Pics/Info: Inside the battery pack

[wk057]

The wire that connects the BMS modules together is quite literally a loop. It appears that the boards need to be all connected to make the chain continuous, also, since half of the connector connects to the previous board and half connects to the next.

I'll see if I can't get a picture of this later.

 

[Cottonwood]

WRT all the BMS boards, there has to be some type of level independent serial bus such that modules referenced at different voltages can all hang out and talk to the main board (422???). The distances and different voltage references rule out a lot of busses.

I'm guessing optically isolated serial buses.

That would make it easy for each BMB to hang out at its own module's Voltages and still provide and easy interconnect.

- - - Updated - - -

The wire that connects the BMS modules together is quite literally a loop. It appears that the boards need to be all connected to make the chain continuous, also, since half of the connector connects to the previous board and half connects to the next.

I'll see if I can't get a picture of this later.

Take a look at what the loop connectors go to on the BMB's. I bet that there are optical isolation devices there.

 

[nlc]

I think they don't daisy chain the TI SPI bus throuht all 16 modules to the master board. SPI bus is a sensitive and high speed bus, not really good for long wiring.
They probably interfaced the BMS modules through a microcontroler, which manage the SPI communication with the TI chip on one side, and communicate with the master board with another kind of bus which can support long distance, as RS485 for example.

At the left of the TI chip there is a 20 pin QFN chip, I can read SIL FS30A, seems to be a Silicon Lab microcontroller FS53x serie.
Pin 16 and pin 17 go the the isolator chip. These pins are P0.4/TX and P0.RX

Thus they communicate through an usart link, thus it's probably RS485 or something like that.
It means all board module are probably in parallel, and the BMS master board communicate to each module with an addressing

- - - Updated - - -

wk057 : you said the wiring go to each module like this 1 --> 2 --> 3 --> 4, etc....
Can you look with the multimeter if wire entering a board are the same signal that wire output which go to the next BMS module ?

 

[yobigd20]

you know, I've seen a few recent Tesla patents publicly published that might help you out. dunno if they were available prior to last month:

Patent application title: FLEXIBLE PRINTED CIRCUIT AS HIGH VOLTAGE INTERCONNECT IN BATTERY MODULES
A System and method for improving on conventional techniques for connecting energy storage elements of a high-voltage battery pack. A tuned flexible printed circuit individually coupled to each electrode of each cell of a matrix of cells of the battery pack provides improved manufacturability and reliability.

details: FLEXIBLE PRINTED CIRCUIT AS HIGH VOLTAGE INTERCONNECT IN BATTERY MODULES - Patent application

Patent application title: POWER ELECTRONICS INTERCONNECTION FOR ELECTRIC MOTOR DRIVES
The bus bar includes a first bus bar layer formed of a first generally uniform thickness of a first bus bar conductor; a first dielectric layer overlying a top surface of the first bus bar layer; and a second bus bar layer formed of a second generally uniform thickness of a second bus bar conductor overlying a top surface of the first dielectric layer and the top surface of the first bus bar layer wherein: the first bus bar layer includes a first via for receipt of a first electrical lead of an electrical component and a second via for receipt of a second electrical lead of the electrical component and wherein: the first dielectric layer and the second bus bar layer each include a via aligned with the first via wherein the first electrical lead is extendable from beneath the first bus bar layer through the first dielectric layer and through the second bus bar layer.


details: POWER ELECTRONICS INTERCONNECTION FOR ELECTRIC MOTOR DRIVES - Patent application

lots more patents: TESLA MOTORS, INC. - Patent applications

 

[kennybobby]

There is no optical bus:

SPI Communications – Device to Host
Device-to-host (D2H) mode is provided on the SPI interface pins for connection to a local host microcontroller, logic, etc. D2H communications operate in voltage mode as a standard SPI interface for ease of connection to the outside world from the bq76PL536Adevice. Standard TTL-compatible logic levels are presented. All relevant SPI timing and performance parameters are met by this interface.

