• State
  • Locked Locked
  • Replies 3 replies
  • Subscribers 35 subscribers
  • Views 306 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

SN65HVD234: Field Failures

Mike Page
Prodigy 10 points
Part Number: SN65HVD234

Application + Implementation

We have designed a product that has two units - a base station and a hand held probe, which communicate with each other using a CAN interface. Both units have SN65HVD234D (or 235) CAN transceivers and T24CAN TVS protection. They are physically connected via a 4 wire curly cable (0V, 12V, CANH, CANL) which is user-detachable. There are no other devices on the CAN Bus.

Customers are now returning the product due to failure of the SN65HVD234D in the hand held probe. We are unable to replicate this fault without implausible abuse and need help to rectify the problem. About 4% of shipped product is affected.

The product is a high voltage breakdown detector, with the hand held probe generating up to 40kV.

Field Returns

  • T24CAN TVS parts are intact and functional.
  • There are no signs of PCB assembly faults.
  • In most cases the IC partially works but the R pin is stuck high.
  • In one case the MCU pin connected to transceiver D pin was damaged.
  • Half a dozen transceivers have craters, most have signs of heating in conformal coating.
  • We have had issues with the 4 way connecting cable failing partially open but it does not appear correlated with transceiver failure.
  • We have had failures on transceivers from two different suppliers.
  • Two MCU types (Renesas and NXP) are in use, both configurations have had transceiver failures.
  • The failures are clustered; some customers have experienced multiple failures.

Tests

All testing has been on the hand-held probe where the failures occur. The transceiver is fed from its own SMPS (== 3V3CAN) while the MCU is fed from a separate low noise LDO. We have tested for the possibility of power sequencing latchup but cannot induce it nor does it fit the evidence or datasheet Fig 28.

Taking 3V3CAN above ~7V causes the transceiver to malfunction but it can recover. Taking 3V3CAN above ~10V causes failure. We are now sending units out with a 3W 4V7 zener (6.1V @ 1.5A) across 3V3CAN and they still fail.

We have tried raising the 12V unit supply to 30V but both regulators take it without complaint. Although it is possible to damage the transceiver with excess voltage on R/D, we do not think this is occurring.

We have tried shorting 12V (and up to 24V) to CANH and CANL, even with D pin stuck low (forced dominant). The thermal protection functions as expected.

We have tried applying transients of 32V PSU and up to 40kV from the HV end of the probe (~50pF, 2uC, 40mJ) without any problems.

We have tried operating the unit with faulty interconnecting cable and without equipment safety ground.

 
Any help would be gratefully received,
Mike Page.

  • 0
    TI__Mastermind 26826 points

    Hi Mike,

    It looks like you and your team have done extensive initial investigation into this issue - thanks for including all of the detail.

    Mike Page said:
    • In most cases the IC partially works but the R pin is stuck high.
    • In one case the MCU pin connected to transceiver D pin was damaged.

    From your description, it sounds like the only misbehaving portion of the component is the logic side of the device, on D/R?

    It also sounds like we've roughly crossed-off the other signal lines as likely contenders. CANH/CANL seem undamaged and the TVS remains operational, and the supply line remains damaged with Zener protection. (Assuming that these Zener diodes also return unbroken for failed units?)

    Is there any observed coupling or injected noise onto the logic lines that can happen during operation? I wouldn't expect the SMPS to be responsible for something like that, but maybe from elsewhere in the probe?

    Best,

    Danny

  • 0 in reply to Danny Bacic
    Prodigy 10 points

    Hi Danny,

    Thanks for your reply. We've reviewed the possibility of noise / transient coupling electrically and magnetically and don't think it's likely. We have the same issues on two different PCB layouts. In both cases the high voltage end is 2-3 inches away and is followed by a passive multiplier even further away.

    Of four damaged ICs measured, we have seen recessive levels on CANH and CANL of 2.7V on one IC and 1.8V on another. The remaining two were 2.3V. This is not a parameter we factory test (nor does the datasheet make any promises!)

    Our inability to repeat the fault (several months trying) is frustrating and we have decided to take a "Kitchen Sink" approach of reinforcing protection of the D,R,CANH and CANL pins as much as practical. Unfortunately we will not know for some further months if this has worked. But for the benefit of all forum users, I will update when I have news.

    Many thanks,
    Mike.

  • 0 in reply to Mike Page
    TI__Mastermind 26826 points

    Mike,

    Just wanting to follow up here (and apologies for the delayed follow-up as I've been traveling for work).

    Any chance there is an inductive element such as a CMC connected as part of the CAN physical layer? There is historical evidence that CMCs can cause transients and overshoots, but this would impact the CANH/CANL lines rather than the digital D/R signals. Only case I could see something impacting the MCU is if the 4-wire connection (hot-plug?) is causing an inductive kick or GND shift that risks damage to the components. This would be both a clustered type of event (which explains the return pattern) and would be difficult to emulate exactly in a lab environment (which could explain your team's difficulty reproducing the damage).

    Best,

    Danny