• State TI Thinks Resolved
  • Locked Locked
  • Replies 4 replies
  • Subscribers 33 subscribers
  • Views 1535 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.

Unpowered LVDS receiver dies

john pease
Prodigy 40 points
Other Parts Discussed in Thread: SN65LVDS3486

Our system uses LVDS to communicate between equipment racks.

The receivers in a turned off rack will blow if the driving rack is still powered.

We are considering a couple of solutions:

1) Use diodes to clamp the receiver inputs to Vcc.  This might damage the drivers on the powered up rack because they will drive a heavy load through the forward biased diodes.  Most driver data sheets show maximum

2) Use clamp diodes as above and series resistors between the termination resistor and the receiver input. This would limit the current the driver would source through the diodes into the turned off reciever rack. But now performance is degraded by the series resistor/receiver capacitance time constant.

 How do we determine the the common mode and normal mode input capacitance of the receiver? 

 Some TI data sheets specify input capacitance to ground but don't specify differential capacitance. Some data sheets like the SN65LVDS3486 do not specify any input capacitance.

How do we calculate the series resistance so that it is high enough to protect the unpowered reciever while connected to the powered driver?

With this resistance determined, we can calculate the frequency response from the input capacitance if it is specified.

Are there better methods to protect LVDS drivers and recievers in the case I described?

 

Thank you

 

  • 0
    TI__Intellectual 1360 points

    John

     

    I would like to understand the failure mechanism a little better, i am interested in trying this in lab as well. You are indicating that if a powered LVDS transmitter is sending data to an unpowered LVDS Receiver, the LVDS receiver is getting damaged?

     

    I will call you and we can try to understand the failure mechanism a little better

  • 0 in reply to Ajinder Singh
    TI__Intellectual 1360 points

    Couple of comments:

     

    Can the customer update the power sequence: that is keep the RX card turned on as well.

     

    second is it possible that customer can isolate the LVDS RX with a relay in the path, the control of the relay can be the VCC on the RX card, so this way if there is no VCC the relay is off and there is no voltage applied to the RX and no possible latchup happening.

    If there is VCC on the RX card, the relay is turned on and the normal operation can take place.

  • 0 in reply to Ajinder Singh
    TI__Intellectual 1360 points

    John

     

    Could you comment if you are satisfied with the response and is the proposed solution of using a zener diode in forward biased when the RX board is off and and reverse when the RX board is on, is it working?

  • 0 in reply to Ajinder Singh
    Prodigy 40 points

    Ajinder,

    Thank you for the advice and consul.

    You suggested that the TI  75lvds390 drive output current should be no more than 15-20 mA steady state during fault conditons. You also thought that the 75lvds391 receiver should not latch up if its  inputs are kept below about Vcc-0.9V when Vcc is turned off

    So this data indicates that a diode clamp may be a suitable solution and perhaps a resistor in series with the diode clamp might be needed to limit driver output current. 

    Consider the Diodes Inc MMBD4448HCQW quad common cathode diode package used to clamp the reveiver inputs to Vcc

    .MODEL DI_MMBD4448HAQW D  ( IS=1.12n RS=84.0m BV=80.0 IBV=100n
    + CJO=2.92p  M=0.333 N=1.70 TT=2.16n )

    The '391 driver sources a maximum of 24 mA into a short circuit and its nominal maximum nominal DC output voltage is 1.37V so the equivalent output impedance is about 57 Ohms.

    When I model a 1.375V voltage driving a grounded '4448 diode through a 57 Ohm resistor, I get a diode voltage of about 0.7V and a diode current of 11 mA. 

     The  grounded clamp diode is a worst case simulation of  turned off Vcc.  Turning off Vcc does not short it to ground.

    Granted there are many simplifications in this analysis, but it indicates that  just the diode clamp alone may be sufficient.

    The '390 driver family has output enable pins. A more robust solution would tie the recevier frame Vcc  to driver output enable. However, my application will not allow that.

    Please let me know if I am on the right track.

     

    John Pease