancer have to support this much current? The balancing wires have a diameter of 22 AWG, so they probably wouldn't support it anyway? What balancing power should I aim at?
Thanks in advance
 
If you only have a balancer then it needs to have the possibility to take the max current from the charger, otherwise one cell could be overcharged.

But if you have a BMS that is also balancing, then 0,1 zo 0,3A balancing current is enough.

The balancer can always be connected to the battery.
 
Some thoughts below (including risk discussion even though you said you're aware of them, because you also said you have very little experience in charging):


I'm currently trying to build an automatic charging module for an autonomous RC vehicle. Before you notice it through my questions, let me say that I have very little experience on the field of charging and that I am aware of the risks.

The battery to be charged is a 4s2p Li-Ion battery with a capacity of 12.4 Ah. The vehicle carrying the battery will steer into a loading dock and power itself down. Now two contacts in the loading dock will connect a Li-Ion charger to the positive and negative pole of the battery and start charging it. The batteries' balancing wires are constantly connected to an active balancing circuit to equalize the cells.
The components to be used:
  • 4s Balance charger
  • A Li-ion Charger with 5-6 A, e.g. this or this (sorry, last one is german only).
  • The batteries are custom made and come without bms.

My questions are the following:
  1. Does this pass a basic sanity check? Did I correctly understand the functions of all parts involved? Is there something that won't work.

For an automated system with no human testing or intervention, you should use something to protect the cells against overcharge (and overdischarge).

For example, even if you have an active shuffle-charge (vs resistive-drain) balancer that is always connected, if anything is wrong with a group of cells, or something goes wrong with the balancer itself or any of it's connections to cells, you can easily end up with cells that get overcharged or overdischarged (or even reversed under load), damaging them in ways that can make them into fire risks from that moment on.

The protection can be a full BMS that actively shuts off the battery power, or just one that monitors cell-level LVC and HVC, and sends a signal to the controller to stop pulling power from the pack and shutdown, or to the charger to stop charging. The signal can be a simple enable/disable line, or it can be serial/etc communication (not usually practical if you've already designed the system).

A normal cheap BMS with FETs can be used without the discharge FETs in the current path, by using the gate signal to them as a signal to turn on/off the controller's "enable" line (via whatever intermediate devices are needed), if it's acceptable for the system to just turn off and stop. (not great for an aerial device, if that's what this is). If it's not, then it can be used to signal the controller there is a problem that requires device shutdown ASAP.

The charge FETs could still be used to directly control the charge current on/off, or the gate signal to those could be used as an enable/disable line to the charger if it supports that (most don't). The former is much simpler and requires no modification to the charger or dock system.



  1. Can the balancer be constantly connected to the balancing wires(during charging and draining). The vehicle can drain the battery with up to 120 A.

10C (120A / 12.4Ah, 60A per cell for 2p) is pretty high drain for a typical battery, and is hard on the cells, causing significant voltage sag and internal heating and aging. As the cells age, they become pretty different from each other (even if they were well-matched from the start, which is rare as packs are not usually built of tested and matched-characteristics cells), which means the balancer has to keep draining the better cells to keep the others equal in voltage.

If you have cells that age in a way that causes them to internally leak, then the balancer will drain the other cells to keep it full, and the whole pack will be drained to try to keep that cell full.

With no cell-level monitoring device to tell you this is happening, your only sign will be that pack life becomes shorter and shorter more and more quickly. If the system is automated, then unless it's programmed to report this, you won't know anything is wrong until the vehicle just stops working suddenly or the pack fails in a more dramatic way (unlikely, but does happen occasionally).


Also, if the balancer doesn't have any smarts or LVC and just keeps draining good cells to keep a really bad cell up, it could drain the whole pack totally dead in the process. So make sure the balancer you use does have a point at which it stops trying to balance the cells.


  1. I want to charge the batteries with 5-6 A max. Does the balancer have to support this much current? The balancing wires have a diameter of 22 AWG, so they probably wouldn't support it anyway? What balancing power should I aim at?
Thanks in advance

With no cell-level monitoring / charge shutoff, the balancer has to support however much current is needed to keep the cells from exceeding HVC during charge under the worst case conditions (badly unbalanced pack with some very low cells and some nearly full at start of charge), if you don't want overcharging to happen.

A shunt-based (or active) balancer built into a BMS doesn't have to deal with much current to avoid this, because when a cell reaches HVC the BMS will stop the charger from supplying any current. Then the balancer can drain down the high cell however it's designed to, until it's below the BMS restart-charge point, and current resumes at normal charge levels, cycling this way over however long it takes, until the cells are generally full enough for charge current to be at or below what the balncer can handle and it just fills and balances concurrently.

