Sorry! I totally forgot to cover the HVIL (High Voltage Interlock Loop) as mentioned at about 1:00. Basically it's a supervisory loop that runs through every HV device and connector, such that if you disconnect a HV connector, it lets the BMS know which then turns off HV to prevent injury or damage. I also neglected to cover isolation detection. Basically the HVP constantly looks at the resistance from the HV system to the car chassis. If it's too low, like in the event of a short, or say if something got wet that shouldn't, the HVP will turn off the HV. UPDATE: Here's the new video on the HVIL: ua-cam.com/video/BC1zCC7CeX8/v-deo.html
Don't think I really agree with that. An EV at it's base can be extreemly simple. I mean my first EV conversion was just a DC motor (1 moving part). Controlled by a home made controller (which is bassicallly a switch that just turns off and on very quickly depending on the speed you want and a home made battery from recycled laptop batteries. Obviously if you want better performance and safety you need to start adding more complexity (Just like a modern fossil fueled car) but at it's heart a basic EV is much simpler than a basic fossil fueled vehicle. Connect a battery to a DC motor and you have forward motion (there's your basic electric vehicle drive unit!) but to run the simplest fossil fueled engine you need to first get motion (by connecting a battery to a dc "starter" motor), pump fuel (at the right pressure), mix it with the precise amount of air, introduce it into the combustion chamber (at the correct time), close the combustion chamber, ignite the fuel (at the precise time), expell the spent gases (at the presise time) and make sure all those complicated systems are lubricated and maintained! That is a lot more complex to do than just getting motion through electricity and magnets. Tesla are pushing the bounds with improving the safety and performance of EV's - that usually involves more complexity. Take the Pyro fuse for example - It has evolved from a basic weak wire link to a fully computer controlled and monitored explosive device!
@@sydneyg007 big lol.... go watch weberauto on tesla teardown, they have really complex and sooo many electronics that are even Software locked....EVs have motor controler, charger/inverter, BMS, high/low voltage protection systems, hvac and many other systems like compressor/hvac... saying that ICE are complex is a joke tbh, there are still ICE on the roads that are over 40 years old.
@@sydneyg007 Indeed, mechanically, they are very simple. But when you consider the BMS, the power electronics, the thermal management system, the magnetics design of the motor, the OBC, the various safety systems and interlocks, it start to be complicated. Of course it's not if you don't design all of these things.
Your videos convinced me to buy a Tesla. Seeing how well made all these PCB designs are just amazes me. The amount of engineering and thought that has going into these is incredible
Very interesting to see the evolution of the BMS from the initial version in the roadster through to this version and the plaid with custom labelled chips and cost reductions. Thanks for a great video.
Really appreciate the insertion of slides like the JTAG and the "bed of nails" makes it easier to understand. I don't mind to pause the video to get more info directly in context!
Thanks, this was my second attempt at editing, it took about 4 hours, but the result was worth it. I've been shooting all my videos in one-take with no scripting, and sometimes I make mistakes. Editing and some better scripting will help make the content better. Please consider supporting the channel! (link in description)
I only recently discovered this channel, but now I'm hooked. I love that you go into excruciatingly deep detail about how this stuff works. It mostly goes over my head, but it's still fun to watch. You're like the "low level" Sandy Munroe, which is great as I continue to wait for his team to get their own Plaid in to do their teardown.
At 10:37 you discuss cell balancing, saying that the BMB boards use the 'Batman'/'Robin' ICs to apply small loads to the cells to balance. That occurs on a different chip & one of the batman/robin (going by the names, my guess is robin), is juts an isoSPI communication line, which is just isolated SPI and used in BMS systems.
Sorry, they absolutely use "wasteful" balancing, only it's not that wasteful because the cells are high quality, and thus, don't need much balancing over their life.
I like a lot. Good Information. It solves my doubts about if the small electronic boards can also balance each battery series cells, using the dump resistors. Thanks so much for this video.
Thanks for your insight once again! Of late I have been watching tear down videos of different manufacturers’ electronics. My general observation is Tesla’s electronics, while being functionally robust, are not over-engineered like the others. Your mention of Tesla dropping robin because of Batman’s good reliability seems to highlight this. The drive inverter, PCS also seem to be remarkably simpler than others. Do you largely agree with this observation?
Tesla is very cost efficient. They have massive data about the fleet performance and seem to use it to remove excess costs. This looks like a typical example of that to me.
@@CafeElectric That definitely seems to be the case and inline with what @ingineerix said in one of his earlier videos- Tesla's on-board diagnostics is unrivaled. But being able to leverage that data to simplify parts is possible only due to vertical integration of the critical components. My guess is asking a supplier to remove "unnecessary" components may not be easy especially given the highly litigious nature of automotive business and its relative nascency. Well, only great engineers can build a bridge that barely stands.
@@Ingineerix Thanks, I take your opinion with considerable weight :). I am a power electronics engineer myself and given the fanfare around Tesla, separating signal from noise and getting a factual understanding of their low-level design and systems engineering from a highly skilled engineer is very hard. And your channel fills that void in a big way. I have much appreciation for your work, amazing skills and contribution! Thank you very much :)
I also tried to remove the Penthouse cover, but I didn't know that the cover makes an interlock loop that instantly opens the main battery contactor. Since the car computer and the screen was still on, it completely discharged my 12V battery (the DC-DC converter was not powered anymore).
@@Ingineerix No worries, I've done this months before your vid ;) I had to access the HV battery terminals for testing purposes. They are incredibly hard to access, you have to unscrew Pentalobe bolts and defeat all interlocks. You have to try very hard to get an electrical shock or access live HV battery wires in a Tesla.
I am an Electrical Engineer. I have been working on and studying Tesla cars since 2014 and EVs since before there were even any production offerings. No, I have never worked for Tesla. They have draconian NDAs, so I would not be able to share all this if I had.
Thanks for the videos! Great job. Will be making a donation today. Could you please upload a video on where are you think we should tap into for a 12 V that will run an inverter or an aftermarket amplifier? This would really help out the audio community as I’m starting to see people tap off that jump post right under the plastic panel that you mentioned in one of your videos will cause damage to the 12 V. I almost had the stereo shop do the same for mine but I have had them put it on hold now until we found a better solution. But you mentioned before that you may know a better spot we could tap directly off of and if you could update that with a new video that would be awesome and potentially save people thousands of dollars and keep customers from getting angry at stereo shops for tapping into the wrong place. Also lots of plaids are having issues with the back of their seat making a squeaking noise I thought maybe you could do a teardown of the driver or passenger seat and show us the microphone array that’s built into the seat for the active noise cancellation. Those videos would be awesome and greatly appreciated. I’m scared to pull the back cover off my seats and fear of breaking it but I want to take it apart to fix the annoying squeaking sound it makes every time I hit a bump. Again thanks a lot for the videos! Subscriber for life!
All hail the algorithm! :-) I hear comments help for that. Thanks for the detailed overview. I'm curious as to what the HV current draw is of the always on power supply when the system is inactive. If you ever happen to measure that... I only saw a glance of it so I'm not sure, but are the cell balance resistors on the back side of the Model 3 cell sensing boards (BMBs)? Are they just one 1/4W resistor per brick? It seems that balancing resistors get smaller every year.
Take a look at this part, I think it's very close to what Tesla is using: www.analog.com/media/en/technical-documentation/data-sheets/adbms1818.pdf No balance resistors at all! Note the 200ma per cell internal balance current capability and 6 µA sleep. The internal BMS power supply shuts off upon sleep command from VCFront, and cannot wake w/o external bootstrap power. The sleep current there is just the sensing dividers.
@@Ingineerix Oh, that is an interesting part. It's been too long since I looked at current parts. Is there some full pack current to keep the pyro sensing even when shut down? I was a bit concerned about that since I have a pack sitting for two years unused, but according the CAN data it has lost less than 2%, so probably a non-issue. So I guess even the internal pyro supply shuts down when VCFront commands, that's nice for those of us that take a long time to finish projects. ;-) Wow! No balance resistors makes for great cost savings. I'm impressed. These design details are yet another reason to HODL the stock.
Another awesome explanation! Would you be able to elaborate on those violet/purple connector headers on the board? From what I've seen on Sandy's videos it's part of an automatic connection assembly process? Interesting for me as I am an Electrical Harness Design Engineer. Thanks!
Very interesting overview, thanks! Just wondering how those parts were corroded if they had conformal coating? Is that a rigid coating or more like a silicone based?
