Tesla High Voltage Battery Pack Information

Repairing The Modules (v1)

Editorial

Most of this tutorial is based on objective experiences and not really much personal opinion and conjecture. I want to spend a minute sharing my OPINION about the nature of most of these battery failures in the Version 1 packs (2012-2015). Admittedly, these are anecdotes based on limited experiences and what I have read from others. They are not scientific studies on large data sets. Only Tesla has the true data on these and they will NEVER share anything that isn't glowing praise. My observation, in short, is most of these early packs fail because of corroded BMBs.

If you get liquid water in your pack through a rusty fuse cover (literally thousands of cars have had has this issue), you will get an internal isolation error of BMS_f123. That is a separate issue. Most of the other internal errors that will take a pack offline are based on what the BMS 'thinks' is happening based on what the BMBs are 'telling' it. It gets data from 3 temperature sensors per module but beyond that, the only data the BMB conveys to the BMS is related to Voltages. Each BMB has circuits to read the 6 unique brick voltages. It is extremely common to see the capacitors that interpret these voltages fail as we will see below. My opinion is that most pack failures are related to failing BMB boards and therefore most packs can be put back in service after repairing or replacing the bad BMBs. A module needing actual replacement or an individual 'failed cell' is the exception, not the rule. Always suspect the BMB first.

I have seen the inside of 2 packs personally. Both were early packs (2013 and 2015) and both had an acute BMB failure and then multiple additional instances of BMBs corroding in earlier stages of failure. I would be shocked to open any pack of this style and not find this corrosion starting somewhere. I'd love to see otherwise and I'll be sure to share and correct this sentiment if I do but I think an unmodified v1 pack without at least some C26/C27 corrosion would be extremely unusual. In the meantime, I am going to assume that there are multiple ticking time bombs inside every pack that should be addressed if the pack is opened for whatever reason. The rest of this section will describe what a colleague and I describe as a 'moisture reset' - how to do it, and why it is necessary.

Getting Started

To do this right, you need to be organized and methodical. You need a clear workbench where you can inspect every part of the module and especially the BMBs. You need to get good at taking the plastic covers off the modules and putting them back on. Modules must be stored with the covers on but you will be taking them on/off a few times each. This is a pretty involved inspection. You will want a jeweler's visor or some other kind of high magnification eye ware and a very bright light. We are looking for sings of corrosion but we are going to coat the known-trouble points with an additional layer of conformal coating even if everything looks good. Conformal coating looks a bit like clear acrylic nail polish and you can get a large bottle of it on Amazon for about $15. Start with your first module...

Inspecting the Module

Look carefully at each set of 'plates' - the shiny metal covers to which the ends of the batteries are fused. You shouldn't see any rust or significant corrosion. The fuse wires should look clean and uniform. Don't touch them. Next, locate the colored wires. These are the voltage sensor wires. There are 3 on one side of the module and 4 on the other. Look carefully at each one and make sure there are no sign of corrosion. If there hasn't been liquid water in the pack, they should be in good shape. You will want to put an additional layer of conformal coating on these sensor wires now so air and moisture won't get to them later.

You might see some yellow gunk around certain cells. Poke it gently with a pick. If it is hard, it is just some extra glue from the assembly process so don't worry about it. If it is crusty, one of the cells could be corroding. The vast majority of your cells should look like the photo to the left. This is from a nearly 10 year old pack that had never been opened.

Take some time to review how the cells are built into bricks with the collector plates I described here. It is a pretty ingenious design!

The sensor wire shown above is from the Version 1 packs made from 2012 through mid-late 2015. Version 2 packs from late 2015 through 2020 have a different sensor wire approach that was intended to provide 'quadruple redundancy' by making contact at 4 points on the plate (good) but the tiny flex cable connections has proven to be very fragile (BAD). As discussed a bit here, I don't have any experience with them but there are more and more stories about them breaking - especially after collisions. The good news is those packs have a different BMB design and don't seem to suffer from the big problem discussed below.