WIth nothing to turn charge current off, the balancer could need to handle the full charge current to keep a cell from exceeding HVC, if the cells are far enough out of balance when charging starts.


If you have something that can monitor for HVC during charge and turn the charge current off, the balancer only needs to handle whatever amount of balance current keeps a badly-unbalanced pack rebalanced in the amount of time you require recharging to happen. (you'd have to calculate this out; for instance if you only have say, 2 hours max to recharge the pack (5-6A charge current and a 12.4Ah pack), and you have a 6Ah imbalance between worst and best cells, the balancer has to be able to move enough current from the highest cell(s) to the lowest during that time to equalize the cell voltage. Probably has to be less than that time, so that it can still fully charge the whole pack within that time.)
 
@amberwolf Thanks, that was VERY comprehensive!

I wanted to keep it generic, but yes, It's a drone. So, low size and weight of any solution are of the essence. It's hard to say where the limits are, though. I can redesign parts of the drone if necessary, but size and weight shouldn't largely exceed the 4s balancer I mentioned.
Having said that, a BMS with active balancing seems to be the best way, but finding one that fits my dimensional requirements is almost impossible. The best I could find is this board, which seems to be a standard component sold by dozens of vendors. However, There isn't much info about it's inner workings and the cheap price isn't reassuring either. Due to the high load current, I would problably have to only use it for charging and not for discharging. Is there any other BMC you could recommend?

So, If I can have an imbalance of at most the cell size (6 Ah) and charge for 2 hours, the balancer needs to constantly balance 3A, right? The charging time is about right and the battery will seldomly be drained more than half, so the mentioned 5A balancer should be sufficient. In conclusion, my initial setup would have the following drawbacks:
  • Cells are not protected against overcharge, if a cell group is damaged or the balancer fails to balance the cells quick enough.
  • The balancer might, in case of a bad cell group, drain other good cell groups to equalize voltage, not noticing that the bad group will never equalize.
Also, when charging with 5A, the AWG22 balancing wires would problably not support the needed balancing current (It's hard to get exact numbers here).

Currently, I'm thinking about giving my loading dock a 7-port connector, run the battery poles and balancing wires out and connect everything to a balancing charger in the loading dock, e.g. this.

By the way: Is it ok to post links to commercial product selling sites and request buying tips on this forum? (On StackExchange, they have you drawn and quartered for that:)).
 
Links to stuff you're going to (or want to know about) use in a project, etc., are fine anywhere. (LInks to stuff you're selling, if you ever do, only go in the for sale subforums).

Some further questions and thoughts:

How fast do you have to charge?

That linked LiPow charger is only going to *draw 100w from it's USB-C supply* and will have less than that available to send to the battery (probably 80-90% of that, depends on it's conversion efficiency and exactly how it works inside, and how close the source voltage is to your pack voltage; the greater the difference the worse the efficiency, usually).

However, that is only at 68F (20C); they don't show the derating chart above that temperature but unless it's always really cold where you are, the charger will not be at that temperature, and won't be able to charge at that rate. Their spec is also only for their specific 3D printed case with fan, presumably designed to ensure proper airflow over the right parts inside it. If your dock is designed to do the same job for the bare board it would then do the same rating...but unless you chill the air going into it, it's likely that it won't be that cold most of the time.

So you might want to count on only say, 50w of charging power. That'd be

Even if you assume it's the low end of the max, for 80w total charging power: 80w / 16.8v = 4.8A max current (which you won't get all the time, only when the pack is far enough discharged to get that much current). For 12.4Ah, assuming totally empty, it'd take at least 12.4 / 4.8 = 2.6hours, but it will really take a lot longer because current tapers off as the cells fill up (and if it has to finish balancing before release, if balancing takes longer because of cell problems...could take hours).

They don't specify balancing current, but many of these things (and most cheap BMS) only do around 50mA of balancing. If this is like that, then if you had a really bad cell group that was say, 3Ah different from the others, then 3000mAh / 50mA = 60 hours to correct the difference assuming ideal conditions. For chargers or BMS that only balance near full charge, you have to add the time to reach that charge level first.



The linked BMS looks typical of most of them; the design of almost all of them is about the same, and they have assorted undesirable failure modes you can't really do much about...but the worst problem with using any of the common ones is that they don't ask the device if it's ok to turn off--they just shut off power whenever a problem is detected, so if your device is aerial, it will just crash and stay there with no power, no way to signal you, etc.