The conformal coating on these is semi-rigid. The coating is designed to reduce damage potential due to condensation, not flooding. The BMB was exposed to liquid water and it is definitely not designed with that in mind. The pack electronics live in a sealed envelope normally.
Conformal coating like this is common on consumer electronics devices that are exposed to the elements of any kind or is in situations where it can be exposed to water and needs to survive it. It's a transparent epoxy with roughly the appearance of acrylic if you were to see a solidified chunk of it. It acts as a barrier to both water (including salt water) and to atmosphere.
@@Ergzay I thought epoxy should not be used as a coating due to thermal expansion of components and the fact that they can crack on heating. But it seems that it is ok, or that is some sort of a semi rigid epoxy
@@tnt9062 This type of coating is somewhat flexible and when used in a thin layer as it is here, it poses no danger to the components. It's almost surely a UV cured Acrylate Polyurethane such as: krayden.com/humiseal-uv-50lv-uv-curable-conformal-coating/
i wonder at what shunt current the pyro is hardware-tripped... and also how long that takes. Sounds like a good-to-know bit of info for folks savaging these battery modules
The HW trip limit is set be resistors on the PCB, and I believe they change these for the different pack sizes as well as the firmware. I'd have to do a fair bit of reversing and bench work to figure out the trip settings, so probably not going to do it anytime soon.
Great explanation! Probably already discussed but how are you able to understand the specific functions of all of the various boards? Are you reverse engineering or do you have a detailed information source?
Hi Phil Do you have some higher quality picture of how they connect the cell interconnections with BMS? Looks like top side laser welding to some metal parts that are somehow connected to BMS PCB. Would be cool to see this up-close. thanks.
Sorry, I don't. I don't have the pack in my possession, I was only allowed one day to study it. I'm trying to raise money to buy one, but it's a tall order.
INGINEERIX... Your thoughts from the EE perspective on general component reliability from your tear downs, etc. Fuzz commentary from Consumer Reports, Edmunds, etc. seem to imply that frequent changes in electrical system suppliers and continuous design changes has resulted in reliability issues in air conditioning, heat pump, center screen, door locking... any relevance to electrical components or software engineering? Thx..
Very complex with praise to all intelligence that makes a Tesla what it is. But at the end of a day if a new revision comes out there is the element of cost savings as every bit possible makes the industry tick in the current state. Pity the shiny coverage could not halt corrosion.
Interesting that there are no BGA or CSP on that PCB, it’s all SOIC and QFP, essentially late 1980’s packaging. Must be for reliability, and easier to repair. BGA have to be under-filled too.
Thanks Ingineerix! Do you have an idea about how they will sense SoC on the LFP vehicles? I heard that the voltage doesn't drop as LFP batteries are discharged
If you are getting a CHG_f012_hwFastChargeDriver code that prevents supercharging, could it be a bad bms board chip signal to the FC contactors? The code is - The fast charger FC contactor drive circuit has shorted to ground while the contactors are requesting to be closed. Seems like it is a chip that drives the mosfet that triggers the FC contactors went bad. Changed onboard chargers on the 2013 MS85 and it charges at home perfectly fine. No other problems with the car.Thanks for this over view.
Yes, this is an internal fault on the BMS board. It's a charge pump driver, not contactors. If you were in the contactor box, it's possible you damaged something while you were in there. You'll need to use Tesla Toolbox to clone the BMS and then install a used one. I would not attempt board-level repair on this board. The FC contactors are not driven by the BMS, they are controlled by the on-board charger, they are not in the battery pack, they are in the HV JB under the back seat. The BMS code has nothing to do with these.
@@Ingineerix toolbox2.1 seems to want a CAN connection to the PT bus to clone the bms. Would using toolbox 3 be the best way to clone the bms? I can’t imagine it’s cloned through the CANbus.
These tiny SMD resistors (mW rating?) are used to dissipate excess charge for balancing? No active balancing which distributes charge from the strong cells to the weak cells?
The batteries and pack construction Tesla uses need very tiny amount of balancing, almost none. The huge numbers of cells cancels out any differences. It's much better than having few large cells like every other EV manufacturer.
In most cases the cells stay well balanced on their own. The amount of balancing normally needed is so minor it's not even a tiny blip. The electronics needed for active balancing would be cost prohibitive and less reliable. It is just not needed.
No, the SMD resistors are not used for cell balancing. Read the linked Analog Devices datasheet for the ***1818 chip which measures the cell voltages for 18 cells. It uses on chip fets to bypass up to 200ma of charge current to achieve cell balance for each of the 18 cells.
I made a video covering some of this years ago: ua-cam.com/video/CLOEGFtFIPA/v-deo.html However Tesla is quite obviously against right-to-repair and does not make this information available to the public. There are some unofficial tools available by third parties, such as "ScanMyTesla", but this is not an endorsement. I really wish Tesla would allow the public access to the basic diagnostic system, but sadly, they have not shown any interest in doing so.
Nice overview, indeed! I would be interested in additional (paid) premium content that actually explains the main HV BMS board and Battery Monitoring Boards (BMB) on the modules on a pinout-level of the ICs. For this, the ICs would have to be explained with their input and outputs, the layers and wires would need to be traced down on the board, and signals to be put on an oscillosope or logic analyzer, and likely a model would be beneficial in some SPICE software. Because quite some statements do not make sense to me at this component level view. Like: why is Batman 64 pins (4x16 pins) and Robin is just (38 = 2x 19 pins)? That is a 26 pin difference. Or on the BMBs: which IC pins are the voltage tabs routed to? How do the two Batman ICs work together? On the earlier version of a BMB: how did the 2 Batman and 2 Robin ICs work together in the older design? The datasheet for the ADBMS1818 describes only an 18-cell BMS in a 64 pin package. Tesla on the other hand needs BMS for 25 cell groups in a Tesla Model 3 larger module. Did they cramp the same into a 64 pin count? How? With time mutiplexing? And stuff like: Why did rain water damage the BMB at 10:17 when there is this glossy protective coating on the board intended to protect from humidity?
I'd definitely love to do it, but the amount of time it would take would be prohibitive. An analysis like this would require hundreds of hours. I am hired to perform this level of reverse-engineering for companies on occasion, and the total bill for it can easily cost more than a whole Tesla does new! In addition, without datasheets on all the parts, it requires a lot of speculation. (or decapping the ASICs and a silicon-level analysis, which just tripled the man-hours!) There are only a few of us that would love this level of detail, but most people here would not, thus there's no way I could afford to do it.
@@Ingineerix So, I will start with some small or most relevant boards like the BMBs and then BMS. And will try to semi-automate this process of reverse engineering: a camera can make images, a positioning system can get the positions, algorithms can detect connections on the PCB and solder points, a classifier algorithm can detect board components like resistors, capacitors, diodes, transistors and whatnot with high likelihood. Thus, a rough model can be generated overnight by some analysis machinery while sleeping. :-D I will just leave the ASIC decapping and silicon-level analysis to other folks with the electron microscope. I had collegues at Infineon doing just that for security ICs. :-) I'm quite aware on the amount of speculation that it requires, but with a nice automated probe tester I'm sure, good (not perfect) results can be received with reasonable effort. And for the large industry-standard processor ICs, the pinout is known from data sheets. This gives a good start. Thank you for all your nice work and inspiration, Phil!
@@koeniglicher Sounds like you should take this on! If you start a UA-cam channel doing this, I'll definitely subscribe! I am running a start-up business and still consulting to pay the bills. I *barely* have time to do the video content I already do.
I found in my tesla model 3 2020 corrosion in the fuse part where it was coated. The mug was green and it seems that the DC voltage did a electro chemical deposition of the metals involved in the soldering composites.
Fascinating to hear about the automatic cell-balancing. So, is the BMS actually able to monitor each cell individually, or are they monitored in groups?