BMB to BMB:


Device-to-Device Vertical Bus (VBUS) Interface
Device-to-device (D2D) communications makes use of a unique, current-mode interface which provides common-mode voltage isolation between successive bq76PL536As. This vertical bus (VBUS) is found on the _N and corresponding _S pins. It provides high-speed I/O for both the SPI bus and the direct I/O pins CONV and DRDY. The current-mode interface minimizes the effects of wiring capacitance on the interface speed.

 

[Cottonwood]

There is no optical bus:

SPI Communications – Device to Host
Device-to-host (D2H) mode is provided on the SPI interface pins for connection to a local host microcontroller, logic, etc. D2H communications operate in voltage mode as a standard SPI interface for ease of connection to the outside world from the bq76PL536Adevice. Standard TTL-compatible logic levels are presented. All relevant SPI timing and performance parameters are met by this interface.

BMB to BMB:


Device-to-Device Vertical Bus (VBUS) Interface
Device-to-device (D2D) communications makes use of a unique, current-mode interface which provides common-mode voltage isolation between successive bq76PL536As. This vertical bus (VBUS) is found on the _N and corresponding _S pins. It provides high-speed I/O for both the SPI bus and the direct I/O pins CONV and DRDY. The current-mode interface minimizes the effects of wiring capacitance on the interface speed.

Sorry for the confusion, but I did not say optical bus, I said optical isolators, or optical couplers that would be on the BMB and the BMB's would be connected by wires. See Optocoupler Tutorial and Optocoupler Application for an explanation. Here is an example of such a device that is less that $0.10 in volume and can handle 1,500 Volts of isolation: Mouser - Transistor Output Optocouplers PTR 10%, 1.5KV. These optical isolators could also be built into some of the IC's that implement higher functions on the BMB.

Optical isolators take a current to run an LED that excites a photo transistor or such, and provide Voltage isolation across that optical path. The optical isolator accepts current in and handles the common mode Voltage between BMB's. From BMB to BMB there is a maximum common mode Voltage of 25 Volts. Because the BMB's are in a circle connecting back to the BMS, there is a 400 Volt potential to jump at the end.

I would bet that at least that one, if not all, the interfaces have optical isolators to accept a current in and take care of common mode Voltage differences.

 

[kennybobby]

Cheers mate, i'll take that bet-- that they don't use optical isolators. Also i'll bet that the BMB cable does not make a circle loop, the top and bottom boards only connect to one other board while the inner BMBs connect to two neighbors.

 

[wk057]

Also i'll bet that the BMB cable does not make a circle loop, the top and bottom boards only connect to one other board while the inner BMBs connect to two neighbors.

It actually does make a full circle...

 

[SirNick]

Hard to say. I don't remember where the OP said the modules connect together -- if it's the white WTB connector (J6), or the bigger black one (J1).

J6 is labeled TS1 & TS2 +/-. The traces from the connector are routed via an inner layer, and I can just make them out on the top-side scan, until they go underneath the passives near the SiLabs F530A (8051-based micro), where I lose track of them. There doesn't look to be optical isolation, though. I suspect that one wasn't the module bus, but maybe temperature sensors (hence, "TS").

But, if that is the module interconnect, then based on the peripherals listed in the micro's datasheet, it might be a LIN bus -- it's kind of the intended application. I'm not very familiar with LIN. It seems typical implementations are single-ended, level-shifted from around 12V to whatever the micro wants, internally. Unless I missed it, the datasheet doesn't specify anything beyond ~5V tolerance, and the +/- silkscreen would kind of imply differential signalling. It doesn't really add up.

Now, if the modules connect via the black connector (J1), then they go into what looks like a SiLabs 8642ED. Can't find much on that part.

From what I can see, there are four pins connected -- two at the top left and right, connected together; and two more that might be TX / RX, or a bi-directional differential pair. There are SOT (D)iodes between the connector and the uC. It looks like the trace on the second position from the top left (near the "PCBA" silkscreen) goes into one of the pins through a 4.7M resistor (R72). It looks like the other pin, near R90/C58 on the bottom right, is driven with a 4.7k resistor. The diodes could be clamps, which, combined with the high-value series resistors, could be limiting the voltage to something the uC can handle. That would make sense with differential signalling, but single-ended would be tricky without a common reference (ground) between boards. I haven't traced all the stuff, and it's a bit hard to tell where some of them go with the parts and silkscreen in the way.

