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.

Changing from two DS90LV001 to crosspoint switch LVDS 455Mbps. Question on integration and parts.

Other Parts Discussed in Thread: DS90LV001, DS90CP22, DS10CP152

Hi,

 

We have a unit that is driving a 15 inch LVDS display at 1024 x 768  - 4 data pairs and 1 clock pair (24 bit LVDS).    Sometimes we want to drive two displays with the same data. The data pairs are at 455 Mbps and the clock pair is about 65 MHz.

 

Computer module -----  LVDS buffer ---- local display   (local display)

                                       |--- LVDS buffer  ---- long cable 25ft ----   LVDS buffer --- display  (remote or Aux display)

 

We are currently using the DS90LV001 for our LVDS buffers. I would like to change the two DS90LV001 on the local display board to a crosspoint switch such as the DS90CP22.

 

We are currently terminating the DS90LV001 at 100 ohm. Our cable has 13 twisted pairs in it with an overall shield.  The cable has a capacitance of 65 pf/m and an impedance of 90 +10/-15(balanced).  No CAT rating.

 

What I need help with is deciding on the terminating resistor(s)/correct chips.

 

In the DS90CP22 data sheet, most of the tests were done with a 75 ohm terminating resistor which would put us around a 350 mV output differential voltage (we are currently around 330 mV looking at the scope and datasheet for DS90LV001). If we used a 100 ohm terminating resistors on the DS90CP22, the output differential would be 400mV according to the graphs on the Datasheet. 

 

If I am going to use these two chips together(DS90CP22 on local board and DS90LV001 on AUX board), would I match up the terminating resistor, or would I try to match up the output differential voltage? Or can I not use these two chips together?

 

From what I have read, I think I should be using a 90 ohm terminating resistor to match our cable impedance; however, I was unclear since we have been using 100 ohm terminating resistors for years. I added 1k resistors in parallel to our 100 ohm terminating resistors but didn't see any noticeable difference.

 

I also looked into DS25CP152 and DS10CP152 but thought the data rate was far more than we needed.

 

Would a chip with PRE-EMPASIS/EQUALIZER be worth using since I am using 25 ft. cable?

 

Thanks for your time,

 

-Thomas

  • Hi Thomas,

    The DS90CP22 will make a good upgrade to a dual DS90LV001 approach.  I agree I think the DS10CP152 is likely more device than you need - either should work quite well.

    I do not think 90 or 100 ohm termination will make much difference.  In theory, 90 ohms should have a slight edge in performance since it matches the cable impedance.  

    A slightly higher VOD from the DS90CP22 will not impact the remote LVDS buffer.

    I do not know the exact attenuation in your cable, a CAT5 cable has ~ 3.6 dB loss for 10 meters at 250MHz.  Since you are only going about 8m at  ~225 MHz losses should be only 3 dB or so.  For differential signals 3 dB is not enough to disrupt data transmission.  The best check is to take a look at the remote LVDS buffer input waveform and see the quality of the eye opening.  

    Regards,

    Lee

  • Hi Lee,

    Thanks for the answers, they are very helpful!

    In regards to the eye diagram, I have looked at the diagram. The eye is shrunk more than I believe it should be( some noise does come across the eye). I don't have the exact percentage of jitter at this time, some data lines are worse than others.

    I only have raw specs on the particular cable, attenuation is not included.

    From your response, it sounds like using PRE-EMPHASIS/EQUALIZER will depend on my cable attenuation. I also looked at the eye diagram on the driver(main) output side and the quality of the diagram is also poor here. I believe this means I need to improve the quality of the drive signal before changing my cable.

    I believe our greatest problem right now lies in our board design - in particular with trace length(1446.5 mil difference in overall trace length between CLK+ and RX0- on the main board alone). After the new boards are designed with the DS90CP22, I am hoping the drive signal quality will be improved.

    Thanks for you time,

    -Thomas