• State Resolved
  • Locked Locked
  • Replies 6 replies
  • Answers 2 answers
  • Subscribers 21 subscribers
  • Views 760 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

LMK04828: Can one LMK04828 DCLKoutX drive another LMK04828 CLKin0 with DC-coupled differential LVDS signaling?

Dan LeVasseur
Prodigy 10 points
Part Number: LMK04828


I have a master LMK04828 that drives a slave LMK04828.
The slave device acts as a buffer and is used to expand the number of JESD REFCLKs and SYSREFs.

I want to connect the master's SDCLKout1+/- to drive the slave's CLKin0+/- as a JESD SYSREF DC-coupled differential LVDS signal.
(SYSREF is a pulsed signal so it must be DC-coupled, not AC-coupled.)
The master's DCLKout0+/- drives the slave's CLKin1+/- as the JESD REFCLK; this can be AC-coupled since it's a clock.

I don't see specs in the datasheet specifying VICM_DC (input common mode voltage for DC-coupling) and VIDIFF (differential input voltage min and max) for the CLKin0+/- input.

Are these specs available?  Can I DC-couple the two devices for a non-periodic SYSREF?

  • 0
    TI__Genius 10995 points

    Hello Dan,

    For bipolar CLKin buffer, the input common mode voltage is around 1.4V (125C): 1.6V (25C): 1.8V (-40C) and maximum diff. swing is 2.4Vpp or SE swing of 1.2V. The CLKout HSDS driver (very power efficient) or LVPECL driver can be DC coupled to bipolar CLKin buffer. For MOS CLKin buffer, the input common mode voltage is around 1.3V~1.4V and maximum diff. swing is 2.4Vpp or SE swing of 1.2V. The  CLKout HSDS driver (very power efficient) or LVPECL driver or LCPECL can be DC coupled to MOS CLKin buffer.

    The LMK04828 CLKout can be DC coupled to LMK04828 CLKin for SYSREF input. Hope this helps!

    Best,

    Andrea

  • 0 in reply to Andrea Vallenilla
    Prodigy 10 points

    Hi Andrea -

    After reading your post and researching other relevant posts, I'll use this configuration:

    For non-periodic SYSREF:
    - SDCLKoutY configured as 8-mA HSDS
    - CLKinX configured as bipolar
    - the two chips connected as DC-coupled with a 100 ohm parallel termination at CLKinX

    For REFCLK, I'll implement the same configuration for the DCLKoutX to CLKinX path to match temperature and transmission path characteristics.

    The bipolar common-mode drift over temperature is a little concerning; my system operating environment will generally be between 25-85C.
    However, with the HSDS 800mV VOD driving VIDCLKin (min) of 250 mV Vpp, the drift should not be a problem.

    Does this sound reasonable?

    I also want to add the following.
    In this post by Derek, regarding CLKin0/0*:

    e2e.ti.com/.../3835156

    ... near the end I found this statement:
    "Bipolar mode is internally AC-coupled with ~1pF"
    Wouldn't that make DC-coupling impossible? Can you clarify this?

    Thanks.

    - Dan

  • 0 in reply to Dan LeVasseur
    TI__Genius 10995 points

    Hello Dan,

    Dan LeVasseur said:
    - CLKinX configured as bipolar

    After reading Derek's reply from the other post, why are you configuring CLKin0 as bipolar and not as MOS? You are correct, AC-coupling should be used when using bipolar mode.

    By using MOS mode, it supports DC-coupling for LVCMOS signals, especially for SYSREF (the comment you mentioned about AC-coupling internally can be ignored), apart from many other capabilities mentioned in that post as well.

    Note if you use MOS mode, you need to set register 0x146 = 1, or CLKin0_TYPE = MOS in TICS Pro (p. 67 of data sheet):

    ...and this (p. 92 of data sheet):

    Hope this helps.

    Best,

    Andrea

  • 0 in reply to Andrea Vallenilla
    Prodigy 10 points

    Thanks Andrea -

    If you can take a look at this post:

    https://e2e.ti.com/support/tools/simulation-hardware-system-design-tools-group/sim-hw-system-design/f/simulation-hardware-system-design-tools-forum/819056/tida-01023-lmk04828-clkin0-dc-coupling-method?tisearch=e2e-sitesearch&keymatch=%22tida-01023%3A%20LMK04828%20CLKin0%20DC-coupling%20method%22#

    Derek answered:

    "Technically this is not disclosed anywhere, but we have observed that LMK04828 HSDS 8mA output common mode and swing matches very well with the CLKinX common mode and swing requirements in bipolar mode. So in this exceptional case, DC coupling the differential input should work. We do not have this use case characterized in the datasheet, so this behavior is not guaranteed. But it is very unlikely that anything in the future will prevent this use case."

