DIY Modular 90V battery

[E-HP]

It's an unfinished sketch, so not really, here's an updated one, still has a few things missing but I'll add things as I go

The balance wire connection between A1 and A2 is similar to charging two lipos using a paralleling board.
When paralleling two packs via the main discharge connectors, I look for 0.1V difference between the total pack voltages before connecting them. But, when paralleling balance connectors, as with lipo parallel charging, each cell group should also be matched before connecting, since the packs may have the same total voltages but differ at the group level if the packs aren’t balanced, similar to charging big lipos with a paralleling board.
Will A1 and A2 always remain paralleled, or will you be mixing and matching, juggling the configuration around?

 

[mateiuskey]

The balance wire connection between A1 and A2 is similar to charging two lipos using a paralleling board.
When paralleling two packs via the main discharge connectors, I look for 0.1V difference between the total pack voltages before connecting them. But, when paralleling balance connectors, as with lipo parallel charging, each cell group should also be matched before connecting, since the packs may have the same total voltages but differ at the group level if the packs aren’t balanced, similar to charging big lipos with a paralleling board.
Will A1 and A2 always remain paralleled, or will you be mixing and matching, juggling the configuration around

Well, my plan was to make a A1, B1 and C1 (column) configuration out of the cells that I have, and then parallel other column groups as I can afford more (each column made of the same cells). While I may do 2 packs of same cells in parallel, a sort of A1 - B1 maybe , I don't plan on mixing older and newer packs in series.
As for the balancing, I'm planning to charge them either with a separate small BMS rated for one pack, or the main BMS, so I doubt the individual groups will be unbalanced. Then I may use a small precise load circuit to make sure all reach the same voltage at the pack level.

 

[mateiuskey]

Yeah, safe connection of packs of varying voltage seems the big limitation.

As long as each parallel pack has its own connection to the BMS B+ and P-/Controller- there will be a large current equalizing the bulk voltage on plugging in. This can be avoided with diodes, but that blocks regenerative braking and the controller feeding surges back into the cells. Also causes voltage drop unless you use ideal diodes (MOSFETs).

Similarly, when plugging the parallel connectors in, each p-group is going to equalize fast. For the Samsung INR2700-50G cells I use the maximum equalization voltage from max charge to min charge would 1.7V over 18.5 mΩ per cell generating 92A, I think. Even more current if many cells are in parallel reducing internal resistance. Meanwhile for 5P packs they could only handle 1.6A charging current x 5 = 8A.

Maybe you could connect the p-groups together with thermistors in the way. Even resistors might work. Sometimes builders use high resistance strips between cells in a p-group anyway to act as fuses. As long as each pack has its own discharge and bulk charge connectors there isn't supposed to be much equalization current between the p-group cells.

Not sure how that affects the BMS attempts to read the voltage of the p-groups and balance via discharge of p-groups during charging, though. Maybe each balance connector has to go to the BMS and not to the parallel pack next to it. That way if you add thermistors to limit inrush current, each parallel pack has the same amount of resistance shoved in front of the voltage measurements instead of that resistance multiplied by how many parallel connectors it goes through.

Definitely starting to make sense why the Daly parallel balance modules don't just have current limiting and a DC-DC converter built in to transfer voltage between parallel packs, but also communicate with the BMS on each pack to decide to turn on or off at all.

That's why one of the biggest concerns is making sure each pack is within 0.4V (or lower) of eachother when connecting. With that rule I can avoid most complications. My main idea for doing that is having a very accurate voltmeter on the charger, and if I cannot regulate it precisely with that, then use a small controllable and variable load with that voltmeter to bring all packs to the lowest one.

 

[E-HP]

That's why one of the biggest concerns is making sure each pack is within 0.05V of eachother when connecting. With that rule I can avoid most complications. My main idea for doing that is having a very accurate voltmeter on the charger, and if I cannot regulate it precisely with that, then use a small controllable and variable load with that voltmeter to bring all packs to the lowest one.

0.05v is a general rule per cell. You have 20s, so 1V, but I generally shoot for 0.5v. Plus I bulk charge, so I’m only doing it once.

 

[mateiuskey]

0.05v is a general rule per cell. You have 20s, so 1V, but I generally shoot for 0.5v. Plus I bulk charge, so I’m only doing it once.

Ah,okay, then I assume for a 8S (just one modular pack) it should be around 0.4V, but if I can be even more precise I suppose it might be better for the cells.