Cell balancing is standard for all BMS's of Li-Ion battery packs, this is not Tesla specific. Tesla Model 3 NCA or NMC packs: Cells are monitored in cell groups, where 31 to 46 cells are all wired in parallel to create one cell group. 96 cell groups in 4 modules of a Model 3: 23s + 25s + 25s + 23s. So the master BMS monitors 96 cell groups. The Battery Monitoring Boards on the modules monitor either 23 cell groups (on the two shorter modules) or 25 cell groups (on the two longer modules). Each cell group is composed from either 31 cells (Standard Range plus), 39 cells (former Mid Range battery pack) or 46 cells (Long Range/Performance) wired in parallel. The cells in one cell group cannot be treated differently, but act as a unit. They get the same balancing input. Tesla LFP packs: cells are monitored individually, because the LFP pack is a 1p design. 102s 1p (23s + 28s + 28s + 23s) educated guess from teardown footage from Munro/One) overall. Pretty large prismatic cells with 180 Ah each. :-)
Read the ADBMS1818 datasheet; there are no balance resistors. There are internal fets that are in parallel with a cell which can divert up to 200ma of charging current and this is controlled by pwm drive signals. So the chip monitors all 18 serial connected cells and keeps their voltages very close together during charging.
They use the bond wires to terminate all the cells in the pack, and it's fully automated, so makes sense to also do this for the connections to the BMB.
Just thought about thos Video a bit and wondered if it would make sense to replace the single remaining Robin Chip with a Batman. Because based on what you said Batman can do everything robin does and more.. Sure, the Part itself might be more expensive (dont think so since you need far more Batmans which gives them a volume advantage), but on the other hand you have one part less in your Parts list and therefore have to keep at hand for Production.
thinking a bit more about it i think its likely they will migrate to Batman only when their already delivered stockpile of Robins is used up maybe? i can imagine Tesla left this one Robin in to still have a way to use them up slowly, and they probably have the Batman only PCB ready for rollout
I bet in future revisions of the HVC, the Robin will be gone. It's only needed to talk to the Robins in the BMBs, and since those are now gone, it's not needed at all.
When you say take extreme care when welding, for example, and that disconnecting the 12V isn't enough to protect against blowing the pyro, what other steps should a reasonable non-electrical person take, before doing anything structural on the car? I realise there won't be exhaust welding due to the lack of emissions..
Once VCFront sleeps the car properly it will shut down the internal power. The only way to be sure without internal access would be to disconnect the connectors on the HVC in the penthouse.
Would you be able to go over variations in boards and modules and comment on ones you feel would be good to avoid on older models? Sort of like the memory cell issue on older ModelS that fail from too many write cycles. Boards that have solved issues or concerns, or just better implementations.
I do often highlight differences, as I did in this video from the old BMBs to the new. I am often asked what Tesla years/models to avoid, here's my quick answer: Don''t buy a Model S before about VIN 60,000 which cooincides with the introduction of Autopilot in Late 2014. Cars made in 2015 or newer were pretty solid, with the exception of the Model X. I would say don't buy/own a Model X unless it's got warranty protection, and even then, be prepared to deal with lots of minor issues. The powertrain in the X is solid (same as S), but all the little things, and some big things, such as the falcon-wing doors give a lot of trouble.
@@Ingineerix Really enjoy your videos, wish we could get this indepth of a review on each and every system to better understand why choices were made and to learn what and how these systems evolve over time. Always interesting! Thanks
Hi Phil, I'm researching using a salvaged UK 2020 Model 3 LR battery pack as the base for a home designed 'powerwall'. Does the BMS function continue to operate when the battery pack is out of a car or does it need something from the car to tell it to be active?
Yes, it can be used this way, and that's the best way, but it needs significant CAN signalling to operate. You system would also have to watch CAN to know state-of-charge, and when to enable and disable charge-discharge. Obviously you'd need to also design your system to use the 300-400v voltage range. The older Model S packs are somewhat less complex to use this way, which is what I'm using for my home backup.
Hi, did you ever manage to get the 'powerwall' project working? I do also plan to use a Model 3 Battery (LFP) as a home energy storage system. As I don't need charging and 12V output when using the pack this way, I'd like to remove the PCS module. Does the rest of the pack still function without the PCS present?
Hi @@Marco-rx4cg I can't comment on anything technical because I don't have the answers. Because of the difficulty in using the electrics that is included within a full pack I have basically resigned myself to using the batteries only and giving/throwing away the circuitry. From my occasional readings I think I am going to see if I can split the 4 nominal 96v packs in half to create 8 nominal 48 volt packs and wire in BMS controllers. The problem is it's near impossible to find 96v equipment in the UK, only in China and frankly I don't want to have to put up with the hassle if something failed! The delay is that I have not yet given up on 96v equipment, hoping someone in the UK will take it on board ... the number of Model 3 96v batteries waiting for support must be building.
Take a look at this part, I think it's very close to what Tesla is using: www.analog.com/media/en/technical-documentation/data-sheets/adbms1818.pdf No external balance resistors needed at all! Note the 200ma per cell internal balance current capability.
@@Ingineerix OK, I read page 72 of the ADBMS1818 data sheet, and even for passive balancing, the discharge resistors are external, just the switching FET could be internal or external, internal up to 200mA. I'm just curious Tesla's implementation of the discharge resistors.
@@gilldo21 The ADBMS1818 is NOT what Tesla used. The Tesla chip is proprietary, so there is no datasheet available. However, you are right and I stand corrected; There are 47 ohm series resistors on the bottom of the BMB PCB. So that puts the balance current at about 90ma peak.
What are pros and cons of isolation transformers vs opto-couplers? Other non-tesla EV manufacturers rely on optical isolation instead of transformers to separate high voltage from low.
Mainly cost and speed. Transformer-based isolation generally allows for a much higher data rate at lower cost. For instance, Ethernet which is used in almost every home now uses transformer-based isolation. Optical isolation at those speeds would be expensive and need many more parts. You need power on both sides of the isolation barrier to use optical, but not usually for magnetic.
Do you have any specific knowledge as to when battery balancing is triggered? There’s a lot of rumor and speculation around how to get the pack to balance but a lot of the info is anecdotal at best. Most speculators don’t have quite the same level of insight you do, and I’m curious if you have that nailed down?
I recommend you allow the car to charge to 100% at least 3-4 times per year. Do not let it sit at 100% for very long though. I usually charge to 75-80% most of the time, unless I need more range. On my 3-4 times per year balance, I take it to 100% in the last 3 hours or so before I know I'm going to leave on a longish trip, so I can get the SoC down fast. This is a good strategy for keeping the pack healthy and balanced. If you never take it to 100%, your pack may drift somewhat out of balance and you'll lost capacity. The occasional 100% charges are a good way to prevent this from happening.
@@Ingineerix This seems anecdotal, is 3 hours really enough time to balance the pack? I think you determined the balance current was only 90 mA in another comment. I'm not sure what the capacity of an individual cell in a Tesla pack is, but it seems like you'd need 12 hours or so to get a good balance (5% of cell capacity?). Other EV's run balancing as needed pretty much full time when the pack is off, I assume Tesla would do the same so your pack would be balanced at 80% if that's what you charge to everyday. Would charging to 100% really improve the balancing or is it a different effect? Seems like it would update the actual pack capacity via coulomb counting more so than to balance it.
@@TechnicalLee The full charge lets the BMS know the pack is out of balance, as the cells don't reveal this when they are in the middle of the dV/SoC curve. This also triggers the string capacity logic to recalculate the individual capacities. Also, The balancing can take place anytime, and doesn't just end after 3 hours. But yes, since I don't have access to the source code, I can only base my theories on what I've observed in the last 7 years on literally thousands of cars. My advice is worth what you paid for it. If you think it's bad, I'm not forcing you to heed it.
3 роки тому
Sorry if anyone has asked before, but how many square feet of circuit board are there actually in a tesla all in all, it seems a lot
No, but luckily in most cases the cells stay well balanced on their own. The amount of balancing needed in most cases is so minor it's not even a tiny blip. The electronics needed for active balancing would be cost prohibitive and less reliable. It is just not needed.
@@Ingineerix pretty disappointing. The Tesla M3 has 46p96s, so I guess when all cells are working, it average out. However, when a single failed cell (fused out), it drops the pack capacity to 45/46 (a little over than 2%). Am I missing anything ?
@@asaftzadok6647 First off a cell failure is very rare, and if so, it doesn't lose the full capacity of one cell per string (implying 96 cells loss for only 1), as it can still use the middle capacity of the remaining cells, just not the ends.
@@Ingineerix I meant individual battery in a group of 46 parallel cells, so instead of having 46 individual cells in parallel, this group will have 45. This means that this group will drain itself faster than others, and will be the limiting factor for the pack.
