• State Resolved
  • Locked Locked
  • Replies 6 replies
  • Subscribers 33 subscribers
  • Views 314 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.

SN75HVD10: Hot switching failure

Michael Cleary
Prodigy 140 points
Part Number: SN75HVD10

RS485 Operation

Part Used:SN75HVD10
Date: 2021-0-27

Issue:
We are using the SN5HVD10 which is a 3.3V RS485 Transceiver for our RS485 communication. The schematic design is very simple as I show in one of the picutres. There
is a 10K pullup on pin 1 (R). The device is powered from a switching regulator at 3.3V. I have a jumper in place to add the 100 ohm termination. The DE and RE lines
are directly connected to my microprocessor for data direction control. The RS485 A and B lines are duplicated on two RJ45 connectors that also contain power and
ground for the board. We want the ability to hot switch the boards. According to the data sheet we should be able to hot switch these transceivers.

In my test setup I have two boards communicating with eachother. This is a half-duplex configuration.

If I start with two new RS485 chips on board1 and board2. I checked the signal on board1 for A and B without board2 connected. Looking at A and B the signals generated
by board1 look good. The waveforms are about 3Vpp and they are what I would expect for a differential signal. This is shown in one of the pictures. I cycled power to
board1 numerous times and signal was fine.

I then hot connected the RJ45 cable to board2 and the signal on A and B looked the same as mentioned above.

In one case I hot switched the board about 10 times and the signals matched the description above.

On the 11th time (this is not significant - it changes) the signal on A looked the same but the signal on B is about 1/4 the amplitude of the signal on A. The signal
is still differential and surprisingly enough the Receiver Input worked. The microprocessor on the receiving side actually worked under these circumstances. I guess
the transceiver is still meeting the 200mV threshold necessary for the communication to work but it's obviously not correct.

I have also seen a case where I experienced the same conditions described above but eventaully B failed where B appeared at 1/4 of the amplitude of A but
it was in phase with A which means the differential signal was not working and the communication failed.

Most of the time I saw the RS485 failure on board1 after connecting board2 but I have seen board2 fail also.

I've replaced numerous chips trying to troubleshoot this issue.

We don't have ESD protection on the circuit. We are not exposing this signal to the outside world. We are within an enclosure but we wanted to be able to hot switch
the boards within the enclosure.

Board1 and board2 are different but the RS485 circuit is exactly the same.

I'm sending a picture of the RS485 circuit and the connector. I can't send the entire schematic. We are using an OKI-78SR-3.3/1.5 switching regulator for 3.3V on the
board. The +24V on the board which is coming from the RJ45 connectors is the input source for the OKI... part.

We are using a CAT5 or CAT6 cable between the boards.

If I don't hotswitch the boards and I perform a controlled power up I really can't recall seeing a problem. Once the system is operational we have never seen a
problem. It has something to do with the hotswitching. It's interesting to me that board1 seems to go bad after I power up board2.

I looked at the power up and power down sequence and monitored A and B during those transitions. I see B having a pulse going down to -6V for about 800ms on power up
but that seems to happen on every power up and the communication was still working.

I really don't think it's a static issue since it's not cold and it's fairly humid in my testing environment although I can't see what could be causing this problem.

Below are some scope pictures. The first scope picture show A (yellow) and B (blue) as an expected RS485 differential signal. After hot switching one of the boards a number of times I'm seeing the B at 1/4 the amplitude of A as is shown in the second and third pictures. The other two pictures are the sections of the schematic for the design of the circuit involving the SN75HVD10.

Thanks

Thanks

  • 0
    TI__Mastermind 25705 points

    Hi,

    When you hot plugged in board2, I understand not only A/B are connected but also 24V supply and GND, correct? How fast is SN75HVD10 on board2 powered up after connection? Do you observe any spike or noise on 24V rail or A/B bus? Given the abs max of A/B pins is -9V and 14V, I'm not sure if the bus voltage could exceed the limit momentarily.  

