[reiderM]

I've been looking all around for a concrete answer about parallel wiring at the module level with Tesla modules and I have gotten all kinds of answers. So I figured I'd make a post about what I've figured out so far and what I'm still trying to figure out.

Let's start with where I started: reaching out to EV West. On the page where they sell Tesla modules, they suggest that they can be arranged "2p5s for a total of 53kWh of energy with a total of 10 modules." So I reached out to them by email and they said that they've had customers that use 5 in their conversions and then want to increase their range so they will parallel another 5 in while still using just one BMS. But then again, EV West often doesn't even use BMSes with lithium batteries, which I find to be honestly idiotic and dangerous.

Next, I decided to contact Orion, who makes the Orion BMS 2 that I intended to use. They pointed me to Evolve Electrics, who sent me these two PDFs: this one on parallel strings and this one on setting up parallel strings with the Orion BMS.

What I know for certain at this point:

1. You need a BMS for each parallel string
2. Each string needs to be very close in voltage to not cause issues

What I don't really understand:

1. In the first document, it recommends using a separate contactor for each string and essentially treating the two parallel strings as separate packs. This seems like it could lead to imbalances at the module level and I would think that this could be potentially more dangerous than just having them both connected to the same contactor.
2. That document also warns against leaving the parallel strings permanently connected, saying, "Never leave two lithium ion strings permanently paralleled or leave multiple strings paralleled without monitoring systems and a means of automatic disconnection. As always with any safety critical circuit, always use multiple redundant and independent shutoff systems." Why would it be dangerous to leave two parallel strings connected? Of course, more safety disconnects is always better but what makes parallel inherently more dangerous than series when left connected?

 

[MattsAwesomeStuff]

But then again, EV West often doesn't even use BMSes with lithium batteries, which I find to be honestly idiotic and dangerous.

On a commercial build with an ignorant owner, absolutely. It's reasonable to expect the car "just work". Then again, EV West's customer service seems to be "We'll take your money but ignore you after that" anyway.

On a DIY project, ehn, not a big deal. Power is flowing in and out of the whole string equally, the BMS is really only there to touch up and adjust microscopic changes in how the cells age over time, and is only relevant at extremes. If you have the skills to manually check the cells once a season, or before a long roadtrip where you'll be using up the entire pack, you're probably fine.

What I know for certain at this point:

1. You need a BMS for each parallel string[/quote]
   
   What you shouldn't do is connect the endpoints of each pack together and presume it will be okay.
   
   But if you also connect a balancing lead for each cell to its corresponding mirror cell, (making the pack in parallel cell-by-cell as well) I don't think you'll ever see a problem running only a single BMS.
   
   There's a debate about it here in the past, it would have to be the most extreme of extreme cases where the wires you connect them with won't be able to supply enough balancing for the amount of power being used and would melt. This would require a massive, sustained amp drain (racing), on already failed or failing cells, that you somehow haven't noticed yet. If cell #13a is weak on one side, and #13b is trying to rapidly add energy to it because of the voltage imbalance, the balancing wire might try to supply more current than the conductor can carry. I see it as all but impossible to be concerned about. The thicker the balancing wire used, the less a concern it would be.
   
   

   Why would it be dangerous to leave two parallel strings connected?

   I guess, suppose one spontaneous thing happens to one cell, self-discharge or BMS failure or something. Then the other pack would discharge into the first pack, that might cause more failure. Versus, if it just happens to one pack, no power is going to flow to cause any further damage and the next time you get to the car you'll see that failure has occurred.
   
   Batteries are already in parallel, so it would have to be some circumstance that is unique to having a second pack there.

 

[reiderM]

On a commercial build with an ignorant owner, absolutely. It's reasonable to expect the car "just work". Then again, EV West's customer service seems to be "We'll take your money but ignore you after that" anyway.

On a DIY project, ehn, not a big deal. Power is flowing in and out of the whole string equally, the BMS is really only there to touch up and adjust microscopic changes in how the cells age over time, and is only relevant at extremes. If you have the skills to manually check the cells once a season, or before a long roadtrip where you'll be using up the entire pack, you're probably fine.

I'd rather just spend the extra couple hundred on a second BMS than having to manually balance the modules from time to time.