If you email me, I might be able to help. I don't physically touch cars though. You can find my email here: (doesn't work on phone) ua-cam.com/users/Ingineerixabout
@@Ingineerix there’s usually another IC to monitor the pack current and also read the DC link voltage. It’s Interesting they don’t have it. How are they measuring this ? I’m assuming they have an opamp and then connects to the HV DSP.
very interesting. but for video "quality" i would recommend laying the board on the table and just slide it around. at 30fps handeling and moving the board brings it out of focus constantly and creates a blurry mess that makes people seasick. just zoom in, keep the board on the table, lock the focus and then start recording and slide the board to the part you are discussing and keep it there until you are done discussing that part, dont move it around too much unless needed. its not a fidget spinner. (i made many demonstration videos so its mistakes i also made.) the first board is like yo uare on a ocean. but you can really notice you are moving it without reason with the second board around the 10 min mark. try not doing that.
@@Ingineerix please note that i am not trying to put you down, its just to help you. i also did what you are doing and never noticed it until it was called out to me. its because as the driver you "know" the motion but the person viewing it does not so people watching are getting car/seasick. it also does not help that the youtube butchers the bitrate so everything becomes are blurry mess. try watching the video and you see that the pcb gets instantly fuzzy when you move it but becomes sharp again when you stop moving. (ignoring the camera desperatly trying to stay in focus)
Yes, but keep in mind it's very tiny amounts, almost all lithium balancers work this way, it would be expensive to make a charge-shuttling system for no real advantage. We're probably talking losses on the order of 0.01% of the pack's capacity worst-case. Any more than that and it would be likely the pack would have severe balance issues. Tesla uses very high-quality cells which are very consistent, so in most cases not much balancing is ever needed.
@@johnaweiss I'm not aware of it used in any mass-market product. It is hugely expensive and complex compared to passive balancing, and a high-quality pack like Tesla uses just doesn't need that much balancing. I posit that active balancing would make a pack less reliable.
@@johnaweiss Because it would require a separate isolated power supply with a lot of components in each for EACH of the strings, so that's somewhere around 100 depending on model. The more parts, the more chance of failure, and if just one dies, then the pack is in jeopardy. You either need to take the 4.2v to full pack voltage for each string, or to some intermediate bus voltage, which will then be taken to pack voltage in a separate stage. This is a LOT of cost and potential reliability hit, and all of it must be sealed in the pack which makes service very difficult. I can assure you, no manufacturer will do this just to save a few cents on your electric bill each full charge. It would have no bearing on range, or driving efficiency, so what's the point?
Effectively, yes. Model 3/Y has 96 and Plaid as 110 series strings of parallel cells, so each string voltage is measured and balanced. Effectively all the parallel cells are just one large cell.
@@Ingineerix I didn't see enough hardware to measure 96 inputs. I guess it is else where isolated and multiplexed. What value resistor do they use for balancing? Years ago I figured each cell had enough energy to move the car about 200ft. Funny way to think about it. But when you do it seems absurd and totally impractical. Been driving Tesla for 7 years, I have and S &Y so I find it quite practical. I marvel at the depth and detail you get from your examinations. Amazing. When your videos pop up among others, I watch yours first. Great Job.
@@johnyoungquist6540 The board shown here is one of 4 used in the 3/Y, as I mentioned and showed in the video. They are daisy-chained back to the HVC. I did not have the boards out of the Plaid pack to show on the bench, but I did show a quick video clip. 3/Y have 96 inputs and Plaid has 110 total. There are no resistors on the board for balancing, it is done by PWM'ing FETs located inside the Batman ASIC.
Since electrical welding uses a lot of current, and it's highly intermittent, it creates a huge amount of EMI and RFI (Electromagnetic and Radio-frequency interference). Welding close by the current shunt or HVC may introduce a transient voltage in those circuits that could be read as an overcurrent fault. It's NEVER a good idea to weld near sensitive electronics unless precautions are taken, even if the circuitry is unpowered.
Electrolytic capacitors, there will be problems with this board in 30 years. When I buy old test equipment always check the electrolytic caps if it does not work.
Lol….. you wanted them to waste money to develop a new powertrain!? This is standard in the industry. Take a normal part and boost it for higher performance
That BATMAN chip is a Ti part, the original was a 16 channel BMS manager maybe it's just been rebranded. Instead of removing the part numbers they playing games with naming.
Why do you think it's TI? The comm protocol is extremely similar to LT, and we know they used LT in early prototyping so I always thought it was LT based.
Not possible! Something like this was what they used on the older S/X. This part does not support the twisted pair comms, and requires external balancing FETs and resistors, none of which are present in Tesla's current design.
@@Ingineerix ah thanks for the reply, twisted pair comms that's a good point... this would mean some CAN bus type of protocol used which the chip doesn't support.
LOL…… that has been standard in the industry for decades. Vw for example uses the same parts in all vehicles from the golf to Audi, Porsche, Bentley, Bugatti……. One engine is also used across multiple vehicles models.
@@carholic-sz3qv I was not talking about using the same parts. Sure the VW is using the same parts for multiple models for example. I was talking about how they decreased the number of chips and simplified the design. Also, they are able to do these improvements very often unlike other automotive companies which are not doing it so often.
@@leonardgrant6876 yes vw has also significantly simplified their designs, they don't just use similar engines but they also tune it for different applications, the golf and transporter or crafter have the same engine but different output.
Such a beautifully coated and no doubt costly board, and then it has those big ugly pins on the end just waiting for a ham fisted technician, a spot of corrosion, or more likely just years of vibration to ruin it.
Tesla technicians never go board level, they replace the whole assembly. The pins are well constrained by epoxy in the housing, I had to heat it up in order to remove them. This is inside a plastic housing, which is then sealed in the pack envelope. It will definitely outlast the car.
Sorry! I totally forgot to cover the HVIL (High Voltage Interlock Loop) as mentioned at about 1:00. Basically it's a supervisory loop that runs through every HV device and connector, such that if you disconnect a HV connector, it lets the BMS know which then turns off HV to prevent injury or damage. I also neglected to cover isolation detection. Basically the HVP constantly looks at the resistance from the HV system to the car chassis. If it's too low, like in the event of a short, or say if something got wet that shouldn't, the HVP will turn off the HV.
UPDATE: Here's the new video on the HVIL: ua-cam.com/video/BC1zCC7CeX8/v-deo.html
Is this the same loop thats for fire departments to cut in case they need to work on the Car or is that a seperate loop?
@@unitrader403 Yes.
Would appreciate a follow up vid on the isolation fault detection
Would love a video about HVIL and HVP specially regarding insulation monitoring on HV paths
I think it would be a good idea - this is one of the things I get the most questions about.
It amuses me when some people says that EVs are very simple, anybody can make them... They totally ignore how much ludicrous engineering goes in them.
I see what you did there! =)
Yep, they have very few moving parts but absolutely millions of electronics here and there.
Don't think I really agree with that. An EV at it's base can be extreemly simple. I mean my first EV conversion was just a DC motor (1 moving part). Controlled by a home made controller (which is bassicallly a switch that just turns off and on very quickly depending on the speed you want and a home made battery from recycled laptop batteries. Obviously if you want better performance and safety you need to start adding more complexity (Just like a modern fossil fueled car) but at it's heart a basic EV is much simpler than a basic fossil fueled vehicle. Connect a battery to a DC motor and you have forward motion (there's your basic electric vehicle drive unit!) but to run the simplest fossil fueled engine you need to first get motion (by connecting a battery to a dc "starter" motor), pump fuel (at the right pressure), mix it with the precise amount of air, introduce it into the combustion chamber (at the correct time), close the combustion chamber, ignite the fuel (at the precise time), expell the spent gases (at the presise time) and make sure all those complicated systems are lubricated and maintained! That is a lot more complex to do than just getting motion through electricity and magnets. Tesla are pushing the bounds with improving the safety and performance of EV's - that usually involves more complexity. Take the Pyro fuse for example - It has evolved from a basic weak wire link to a fully computer controlled and monitored explosive device!
@@sydneyg007 big lol.... go watch weberauto on tesla teardown, they have really complex and sooo many electronics that are even Software locked....EVs have motor controler, charger/inverter, BMS, high/low voltage protection systems, hvac and many other systems like compressor/hvac... saying that ICE are complex is a joke tbh, there are still ICE on the roads that are over 40 years old.
@@sydneyg007 Indeed, mechanically, they are very simple. But when you consider the BMS, the power electronics, the thermal management system, the magnetics design of the motor, the OBC, the various safety systems and interlocks, it start to be complicated. Of course it's not if you don't design all of these things.