- - - Updated - - -

Oh, also, based on the layout, and nearby "VBAT" silk, U2 looks like it might be a voltage regulator. If I were going to hack this board, I would probably start there.

 

[hcsharp]

... I would bet that at least that one, if not all, the interfaces have optical isolators to accept a current in and take care of common mode Voltage differences.

Cheers mate, i'll take that bet-- that they don't use optical isolators. Also i'll bet that the BMB cable does not make a circle loop, the top and bottom boards only connect to one other board while the inner BMBs connect to two neighbors.

Hah! You're both wrong! It uses newer tech than opto-isolators. It uses RF isolators. Don't worry if you hadn't heard of it before now. I hadn't either! From the isolator chip datasheet:

The operation of an Si864x channel is analogous to that of an opto coupler, except an RF carrier is modulated
instead of light.

I don't take credit for figuring this out. The Si8642ED-B-IS chip being used was posted earlier in the thread, and somebody else even identified it as a radio-isolator, which I'd never heard of before so I read the data sheet. You learn something new every day!

 

[rabar10]

Lots of good info and links to previously-identified ICs were posted earlier in-thread:

On that board

U4 - Silabs isolator (http://uk.mouser.com/ProductDetail/Silicon-Labs/Si8642ED-B-IS/?qs=NVLsoTRMv1oBDPBFWNUX/g==) for the interface back to the BMS board
U100 - Silabs MCU (http://www.ko4bb.com/Manuals/12)_Components_Specs&Datasheets/Silabs/C8051F520A-F530A.pdf) very basic - probably only interfaces with TI chip
U1 - TI battery management IC, part number seems custom or possibly re-marked (it seems unlikely Tesla would be worth spinning an IC for - but maybe since this is a later revision it is?)
H
Interesting that they have a huge isolation barrier for the pack board. I wonder if the main BMS board is completely isolated from the pack, as I wouldn't have thought something like 8mm isolation would otherwise be necessary.

edit: nvm - the TI IC is definitely not custom or re-marked, it's this:
http://www.ti.com/product/bq76pl536a

designed specifically for EV/Hybrid, $8.00 a pop so Tesla must think it's simpler than a lot of diff-amps and a faster MCU with multiple channels. (it probably is in the end.)

Tesla use -Q1 variant "zero defect".

From the link posted above, the SI8642ED is an isolation chip, used between the individual BMB and the main BMS board. Each side of the chip needs its own power source. It supports passing 2 channels of high rate serial data in each direction (4-ch total).

(edit: and hcsharp beat me to it)

 

[SirNick]

Ohhh... I missed that. Assumed it was another micro. Oops. Man... that's esoteric stuff. Wonder why one would choose that over optical? Current draw seems low, but not super low, at ~1-5mA per channel.

 

[spaceballs]

Just testing a bit while I wait around for my inverter...

The module side of the module BMS board's isolation IC's power input pin reads 5V, so that portion of the board is active in some fashion still running off of thr module itself.

Since I'm not sure what, if anything, this is doing in my configuration I'm going to finish disconnecting the remaining boards from my modules...

btw do the LED(s) blink one about a minute? the Roadsters BMBs do (even when not connect to the CAN bus).

 

[tom66]

Although the TI chips support a method of intermodule communication, Tesla does not use this interface. My guess is it's possibly because they need sufficient isolation between modules and that has none at all. Or maybe the noise immunity is not great, which is important in a safety critical system like this. It's a current mode interface.

It wouldn't be hard to figure out the interface on my opinion. I would expect the BMB and BMS to operate independently from the vehicle systems. So if you give it 12V and GND it will probably fire into life. You should then be able to see the BMS<->BMB communication on a scope. If it is RS485, many scopes will do serial decode on the data.

