TCAN337: TCAN337 IC Failure and burned out

Hi Jit,

Can you send over a schematic of the CAN transceiver that burned out. It is possible there was a fault event that caused an excessive amount of current to run through the device and that could cause the device to internally burn. However, this device has a bus fault protection of up to +/- 14V. Is it possible that a high voltage line was running close to CANH and CANL bus? Could a high voltage line been accidentally connected to the CAN bus?

Once I see the schematic we can make a better judgement on what might have happened.

Best,

Chris

Dear Chris,

Thanks for your reply. I am attaching the CAN section of our CAN Boards.

There are several such boards and each board serves as one CAN node.

Each card has two CAN connectors (CN4 & CN5) out of which one is for input CAN connector and second is for output CAN connection.

Each connector for CAN has got 6 wires - One for 12V supply, One for Ground, One of CANH, One for CANL, One for CANH-LOOPBACK and one for CANL-LOOPBACK. The CAN connection running into this board enters through CANH and CANL. Then it is looped back to the next CAN board using a CANH-LOOPBACK and CANL-LOOPBACK. This way we can daisy-chain all the nodes in our CAN network. The last board does not use the LOOPBACK connection but instead uses a 120Ohms termination resistor. The resistors R8 and R11 are not mounted on CAN boards except for the first controller board which has a microcontroller controlling the entire system operations. There is a shield above these 6 wires, but the shield is left unterminated on both ends. Also this bunch of 6 wires run close to 48V live and return wires running from battery to the motors. The connectors CN1 and CN2 are mounted on boards and are for debugging purposes only and do not have any harness in actual operation. 

Kindly let me know if you need further details. Thank you for your support and valuable guidance.

Hi Jit,

I'll look through this information and get back to you tomorrow.

Best,

Chris

Hi Jit,

Have you double checked to the wiring and connectors to make sure that they are connected correctly. You might be shorting the 12 V rail to GND on the device that burned.

I would start off by checking all your psychical connections. Also can you send me a picture of the device that got burned? If possible are you able to tell which pin got damaged?

My only question about the schematic is the VIO pin. The TCAN337 does not have a VIO pin. Pin 5 is a FAULT pin. Can you confirm which device you are using?

Best,

Chris

Hi Chris,

Thank you for your reply. Yes we have double checked that there is no shorting between 12V rail to GND as our wiring is fixed and our connectors can't be mated in opposite direction. 

We have measured the impedances of all pins of the faulty TCAN337. Here is a summary of our impedance measurements.

Impedance                --------System-1----------------             ---------System-2------------      Good Device

                                  Burnt IC-1          Failed IC-1             Burnt IC-2      Failed IC-2

CANL to GND           2.3M                   83K                        28K                70K                  2.3M

CANH to GND          2.3M                   27K                        42K                1.3K                 2.3M

CANH to CANL         56K                    46K                        70K                 42K                 56K

VCC to GND             32 Ohms            1.9K                       25 Ohms         32K                 233K

RX to GND               4.4 Ohms            OPEN                    80 Ohms         OPEN             OPEN

The VIO pin is actually a pull up for the Fault pin and it is just named as VIO and not used. 

I am sharing the photos of burnt IC for your reference.

Regards,

Jit

Jit,

Are these two pins the VCC and GND pins?

Did you also take a measurement of TX to ground? I'm sorry if you said this before but what is the difference between the burnt IC and the failed IC? Are they on the same board and located near each other? It looks like the VCC rail suffered a strong short to GND. Are all your CAN transceivers on the same board or do they have their own induvial boards that connect to each other.

I think the next step here would be to double check the design files. If you are comfortable sharing the project with me let me know.

Best,

Chris

Dear Chris,

Thank you for your prompt reply. 

The two pins are not VCC and GND but are CANH and CANL pins.

I have took the measurement of TX to gnd but all are showing open for good and bad ICs and hence did not mention it here.

We had failures in two separate systems. In both failures, two CAN boards have failed. In each system, one CAN board has got a severely damaged (burnt) TCAN337 IC and a second TCAN337 IC which is not burnt or damaged externally but is not working. 

TCAN337 ICs are located on different boards and boards are connected through the CAN wiring connectors (CN4 and CN5 shown above).

In summary we have a system of two CAN buses CAN1 and CAN2. CAN-1 has 5 separate boards each having one TCAN337 IC. CAN-2 has 4 separate boards each having one TCAN337 IC. We have observed that two TCAN337 ICs in two boards of CAN-2 have failed, where one is literally burnt out and second is visually OK but not functional. We have had two such failure observations in two systems (Totally 4 TCAN ICs have failed (2 burnt and 2 faulty)).

Hope that this clarifies the system.

Regarding sharing of project I will have to check with my system admins. Meanwhile do let me know if you can have any clues regarding this failures.

Thanks and regards,

Jit 

Jit,

Do all the boards share the same PCB design? If you populate a different IC onto the board that burned the last one does the same board burn the new IC?

I am going to check with design to see if I can find out what part of die is located under that melted mark. Let me know if you can share the PCB design.

Best,

Chris

Dear Chris,

Sorry for the late reply.

All boards share the same PCB design. Also I checked if there is any component on bottom side of transreceiver but there is nothing exactly below CANH & CANL lines. There is a partial area cover by the connector on the top side which has a 12V net (NETCN2_4). (See figure)

Yes we replaced the CAN transreceiver and TVS diode in cards with Burnt IC as well as Faulty IC. We found that the cards which had faulty ICs are now up and working fine. But the cards which had burnt IC have now normal currents, but the CAN node still can't be detected after IC replacement. We probed the CANH/CANL lines to see the behaviour and found that in absence of any CAN node the CANH and CANL waveforms of the Card (having the burnt IC replaced with a new one) are OK (complimentary data patterns visible on scope), but the moment we connect it to another good CAN node, the CANH/CANL waveforms are no longer normal (infact we see same polarity swings in CANH and CANL). Still trying to understand this behaviour.

Regarding full PCB, we can share to you directly in case you are willing for an NDA. 

Thanks for your support and do let me know if you can figure out anything.

Jit,

In both cases, it seems like electrical overstress is occurring. The burnt out IC is obvious because it catastrophic damage caused by heat aka excessive power. The burnt ICs show a low impedance short on VCC to GND, which would indicate damage along that path, and the failed ICs show a low impedance short on CANH and CANL, which is typically of overvoltage damage on the CAN bus. Also the fact that the TVS diode is burnt out is just more evidence that electrical overstress has occurred.

Can any waveforms be captured of the system? I know it's difficult to capture random damage occurring, but all the information we can get will be helpful.

Regards,

Eric Hackett