Your videos convinced me to buy a Tesla. Seeing how well made all these PCB designs are just amazes me. The amount of engineering and thought that has going into these is incredible
Only Elon and Tesla would allow pieces to be called Batman & Robin !! This is what sets them apart from crowd, so cool.
Tesla is so good at this way ahead of everyone else.
Very interesting to see the evolution of the BMS from the initial version in the roadster through to this version and the plaid with custom labelled chips and cost reductions. Thanks for a great video.
your editing is getting good... it's always relevant and well timed, and it doesn't disrupt the flow of your voice
I appreciate that!
Really appreciate the insertion of slides like the JTAG and the "bed of nails" makes it easier to understand. I don't mind to pause the video to get more info directly in context!
Thanks, this was my second attempt at editing, it took about 4 hours, but the result was worth it. I've been shooting all my videos in one-take with no scripting, and sometimes I make mistakes. Editing and some better scripting will help make the content better. Please consider supporting the channel! (link in description)
Tesla electronic engineering is IMPRESSIVE.
I like the community here. More technical and less joking around.
I only recently discovered this channel, but now I'm hooked. I love that you go into excruciatingly deep detail about how this stuff works. It mostly goes over my head, but it's still fun to watch. You're like the "low level" Sandy Munroe, which is great as I continue to wait for his team to get their own Plaid in to do their teardown.
Welcome aboard! Please consider supporting my channel! (link in description)
@@Ingineerix So much better than Sandy, you can tell he's not an in-depth engineer like Ingineerix.
At 10:37 you discuss cell balancing, saying that the BMB boards use the 'Batman'/'Robin' ICs to apply small loads to the cells to balance. That occurs on a different chip & one of the batman/robin (going by the names, my guess is robin), is juts an isoSPI communication line, which is just isolated SPI and used in BMS systems.
Sorry, they absolutely use "wasteful" balancing, only it's not that wasteful because the cells are high quality, and thus, don't need much balancing over their life.
A list of things I didn’t expect in this video:
• A reference to DC Comics.
A list of things I expected in this question:
• More than one item in the list! =)
I like a lot. Good Information. It solves my doubts about if the small electronic boards can also balance each battery series cells, using the dump resistors. Thanks so much for this video.
Thanks for your insight once again! Of late I have been watching tear down videos of different manufacturers’ electronics. My general observation is Tesla’s electronics, while being functionally robust, are not over-engineered like the others. Your mention of Tesla dropping robin because of Batman’s good reliability seems to highlight this. The drive inverter, PCS also seem to be remarkably simpler than others. Do you largely agree with this observation?
Tesla is very cost efficient. They have massive data about the fleet performance and seem to use it to remove excess costs. This looks like a typical example of that to me.
@@CafeElectric That definitely seems to be the case and inline with what @ingineerix said in one of his earlier videos- Tesla's on-board diagnostics is unrivaled. But being able to leverage that data to simplify parts is possible only due to vertical integration of the critical components. My guess is asking a supplier to remove "unnecessary" components may not be easy especially given the highly litigious nature of automotive business and its relative nascency. Well, only great engineers can build a bridge that barely stands.
Yes I do. Tesla is engineering porn! =)
@@Ingineerix "Any idiot can build a bridge that stands, but it takes an engineer to build a bridge that barely stands."
@@Ingineerix Thanks, I take your opinion with considerable weight :). I am a power electronics engineer myself and given the fanfare around Tesla, separating signal from noise and getting a factual understanding of their low-level design and systems engineering from a highly skilled engineer is very hard. And your channel fills that void in a big way. I have much appreciation for your work, amazing skills and contribution! Thank you very much :)
I also tried to remove the Penthouse cover, but I didn't know that the cover makes an interlock loop that instantly opens the main battery contactor. Since the car computer and the screen was still on, it completely discharged my 12V battery (the DC-DC converter was not powered anymore).
Yes, I should have mentioned this, but I totally forgot to cover the HVIL. Maybe that warrants a follow-up?
@@Ingineerix No worries, I've done this months before your vid ;) I had to access the HV battery terminals for testing purposes. They are incredibly hard to access, you have to unscrew Pentalobe bolts and defeat all interlocks. You have to try very hard to get an electrical shock or access live HV battery wires in a Tesla.
Great content and information, thanks! Please consider using a pointer 😉
Thanks, I am learning slowly how to improve the videography. Support of the channel means I can buy better equipment too.
do you expect lifetime issues due to the installed electrolytic capacitors?
New to your channel.
Have you ever said what your background is?
Very interested in how you know so much about the Tesla’s
I am an Electrical Engineer. I have been working on and studying Tesla cars since 2014 and EVs since before there were even any production offerings. No, I have never worked for Tesla. They have draconian NDAs, so I would not be able to share all this if I had.
Thanks for all the amazing insights
Thanks for the videos! Great job. Will be making a donation today. Could you please upload a video on where are you think we should tap into for a 12 V that will run an inverter or an aftermarket amplifier? This would really help out the audio community as I’m starting to see people tap off that jump post right under the plastic panel that you mentioned in one of your videos will cause damage to the 12 V. I almost had the stereo shop do the same for mine but I have had them put it on hold now until we found a better solution. But you mentioned before that you may know a better spot we could tap directly off of and if you could update that with a new video that would be awesome and potentially save people thousands of dollars and keep customers from getting angry at stereo shops for tapping into the wrong place. Also lots of plaids are having issues with the back of their seat making a squeaking noise I thought maybe you could do a teardown of the driver or passenger seat and show us the microphone array that’s built into the seat for the active noise cancellation. Those videos would be awesome and greatly appreciated. I’m scared to pull the back cover off my seats and fear of breaking it but I want to take it apart to fix the annoying squeaking sound it makes every time I hit a bump. Again thanks a lot for the videos! Subscriber for life!
When you say be careful welding, do you mean on the car itself or simply nearby? Is there a possibility of simply a nearby EMF setting off the Pyro?
No, you'd need to be welding really close to the penthouse. I know this because someone working on a conversion did so and blew his Pyro.
All hail the algorithm! :-) I hear comments help for that.
Thanks for the detailed overview.
I'm curious as to what the HV current draw is of the always on power supply when the system is inactive. If you ever happen to measure that...
I only saw a glance of it so I'm not sure, but are the cell balance resistors on the back side of the Model 3 cell sensing boards (BMBs)? Are they just one 1/4W resistor per brick? It seems that balancing resistors get smaller every year.
Take a look at this part, I think it's very close to what Tesla is using:
www.analog.com/media/en/technical-documentation/data-sheets/adbms1818.pdf
No balance resistors at all! Note the 200ma per cell internal balance current capability and 6 µA sleep. The internal BMS power supply shuts off upon sleep command from VCFront, and cannot wake w/o external bootstrap power. The sleep current there is just the sensing dividers.
@@Ingineerix Oh, that is an interesting part. It's been too long since I looked at current parts.
Is there some full pack current to keep the pyro sensing even when shut down?
I was a bit concerned about that since I have a pack sitting for two years unused, but according the CAN data it has lost less than 2%, so probably a non-issue. So I guess even the internal pyro supply shuts down when VCFront commands, that's nice for those of us that take a long time to finish projects. ;-)
Wow! No balance resistors makes for great cost savings. I'm impressed. These design details are yet another reason to HODL the stock.
Another awesome explanation! Would you be able to elaborate on those violet/purple connector headers on the board? From what I've seen on Sandy's videos it's part of an automatic connection assembly process? Interesting for me as I am an Electrical Harness Design Engineer. Thanks!
Very interesting overview, thanks!
Just wondering how those parts were corroded if they had conformal coating?
Is that a rigid coating or more like a silicone based?
The conformal coating on these is semi-rigid. The coating is designed to reduce damage potential due to condensation, not flooding. The BMB was exposed to liquid water and it is definitely not designed with that in mind. The pack electronics live in a sealed envelope normally.
Conformal coating like this is common on consumer electronics devices that are exposed to the elements of any kind or is in situations where it can be exposed to water and needs to survive it. It's a transparent epoxy with roughly the appearance of acrylic if you were to see a solidified chunk of it. It acts as a barrier to both water (including salt water) and to atmosphere.