So here's my idea: two separate strings of modules, each has 5 tesla modules in series. Each string has its own BMS (the orion BMS has a paralleling functionality built in that should make this straightforward). Then, the positive lead of each series string has it's own contactor. Both of these leads connect to the B+ bus which everything else is connected (inverter/controller, DCDC, charger). So that would guarantee that the two strings are balanced when charging and discharging. When sitting idle, the two strings are not connected since both contactors are not being powered and thus in the open position. Since lithiums don't really self-discharge, this should mean that they remain balanced.

However, in the event that they become unbalanced, I need to make sure that things don't spark and I don't accidentally cook one of the strings.

So then the thing that I need to figure out is a way to only allow both strings to contact when the voltage is very close (not sure how close is needed, maybe <0.05v difference between the strings). If this is triggered, then it won't let me close either contactor unless I manually override this and only operate off of one string.

On the negative, they can all share a terminal since no circuit will be completed without the contactors being closed.

Another idea is to run them as two separate packs, but I don't know about the implications of that for charging and discharging.

 

[MattsAwesomeStuff]

I'd rather just spend the extra couple hundred on a second BMS

Looks like a lot more than just a couple, but, yeah, informed decisions and all that.

So then the thing that I need to figure out is a way to only allow both strings to contact when the voltage is very close (not sure how close is needed, maybe <0.05v difference between the strings). If this is triggered, then it won't let me close either contactor unless I manually override this and only operate off of one string.

From the document you linked:

"Besides just calculating specific information and controlling the application, the master node is also responsible for determining when / if slave string nodes are permitted to connect (engage) with the DC bus by looking for a number of criteria.

- Verifying that the slave pack voltage is close to the active DC bus voltage to prevent large inrush current if the slave node were to be engaged (the exact maximum range it is looking for is programmable).

The slave node itself will also validate that the above criteria are satisfied before requesting permission from the master node to join the DC bus. Both the slave node itself and the master node must agree in order for the slave node to engage. Once the negotiation is successfully completed, the slave will engage its contactor to join the DC bus."

...

Looks like they've got it handled internally quite thoroughly.

Also.. backing up to your previous post:

Why would it be dangerous to leave two parallel strings connected?

The next several pages immediately after this warning describes exactly the scenarios that it thinks would be detrimental or dangerous. One of which is what I mentioned earlier. It's quite thorough.

 

[OR-Carl]

I am in a similar boat, as I am also going to run a parallel setup of Tesla modules. I considered the idea of trying to bridge the cells as described above - and I concluded that the way Tesla makes its packs, this is not ideal. I think there was a guy who was going to give it a go - in theory I think it would probably be okay, but I decided I didnt want to be around to find out if I was wrong.

I think a lot of concern with parallel packs is based on one key assumption - that your parallel packs are made up of large format cells. Quoted from above:

"Never leave two lithium ion strings permanently paralleled or leave multiple strings paralleled without monitoring systems and a means of automatic disconnection.

Automatic disconnection seems like a great way to describe the tiny fuses on every last one of the tesla cells. If something goes wrong with a cell anywhere in the parallel arrangement, the fuse will blow, and the problem will be solved. With that level of redundancy, I dont see that it matters if the strings are connected at all times or not. The cells should not be drifting apart so much that connecting them together should cause any cause for concern, but leaving them connected should keep them in over-all balance.

I am going to run a full BMS - covering both strings at the cell level. The strings are then going to be connected together via 200Amp fuses, and feed into a single Positive contactor. The Negative will also be common, and run though its own contactor to give me redundancy in case of a welded contact. I do not anticipate this should present any problems, but I have yet to actually hook it all up, so I will keep you posted if I run into problems.

 

[reiderM]

I am in a similar boat, as I am also going to run a parallel setup of Tesla modules. I considered the idea of trying to bridge the cells as described above - and I concluded that the way Tesla makes its packs, this is not ideal. I think there was a guy who was going to give it a go - in theory I think it would probably be okay, but I decided I didnt want to be around to find out if I was wrong.

I think a lot of concern with parallel packs is based on one key assumption - that your parallel packs are made up of large format cells. Quoted from above:

Automatic disconnection seems like a great way to describe the tiny fuses on every last one of the tesla cells. If something goes wrong with a cell anywhere in the parallel arrangement, the fuse will blow, and the problem will be solved. With that level of redundancy, I dont see that it matters if the strings are connected at all times or not. The cells should not be drifting apart so much that connecting them together should cause any cause for concern, but leaving them connected should keep them in over-all balance.

