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.

CDCI6214: AC-LVPECL source side termination

Part Number: CDCI6214

Hi,

We are using a CDCI6214 clock output (AC-LVPECL @100MHz, 2.5V supply) to feed the serdes differential clock (SD1_REF_CLK_P/N) of NXP T1024 CPU, with max 0.8V differential allowed.

We are having some PLL sync problems with this clock.

We connected that clock as diagram below:

[CDCI6214 AC-LVPECL clk source P] --- [TX line 50ohm] ---- [22ohm*] --- [100nF] --- [SD1_REF_CLK_P, on T1024]

[CDCI6214 AC-LVPECL clk source N] --- [TX line 50ohm] ---- [22ohm*] --- [100nF] --- [SD1_REF_CLK_N, on T1024]

* = the 22ohm resistor is put because AC-LVPECL has Vpkpk=0.9V while the CPU accepts only 0.8V so that resistor gets a 0.7 moltiplication factor.

I know that AC-LVPECL is not standard: I wonder if using CDCI6214 with output clock differential AC-LVPECL @100MHz we'd better to put also a 150ohm to ground at source side.

My doubt originates because figure 8 in datasheet doesn't show any 150ohm termination at source side (so we didn't put this resistor) but generally a LVPECL needs a 130-150ohm to GND.

Moreover measuring the clock at source side with a FET probe with 150ohm to GND on both clock signals we obtain 1.8Vpk-pk instead that expected 0.9Vpk-pk.

Thank you,

Cheers,

Giacomo.

  • Hi Giacomo,

    The AC-LVPECL is an LVPECL-like driver, but AC-LVPECL is only supported in AC-coupled mode. The AC-LVPECL output must be AC-coupled to 100ohm differential input, or 50ohms single ended inputs, like datasheet Figure 8.

    When you measure with the proper termination, you can see that VOD of 900mV is the VOH - VOL of the single ended waveforms, and 2 * VOD is the amplitude of the differential signal. This is illustrated in datasheet Figures 8 and 9

    Kind regards,
    Lane

  • We are measuring with a 1Mohm input FET probe (AP034 by Lecroy), but the transmission line is terminated both sides with 100ohm differential (inside CPU and inside clock generator).

    The circuit is in the picture below:

    I've done further measures:

    case

    2*VOD (differential voltage pk-pk)

    measure point

    R262,R263=0ohm

    R163,R168=150ohm

    0.91V (avg) CPU side (orange lines in the picture)

    (as above)

    1.01V (avg) clock gen side (yellow lines in the picture)

    R262,R263=0ohm

    R163,R168=OL
    0.90V (avg) CPU side (orange lines in the picture)
    (as above) 1.71V (avg) clock gen side (yellow lines in the picture)

    With 150ohm resistors: 2*VOD seems wrong but similar at both side of transmission line.

    Without 150ohm resistors: 2*VOD seems correct at clock generator side but it's half at CPU side.

    Did you expected this values?

    Which resistor configuration (if any) is the correct one?

    Thank you.

    Giacomo.

  • Hi Giacomo,

    Without the 150ohm resistors is the correct configuration. One thing that isn't clear is why you are measuring half the swing with R163 and R168 depopulated based on the measurement location. I would expect 2*VOD for both measurements because these are effectively measuring the same node due to R262 and R263 being shorted. Would you agree?

    Kind regards,
    Lane

  • Thank you Lane,

    rechecking the measurement the voltage was ok (=2*Vod): perhaps a bad contact on test point.

    We'll remove 150ohm to ground as your advice.

    Have a nice day,

    Giacomo.