@@Ergzay I thought epoxy should not be used as a coating due to thermal expansion of components and the fact that they can crack on heating. But it seems that it is ok, or that is some sort of a semi rigid epoxy
@@tnt9062 This type of coating is somewhat flexible and when used in a thin layer as it is here, it poses no danger to the components. It's almost surely a UV cured Acrylate Polyurethane such as: krayden.com/humiseal-uv-50lv-uv-curable-conformal-coating/
i wonder at what shunt current the pyro is hardware-tripped... and also how long that takes. Sounds like a good-to-know bit of info for folks savaging these battery modules
The HW trip limit is set be resistors on the PCB, and I believe they change these for the different pack sizes as well as the firmware. I'd have to do a fair bit of reversing and bench work to figure out the trip settings, so probably not going to do it anytime soon.
Great explanation! Probably already discussed but how are you able to understand the specific functions of all of the various boards? Are you reverse engineering or do you have a detailed information source?
Yes, Observation and Reverse-Engineering as well as a lot of experience with these both for repair and re-use in other applications.
Hi Phil
Do you have some higher quality picture of how they connect the cell interconnections with BMS? Looks like top side laser welding to some metal parts that are somehow connected to BMS PCB. Would be cool to see this up-close. thanks.
Sorry, I don't. I don't have the pack in my possession, I was only allowed one day to study it. I'm trying to raise money to buy one, but it's a tall order.
@@Ingineerix No worries. You already provided plenty of information in your videos. Thank you for that!
INGINEERIX... Your thoughts from the EE perspective on general component reliability from your tear downs, etc. Fuzz commentary from Consumer Reports, Edmunds, etc. seem to imply that frequent changes in electrical system suppliers and continuous design changes has resulted in reliability issues in air conditioning, heat pump, center screen, door locking... any relevance to electrical components or software engineering? Thx..
Thanks for your work!
Thanks for the Thanks! Consider supporting my channel! (link in description)
Very complex with praise to all intelligence that makes a Tesla what it is. But at the end of a day if a new revision comes out there is the element of cost savings as every bit possible makes the industry tick in the current state. Pity the shiny coverage could not halt corrosion.
Interesting that there are no BGA or CSP on that PCB, it’s all SOIC and QFP, essentially late 1980’s packaging. Must be for reliability, and easier to repair. BGA have to be under-filled too.
Thanks Ingineerix! Do you have an idea about how they will sense SoC on the LFP vehicles? I heard that the voltage doesn't drop as LFP batteries are discharged
If you are getting a CHG_f012_hwFastChargeDriver code that prevents supercharging, could it be a bad bms board chip signal to the FC contactors? The code is - The fast charger FC contactor drive circuit has shorted to ground while the contactors are requesting to be closed. Seems like it is a chip that drives the mosfet that triggers the FC contactors went bad. Changed onboard chargers on the 2013 MS85 and it charges at home perfectly fine. No other problems with the car.Thanks for this over view.
Yes, this is an internal fault on the BMS board. It's a charge pump driver, not contactors. If you were in the contactor box, it's possible you damaged something while you were in there. You'll need to use Tesla Toolbox to clone the BMS and then install a used one. I would not attempt board-level repair on this board. The FC contactors are not driven by the BMS, they are controlled by the on-board charger, they are not in the battery pack, they are in the HV JB under the back seat. The BMS code has nothing to do with these.
@@Ingineerix toolbox2.1 seems to want a CAN connection to the PT bus to clone the bms. Would using toolbox 3 be the best way to clone the bms?
I can’t imagine it’s cloned through the CANbus.
These tiny SMD resistors (mW rating?) are used to dissipate excess charge for balancing? No active balancing which distributes charge from the strong cells to the weak cells?
The batteries and pack construction Tesla uses need very tiny amount of balancing, almost none. The huge numbers of cells cancels out any differences. It's much better than having few large cells like every other EV manufacturer.
In most cases the cells stay well balanced on their own. The amount of balancing normally needed is so minor it's not even a tiny blip. The electronics needed for active balancing would be cost prohibitive and less reliable. It is just not needed.
No, the SMD resistors are not used for cell balancing. Read the linked Analog Devices datasheet for the ***1818 chip which measures the cell voltages for 18 cells. It uses on chip fets to bypass up to 200ma of charge current to achieve cell balance for each of the 18 cells.
Thanks for the very interesting video. Could you please show how the values of the BMS can be looked at in a M3? Is there some DIY diagnostic tool?
I made a video covering some of this years ago: ua-cam.com/video/CLOEGFtFIPA/v-deo.html
However Tesla is quite obviously against right-to-repair and does not make this information available to the public. There are some unofficial tools available by third parties, such as "ScanMyTesla", but this is not an endorsement. I really wish Tesla would allow the public access to the basic diagnostic system, but sadly, they have not shown any interest in doing so.
I wonder how the main processor MCP5746A talks to the little guy the TMS570LSx32?
SPI, I2C, CAN? Do they even talk to each other?
Great question! There is a dedicated CAN link.
Thank you!
Nice overview, indeed! I would be interested in additional (paid) premium content that actually explains the main HV BMS board and Battery Monitoring Boards (BMB) on the modules on a pinout-level of the ICs. For this, the ICs would have to be explained with their input and outputs, the layers and wires would need to be traced down on the board, and signals to be put on an oscillosope or logic analyzer, and likely a model would be beneficial in some SPICE software. Because quite some statements do not make sense to me at this component level view. Like: why is Batman 64 pins (4x16 pins) and Robin is just (38 = 2x 19 pins)? That is a 26 pin difference. Or on the BMBs: which IC pins are the voltage tabs routed to? How do the two Batman ICs work together? On the earlier version of a BMB: how did the 2 Batman and 2 Robin ICs work together in the older design? The datasheet for the ADBMS1818 describes only an 18-cell BMS in a 64 pin package. Tesla on the other hand needs BMS for 25 cell groups in a Tesla Model 3 larger module. Did they cramp the same into a 64 pin count? How? With time mutiplexing? And stuff like: Why did rain water damage the BMB at 10:17 when there is this glossy protective coating on the board intended to protect from humidity?
I'd definitely love to do it, but the amount of time it would take would be prohibitive. An analysis like this would require hundreds of hours. I am hired to perform this level of reverse-engineering for companies on occasion, and the total bill for it can easily cost more than a whole Tesla does new! In addition, without datasheets on all the parts, it requires a lot of speculation. (or decapping the ASICs and a silicon-level analysis, which just tripled the man-hours!) There are only a few of us that would love this level of detail, but most people here would not, thus there's no way I could afford to do it.
@@Ingineerix So, I will start with some small or most relevant boards like the BMBs and then BMS. And will try to semi-automate this process of reverse engineering: a camera can make images, a positioning system can get the positions, algorithms can detect connections on the PCB and solder points, a classifier algorithm can detect board components like resistors, capacitors, diodes, transistors and whatnot with high likelihood. Thus, a rough model can be generated overnight by some analysis machinery while sleeping. :-D
I will just leave the ASIC decapping and silicon-level analysis to other folks with the electron microscope. I had collegues at Infineon doing just that for security ICs. :-) I'm quite aware on the amount of speculation that it requires, but with a nice automated probe tester I'm sure, good (not perfect) results can be received with reasonable effort. And for the large industry-standard processor ICs, the pinout is known from data sheets. This gives a good start. Thank you for all your nice work and inspiration, Phil!
@@koeniglicher Sounds like you should take this on! If you start a UA-cam channel doing this, I'll definitely subscribe! I am running a start-up business and still consulting to pay the bills. I *barely* have time to do the video content I already do.
I found in my tesla model 3 2020 corrosion in the fuse part where it was coated. The mug was green and it seems that the DC voltage did a electro chemical deposition of the metals involved in the soldering composites.
Fascinating to hear about the automatic cell-balancing. So, is the BMS actually able to monitor each cell individually, or are they monitored in groups?
Cell balancing is standard for all BMS's of Li-Ion battery packs, this is not Tesla specific.
Tesla Model 3 NCA or NMC packs: Cells are monitored in cell groups, where 31 to 46 cells are all wired in parallel to create one cell group. 96 cell groups in 4 modules of a Model 3: 23s + 25s + 25s + 23s. So the master BMS monitors 96 cell groups. The Battery Monitoring Boards on the modules monitor either 23 cell groups (on the two shorter modules) or 25 cell groups (on the two longer modules). Each cell group is composed from either 31 cells (Standard Range plus), 39 cells (former Mid Range battery pack) or 46 cells (Long Range/Performance) wired in parallel. The cells in one cell group cannot be treated differently, but act as a unit. They get the same balancing input.