 

[henderrj]

I do love this thread – thank you wk057! I’ve read every word and searched other places and still cannot find an answer to a couple of questions. I hope you might find time to answer them for me.

You show a couple of pics of the high voltage connector (captions “Top view of the high voltage connector” and “Close up view of the external HV connector …”). Could you give me their size? I would love to know the size of the blades but I realize that information is not available to you.

The second issue is about the modules. First, I can guess the size but could you give us a more accurate measurement? Part B – Would it seem that the modules could be alternatively arranged (i.e. four each stacked in four rows)? Could they be separated (front and back of vehicle)? As I consider it the three connection types are low voltage (communication) wiring, high voltage (power) connections, and the coolant. Am I missing anything?

Again, thanks to you and all the contributors. Amazingly informative.

 

[Cottonwood]

From what I have learned on this thread, the 85 battery pack has 16 modules with each Module having 6x74 cells, 6 series/74 parallel. The 16 modules are connected in series to get a total of 96 cells in series, 96*74=7,104 cells/pack.

I am curious about the 60 pack.

Is it true that the 60 battery pack uses the same modules? How many modules are in series in a 60 pack?

Thanks ahead of time for the info.

 

[scaesare]

From what I have learned on this thread, the 85 battery pack has 16 modules with each Module having 6x74 cells, 6 series/74 parallel. The 16 modules are connected in series to get a total of 96 cells in series, 96*74=7,104 cells/pack.

I am curious about the 60 pack.

Is it true that the 60 battery pack uses the same modules? How many modules are in series in a 60 pack?

Thanks ahead of time for the info.

This very thread has some pics of the 60kWh pack and some counts of the cell totals...

 

[wk057]

I do love this thread – thank you wk057! I’ve read every word and searched other places and still cannot find an answer to a couple of questions. I hope you might find time to answer them for me.

You show a couple of pics of the high voltage connector (captions “Top view of the high voltage connector” and “Close up view of the external HV connector …”). Could you give me their size? I would love to know the size of the blades but I realize that information is not available to you.

The second issue is about the modules. First, I can guess the size but could you give us a more accurate measurement? Part B – Would it seem that the modules could be alternatively arranged (i.e. four each stacked in four rows)? Could they be separated (front and back of vehicle)? As I consider it the three connection types are low voltage (communication) wiring, high voltage (power) connections, and the coolant. Am I missing anything?

Again, thanks to you and all the contributors. Amazingly informative.

The female side of the HV connector has two blade slots that are almost exactly 1" wide. They go about 1.5" in, and the blade would have to be at least 1/8" thick, probably more. There is also a rod female slot that is for chassis ground that is about 1/4" diameter. There is a smaller rod connector on the HV connector that is wired with two connectors to the internal LV harness on the main PCB... presumably some kind of connection sense since if a solid rod were inserted it would close the connection.

There is a temperature sensor (wired into the internal LV harness) attached to one of the internal HV bus bars close to the HV connector.

The internal low voltage wiring harness is labeled "BIZLINK".

I don't have exact measurements on the modules, but they are approximately 28" long by 12" wide by around 3-4" tall... I'll have to measure one more accurately one of these days.

Given the dimensions of the housing, I don't see any usable alternative arrangements possible. The stock configuration would seem to make the most sense in any case. (I've rearranged the modules in a 44.4V configuration for my offgrid solar project, but that is outside of the original housing.)

There are three connectors, the HV, LV, and coolant connectors. Correct.

 

[henderrj]

Thank you - this is exactly what I needed.

 

[jumpjack]

Removed all screws from the top of the pack and starting to peel away the top cover. The thing is held on with so much adhesive/sealant that it took us nearly 40 minutes to get to this point.

This battery package is a maintainace nightmare: how is Tesla supposed to repair a defective battery, if it has to destroy the glued cover to access the cells?!?
There must be another method.
I also wonder if there is some redundancy in such an high complexity "part", made up of around 10.000 cells; if no redundancy is present, just one failure in 10.000 will make the whole 85 kWh battery unusable and un-repairable: a 10$ failure which leads to a 30.000$ unrecovrable failure?!?
It is just not possible.