In broad terms, if you have a fuse cover on the bottom, you have v2 modules and if you have the fuse cover on the top, you have v1. There are a few extremely unlucky souls that have the top fuse cover (bad because it likes to rust) AND the fragile sense wires because the new modules were being built before the down-facing fuse v2 cases were ready. At least they have the better BMB design.

The Billion Dollar Problem

Have you ever wondered why C26 and C27 are always the capacitors that fail? Somehow, 12 years into this horrible design, very few people speak matter-of-factly about it. The solution is a simple modification that MUST happen if you are working this far into the pack and I believe it was the original design intention to do exactly what I describe here...

Look closely at the end of a module with the BMB with both covers on. The lower cover goes up the front of the BMB and basically covers it. It is even shaped like the BMB. The top cover goes over the top of the BMB and over the lower cover and comes down to just below the perforated line on the lower cover. The covers overlap a couple of inches. When the yellow tape is applied to hold the top and bottom cover together, it squeezes the bottom cover in and it rests on C26 and C27. This contact, years and years of tiny vibrations, and the presence of humidity from the breathers at the front of the pack combine to wear off the conformal coating on C26 and C27 until they soak up that moisture, corrode, and eventually stop functioning. If you learn nothing else from this tutorial, please at least understand this basic FACT. The double-thick cover should have never been there.

If you are discussing a pack repair with a shop and they don't know about this, inform them. If they say it isn't a big deal or they 'haven't seen it be a problem' and won't address it with your pack, take your car to a different shop. THIS is the 800# Gorilla. THIS is the monster under the bed. Sorry to be dramatic, but I'm right about this. [\discussion]

Modifying The Cover

There is a perforated line on the bottom cover. When you have the bottom cover off inspecting those sense wires, simply fold it back and forth a couple times on the perforations and it will fatigue and snap off. I believe this tab was supposed to be removed during the original assembly process but it took an extra 20 seconds per module so they decided to leave it. Why would you design a perforation like that if it wasn't a fold or snap line? It certainly isn't folding anywhere! I seriously doubt Tesla understood the implications of cutting this corner but the damage is done - a billion dollars worth of packs replaced and that is a conservative estimate. Thankfully most have been under warranty but this generation of packs is all out of warranty now so we need to make sure everyone understand the problem.

I took the process a step farther by installing a longer screw circled in the image to the right and backed it up by a few tiny nylon washers stacked. The screw is stainless and serves as the sacrificial high point on the BMB - a standoff to keep the plastic even farther from the board. I think it is overkill in hindsight but these BMBs will survive for SIGNIFICANTLY longer than they did with the plastic shield rubbing on them.

Correct Application

This is what the assembly should look like when you are done. The bottom and top covers still overlap by a couple of mm but there is now a couple of mm of air gap to those tortured capacitors at the top of the BMB. You will notice that if you overtighten the yellow tape, it can still compress the covers in a bit. Since these were all assembled by hand for the first few years, you can see how modules from one worker may have failed more quickly than modules from another worker but in either case, the failure mode was years down the line so nobody even knew to follow up.

Removing the BMB

Each BMB is held in with 4 tiny plastic fasteners. Pull the pin out of the center and then the rest of the fastener can be wiggled out with fingernails. See if you can do 16 BMBs without losing one of the little black pins!

The Assembly Line

I checked the sense wires on 6 modules at a time, recoated with conformal coating and trimmed the lower cover. I removed each BMB and inspected them for any signs of corrosion. In my first pack, there were five showing signs of corrosion in various stages. In the most recent pack, there were eight. Remember, in each case, only one failed acutely so if my 'repair' had been to just fix the acute failure, each of these packs would have died again within a year or two. This is why refurbished packs from Tesla die so quickly - they do it fast and cut corners.

I recoated C26 and C27 on all of the BMBs that were not showing signs of corrosion and replaced the ones that were (also recoated those two caps) on those. It is critical to make sure the coating is intact, but removing the extra tab on the bottom shield is where the real magic happens.