Tesla LFP packs: cells are monitored individually, because the LFP pack is a 1p design. 102s 1p (23s + 28s + 28s + 23s) educated guess from teardown footage from Munro/One) overall. Pretty large prismatic cells with 180 Ah each. :-)
First! And the uploads…….. I LOVE IT!😍
Is it possible to know whether the balance resistors are active all the time? Or only active over a certain string voltage for example? Great video 🤙🏻
Read the ADBMS1818 datasheet; there are no balance resistors. There are internal fets that are in parallel with a cell which can divert up to 200ma of charging current and this is controlled by pwm drive signals. So the chip monitors all 18 serial connected cells and keeps their voltages very close together during charging.
Did you also dissect other EVS BMS? How many years ahead is Tesla?
Why did they use pairs of bond wires to connect the board instead of more traditional connectors ?
They use the bond wires to terminate all the cells in the pack, and it's fully automated, so makes sense to also do this for the connections to the BMB.
hi, can you tell me what is the circuit responsible for the 20 mAh HVIL signal? Thank you
Awesome
Thanks! This one was fun to make, a lot more editing than I've ever done.
Just thought about thos Video a bit and wondered if it would make sense to replace the single remaining Robin Chip with a Batman. Because based on what you said Batman can do everything robin does and more..
Sure, the Part itself might be more expensive (dont think so since you need far more Batmans which gives them a volume advantage), but on the other hand you have one part less in your Parts list and therefore have to keep at hand for Production.
thinking a bit more about it i think its likely they will migrate to Batman only when their already delivered stockpile of Robins is used up maybe? i can imagine Tesla left this one Robin in to still have a way to use them up slowly, and they probably have the Batman only PCB ready for rollout
I bet in future revisions of the HVC, the Robin will be gone. It's only needed to talk to the Robins in the BMBs, and since those are now gone, it's not needed at all.
When you say take extreme care when welding, for example, and that disconnecting the 12V isn't enough to protect against blowing the pyro, what other steps should a reasonable non-electrical person take, before doing anything structural on the car? I realise there won't be exhaust welding due to the lack of emissions..
Once VCFront sleeps the car properly it will shut down the internal power. The only way to be sure without internal access would be to disconnect the connectors on the HVC in the penthouse.
Would you be able to go over variations in boards and modules and comment on ones you feel would be good to avoid on older models? Sort of like the memory cell issue on older ModelS that fail from too many write cycles. Boards that have solved issues or concerns, or just better implementations.
I do often highlight differences, as I did in this video from the old BMBs to the new. I am often asked what Tesla years/models to avoid, here's my quick answer: Don''t buy a Model S before about VIN 60,000 which cooincides with the introduction of Autopilot in Late 2014. Cars made in 2015 or newer were pretty solid, with the exception of the Model X. I would say don't buy/own a Model X unless it's got warranty protection, and even then, be prepared to deal with lots of minor issues. The powertrain in the X is solid (same as S), but all the little things, and some big things, such as the falcon-wing doors give a lot of trouble.
@@Ingineerix Really enjoy your videos, wish we could get this indepth of a review on each and every system to better understand why choices were made and to learn what and how these systems evolve over time. Always interesting! Thanks
Are the BMS master board or battery monitoring boards any different on the LFP packs and modules?
I haven't seen one yet, but I suspect no. Maybe slight differences.
Hi Phil, I'm researching using a salvaged UK 2020 Model 3 LR battery pack as the base for a home designed 'powerwall'. Does the BMS function continue to operate when the battery pack is out of a car or does it need something from the car to tell it to be active?
Yes, it can be used this way, and that's the best way, but it needs significant CAN signalling to operate. You system would also have to watch CAN to know state-of-charge, and when to enable and disable charge-discharge. Obviously you'd need to also design your system to use the 300-400v voltage range. The older Model S packs are somewhat less complex to use this way, which is what I'm using for my home backup.
Hi, did you ever manage to get the 'powerwall' project working?
I do also plan to use a Model 3 Battery (LFP) as a home energy storage system.
As I don't need charging and 12V output when using the pack this way, I'd like to remove the PCS module.
Does the rest of the pack still function without the PCS present?
Hi @@Marco-rx4cg I can't comment on anything technical because I don't have the answers. Because of the difficulty in using the electrics that is included within a full pack I have basically resigned myself to using the batteries only and giving/throwing away the circuitry. From my occasional readings I think I am going to see if I can split the 4 nominal 96v packs in half to create 8 nominal 48 volt packs and wire in BMS controllers. The problem is it's near impossible to find 96v equipment in the UK, only in China and frankly I don't want to have to put up with the hassle if something failed! The delay is that I have not yet given up on 96v equipment, hoping someone in the UK will take it on board ... the number of Model 3 96v batteries waiting for support must be building.
Can you show the BMB bleed resistors? How much power can they dissipate from a parallel cell group?
Take a look at this part, I think it's very close to what Tesla is using:
www.analog.com/media/en/technical-documentation/data-sheets/adbms1818.pdf
No external balance resistors needed at all! Note the 200ma per cell internal balance current capability.
@@Ingineerix does that mean they are using active cell balancing?
@@gilldo21 No, it's active in the sense that it's digitally controlled, but if you mean "charge shuttling", no, it does not do this.
@@Ingineerix OK, I read page 72 of the ADBMS1818 data sheet, and even for passive balancing, the discharge resistors are external, just the switching FET could be internal or external, internal up to 200mA. I'm just curious Tesla's implementation of the discharge resistors.
@@gilldo21 The ADBMS1818 is NOT what Tesla used. The Tesla chip is proprietary, so there is no datasheet available. However, you are right and I stand corrected; There are 47 ohm series resistors on the bottom of the BMB PCB. So that puts the balance current at about 90ma peak.
Thanks for the video. Great to see the USA flag!
What are pros and cons of isolation transformers vs opto-couplers? Other non-tesla EV manufacturers rely on optical isolation instead of transformers to separate high voltage from low.
Mainly cost and speed. Transformer-based isolation generally allows for a much higher data rate at lower cost. For instance, Ethernet which is used in almost every home now uses transformer-based isolation. Optical isolation at those speeds would be expensive and need many more parts. You need power on both sides of the isolation barrier to use optical, but not usually for magnetic.
Bed of nails is for end of fab line testing . Not sure why you reference it for debugging
At 8:27, I said "Testing and Programming", I later mentioned "development".
Do you have any specific knowledge as to when battery balancing is triggered? There’s a lot of rumor and speculation around how to get the pack to balance but a lot of the info is anecdotal at best. Most speculators don’t have quite the same level of insight you do, and I’m curious if you have that nailed down?
I recommend you allow the car to charge to 100% at least 3-4 times per year. Do not let it sit at 100% for very long though. I usually charge to 75-80% most of the time, unless I need more range. On my 3-4 times per year balance, I take it to 100% in the last 3 hours or so before I know I'm going to leave on a longish trip, so I can get the SoC down fast. This is a good strategy for keeping the pack healthy and balanced. If you never take it to 100%, your pack may drift somewhat out of balance and you'll lost capacity. The occasional 100% charges are a good way to prevent this from happening.
@@Ingineerix This is basically to reset the coulomb counters and the max voltage per cell?
@@supernumex It's to trigger a full balance. Coulomb counter is reset by a map if it drifts too much, so that's not any issue.
@@Ingineerix This seems anecdotal, is 3 hours really enough time to balance the pack? I think you determined the balance current was only 90 mA in another comment. I'm not sure what the capacity of an individual cell in a Tesla pack is, but it seems like you'd need 12 hours or so to get a good balance (5% of cell capacity?). Other EV's run balancing as needed pretty much full time when the pack is off, I assume Tesla would do the same so your pack would be balanced at 80% if that's what you charge to everyday. Would charging to 100% really improve the balancing or is it a different effect? Seems like it would update the actual pack capacity via coulomb counting more so than to balance it.
@@TechnicalLee The full charge lets the BMS know the pack is out of balance, as the cells don't reveal this when they are in the middle of the dV/SoC curve. This also triggers the string capacity logic to recalculate the individual capacities. Also, The balancing can take place anytime, and doesn't just end after 3 hours. But yes, since I don't have access to the source code, I can only base my theories on what I've observed in the last 7 years on literally thousands of cars. My advice is worth what you paid for it. If you think it's bad, I'm not forcing you to heed it.