  • 0 in reply to Hao L
    Prodigy 140 points

    Hi,

    Yes 24V and GND are also on the RJ45 along with A and B for the RS485. I have pictures of the power up showing +24V and the 3.3V. I was concerned about this issue also but I did not see a spike or a level below -9V. I saw a spike of about -6V on B and a level of -4V on A. The scope captures on power up were fairly consistent with the pictures. When I observed these conditions on power up the communication was working and I really could not correlate what I saw with the power up process with the abnormal behavior of the transceiver.

    I'm attaching two pictures. In both pictures I'm capturing A (yellow) and B (blue) when I plug in the RJ45 to power up the board. In the first picture the pink is the 3.3V. In the second picture the pink is the +24V. The pattern is somewhat consistent. One difference is the -6V level on the A line. Sometimes I would see that and other times I would not. I really can't explain why that happened sometimes and not other times. I don't think it had any connection to the transceiver failing. The failure mode seems to be on B, not A but sometimes I would see the -6V level on A. Not sure if that is a problem. I still believe all the levels I'm seeing are well within the -9V to +14V acceptable range.

  • 0 in reply to Michael Cleary
    TI__Mastermind 25705 points

    Before plug-in, bus pins of board2 were floating. Do you think it's possible that some potential had existed here? What you captured is the pin voltage wrt to the ground of board1. I was wondering how it looks like to the board2 ground. Do the units on board1 only behave weirdly or the units on both boards do?   

  • 0
    Prodigy 140 points

    Yes the pins on Board2 are floating. Board2 is completely unpowered when I remove the RJ45 cable. The scope capture of the startup was actually recorded at Board2, not Board1. I have a few short jumpers soldered to the RJ45 connector on Board2 to capture the A, B, +24V, and GND.

    One point you mentioned is whether there was some potential on Board2. When I did my earlier testing I was mating and de-mating the RJ45 connector every couple of seconds. I was doing in fairly quickly when it failed. I was in the process of replacing the SN75HVD10 chip on Board1 and I was going to mate and de-mate Board2 at a much slower rate to see if any residual potential had any affect when I received your update on the E2E forum. Maybe if there was some residual charge on Board2 when Board1 is mated to Board2, it could create a problem. I'm going to monitor the A, B, 24V, and GND and verify there is no residual charge in the circuit before mating Board2 to Board1.

    It seems like the chip on Board1 seems to go bad most, if not all the time. I've been replacing the chip on Board1 all the time lately. I have replaced the chip on Board2 but initially I wasn't sure where the failure was so I replacing chips trying to figure out what was happening. Over the last few days I've only replaced the chip on Board1.

    I had stated earlier that I only saw the 1/4 amplitude signal on B. Yesterday I had an incident where I had the 1/4 amplitude on A. Now I've seen the issue on both A and B. 

  • 0 in reply to Michael Cleary
    Prodigy 140 points

    Hi,

    I did some more testing and I got the same issue on A which was Channel 1 on my scope. I wanted to keep the setup the same to have consistency during my testing but I decided to change it up. I swapped the scope probes and the reduced signal stayed on Channel A. Then I kept swapping the probes and the reduced signal stayed on Channel A. That made no sense at all. I wiggled the connected on Channel A and the scope V/div changed from 2V to 20V. I changed the setting to 2V and now my signals looked the same. The scope was doing a divide by 10 on it's own it was not reporting it on the screen. I removed the probes and re-attached them. I ran 30 hot switches monitoring the startup of Board2 through the initialization and running of Board2 and it worked every time. Even when the voltage appeared to be low on the scope the circuit was working to my surprise but it was the scope that was showing deceptive results. 

    I really want to apologize for wasting any of your time. I really had no idea why I was seeing these results. It really helped to switch the probes which pointed out something was not right. The scope is only a few years old and it is a good one but in the future I have to make sure the results on the scope are what I expect to see.

    I'm going to continue testing but I'm hoping that I will have much better results moving forward.

    It seems like the post was a ghost hunt. I don't know if it should be removed. I would hate to lead someone else down an incorrect path.

    Thanks

  • 0 in reply to Michael Cleary
    Prodigy 140 points

    The issue has been resolved.