• State TI Thinks Resolved
  • Locked Locked
  • Replies 3 replies
  • Answers 1 answer
  • Subscribers 33 subscribers
  • Views 1005 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.

SN65HVD72: Unit load test for faulty transceivers

Muhammad Naeem1
Prodigy 50 points
Part Number: SN65HVD72
Other Parts Discussed in Thread: THVD1520

Hi,

We were recently troubleshooting an RS485 Modbus RTU network which had approx 40 x SN65HVD72 Slave nodes, the problem started when the Master node (SN65HVD72) created a lot of network errors (i.e. could not make sense of the data on the network). We started disconnecting the slave nodes from the network one at a time and eventually found the culprit node which was bringing the whole network down. We replaced the faulty node with another node and everything worked fine thereafter. Unfortunately, we were not able to bring the faulty node back from the plant and it is still with the plant engineer.

Going forward, I was wondering if there is a unit load test that we can perform on the slave nodes at the production stage.

As potentially we can connect up to 200 SN65HVD72 transceivers, the idea is to check the integrity of every single SN65HVD72 transceiver by simulating198 slave nodes and one master node (SN65HVD72), the master node would communicate with the SN65HVD72 slave node under test and if the device communicates without any errors this would be considered a pass.

My questions are:

Is the above-discussed test a fair test?

What sort of setup do we need to simulate 198 additional nodes (SN65HVD72)?

Any help would be much appreciated, thanks in advance.

  • 0
    TI__Guru 62920 points
    Hi Muhammed,

    Yes, it is possible to model an inactive transceiver as just a resistance (whose value would be chosen to correlate with the bus input leakage current specifications for a given device). I can get you more information on this once I get back to the office later this week.

    Do you have a chance to get any more information on the failing slave? Understanding that failure better would help in developing a screening procedure. (For example, if there were some other issue that caused the slave to switch into transmit mode, then simply testing it with what looks like a fully loaded network may not successfully recreate the failure.)

    Max
  • 0 in reply to Miguel Robertson
    Prodigy 50 points
    Hi Max,

    Thanks for your reply. I will request the return of faulty node, but it may take a while.

    Meanwhile, I will look forward to reveiving further informtaion from you.
    Best Regards
  • 0 in reply to Muhammad Naeem1
    TI__Guru 62920 points

    Hi Muhammad,

    Sorry for the delay.

    To expand on what I mentioned previously, you could use the bus input current specifications of a device to determine what kind of load (in terms of resistance) it presents on a network.  In the case of SN65HVD72, you can see that its maximum leakage current is 150 uA for an input voltage of 12 V and 100 uA for an input voltage of -7 V.  Looking at the ratio V/I, this would give resistance values of 80 kOhm and 70 kOhm, respectively.  For a worst-case analysis, then, I would use 70 kOhms.  If you had 40 such nodes, then the total loading on each RS-485 line to ground would then be equivalent to 70 kOhms / 40 = 1750 Ohms.

    In general this does not seem like an extreme degree of loading.  In fact, the RS-485 standard allows for up to 32 "unit loads," each one roughly equivalent to 12 kOhms.  You can read more about this concept here:

    RS-485 basics: how to calculate unit loads and the maximum number of nodes on your network
    Other Parts Discussed in Post: THVD1520 As my colleague Bart Stiller mentioned in the first installment of this series, RS-485 is a multipoint differential bus, shown in Figure 1. This means that all of…
    By in Technical articles > Analog

    The HVD72 device was designed to support this "fully loaded" (32 unit loads) configuration, which is why the driver circuit is characterized with a test circuit with additional loading resistors present (see Figure 10 in the datasheet).  As long as you don't go beyond this loading I would not expect any issues.  My guess, then, is that there was a more extreme issue with the node that was bringing down the network.

    FYI, another thing to consider when implementing high-node-count networks (outside the leakage currents) is the cabling used.  If there are long, unterminated branches then the resultant reflections that occur at the end of the line may degrade signal integrity for the other nodes.  This is generally an effect that is not so pronounced at lower data rates, but it of course depends on the cabling involved and distances between nodes, termination locations, etc.  You should also take care that the total bus capacitance is not so high that the signals become heavily filtered and that only two nodes at most contain termination resistances.

    If there were any RS-485 signal waveforms captured when the issue occurred it would be good to review those - they might help give some indication of what is potentially going wrong.

    Regards,
    Max