Sorry if anyone has asked before, but how many square feet of circuit board are there actually in a tesla all in all, it seems a lot
Good question, but I don't have that answer. Probably under a square meter though.
Wow, these cars are so complicated.
Complicated is underrated mate.
The balancing is done via draining cells ?
I thought that they move charge from a higher capacity cell to a lower capacity.
No, but luckily in most cases the cells stay well balanced on their own. The amount of balancing needed in most cases is so minor it's not even a tiny blip. The electronics needed for active balancing would be cost prohibitive and less reliable. It is just not needed.
@@Ingineerix pretty disappointing. The Tesla M3 has 46p96s, so I guess when all cells are working, it average out. However, when a single failed cell (fused out), it drops the pack capacity to 45/46 (a little over than 2%). Am I missing anything ?
@@asaftzadok6647 First off a cell failure is very rare, and if so, it doesn't lose the full capacity of one cell per string (implying 96 cells loss for only 1), as it can still use the middle capacity of the remaining cells, just not the ends.
@@Ingineerix I meant individual battery in a group of 46 parallel cells, so instead of having 46 individual cells in parallel, this group will have 45. This means that this group will drain itself faster than others, and will be the limiting factor for the pack.
Does Phil still offer services for repairing model 3
If you email me, I might be able to help. I don't physically touch cars though. You can find my email here: (doesn't work on phone) ua-cam.com/users/Ingineerixabout
Hi, did not find the battery current monitoring IC on this board? Is it not present on this board?
No, there is an external shunt.
@@Ingineerix there’s usually another IC to monitor the pack current and also read the DC link voltage. It’s Interesting they don’t have it. How are they measuring this ? I’m assuming they have an opamp and then connects to the HV DSP.
@@DivineSoundsMaster-nm6re Sorry, This is beyond a youtube comment discussion. You'd have to hire me for engineering research.
very interesting. but for video "quality" i would recommend laying the board on the table and just slide it around. at 30fps handeling and moving the board brings it out of focus constantly and creates a blurry mess that makes people seasick. just zoom in, keep the board on the table, lock the focus and then start recording and slide the board to the part you are discussing and keep it there until you are done discussing that part, dont move it around too much unless needed. its not a fidget spinner. (i made many demonstration videos so its mistakes i also made.) the first board is like yo uare on a ocean. but you can really notice you are moving it without reason with the second board around the 10 min mark. try not doing that.
Thanks for the tip, I'll try to improve it in the future!
@@Ingineerix please note that i am not trying to put you down, its just to help you. i also did what you are doing and never noticed it until it was called out to me. its because as the driver you "know" the motion but the person viewing it does not so people watching are getting car/seasick.
it also does not help that the youtube butchers the bitrate so everything becomes are blurry mess. try watching the video and you see that the pcb gets instantly fuzzy when you move it but becomes sharp again when you stop moving. (ignoring the camera desperatly trying to stay in focus)
Where are the balancing resistors?
Does the balancer just bleed off the fuller cells as heat? Or does it pass the extra charge to the emptier cells?
Yes, but keep in mind it's very tiny amounts, almost all lithium balancers work this way, it would be expensive to make a charge-shuttling system for no real advantage. We're probably talking losses on the order of 0.01% of the pack's capacity worst-case. Any more than that and it would be likely the pack would have severe balance issues. Tesla uses very high-quality cells which are very consistent, so in most cases not much balancing is ever needed.
@@Ingineerix I thought active balancers are becoming common.
@@johnaweiss I'm not aware of it used in any mass-market product. It is hugely expensive and complex compared to passive balancing, and a high-quality pack like Tesla uses just doesn't need that much balancing. I posit that active balancing would make a pack less reliable.
@@Ingineerix I'm not arguing that Tesla should use active balancing, just learning. Why would it make the pack less reliable? Thx
@@johnaweiss Because it would require a separate isolated power supply with a lot of components in each for EACH of the strings, so that's somewhere around 100 depending on model. The more parts, the more chance of failure, and if just one dies, then the pack is in jeopardy. You either need to take the 4.2v to full pack voltage for each string, or to some intermediate bus voltage, which will then be taken to pack voltage in a separate stage. This is a LOT of cost and potential reliability hit, and all of it must be sealed in the pack which makes service very difficult. I can assure you, no manufacturer will do this just to save a few cents on your electric bill each full charge. It would have no bearing on range, or driving efficiency, so what's the point?
Do they measure each every cell voltage?
Effectively, yes. Model 3/Y has 96 and Plaid as 110 series strings of parallel cells, so each string voltage is measured and balanced. Effectively all the parallel cells are just one large cell.
@@Ingineerix I didn't see enough hardware to measure 96 inputs. I guess it is else where isolated and multiplexed. What value resistor do they use for balancing? Years ago I figured each cell had enough energy to move the car about 200ft. Funny way to think about it. But when you do it seems absurd and totally impractical. Been driving Tesla for 7 years, I have and S &Y so I find it quite practical. I marvel at the depth and detail you get from your examinations. Amazing. When your videos pop up among others, I watch yours first. Great Job.
@@johnyoungquist6540 The board shown here is one of 4 used in the 3/Y, as I mentioned and showed in the video. They are daisy-chained back to the HVC. I did not have the boards out of the Plaid pack to show on the bench, but I did show a quick video clip. 3/Y have 96 inputs and Plaid has 110 total. There are no resistors on the board for balancing, it is done by PWM'ing FETs located inside the Batman ASIC.
Why would welding cause the pyro fuse to fire?
Since electrical welding uses a lot of current, and it's highly intermittent, it creates a huge amount of EMI and RFI (Electromagnetic and Radio-frequency interference). Welding close by the current shunt or HVC may introduce a transient voltage in those circuits that could be read as an overcurrent fault. It's NEVER a good idea to weld near sensitive electronics unless precautions are taken, even if the circuitry is unpowered.
👍👍👍👍
thx
Electrolytic capacitors, there will be problems with this board in 30 years. When I buy old test equipment always check the electrolytic caps if it does not work.
those capacitors will outlive the lithium cells, its a non-issue
@@MarcosChaparro
For sure, in Canada with the salt we put on the roads they will out live 2 care bodies.
I like how all this Model 3 technology trickled up to the Model S.
Lol….. you wanted them to waste money to develop a new powertrain!? This is standard in the industry. Take a normal part and boost it for higher performance
Tesla is really on a different level than all of legacy auto.
Lol……
Model 3 parts genius!!
Not genius but just common sense….
That BATMAN chip is a Ti part, the original was a 16 channel BMS manager maybe it's just been rebranded. Instead of removing the part numbers they playing games with naming.
Why do you think it's TI? The comm protocol is extremely similar to LT, and we know they used LT in early prototyping so I always thought it was LT based.
@@CafeElectric The previous design was a Ti BMS Manager, forgot the part number. They probably trying to hide these factors going forward.
Texas Instruments bq76PL536AQ1
Not possible! Something like this was what they used on the older S/X. This part does not support the twisted pair comms, and requires external balancing FETs and resistors, none of which are present in Tesla's current design.
@@Ingineerix ah thanks for the reply, twisted pair comms that's a good point... this would mean some CAN bus type of protocol used which the chip doesn't support.
Tesla is really doing a great job in order to cut the cost, impressive no wonder the Tesla stocks are going up.
LOL…… that has been standard in the industry for decades. Vw for example uses the same parts in all vehicles from the golf to Audi, Porsche, Bentley, Bugatti……. One engine is also used across multiple vehicles models.
@@carholic-sz3qv I was not talking about using the same parts. Sure the VW is using the same parts for multiple models for example. I was talking about how they decreased the number of chips and simplified the design. Also, they are able to do these improvements very often unlike other automotive companies which are not doing it so often.
@@leonardgrant6876 yes vw has also significantly simplified their designs, they don't just use similar engines but they also tune it for different applications, the golf and transporter or crafter have the same engine but different output.
Why aren't the entire boards conformity coated?
Cost. Additionally you can't coat connectors or areas that need to conduct, such as mounting holes.
Such a beautifully coated and no doubt costly board, and then it has those big ugly pins on the end just waiting for a ham fisted technician, a spot of corrosion, or more likely just years of vibration to ruin it.
Tesla technicians never go board level, they replace the whole assembly. The pins are well constrained by epoxy in the housing, I had to heat it up in order to remove them. This is inside a plastic housing, which is then sealed in the pack envelope. It will definitely outlast the car.
And an American Flag
Looks like this stuff is impossible to repair by a 3'rd party.