 

[E-HP]

Ah,okay, then I assume for a 8S (just one modular pack) it should be around 0.4V, but if I can be even more precise I suppose it might be better for the cells.

Not really. It’s similar to voltage sag, but in reverse, or like regen or plugging in your charger. The voltage will rise for the low pack and sag for the higher pack and meet in the middle with little actual current flowing between the packs for a millisecond.

 

[mateiuskey]

Not really. It’s similar to voltage sag, but in reverse, or like regen or plugging in your charger. The voltage will rise for the low pack and sag for the higher pack and meet in the middle with little actual current flowing between the packs for a millisecond.

Yes, but doesn't the 0.05V per cell rule change with cell specs? Wouldn't it be better to just calculate the current at certain voltage differences (and with the resistance of the cells) to find a reasonable limit (factoring in any resistance changes over time) ?

 

[E-HP]

Yes, but doesn't the 0.05V per cell rule change with cell specs? Wouldn't it be better to just calculate the current at certain voltage differences (and with the resistance of the cells) to find a reasonable limit (factoring in any resistance changes over time) ?

Sure go for it. It’s always best and safest to have a healthy dose of paranoia when combining packs, or with any form of lithium battery in general; and of course, never charge unattended.

 

[magnie]

I recently discovered Ohm's Law and how it relates to current, so I am still learning, but..

The M50L's datasheet lists the DCIR as 23±6 mΩ (they use the term "DC internal impedance" which I think is a typo). So at .023Ω * 20s1p pack = .46Ω. 1v / .46Ω = 2.17a. So a 20s battery pack in parallel with another 20s1p battery pack with a 1v difference will "balance" at 2.17a.

With 8s1p battery packs and a .5v difference: .023Ω * 8s1p = .184Ω, 0.5v / .184Ω = 2.71a.

When I connected two 88v50ah batteries with a .08v difference, I was getting 2-3a of current flowing between the two:

I don't know the internal resistance of the cells I have though. That being said, I guess I could calculate it now..

 

[docw009]

A concern with series packs is that the switch transistors in each BMS need to withstand the total voltage of the entire string, if that BMS were to shut down. What would cause it to shut down. Well, it could simply run down to LVC on its own. Or it could be unbalanced, If you overvolt a BMS, the transistors either short out or blow open. SHorts are the problem.

It does happen. There was a fellow that put two 13S packs in series. He did ask here, was advised of the risks and he did it anyway. He later posted that one of the batteries caught fire while riding. What probably happened something like a bad cell caused one battery to shut down and it shorted out the BMS. Then the battery was forced to charge and recharge that weak cell and it eventually degraded and caught on fire.

It is hard to find high voltage transistors in a lower voltage BMS. The designer is not going to spend the extra money when it's not needed. An electrically skilled person could install higher voltage transistors or use the BMS to drive a relay,

 

[lnanek]

Yeah, most BMS I've seen have FETs that break down at 63V. I've seen some BMS that allow attaching a variable number of p-groups, though. E.g. anywhere from 16 to 20s. The instructions for wiring vary. Sometimes all the end balance wires go to the same, last p-group. Sometimes they are left unattached. Might be able to get over-spec'ed FETs that way.

 

[mateiuskey]

Yeah, most BMS I've seen have FETs that break down at 63V. I've seen some BMS that allow attaching a variable number of p-groups, though. E.g. anywhere from 16 to 20s. The instructions for wiring vary. Sometimes all the end balance wires go to the same, last p-group. Sometimes they are left unattached. Might be able to get over-spec'ed FETs that way.

Well, the BMS that I will use is exactly that type, it goes from 8S to 24S, and the balance wires have JST connectors that attach to it (refer to my last drawing). Otherwise I would've had to add custom MOSFETs and have several BMS, which is tedious and error-prone.

 

[mateiuskey]

Hello! Back on the thread with some updates:
I decided to only make 3 modules of 8s5p, and use them in a 24s5p configuration, until I get some more experience in this field. I've acquired 8 XT90 connectors and looking to get myself a meter of 35 mm2 wire (slightly bigger than 2 AWG) to make the connections.
I then started the very scary task of cutting the 16s8p unfinished battery that I bought into two 8s8p, and then the infinitely more dangerous and tedious task of cutting them into 8s5p. The worst part will come when I'll have to join the remains for the third module. This led to many sparks, close calls and a lot of stress, and it's not over, hopefully I'll finish by the end of the month.