There are ways around that by modifying the BMS, taking off the discharge FETs and wiring across the D-S pads where they had been. and using their gate signal to tell the device that there's a battery problem, and let the device decide when to turn off--like go into parking mode, land, and signal you it's in trouble. Since it's not using high current at that point, it owuldn't need to shutdown completely and could keep pinging you with status and location.


If you're using RC LiPo then the typical BMS (including this one) goes too low for LVC, and higher HVC than you might want to use.

I don't have any specific BMS recommendations, though.


If you're still designing the electronics, you could just get the chips used on these BMS, put it on your control PCB with everything else, and have it monitor the LVC, and signal your controls as noted above so the device chooses how to proceed or not. Some of the chips are externally programmable for LVC, HVC, etc., and you can set them to whatever values you want, instead of being stuck with the values in there already. This is a page I don't know the age of the data on, but shows you a few

Then your dock could do all the balancing during charge, or you could keep the already-planned active balancer, etc. Personally I'd keep that all in the dock if you can add the pack protection functions to the device itself. Makes it lighter, with less stuff to go wrong and damage the pack.



Are you using RC LiPo packs? Or some other kind of cells?

The former are rarely good quality and even rarer using matched cells (never, in my little experience with them, and based on posts by others over the years, but I'm sure there are exceptions). Used hard, they tend to age fast, compounding the problems of unmatched cells, requiring more balancing more frequently.

There are better cells out there, especially EV-grade stuff out of used EV modules, that are well-matched to each other, and dont' require any balancing until they have aged significantly (I have some ancient (for battery technology) EIG C020 NMC that was used when I got it, and is now well over a decade old, used every day on my SB Cruiser heavy-cargo trike, and still hasn't reached the point of requiring balancing, with only one cell (on an earlier bike) having failed...I'm sure newer EV-battery technology is even better). At least some of these cells are a high enough C-rate for your needs; my 20Ah cells are 5C continuous 10C burst for 10sec (IIRC), so if they were still available they'd do the job if you didn't need 120A for very long.
 
How fast do you have to charge?

That linked LiPow charger is only going to *draw 100w from it's USB-C supply* and will have less than that available to send to the battery (probably 80-90% of that, depends on it's conversion efficiency and exactly how it works inside, and how close the source voltage is to your pack voltage; the greater the difference the worse the efficiency, usually).

However, that is only at 68F (20C); they don't show the derating chart above that temperature but unless it's always really cold where you are, the charger will not be at that temperature, and won't be able to charge at that rate. Their spec is also only for their specific 3D printed case with fan, presumably designed to ensure proper airflow over the right parts inside it. If your dock is designed to do the same job for the bare board it would then do the same rating...but unless you chill the air going into it, it's likely that it won't be that cold most of the time.

So you might want to count on only say, 50w of charging power. That'd be

Even if you assume it's the low end of the max, for 80w total charging power: 80w / 16.8v = 4.8A max current (which you won't get all the time, only when the pack is far enough discharged to get that much current). For 12.4Ah, assuming totally empty, it'd take at least 12.4 / 4.8 = 2.6hours, but it will really take a lot longer because current tapers off as the cells fill up (and if it has to finish balancing before release, if balancing takes longer because of cell problems...could take hours).

They don't specify balancing current, but many of these things (and most cheap BMS) only do around 50mA of balancing. If this is like that, then if you had a really bad cell group that was say, 3Ah different from the others, then 3000mAh / 50mA = 60 hours to correct the difference assuming ideal conditions. For chargers or BMS that only balance near full charge, you have to add the time to reach that charge level first.
I agree, this may not be the best charger. I could even add some points, e.g. the sparse documentation and the missing API to integrate it into your projects.
However, it seems to offer a decent amount of power (even if it doesn't reach the maximum) and, most importantly, it runs without user interaction. Especially the last part is a rare trait. Also, I could connect to it via UART, and, with a little effort, query all important values directly in software.
The linked BMS looks typical of most of them; the design of almost all of them is about the same, and they have assorted undesirable failure modes you can't really do much about...but the worst problem with using any of the common ones is that they don't ask the device if it's ok to turn off--they just shut off power whenever a problem is detected, so if your device is aerial, it will just crash and stay there with no power, no way to signal you, etc.
I'd like to use the BMS just for charging and not for routing discharge power through it. A BMS that can withstand 120A at 4s would be way too big and heavy. I've never seen a self build drone using any kind of BMS, problably for that very reason. The balancer would have to clean up whatever mess the discharge made during the charging phase.
There are ways around that by modifying the BMS, taking off the discharge FETs and wiring across the D-S pads where they had been. and using their gate signal to tell the device that there's a battery problem, and let the device decide when to turn off--like go into parking mode, land, and signal you it's in trouble. Since it's not using high current at that point, it owuldn't need to shutdown completely and could keep pinging you with status and location.