Pro Tip: My wife was out of town to allow me this nice workshop space. Proceed carefully if yours is home!

C26 and C27 is not the Only Place to Corrode

One of my BMBs also had really bad corrosion on the pins to the sensor harness. This area also makes contact with the extra tab of the bottom shield so on this one, the corrosion started here in addition to the C26/C27 corrosion. You need to check everything, but these are the two usual suspects for v1 BMBs. Recoat these areas also.

Check Voltages Manually

I do recommend you check the voltage of each module and make notes. The six bricks together should be within 10mV which means you will see something like 23.62 or 23.63 (.01 variance) on the main lugs. These values will obviously vary at your specific state of charge so you are looking for consistency at whatever values you have. If you get one that is lower than the others, you should read the individual voltages manually. Use a voltmeter than gets down to the millivolt (mV) which means 3 decimal places. You can use the pins on the BMB to measure this as long as you haven't dumped too much conformal coating on them yet. There are 7 pins labeled 0 through 6. Put the negative probe on zero and the positive on 1. Write it down. That is brick #1 of the module. Put the - on 1 and the + on 2 and write it down. - on 2 and the + on 3, etc... You can also probe combinations of pins to measure 2 or more bricks together. The - on zero and the + on 6 should be the same value as the two lugs (the full module). You can do the same thing with probes on different combinations of the plates.

What If One is Low?

When the BMB is failing, it is sending bogus data to the BMS so the BMS might instruct the BMB to fire off the bleed resistors to lower the voltage of a brick that doesn't need voltage removed. If this happens for long enough, you could have a brick that is actually lower than you want. In this case, you can charge up an individual brick manually using a bench charger. To do this, you need to attach the + and - to the correct sections of the plate. Please note - this is how Gruber burned his shop down... twice. If you aren't sure what sections of the plate to connect, you should get some help from someone with more experience. Do not leave the charger unattended and do not hit the brick with crazy amounts of current. I had a brick to bring up and set the charger at 4.1V at 1 Amp and watched it closely for several hours to bring it up. If you let it run overnight, it could easily overcharge and catch the cells on fire. A bench charger has NONE of the safety or protections built in that the BMS/BMB has. Please don't gloss over this safety concern.

What if One is High?

If your actual brick voltage is too high (see below), you will need to bleed off some energy. Search Amazon for something like 'Adjustable Constant Current Electronic Load' and make sure it will work in the voltage range of our bricks. Like charging an individual brick, discharging one too quickly can generate heat and be dangerous so don't do too much current (Amps). If you overcharge a brick with a manual charger, you will burn down your garage (and the home it is attached to if applicable). If you over-discharge a Brick, it will be a... brick. Don't rush this process.

Manual Balancing

The BMS and functioning BMBs do a good job of balancing the voltages across a module with the bleed resistors as long as they are close but they seem be unable to lower multiple modules to match a neighboring module that is low. One of my packs had a module that was consistently 4-6 mV per brick lower than the averages in the other areas of the pack and I thought fresh BMBs would correct that but as I observed the data in SMT over the next several months, that module remained consistently low. It wasn't a case of capacity because the same module was low at high SOC and at low SOC. In hindsight, I wish I had charged that module up while I had it out of the car. I think if I had, it would have stayed even with its neighbors

Ultimately, it wasn't hurting anything but at the extremely high SOC, the highest brick will limit charging the other 95 bricks and at extremely low SOC, the lowest brick will tell the car where 0% range is and ultimately when to shut things off. 20 extra mV manually placed in that low module (combined across 6 bricks) might have allowed me to use 20 extra mV from all 16 modules at an extremely low SOC. That would have been worth almost an extra mile of range at 400Wh/mi.

The takeaway is if you see a module that is consistently low on SMT, consider bumping it up to match the others while it is out of the car.