I've also started designing a 3D printed shell for the modules, I'd like to make some way for them to connect securely (especially the XT90 connectors). I'll post an Onshape link here, I'd be very thankful for some feedback: Onshape

My questions would be:
1. How would one make an AntiSpark with normal X90 connectors? Do I need one with the JK BMS?
2. Any feedback on the 3D printed shell?
3. What are some additional protection features I could add?
4. Should I add more nickel strip? The battery I have has these strange parallel connections on one side that are just 2 copper wires soldered to the strip with copious amounts of solder, I'm thinking a DIY trick for some better current management.
5. What is the meaning of life?
And anything related to these ...

 

[lnanek]

Ali has been starting to offer some neat bolt together cases:

https://a.aliexpress.com/_mtqTiaL

Probably pretty easy to 3D print.

It's tough to turn a regular XT90 connector into an XT90S. The later has an internal resistor that connects when the plug is halfway. Probably easier to just wire two connectors to the motor, one with a resistor inline. Always plug the resistor one in first to charge the capacitors and you'd be fine. You could even plug it in, unplug, then plug in the regular one if you want fewer plugs. The capacitors in the controller don't drain particularly fast.

Some BMS have a two pin connector for an on off switch. If yours does, you don't need an anti-spark connector so much. Just attach a switch, turn the pack off, then connect everything up, then turn the pack on. Most sparks come from the capacitors charging up immediately and the connector being half plugged. So if the pack is off when you plug in, then you connect fully while it's off, I've found there isn't much fireworks when you turn on the path to the motor controller.

For connecting individual packs, that should only ever be done when they are at exactly the same voltage state of charge, so again, there shouldn't be sparks. A fuse slightly above your expected max discharge is a good idea, though. And making sure every connector is keyed so you can never plug one in in reverse polarity.

 

[mateiuskey]

Ali has been starting to offer some neat bolt together cases:
View attachment 392076

https://a.aliexpress.com/_mtqTiaL

Probably pretty easy to 3D print.

It's tough to turn a regular XT90 connector into an XT90S. The later has an internal resistor that connects when the plug is halfway. Probably easier to just wire two connectors to the motor, one with a resistor inline. Always plug the resistor one in first to charge the capacitors and you'd be fine. You could even plug it in, unplug, then plug in the regular one if you want fewer plugs. The capacitors in the controller don't drain particularly fast.

Some BMS have a two pin connector for an on off switch. If yours does, you don't need an anti-spark connector so much. Just attach a switch, turn the pack off, then connect everything up, then turn the pack on. Most sparks come from the capacitors charging up immediately and the connector being half plugged. So if the pack is off when you plug in, then you connect fully while it's off, I've found there isn't much fireworks when you turn on the path to the motor controller.

For connecting individual packs, that should only ever be done when they are at exactly the same voltage state of charge, so again, there shouldn't be sparks. A fuse slightly above your expected max discharge is a good idea, though. And making sure every connector is keyed so you can never plug one in in reverse polarity.

Unfortunately my battery was already welded (into a 16s8p) so there wasn't much I could do with it. Also I'd doubt I could find my specific shape and s count(8s5p parallelogram) so there's that. I will model the XT90s holder, a voltage and maybe amperage reader, and whatever I could think of for protection. I'm mostly curious what do people do for fireproofing.

 

[lnanek]

For fire proofing when charging I use a metal ammo can. Others use an old BBQ. I never wanted to commit to the extra weight to encase the battery in steel on my bike, though. I just try to make sure there's a chance I can get it off to burn by itself and not take the bike with it. There's a bunch of silly "fireproof" lipo bags out there that have been proven to do nothing.

People do fully pot their batteries to help, I guess. Can't have something vibrate off and short that way. And use low AWG wires and thick interconnects that stay cool under the amps. There's phase change material you can use to help cooling, but nothing can stop a runaway lithium pack from burning in the end. There's just too much energy being released.

Unfortunately for us DIYers, best first proofing is likely to buy a single battery pack for each use case with sufficient capacity that it can always be used by itself without any hacky parallel or serial strings, and make sure it's built out of brand name cells, like from Samsung, with safety vents into a UL listed pack. I label every wire B+ and B-, and try to triple check, but over a thousand connects, I did mess it up once when the little flags I put on the wires were a little too close and curved together and I hadn't staggered the splices.