    This is why I was leaning toward bipolar.  In any event, if I connect SDCLKoutY from the master LMK to CLKinX of the slave LMK differentially with a 100 ohm parallel terminator (resistor across the differential lines at the receiver) I would be able to switch modes of the driver and receiver via register settings.

    I believe the datasheet is very confusing.  In Table 40, CLKinx_TYPE, it says “Bipolar is recommended for differential inputs such as LVDS”, but also says “When using CMOS, CLKinX and CLKinX* may be AC or DC coupled if the input signal is differential”.

    Then in section 10.3.1 it says “TI recommends that the CLKin input mode be set to bipolar (CLKinX_TYPE = 0) when using differential reference clocks.”  It also says “The LMK0482x family internally biases the input pins, thus the differential interface should be AC coupled”.

    Yet again, in section 10.4.2, it says “In cases where the common mode output voltage of the LMK0482x family LVDS matches the common mode input voltage of the LVDS receiver, DC coupling can be used”.

    I don't see a use case in the table 7.5 INPUT CLOCK SPECIFICATIONS for DC-coupled differential inputs, but there's a use case for single-ended DC-coupled (with the MOS setting).  If I configure the slave LMK CLKinX input as MOS, how would I configure the master LMK SDCLKoutY output to be compatible?  There's no LVCMOS/LVTTL output mode for SDCLKoutY.

    - Dan

  • +1 in reply to Dan LeVasseur
    TI__Genius 10995 points

    Hello Dan,

    Dan LeVasseur said:
    This is why I was leaning toward bipolar.  In any event, if I connect SDCLKoutY from the master LMK to CLKinX of the slave LMK differentially with a 100 ohm parallel terminator (resistor across the differential lines at the receiver) I would be able to switch modes of the driver and receiver via register settings.

    Thank you for this reply, then based on the previous E2E you referenced, it mentions the positives and negatives of MOS mode and you can make a decision from there, but do note that in general we do recommend bipolar mode. I suggested MOS mode because of the first E2E you reference in this entire E2E post.

    I can see your confusion with the data sheet. In summary, what it should be understood from all those different excerpts is that bipolar mode is preferred and when using bipolar mode, you should AC-couple because of the internal bias of the input pins (which is not fully true because both Derek and our design engineer said it could be DC-coupled, but not recommended since we did not run tests on this setup). Also, if you have a differential output swing, then bipolar mode is preferred as well. When dealing with a CMOS or single-ended signal, then bipolar mode is not heavily suggested and MOS mode can be used, which we have tested (hence why it was listed in the data sheet). 

    Dan LeVasseur said:
    “In cases where the common mode output voltage of the LMK0482x family LVDS matches the common mode input voltage of the LVDS receiver, DC coupling can be used”.

    I can see the confusion mainly generated here and do agree with you that we are saying to use bipolar mode for LVDS and that you HAVE to AC-couple, yet here we mention that DC-coupling can be used. I'll add this in our system for our next data sheet update.

    Dan LeVasseur said:
    I don't see a use case in the table 7.5 INPUT CLOCK SPECIFICATIONS for DC-coupled differential inputs, but there's a use case for single-ended DC-coupled (with the MOS setting). 

    I will also add this to be changed for the next data sheet update. Thank you for bringing this up.

    Dan LeVasseur said:
    If I configure the slave LMK CLKinX input as MOS, how would I configure the master LMK SDCLKoutY output to be compatible? 

    If you decide to use MOS, then you can connect your unused pin through a 50-ohm resistor to GND. Hope this helps!

    Best,

    Andrea

  • 0 in reply to Andrea Vallenilla
    Prodigy 10 points

    OK thanks Andrea, I'm glad you connected with Derek and got to the bottom of this.  I'll go with DC-coupling as described above.

    A minor comment is that I believe AC-coupling is recommended (according to the datasheet) when using bipolar input mode because the chip provides a mild internal bias, but DC-driving the inputs easily overpowers the bias (I saw this in one of Derek's posts).  If this is true I think it should be clarified in the datasheet as well.

    - Dan