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.

Cable Length with DS90C387A

Other Parts Discussed in Thread: DS90C387A, DS90C387

We have a design utilizing the DS90C387A to transmit parallel 24-bit video to an LCD panel in single-to-dual pixel conversion at an input clock rate of 148.8MHz for an effective LVDS clock rate of 74.4MHz. Pre-emphasis is set to 100%.

We are connecting the board using the DS90C387A to the display with a micro-coax 100Ohm cable. At cable lengths less than 9-inches, things work great and the video is crisp. Using a 4-foot cable, black bars appear which get progressively larger the further into the frame the video gets. Using a 9-foot cable produces a black screen with an occasional short lived flicker.

The data sheet for the DS90C387 states that for an LVDS clock rate of 65MHz, a cable length of 10 meters can be used and for a clock rate of 80MHz, a cable of 7 meters can be used with 30% pre-emphasis and a cable of 2 meters can be used with 25% pre-emphasis. Note 19 in the data sheet states that a shielded twisted pair cable was used. Can anybody provide any further information on this test setup? What the gauge of the cable is and how much error is acceptable in these tests? Also, there is no test data for setting the pre-emphasis higher than 50%. What is the consequence of a higher pre-emphasis? Is there a reason that I am unable to achieve success with a 4-foot cable?

  • Hi Bryce,

    Thanks for your post, and our apologies for the response delay. I will look into this and see what we can provide from our archives regarding this project. I will get back to you in the next day or two with a response.

    Regards,

    Michael

  • Hi Bryce,

    I was unable to find specific details about the cables used to generate the datasheet plots. However, since cable performance varies with length, manufacturer, and operating frequency, it is much easier and more systematic to step through the issue one component at a time. In most systems, a BER < E-12 is desired to conclude a system is working error-free.

    Pre-emphasis is a form of Tx equalization where the high frequency components are given an extra boost of energy by pulling more drain current at signal transitions. This is helpful when anticipating significant attenuation losses following the transmitter. Therefore, more pre-emphasis is associated with greater anticipated attenuation after the DS90C387A. However, more pre-emphasis corresponds to more overshoot from the original VOD, and this can potentially lead to increased power dissipation and EMI. Moreover, there is usually some EQ Rx post processing anticipated at the receiver. Therefore, though pre-emphasis is beneficial for aiding in Tx EQ, it is best to keep it at a minimum when possible.

    From your description, it seems like the cables you are using may incorporate a high amount of intrinsic jitter. Another possibility is that the LVDS clock has a noticeable amount of jitter.

    Can you send us a few things so we can get a better understanding of the system?

    - Scopeshots or eye diagrams from the DS90C387A transmitter LVDS output and clock signal with 0% pre-emphasis. We are interested to see the signal quality at the transmit launch before entering any media.

    - Do you have insertion loss curves or details that you can share for the various cables you are using? If the cables are particularly lossy, then we will be much more limited than what is seen on p. 15 of the datasheet. Other quantitative information such as cable pair-to-pair deskew would be helpful to know as well.

    - Any schematics or block diagrams for how you are setting up the device will be helpful as well.

    Thanks,

    Michael