Adding a switch to (an already switched) BMS?

[Watchamacallit]

Hi all.

This is going to be quite a long winded, poorly written post (Sorry, I'm not the greatest at explaining things!) Patience with me is hugely appreciated.

Long story short; i've had months worth of deeply frustrating BMS troubles. Largely stemming from my Hailong case not accomodating their sizes, and requiring me to remove their heatsinks in order to get them to work/fit inside the case.

After many tests and variations, i finally found a BMS that works perfectly for my limited space needs due to its very small form factor. It's this one (35a version, although i only continually run around 25):

https://www.aliexpress.com/item/32862984213.html

It's a reasonably cheap aliexpress deal, and removing it's alumininum casing reveals a tiny PCB ideal for my space requirements. Testing it with the case/heatsink off under load (1100w continuous) and charge scenarios (5a) also shows it runs very cool (tops at about 60/70 degree's) makes it perfect. PCB Pictures below:


https://imgur.com/a/GnPxXJs


However, sods law and all that, it never goes that smoothly....


Another requirement i have is needing to switch the battery on and off, This BMS comes with a wire you can cut to then solder on your own switch (White wire) - perfect. Just what i need.

However, over time the switch would fail and the BMS is stuck "on" - this is a big problem for me, as i dont feel it's very safe and i like to be able to control power delivery.

I bought three more BMS's of this exact make, and they all eventually displayed the EXACT same problem - the BMS would appear to switch on and off fine, but then suddenly without warning, the switch functionality would fail and it would be stuck on. One actually did this the moment i wired it all up. No switch functionality at all.

I tested it on at least 5 new hailong case switches too; the problem still persisted. Absolutely was NOT the switch itself. I also verified countless times it wasn't an issue with solder points bridged or wires making contact.

Now, as i cannot really replace these BMS's easily due to my specific requirements; is it possible for me to rig up the original hailong case switch to work with them? as the original spot for a switch clearly doesn't function reliabily, could i maybe:


1. use the thermal probe contact as a switch instead? (It's the black cable, solder points below switch ones) ie; fooling the bms it's overheating, and cutting off.

I did wire the switch to the probes contacts but it had no effect. The thermal probe DOES function as intended, and switches OFF the bms's when triggered (tested it) and does so for all of the them.

2. Ive seen some BMS's have a switch soldered to their MOSFETS, no idea if this is possible? I Dont understand mosfets enough to try.

3. trick the BMS via the balance wires in some way? (thinking there's a balance wire break) Although i fear this might not work as it'll make the BMS think there's a balance fault, meaning it wont switch back on unless i have a charger to hand to "wake" it back up. Not feasible.

Option 1 is probably the best (only) method, even if it means i lose the thermal protection functionality, i absolutely must have a switch. I'm just not sure if it's doable as my experience is limited, so i'm here to consult you guys in the desperate hope you'll share your wisdom and experience my way.


Any help/asisstance is gratefully recieved. Thank you for your patience.

 

[amberwolf]

Notes and considerations are between quotes of your post sections.

Long story short; i've had months worth of deeply frustrating BMS troubles. Largely stemming from my Hailong case not accomodating their sizes, and requiring me to remove their heatsinks in order to get them to work/fit inside the case.

Note that when you remove heatsinks (which they wouldn't spend money installing if they werent' needed), you remove the ability of the device to correctly get rid of the heat from the components it is there to do so for.

The device will still operate, but it will not be able to handle as much power in the heat-shedding devices as it was designed and specified for without the heatsink. How much less you would only find out if/when it fails. (You might be able to find specs on the parts involved and if the spec sheet includes thermal calculation information you may be able to determine the non-heatsinked limits in the specific application they're in).

Depending on specific usage and conditions, and the design of the system (including how well the designers overspecified the parts), it could be as little as half the power or even less, especially since it is inside a closed battery casing with no airflow (and the heatsink no longer present to spread the component heat out over a much larger surface area for conduction / radiation away from those components).

Some things work fine without the heatsink, some don't, depends on things noted above.

On some devices the metal plates aren't just heatsinks, but may be intended to retard EMI from all the noise sources within a motor system; without them there may be more risk of unintended/undesired operation.

Testing it with the case/heatsink off under load (1100w continuous) and charge scenarios (5a) also shows it runs very cool (tops at about 60/70 degree's) makes it perfect.

Is that testing done within the battery case under the same conditions it would be used in? If not, you may wish to do that testing to ensure it will still operate as expected when there is no airflow, just in case.

Another requirement i have is needing to switch the battery on and off, This BMS comes with a wire you can cut to then solder on your own switch (White wire) - perfect. Just what i need.

However, over time the switch would fail and the BMS is stuck "on" - this is a big problem for me, as i dont feel it's very safe and i like to be able to control power delivery.

If the switch itself has actually failed, and measures shorted across when disconnected from everything else, then I would recommend using a "better" switch.

What is the specific way the switches fail in your usage? Is it mechanical, or electrical or environmental? Switches exist for just about any application, that will help prevent most issues, but they do all have a finite lifespan, just so many switching cycles guaranteed, so they do eventually have to be replaced regardless.

