• State TI Thinks Resolved
  • Locked Locked
  • Replies 13 replies
  • Answers 4 answers
  • Subscribers 36 subscribers
  • Views 1063 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.

LVTTL 50 ohms Driver

Louis Thuilliez
Prodigy 100 points

Other Parts Discussed in Thread: DSLVDS1047, DSLVDS1048, TLV3602, LMH6703, OPA693, SN74LXC8T245, SN74AVC2T45

Hello,

I would like to use the DSLVDS1048 and DSLVDS1047 components but with a 50 ohm adaptation on the LVTTL lines as follows:

I'd like to be able to use the same buffer / level shiffter on input and output to keep the same propagation time on the lines.
Do you have an idea of a component that could match?

thanks in advance,

Best Regards,

Louis

  • 0
    Prodigy 100 points

    I would like to add the following information:

    - The maximum operating speed is 400MBps.
    - Drivers must be able to supply the current required to load 100ohms (50 + 50) at 3V3, -> 33mA minimum.

  • 0
    Guru 320870 points

    33 mA would be possible by paralleling multiple buffers, but there is no logic device that is guaranteed to handle 400 Mbps at 3.3 V. (LVC could do it at 5 V, or AUC at 2.5 V.) The TLV3602 comparator would be fast enough, but there are not enough parallel outputs in the same package.

    The TLV3602 would be fast enough for the level shifting, but the same could be done by using an LVDS buffer (whose inputs work like a comparator).

  • 0 in reply to Clemens Ladisch
    Prodigy 100 points

    Hi,

    Thank you for your reply. I don't really want to parralelize multiple buffers.

    I'm currently looking for an AOP that can supply at least 33mA at 400Mbps (e.g. OPA 659). But I doubt it's the best and most effective solution.

  • 0 in reply to Louis Thuilliez
    TI__Guru 74302 points

    Louis,

    Thanks for bringing this to E2E, our LVDS experts will get back to you and have a response end of day 01/17/2024.

    Clemens,

    Thanks for your help in the meantime.

    Regards,

    Eric Hackett 

  • 0 in reply to Louis Thuilliez
    TI__Guru* 78531 points

    Hi Louis,

    The TTL outputs on our LVDS devices weren't designed specifically for what you are asking. Generally the outputs (ROUT) are tied directly to a receiver pin. Looking at the datasheet for DSLVDS1048 for example, it looks like this device was mainly tested using a 15pF load without any kind of resistive loading.

    Adding a resistive load on the pin like what you are asking would likely force the IC designer to make much larger FETs to support the current load. I suspect if you tried to tie ROUT to a 100 ohm load to GND, when ROUT swings HIGH it would likely saturate the FET (this is my guess on what would happen). If the FET did saturate, then would have some kind of collapse of VoH and your node between the 50 ohm up and 50 ohm down would likely get distorted. (Imagine VoH gets stuck at 2V for example, then the receiver between the 50 ohm resistors would see around 1V as a logic high).

    If the layout designers for the chip did not anticipate ~33mA current could be sourced from this pin then the device could likely weld/melt somewhere in the layout path between Vcc to the ROUT pin. 

    Since it wasn't designed for this kind of loading, you may end up seeing failures in time increase as well. 

    Having a parallel connection like Clemen's pointed out would likely help since the current draw from the device would be divided into two instead of one, though you might need more than two if what I expect may happen does.... I think you would be better off trying to connect a signal redriver/buffer infront of the ROUT pin inorder to make that device handle the resistive loading instead of forcing it on the LVDS TTL output likely wouldn't be able to handle it. 

    I know this isn't the answer you wanted but I think you will likely need to put something infront of the ROUT output to make a 100 ohm resistive load set up work. 

    I've reassigned the thread to a different team in TI who may be able to suggest a potential buffer that could fit. 

    -Bobby

  • 0 in reply to BOBBY
    Prodigy 100 points

    Hi Bobby,

    Thanks a lot for your reply.

    I agree with adding a buffer/driver at the ROUT output to feed the matched 50 ohms.

    We're currently looking for the component that might match this.

  • 0 in reply to Louis Thuilliez
    TI__Mastermind 43795 points

    Hi Louis,

    Let me discuss this with our team and we will see what part we can recommend you.

  • 0 in reply to Louis Thuilliez
    TI__Mastermind 43795 points

    Hi Louis,

    Unfortunately we do not have a device that will be fast enough for this use case. 

  • 0 in reply to Albert Xu1
    Prodigy 100 points

    Hi,

    Are there any buffer-style components that can be paralleled ?

  • 0 in reply to Louis Thuilliez
    Guru 320870 points

    A single OPA693 or LMH6703 might work (if you have wider supplies).

  • 0 in reply to Clemens Ladisch
    Prodigy 100 points

    Ok, thanks for your feedback.

    In simulation, the OPA693 doesn't seem to adequately handle the required 200 MHz bandwidth. We have not yet tested the LMH6703.

    Would a parallel level shifter setup like the SN74LXC8T245 or SN74AVC2T45 be suitable?

  • 0 in reply to Louis Thuilliez
    Guru 320870 points

    AVC level shifters indeed might be fast enough.