I am going to run a full BMS - covering both strings at the cell level. The strings are then going to be connected together via 200Amp fuses, and feed into a single Positive contactor. The Negative will also be common, and run though its own contactor to give me redundancy in case of a welded contact. I do not anticipate this should present any problems, but I have yet to actually hook it all up, so I will keep you posted if I run into problems.

What BMS are you using? The Orion BMS recommends running a contactor for each string and leaving contactor control up to the BMS. It will verify that the strings are close enough in voltage that it's OK to connect. So I think I'm just going to go that route. Then, I can decide to run the two packs entirely separately if I decide to do that at some point. Will just need to only power one contactor at a time for that.

Might be a good call on the contactor for the negative though. One thing I still don't totally understand is pre-charge circuitry and how to configure that to avoid a welded contactor.

 

[OR-Carl]

I am using a thunderstruck BMS, and it does not interface with the contactors. Their documentation does not call for two contactors, but treats the parallel strings as one big pack where you are simply monitoring all the cells. Isolating the strings seems like an added complexity, and with fused cells, I am not sure that it actually adds any safety margin. There are only 2 reasons I can think of that current will want to flow from one string to another;

1) One string is lower voltage than the other - which would never become a problem if they were just always connected together, as being in parallel will force them to maintain the same voltage.
2) A short develops in one string, so that the full voltage of the other string tries to flow through the shorted cell, using the second string as a conductor. This would be a nightmare with big pouch cells, and I would not like to bet that the main fuse on each string would save you from a fire if that happened - but with Tesla packs, any cell that shorts will blow its fuse right away regardless, since they are already parallel in the module.

Am I missing something?

As for precharging: the controllers have big banks of capacitors in them, so if you just connect it up to the battery, the caps act like a short circuit. Since V = IR; if you have 170 volts, and the resistance is something like 50mOhms (or whatever, it will be very small since you are dealing with big 2/0 cables), the battery will try and push 3400 amps for the fraction of a second it takes to fill those capacitors. This current will try and jump the gap in the contactor as it closes, potentially welding it shut. To fix that problem, you want to charge up the caps slowly, using a resistor. If you try and push 170volts through say 100ohms, you only push 1.7 amps. Now the caps take a few seconds to fill, but once full, there is no longer 170 volts across the contactor.

Zeva makes a module that I think would make for a really failsafe solution, but plenty of people have made it work with simpler arrangements.

https://www.zeva.com.au/Products/datasheets/SmartPrechargerV1-3.pdf

 

[reiderM]

As for precharging: the controllers have big banks of capacitors in them, so if you just connect it up to the battery, the caps act like a short circuit. Since V = IR; if you have 170 volts, and the resistance is something like 50mOhms (or whatever, it will be very small since you are dealing with big 2/0 cables), the battery will try and push 3400 amps for the fraction of a second it takes to fill those capacitors. This current will try and jump the gap in the contactor as it closes, potentially welding it shut. To fix that problem, you want to charge up the caps slowly, using a resistor. If you try and push 170volts through say 100ohms, you only push 1.7 amps. Now the caps take a few seconds to fill, but once full, there is no longer 170 volts across the contactor.

Zeva makes a module that I think would make for a really failsafe solution, but plenty of people have made it work with simpler arrangements.

https://www.zeva.com.au/Products/datasheets/SmartPrechargerV1-3.pdf

Now that I think about it more, I remember the controller that comes with the Hyper 9 has an integrated precharge circuit. And I doubt anything else on the B+ bus would pull enough current to make precharge necessary, so I might be fine in that regard. Will have to double check though.

I am using a thunderstruck BMS, and it does not interface with the contactors. Their documentation does not call for two contactors, but treats the parallel strings as one big pack where you are simply monitoring all the cells. Isolating the strings seems like an added complexity, and with fused cells, I am not sure that it actually adds any safety margin. There are only 2 reasons I can think of that current will want to flow from one string to another;

1) One string is lower voltage than the other - which would never become a problem if they were just always connected together, as being in parallel will force them to maintain the same voltage.
2) A short develops in one string, so that the full voltage of the other string tries to flow through the shorted cell, using the second string as a conductor. This would be a nightmare with big pouch cells, and I would not like to bet that the main fuse on each string would save you from a fire if that happened - but with Tesla packs, any cell that shorts will blow its fuse right away regardless, since they are already parallel in the module.

Am I missing something?

I don't think you're missing anything. So safety-wise, with tesla modules it is probably alright to keep them connected with the cell-level fusing. But you're risking your entire pack if something goes wrong, instead of just one of the strings. If the cell fuses blow, I don't think it's an easy fix.