If you're using RC LiPo then the typical BMS (including this one) goes too low for LVC, and higher HVC than you might want to use.

I don't have any specific BMS recommendations, though.


If you're still designing the electronics, you could just get the chips used on these BMS, put it on your control PCB with everything else, and have it monitor the LVC, and signal your controls as noted above so the device chooses how to proceed or not. Some of the chips are externally programmable for LVC, HVC, etc., and you can set them to whatever values you want, instead of being stuck with the values in there already. This is a page I don't know the age of the data on, but shows you a few

Then your dock could do all the balancing during charge, or you could keep the already-planned active balancer, etc. Personally I'd keep that all in the dock if you can add the pack protection functions to the device itself. Makes it lighter, with less stuff to go wrong and damage the pack.
I'm a computer scientist. I know a thing or two about electronics, but I have little engineering experience. Changing a FET for a wire might work, but designing the electronics myself is above my level.
Are you using RC LiPo packs? Or some other kind of cells?

The former are rarely good quality and even rarer using matched cells (never, in my little experience with them, and based on posts by others over the years, but I'm sure there are exceptions). Used hard, they tend to age fast, compounding the problems of unmatched cells, requiring more balancing more frequently.

There are better cells out there, especially EV-grade stuff out of used EV modules, that are well-matched to each other, and dont' require any balancing until they have aged significantly (I have some ancient (for battery technology) EIG C020 NMC that was used when I got it, and is now well over a decade old, used every day on my SB Cruiser heavy-cargo trike, and still hasn't reached the point of requiring balancing, with only one cell (on an earlier bike) having failed...I'm sure newer EV-battery technology is even better). At least some of these cells are a high enough C-rate for your needs; my 20Ah cells are 5C continuous 10C burst for 10sec (IIRC), so if they were still available they'd do the job if you didn't need 120A for very long.
The packs are custom made by the people that build our drone. Afaik, they're using standard Li-Ion cells. I can't say much about their quality.
 
I'll echo amberwolf's balancing through the dock idea. Extra connections to the charger but less weight on the drone.
Only downside is if one cell goes bad, does the drone fall out of the sky? Might be worth having cell monitoring on the drone side to watch battery condition if it doesn't already.

Some of the linked electronics are quite pricey for what they are and others are likely of questionable quality for a commercial offering. Ok for a prototype but nothing you want to go to production with. You could hire an ee for a pre-production version's electronics if that's what you're looking to do.
 
I'll echo amberwolf's balancing through the dock idea. Extra connections to the charger but less weight on the drone.
Only downside is if one cell goes bad, does the drone fall out of the sky? Might be worth having cell monitoring on the drone side to watch battery condition if it doesn't already.
The flight controller doesn't monitor the individual cells, but it keeps track of the battery's overall voltage. If said voltage continuously drops below a certain threshold, safeguards are activated and the drone lands. I don't think I could easily provide a better monitoring by myself.
Some of the linked electronics are quite pricey for what they are and others are likely of questionable quality for a commercial offering. Ok for a prototype but nothing you want to go to production with.
Correct, but my requirements are tough and leave only few possibilities. The fact that the charger must be able to charge without user interaction (e.g. pressing a start-button or navigating a menu) rules out 95% of all products (at least).
Also, the (problably low quality) product I mentioned seems to be one of the few available. It is sold in different designs by dozens of vendors. I couldn't find anything that looks any better and fits my cause.
 
Last edited:
Correct, but my requirements are tough and leave only few possibilities. The fact that the charger must be able to charge without user interaction (e.g. pressing a start-button or navigating a menu) rules out 95% of all products (at least).
Do you have an extra IO that could trigger the charger to start charging?
 
Do you have an extra IO that could trigger the charger to start charging?
Yes. I'm using a Raspberry Pi to control the dock, which potentially gives me all the IO I want.
However, I'm not too fond of opening a bought charger and soldering wires to undocumented places, If that's what you mean. The overall system should at least make the impression of a professional setup and all components should be used as intended by the manufacturer.
 
Back
Top