One major thing to be sure of is to always use a switch that is rated for *at least* the DC voltage it must disconnect, and *at least* the maximum current that will ever flow thru it under any conditions. If it can't, it may weld the contacts together during switching and stick on.

Are the wires it is switching at battery voltage, or logic-level, or something else? Do they carry any current, or are they just turning a control signal on and off?

By "control power delivery", what specifically do you mean? Just being able to turn off the output of the battery in some way? Or something else?

I bought three more BMS's of this exact make, and they all eventually displayed the EXACT same problem - the BMS would appear to switch on and off fine, but then suddenly without warning, the switch functionality would fail and it would be stuck on. One actually did this the moment i wired it all up. No switch functionality at all.

I tested it on at least 5 new hailong case switches too; the problem still persisted. Absolutely was NOT the switch itself. I also verified countless times it wasn't an issue with solder points bridged or wires making contact.

This is almost certain not a problem with your switches, but simply a problem with the FETs failing stuck on (shorted), It is a common problem when they overheat (which can be a momentary spike of heat you may not even measure dependign on your specific test instrumentation and setup), or when the current flow thru them or the voltage across them exceeds their ratings even for a moment,

Since it's not that uncommon for cheap stuff to use really cheap parts (which may not even be the parts they say they are on their labels; counterfeit parts are extremely common), the FETs may not be able to handle even what the BMS is rated for.

Now, as i cannot really replace these BMS's easily due to my specific requirements; is it possible for me to rig up the original hailong case switch to work with them? as the original spot for a switch clearly doesn't function reliabily, could i maybe:

1. use the thermal probe contact as a switch instead? (It's the black cable, solder points below switch ones) ie; fooling the bms it's overheating, and cutting off.

If the other switch you're using is not actually failing, but the BMS is not turning off regardless of that switch state, then the FETs are failed shorted (stuck on) so it doesn't matter what you do to switch the BMS itself off--it has already commanded the FETs to do so but they no longer function and so cannot turn off.

If the problem is actually with your other switch failing stuck on, then causing the thermal probe signal to engage by whatever means (switch that either opens it completely, or shorts across it) would command the BMS to shut off...as long as the FETs are still working.

2. Ive seen some BMS's have a switch soldered to their MOSFETS, no idea if this is possible? I Dont understand mosfets enough to try.

You'd need to specify the actual circuit used--what on the FETs they are switching, to say if it would help.

A shorted stuck-on FET cannot be turned off at all by any normal means, so a switch that turns off the gate signal wouldn't work.

A switch that completely cuts the current flow thru the FETs would work, but that switch has to carry the full current of the battery, *and* has to be able to handle the full voltage of the system across it. That is generally a large switch, and will not fit within your case. You would have to put it external to the battery case, and if it is not waterproof (if waterproofing is a requirement for your system) you'd need to waterproof it, too.

You may be able to use solar breakers or switches for this, if they're rated for the voltage and current of your system. Regular AC-style breakers aren't designed for DC so shouldn't be used (no guarantee they will disconnect the DC current flow under any load). Some people have used marine breakers or battery disconnects, but these are not rated very high voltage, so there's no guarantee they'll disconnect under any load--they *should* still work when no load is present (like if the controller is turned off, motor not operating, etc).

3. trick the BMS via the balance wires in some way? (thinking there's a balance wire break) Although i fear this might not work as it'll make the BMS think there's a balance fault, meaning it wont switch back on unless i have a charger to hand to "wake" it back up. Not feasible.

This still wont' work if the BMS is actually stuck on, meaning it doesn't respond to the first switch, same as it wont' respond to the thermal probe input. (becuase it already responded to teh first switch, the FETs just dont' work anymore and can't resppond to BMS commands).



So...you'd need to first determine what is actually failing before you can either prevent the failure, fix the failure, or provide a backup against that failure.

It's almost certainly the FETs failed stuck on, so you can easily test that with your multimeter set to 200ohms or continuity.

There are two sets of FETs on a BMS. The one with more FETs is the discharge set, and that's the one you're having a problem with. The one with less is the charge set, and it's likely that still works.

Take a dead BMS that's not attached to anything else, and use the black lead on a discharge FET heatsink tab. Red lead to each of the three pins on the FET. It will measure a dead short on one pin, usually center, as that's connected to the tab. If the FET is not failed shorted, it will read open circuit on the other two. If it reads nearly zero ohms or less on either of the other two, it's shorted, stuck on.

If one FET is stuck on, it wont' matter if any others work because they're all in parallel. But it's not uncommon for them to all fail about the same time from the same conditions.

If there are stuck FETs, you can replace them with new ones, probalby better ones. They must be rated for more than the maximum voltage (Vds) that would ever be placed across them (including by any regen braking your system could create, which sometimes is significantly higher than the battery voltage, and is sometimes the cause of BMS failures, if the BMS tried to turn off for whatever reason during regen braking and thus placed the full regen voltage across FETs not rated for that voltage).