BMB Failure Anecdote

BMB Failure Example

The most recent pack I've been helping with had a BMS_u018 error which is supposed to be one of the 'bad' ones where you may have an individual cell that has gone bad requiring a

Cell replacement - The Eastern Europeans say they can do it but I've never seen it documented,
Cell removal - Certain famous YouTube personalities say this is super easy but doesn't say how the pack will compensate for the drop in capacity,
Module Replacement - Some say this is easy but almost nobody can describe how to find a module with an identical discharge curve. Read more about this bad option here.
Module Removal - Taking your 16 Module 400V pack and turning it into a 14 Module 350V pack which does chop off 12.5% of your capacity and range by basic math...

Options #1 and #2 also require you to somehow be able to identify which cell is causing the problem - challenging for the DIYer with limited tools and experience.

The good news on this '018' pack is that it was all BS. Take a look at Cells 71 and 72 in the image to the left - There is almost a FULL Volt (964mV) of imbalance there! Of course the car shut down! If #71 was at 3.158V and #72 matched all the rest, it would be very concerning but how do you suppose #72 got all that energy stuffed into it without catching fire? The short answer is it didn't. The numbers are bogus.

Major Confusion and Guessing

The table to the left is what was really happening with Module #12. Compared to the SMT data above, the numbers are puzzling. Notice that the BMB kept the TOTAL voltage at the same as #11 but didn’t know exactly what was actually happening inside. It also reported a pretty significant difference on the module total from what there was in reality. It bled some energy from the first 4 Bricks and left the last two charged up a little more. It 'knew' there was a problem with the voltage readings so tried to use the Module total and work backwards working through the math. This is a cautionary tale that the data visible in SMT isn't always accurate even though you can see something fishy going on.

A Real Failure?

What about the 'resistive' cell that bleeds energy out of a brick? Do I have a 'bad' module? Maybe... but the only way to tell is to read the voltages and compare them to what SMT reported. If they match SMT (i.e. the BMB was telling the truth), you may have a genuinely bad cell in the module but there is still more testing to do. Balance the bricks out manually, let them stabilize and then let them sit for a couple of days. If the voltage drops in a brick, charge it back up again and disconnect the BMB to make sure the BMB isn't what is bleeding off the energy with a stuck bleed resistor. If the voltage still drops in a brick/cell, you have a problem. If it is stable with the BMB disconnected but drops with it connected, you have a stuck bleed resistor - repair or replace the BMB.

In the example to the left, we see a huge voltage drop in just one brick. When they happen in adjacent pairs, the issue is almost always a BMB. When it happens as a lone-wolf like this, it is usually a bad cell but it could also be a stuck bleed resistor. Check every possibility before you say goodbye to a module.

Best practice for any troubleshooting any module involves charging up the low bricks to match their neighbors, unplugging the BMB so you know it isn't a bleed resistor or some funky malfunctioning circuit on the BMB causing the problem, and watching over the next 24 hours to see if the bricks bleeds down again. Always eliminate the BMB as the source of the problem.

Repairing the BMB

When I fixed my first pack in late 2023, I bought 5 BMBs on eBay for $15 each. The only ones I see on eBay now are $195. Fortunately, they are very easy to repair for a professional board tech. It is important to note, that while the repair is 'easy' for the professional, it is not really a good DIY repair. A hobbyist with a home soldering station is likely to damage the board without the correct experience and equipment. It is strongly recommended that you leave this part of the repair to a professional.

At this time, we are still looking for a vendor or two to systematize these repairs and make refurbished boards available for a small fee with exchange.

Here are the capacitors that need to be replaced.

Nobody should ever throw away a v1 BMB, even if they have failed. They aren't being made anymore so we need to keep them running for these repairs over the long haul!

End Goal

At the end of the refurbishment process, you want 96 bricks matched as closely as possible. A couple mV up or down is fine but if you have a random brick(s) with a 10+ mV delta, get them balanced and make sure they stay that way for a couple of days. Make sure you have thoroughly investigated the BMBs. So much of this proper diagnosis just takes time - time that is very hard for a shop to take. A DIY owner can be so much more thorough.

Page updated

Google Sites

Report abuse