I'm using an Orion BMS, and they suggest having two separate contactors arranged like this:

So I think I'm not going to try to outsmart Orion and just take their advice. You can see that they don't have a contactor on the negative side.

 

[piotrsko]

My $0.02 USD: on a tesla, when you fry a cell fuse you've just hard unbalanced the pack and if the packs have a heavy enough wiring it goes downhill from there. Fry a couple fuses and you have an incident started.

 

[OR-Carl]

when you fry a cell fuse you've just hard unbalanced the pack

Unbalanced would suggest that the voltage would change - which removing a single parallel element will not do. What will have happened is that you have changed the capacity of that cell-group. When you go to charge up, that group will hit High Voltage Disconnect before all the rest. It should not affect the lifespan, but it is sort of like removing a cell from each other cell-group in the string. The impact on capacity then is quite a lot more than just losing one cell.

 

[brian_]

My $0.02 USD: on a tesla, when you fry a cell fuse you've just hard unbalanced the pack and if the packs have a heavy enough wiring it goes downhill from there.

Unbalanced would suggest that the voltage would change - which removing a single parallel element will not do. What will have happened is that you have changed the capacity of that cell-group. When you go to charge up, that group will hit High Voltage Disconnect before all the rest...

But mismatched capacity means the pack will become unbalanced in voltage as soon as the state of charge is changed. With only one cell open-circuited out of 74, that's probably not a severe problem.

 

[OR-Carl]

Yes, this is a valid point. Luckily with lithium batteries, the discharge curve is pretty flat, so even a measurable difference in overall state of charge between two cells is going to yield a fairly small change in the voltage. I suspect if there was a tendency for Tesla packs to experience some sort of cascading failure if a single cell fuse blows, it would have turned up by now. Then again, i dont know how common it is for one of those fuses to blow in the first place. It seems like too much work to include thousands of fuses per battery if they didnt have a very good reason for doing so.

 

[MattsAwesomeStuff]

The cell-level fuses are to protect one cells that fails shorted, from shorting the whole rest of the parallel group through it. It deletes itself from the circuit.

It's not to prevent that one cell from overpowering the rest of its neighbors. That's impossible.

If you lose 1/74, then, weakest link, you've lost 1/74 of your total capacity of the car, not just of that parallel group. (Well, plus whatever the BMS can recover on the fly, which isn't going to be much).

For practical purposes, how often are you using the full power of the pack? Or even half of it? Unless you're drag racing it's doubtful you'll push much past, say, 150hp. So there's plenty of overhead (~2-300%) in a Tesla for even zippy driving. Meaning you could probably blow 35-50 cells before you'd start lighting fuses like chinese firecrackers one after the next. If you get anywhere close to that point, you're in for a looming disaster regardless.

 

[reiderM]

The cell-level fuses are to protect one cells that fails shorted, from shorting the whole rest of the parallel group through it. It deletes itself from the circuit.

It's not to prevent that one cell from overpowering the rest of its neighbors. That's impossible.

If you lose 1/74, then, weakest link, you've lost 1/74 of your total capacity of the car, not just of that parallel group. (Well, plus whatever the BMS can recover on the fly, which isn't going to be much).

For practical purposes, how often are you using the full power of the pack? Or even half of it? Unless you're drag racing it's doubtful you'll push much past, say, 150hp. So there's plenty of overhead (~2-300%) in a Tesla for even zippy driving. Meaning you could probably blow 35-50 cells before you'd start lighting fuses like chinese firecrackers one after the next. If you get anywhere close to that point, you're in for a looming disaster regardless.

Yeah I don't think I'll ever pull so much current that it would blow a cell fuse. But do cells ever fail randomly?

 

[bawfuls]

But then again, EV West often doesn't even use BMSes with lithium batteries, which I find to be honestly idiotic and dangerous.

This is simply not true. Years ago they did not feel any available BMS products were satisfactory and so they relied on periodic manual rebalancing as many diy people did at the time. But that’s not been the case for awhile now. When I started my build 3 years ago they emphatically suggested I use a BMS and I bought the Dilithium Designs one from them (along with most of my other EV components). I actually miss-wired it and blew a component on the main board, and EV West handled the return and repair of the BMS for me. I’ve always found EV West extremely helpful and open to questions/discussion. Then again I live in the area so sometimes I’ll stop by in person when I want their advice.