• State Resolved
  • Replies 2 replies
  • Subscribers 33 subscribers
  • Views 65 views
  • Users 0 members are here
Support feedback
Options
Options
Related

Original question:

[FAQ] CAN Transceiver Resources

TCAN1145-Q1: Problem with CAN transceiver TCAN1145-Q1

Jan Thorstensson
Prodigy 10 points

Part Number: TCAN1145-Q1
Other Parts Discussed in Thread: ESD2CAN24

We have experienced a couple of TCAN1145-Q1 failing when exposed to a wiring short circuit between CANL and GND.
After failing, the DC-resistance between CANL and GND was found to be approx. 5ohm when non-powered. Similar on both IC:s, obviously EOS damage.

The TCAN1145-Q1 is used combined with a Murata DLW32SH101XF2 common-mode choke in this application.

Both simulations and bench tests indicate that the CM choke causes negative transients at the transceiver CANL port at the trailing edge of every CAN bit sent.
Those transients are being clamped to approx. -60V by internal transceiver TVS

However, the energy absorbed by the TVS during these pulses is very small, less than 700nJ per pulse.
Even when using extremely unfavourable conditions (such as 25V GND voltage offset), energy never exceeds 1.5uJ.

The datasheet for TCAN1145 indirectly suggest there are internal TVS diodes on CANL and CANH clamping at approx. ±60V. This has also been confirmed in bench tests.
The 8kV CD ESD-spec implies that these TVS diodes should be able to handle at least 83uJ
ISO 7637-2 pulse 1 appears to subject internal TVS diodes to approx. 5mJ, pulse 2 approx. 1mJ



TI Application Report “SLLA271 - January 2008”
This report suggests that a CM choke may cause dangerous transient voltages whenever the CAN bus is shorted to VBAT or GND
The recommended solution is to add external TVS diodes between CM-choke and transceiver.

Question 1: Is such external TVS diodes really necessary for TCAN1145-Q1, which apparently has internal TVS diodes able to clamp pulses with 1000x higher energy than typically caused by the CM choke. I.e. is it highly recommended to design according to the Application Report SLLA271 - January 2008?
Question 2: Is the transceivers ability to survive ± 58V fully tested in production? Could there be individuals that may fail before the internal clamping goes into action?
Question 3: Is the internal clamping function fully tested in production? Or could the clamping simply be missing on certain individuals?

Question 4: Is the indirect estimation of TVS energy limits, based on ESD and ISO7637-ratings,  reasonable?

slla271.pdf

  • 0
    TI__Mastermind 27302 points

    Hi Jan,

    Thanks for reaching out on E2E. 

    So you are saying that when CANL is shorted to GND, TCAN1145 is permanently damaged? Did you check to see if CANH is shorted to GND well?

    Yes, shorting any supply at or above 12V to CANH/L can be a problem with a CMC as that app note shows. However, shorting CANL to GND with a CMC should not be a problem at all.

    Jan Thorstensson said:
    Both simulations and bench tests indicate that the CM choke causes negative transients at the transceiver CANL port at the trailing edge of every CAN bit sent.

    During those tests, was CANL shorted to GND? Or was CANH and CANL in normal operating conditions? Dose every trailing edge show this?

    Q1. Yes, I would still recommend external ESD diodes (especially in between the CMC and the transceiver if you suspect that there is a potential issue). The main reason is the transient created by the CMC may not match the IEC 61000-4-2 or other ESD standards. The real-world ESD strikes can also be much higher than 8kV, and the internal diodes don't protect against any other forms of longer ESD strikes (like IEC 61000-4-5 surge). Our recommended diode is the ESD2CAN24, which supports up to 30kV IEC 61000-4-2 and ~5A of surge. This diode clamps at ~24V which will also help prevent voltage spikes from damaging the CAN transceiver pins (i.e. going above 58V). If those internal diodes clamp any higher than 60V (like in the case of a large surge event), they could potentially cause that EOS. 

    Q2. Yes, this is tested. However, this is not tested with a CMC. The issue of clamping does not apply below 58V. This is because the transceiver can sustain 58V indefinitely. This would not be a transient. Hence, why the internal diodes clamp just above 58V. 

    Q3. Yes, the clamping is always present since it is part of the TCAN1145 die. It is not a separate piece of silicon. 

    Q4. Your calculations are accurate, but they do not really apply as the issue you are experiencing likely does not line up with the exact ESD standard waveform. So the voltage, amperage, and time will be different. 

    Best,

    Ethan

  • 0 in reply to Ethan Sempsrott
    Prodigy 10 points

    Thank you Ethan for the good answers.

    Ethan Sempsrott said:
    During those tests, was CANL shorted to GND? Or was CANH and CANL in normal operating conditions? Dose every trailing edge show this?

    Yes, CANL was shorted to GND during those tests. Yes every trailing edge.