How to use AD9557 for Synchronous Ethernet?

Hello: we plan to recover the clock from the Synchronous Ethernet (SE) using MAX24288, whose output clock jitter is quite bad (a nanosecond!). So we need to clean that clock. Our jitter requirement is pretty modest. A few tens of ps will be perfect. We do not need to go to single picoseconds.

Our candidate clock cleaner could be AD9557, whose data sheet mentions SE as one of the applications on page 1. Unfortunately, besides mentioning SE, there is no other reference to SE in the AD9557 Data Sheet. (A very untypical situation for Analog Devices, whose Data Sheets are usually very good.) The AD9557 press release says that the part can be pin-strapped for SE, but there is no further clue. Furthermore, when looking at the Eval Board, there is BOM and the schematics, but there seems to be no software to drive it, even though the eval board has a processor onboard.

Questions:

1. Are there any App Notes which would illustrate how to use the part for SE clock cleaning?

2. What is the jitter performance of the cleaned clock?

3. How to configure the part? How to pin strap the part for the SE?

4. Is it the best / simplest part? Are there better / simpler clock cleaners? (Ideally no programming!)

5. Any gotchas?

6. Any sugestions whether ADI is offering an alternative to MAX24288?

Thank you! -- Wojtek

Parents
  • 0
    •  Analog Employees 
    on Apr 9, 2012 5:54 PM

    Hi Wojtek,

         Typical ethernet applications use a +/- 100 ppm frequency source to generate their local timing, usually an XTAL. Each ethernet node which generates its local timing from a local source is asynchronous with other ethernet nodes. The idea behind synchronous ethernet is to distribute a high quality time base to all nodes to create a synchronous system. This requires the use of a CDR to recover the transmitted clock and a PLL to translate and distribute it. ITU-T G.8262 describes requirements of such a PLL and the AD9557 is compliant with the specifications laid out by ITU for synchronous ethernet for both an EEC (Ethernet Equipment Clock) option 1 and EEC option 2. In particular, jitter/noise generation, filtering bandwidth requirements, free-run accuracy and holdover stability.

    Questions/Answers:

    1. Are there any App Notes which would illustrate how to use the part for SE clock cleaning?

          To my knowledge, there are currently no ADI app notes specifically targeted at setting up the AD9557 for use with synchronous ethernet.

    2. What is the jitter performance of the cleaned clock?

         This is dependent upon many factors, including the jitter bandwidth of interest, but typically one can expect ~300 fsec of jitter (integrated phase noise) on a 12kHz - 20 MHz bandwidth. Please see table 17 in the AD9557 datasheet

    3. How to configure the part? How to pin strap the part for the SE?

         The AD9557 is easily configured using the integrated wizard in the evaluation software found here. You must select the "software and tools" tab on the AD9557 evaluation board page. To learn about pin strapping the part, please see the AD9557 datasheet, under the section "Hard Pin Programming Mode".

    4. Is it the best / simplest part? Are there better / simpler clock cleaners? (Ideally no programming!)

         This is a highly configurable, extremely versatile, high performance device. Try looking into the AD957x family or the AD9551, AD9552, AD9553 for power on ready devices.

    5. Any gotchas?

         The free run accuracy and holdover stability of the AD9557 are directly derived from the quality of your system clock oscillator.

    6. Any suggestions whether ADI is offering an alternative to MAX24288?

         Currently, we have no single IC, full solution to implement an IEEE 1588 solution. However, the AD9548 can be used to clean up, translate, and distribute an IEEE 1588 generated 1 pps signal.

Reply
  • 0
    •  Analog Employees 
    on Apr 9, 2012 5:54 PM

    Hi Wojtek,

         Typical ethernet applications use a +/- 100 ppm frequency source to generate their local timing, usually an XTAL. Each ethernet node which generates its local timing from a local source is asynchronous with other ethernet nodes. The idea behind synchronous ethernet is to distribute a high quality time base to all nodes to create a synchronous system. This requires the use of a CDR to recover the transmitted clock and a PLL to translate and distribute it. ITU-T G.8262 describes requirements of such a PLL and the AD9557 is compliant with the specifications laid out by ITU for synchronous ethernet for both an EEC (Ethernet Equipment Clock) option 1 and EEC option 2. In particular, jitter/noise generation, filtering bandwidth requirements, free-run accuracy and holdover stability.

    Questions/Answers:

    1. Are there any App Notes which would illustrate how to use the part for SE clock cleaning?

          To my knowledge, there are currently no ADI app notes specifically targeted at setting up the AD9557 for use with synchronous ethernet.

    2. What is the jitter performance of the cleaned clock?

         This is dependent upon many factors, including the jitter bandwidth of interest, but typically one can expect ~300 fsec of jitter (integrated phase noise) on a 12kHz - 20 MHz bandwidth. Please see table 17 in the AD9557 datasheet

    3. How to configure the part? How to pin strap the part for the SE?

         The AD9557 is easily configured using the integrated wizard in the evaluation software found here. You must select the "software and tools" tab on the AD9557 evaluation board page. To learn about pin strapping the part, please see the AD9557 datasheet, under the section "Hard Pin Programming Mode".

    4. Is it the best / simplest part? Are there better / simpler clock cleaners? (Ideally no programming!)

         This is a highly configurable, extremely versatile, high performance device. Try looking into the AD957x family or the AD9551, AD9552, AD9553 for power on ready devices.

    5. Any gotchas?

         The free run accuracy and holdover stability of the AD9557 are directly derived from the quality of your system clock oscillator.

    6. Any suggestions whether ADI is offering an alternative to MAX24288?

         Currently, we have no single IC, full solution to implement an IEEE 1588 solution. However, the AD9548 can be used to clean up, translate, and distribute an IEEE 1588 generated 1 pps signal.

Children
No Data