They must also be rated for at least the worst-case current (Ids) that the system will ever draw from the battery, and preferably for the lowest possible RdsOn resistance (for the least heating per amp).

They need to match the control voltage level (Vgs) used by the BMS, too; you can just make sure it's the same kind of control as the existing FET uses (5v, 12v, 15v, etc).

 

[Watchamacallit]

What is the specific way the switches fail in your usage? Is it mechanical, or electrical or environmental? Switches exist for just about any application, that will help prevent most issues, but they do all have a finite lifespan, just so many switching cycles guaranteed, so they do eventually have to be replaced regardless.


Are the wires it is switching at battery voltage, or logic-level, or something else? Do they carry any current, or are they just turning a control signal on and off?

I should specify that the BMS's are failing "on" almost immediately after install. Even before putting them under any load (one even before testing) - i've not been able to ride the bicycle yet.

As for how they're actually turning off the BMS, i've no idea. I'm assuming it's a low current setup as this is surely what it's designed for - a low power/current switch to turn off a high current circuit. the wires it originally came with are quite thin guage.

This is almost certain not a problem with your switches, but simply a problem with the FETs failing stuck on (shorted), It is a common problem when they overheat (which can be a momentary spike of heat you may not even measure dependign on your specific test instrumentation and setup), or when the current flow thru them or the voltage across them exceeds their ratings even for a moment,

I'm 100% positive it's not a fet being stuck on or failing, As this would mean all three of my BMS's have failed on before i've even managed to test them. (edit - just used a multimeter on diode test and all the fet's behave as intended)

Under test scenarios they operate fine and well within thermal limits (their "junction temperature" is apparently around 150C). nothing else on the board bar the fets makes contact with the heatsink.

Testing consists of assembling them within the battery, putting the hailong lid on with a small hole drilled in it to allow a temperatre probe in to monitor them, while they're placed on a continual load of 1100w to emulate my peak consumption of the ebike, which would never happen continually anyway (It's just to see if i have any headroom for exceptionally hot days with huge hills lol) I closely monitored their temperature under different scenarios to ensure they'd operate fine without the heatsink.

Also, if the FET's were stuck on, surely the temperature sensor wouldn't work? (It works fine on all three BMS's). Applying some heat to the black temperature probe on the board makes them switch off momentarily until it's cooled enough to turn back on.

Surely if the FET's were failed on, this wouldn't occur at all?

I'm convinced it's the switch circuitry, that for whatever reason is tempermental and jank.

Speaking of the temperature probe, would it be possible to use that as a mechanical switch instead of the temperature probe setup? Although not ideal as I'll lose temperature sensing capabilities, i'd rather have a manual switch then none at all.

I was hoping you'd reply, thank you for your detailed knowledge and response.

 

[amberwolf]

What is the specific way the switches fail in your usage? Is it mechanical, or electrical or environmental? Switches exist for just about any application, that will help prevent most issues, but they do all have a finite lifespan, just so many switching cycles guaranteed, so they do eventually have to be replaced regardless.


Are the wires it is switching at battery voltage, or logic-level, or something else? Do they carry any current, or are they just turning a control signal on and off?

I should specify that the BMS's are failing "on" almost immediately after install. Even before putting them under any load (one even before testing) - i've not been able to ride the bicycle yet.

As for how they're actually turning off the BMS, i've no idea. I'm assuming it's a low current setup as this is surely what it's designed for - a low power/current switch to turn off a high current circuit. the wires it originally came with are quite thin guage.

So, it isn't the switches that are failing? It is the BMS itself that stops responding to the switch input? Meaning, manually shorting or disconnecting the wires to the switch no longer changes the BMS state, regardless of the presence of the switch or lack thereof?

Also, if the FET's were stuck on, surely the temperature sensor wouldn't work? (It works fine on all three BMS's). Applying some heat to the black temperature probe on the board makes them switch off momentarily until it's cooled enough to turn back on.

Surely if the FET's were failed on, this wouldn't occur at all?

That's correct. If they were failed on, the BMS could not shut off the output (or input if that was the damaged FETs).

Speaking of the temperature probe, would it be possible to use that as a mechanical switch instead of the temperature probe setup? Although not ideal as I'll lose temperature sensing capabilities, i'd rather have a manual switch then none at all.

If the temperature input does the same job, of disabling the BMS output and/or input, that you want the switch to do, then you can simply wire your switch in series with the temperature probe, because the temperature probe in this case must only be a thermal switch. Thermal switches often look like this, sometimes in metal, sometimes in ceramic, etc:
https://www.amazon.com/a12080600ux0947-Bimetal-Temperature-Control-Thermostat/dp/B0094GA9GG
and will have their switching temperature printed on the outside. They are usually normally-closed, meaning they are shorted like a switch that is turned on until they reach that temperature.

So wiring them in series with your switch means that either the switch or the probe will turn the BMS off, so you won't lose temperature sensing.

If you have an actual temperature sensor that the BMS can monitor and report the temperature of, then it is not a switch type input, and I don't know what behavior it will have if used with a switch.