What is jitter of ADN2816 output in ps?

Hello,

I am wanting to use an ADN2816 clock and data recovery chip as the clock source for an ADC. I need 0.4 ps jitter. The ADN2816 makes great play of using a dual PLL architecture and filter to achieve very low jitter generation and good jitter attenuation above the cutoff frequency.

However all the jitter specs are in UI (unit intervals), based on a particular clock frequency. This suggests that generated jitter is proportional to operating frequency. Looking at other similar devices jitter plots this might not be so (for example a MAX3872 CDR also specified in UI, shows 1.6ps rms jitter at 2.488 Gbps and 2.2ps rms at 622 Mbps. Jitter clearly does not scale with frequency  - maybe more with slew rate).

So I would be very grateful if someone could give me the jitter in absolute ps, not some relative value for the magical ADN2816. I'd be thinking that the recovered clock jitter plots shown in the Maxim data sheet would be very useful in the AD data sheets as well.

Thanks for your help!

- Bart Schroder

Parents
  • 0
    •  Analog Employees 
    on Aug 10, 2011 9:30 PM

    Hi Bart,

    >However all the jitter specs are in UI (unit intervals), based on a  particular clock frequency.

    We did it this way so that we're consistent with the Telcordia GR-253 core spec. Here's an excerpt:

    > This suggests that generated jitter is  proportional to operating frequency.

    Yes and no. The jitter limits (in ps) in GR-253 scale with UI, which is inversely proportional with operating freqeuncy.

    > Looking at other similar devices  jitter plots this might not be so (for example a MAX3872 CDR

    > also  specified in UI, shows 1.6ps rms jitter at 2.488 Gbps and 2.2ps rms at  622 Mbps.

    > Jitter clearly does not scale with frequency  - maybe more  with slew rate).

    There are a lot of factors involved there. Yes, it's possible if the part is starting to be limited by (as you mention) slew rate, etc.

    > So I would be very grateful if someone could give me the jitter in  absolute ps,

    > not some relative value for the magical ADN2816.

    1 UI at OC-3 is 1/155Mbps = 6.43ns, so 1 mUI = 6.43 ps.

    1 UI at OC-12 is 1/622Mbps ~ 1.6ns, so 1 mUI = 1.6 ps.

    In Table 2 of the datasheet, we have:

    Standard     Typ     Max         

    OC-3          1 mUI     2 mUI

    OC-12        1 mUI     3 mUI

    I think the typical numbers mentioned above are probably rounded as they exceed the specifications (usually 10 mUI) by such a large margin.

    Standard    Typ          Max         

    OC-3          6.43ps     12.8ps, RMS (integrated from 12kHz to 1.3 MHz.)

    OC-12        1.6ps         5.4ps, RMS (integrated from 12kHz to 5 MHz.)

    As a comparison, at the OC12 data rate: 622.08Mbps, our ADN2816 jitter generation is about 1.6pS, rms, which is much better than the 2.2pS, rms specified for MAX3872.

    > I'd be  thinking that the recovered clock jitter plots shown in the Maxim data  sheet would be very useful in the AD data sheets as well.

    I'll pass along the request about recovered jitter plots to the applications team.

    -Paul Kern

Reply
  • 0
    •  Analog Employees 
    on Aug 10, 2011 9:30 PM

    Hi Bart,

    >However all the jitter specs are in UI (unit intervals), based on a  particular clock frequency.

    We did it this way so that we're consistent with the Telcordia GR-253 core spec. Here's an excerpt:

    > This suggests that generated jitter is  proportional to operating frequency.

    Yes and no. The jitter limits (in ps) in GR-253 scale with UI, which is inversely proportional with operating freqeuncy.

    > Looking at other similar devices  jitter plots this might not be so (for example a MAX3872 CDR

    > also  specified in UI, shows 1.6ps rms jitter at 2.488 Gbps and 2.2ps rms at  622 Mbps.

    > Jitter clearly does not scale with frequency  - maybe more  with slew rate).

    There are a lot of factors involved there. Yes, it's possible if the part is starting to be limited by (as you mention) slew rate, etc.

    > So I would be very grateful if someone could give me the jitter in  absolute ps,

    > not some relative value for the magical ADN2816.

    1 UI at OC-3 is 1/155Mbps = 6.43ns, so 1 mUI = 6.43 ps.

    1 UI at OC-12 is 1/622Mbps ~ 1.6ns, so 1 mUI = 1.6 ps.

    In Table 2 of the datasheet, we have:

    Standard     Typ     Max         

    OC-3          1 mUI     2 mUI

    OC-12        1 mUI     3 mUI

    I think the typical numbers mentioned above are probably rounded as they exceed the specifications (usually 10 mUI) by such a large margin.

    Standard    Typ          Max         

    OC-3          6.43ps     12.8ps, RMS (integrated from 12kHz to 1.3 MHz.)

    OC-12        1.6ps         5.4ps, RMS (integrated from 12kHz to 5 MHz.)

    As a comparison, at the OC12 data rate: 622.08Mbps, our ADN2816 jitter generation is about 1.6pS, rms, which is much better than the 2.2pS, rms specified for MAX3872.

    > I'd be  thinking that the recovered clock jitter plots shown in the Maxim data  sheet would be very useful in the AD data sheets as well.

    I'll pass along the request about recovered jitter plots to the applications team.

    -Paul Kern

Children
No Data