• State Resolved
  • Locked Locked
  • Replies 8 replies
  • Answers 1 answer
  • Subscribers 22 subscribers
  • Views 256 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.

LMK04832: LMK04832: LMK04832 PECL Output [DC Termination Versus AC Termination]

Dallas Pace
Prodigy 120 points
Part Number: LMK04832
Other Parts Discussed in Thread: DS250DF410

Tool/software:

I'm trying to use the LMK04832 to generate a 1.611328125GHz Clock.

I'm targeting using LVPECL 1600 Format at the Output of the LMK04832.

The 1.611328125GHz Clock {LMK Output] is AC-Couple to a DS250DF410's Rx Input Port.

Data Sheet for DS250DF410 shows Pin Function Table, noting 100Ohm Termination and AC-Coupling Requirements.

The LMK04832 Data Sheet shows the Output Voltage to be 800mV when DC-Coupling is used:

The LMK04832 Data shows that the Output Voltage "Varies" as a Function of Frequency when AC-Coupling is used:

At 250MHz, Vod = 760mV.

At 2.5GHz, Vod = 510mV.

I've tried to analyze that these are the "Expected" Voltages at the Output of the LMK04832 when AC-Coupling is used, but I'm not getting consistent results.

A). Analysis

The AC-Coupling Capacitor and the 100Ohm Internal Termination of the DS250F410, form an Active Voltage Divider that can be modeled as shown:

If I select C = 220pF and f = 250MHz, I get:

Xc = [(1)/(2)*(Pi)*(250MHz)*(220pF)] = 2.8937

So, V2 = V1 * [(50)/(50 + 2.8937)]

When V1 = 2.3V [Voh from LMK], then V2 = 2.174V

When V1 = 1.5V [Vol from LMK], then V2 = 1.418V

So, V2 = 2.174V - 1.418V = 756mV [Data Sheet says 760mV].

If I select C = 220pF and f = 2.5GHz, I get:

Xc = [(1)/(2)*(Pi)*(2.5GHz)*(220pF)] = 0.28937

So, V2 = V1 * [(50)/(50 + 0.2897)]

When V1 = 2.3V [Voh from LMK], then V2 = 2.287V

When V1 = 1.5V [Vol from LMK], then V2 = 1.491V

So, V2 = 2.287V - 1.491V = 796mV [Data Sheet says 510mV].

As you can see, I'm getting results that match the Data Sheet at 250MHz Output [With C = 200pF], but 'Not' at 2.5GHz!

Question: Could you provide the Math Steps used to obtain the Vod Values in the LMK04832 Data Sheet when AC-Coupling is used with a 100Ohm Termination?

  • 0
    TI__Mastermind 29172 points

    Hi Dallas, 

    Please note these are also typical values, these are not guarantees with any min or max. 

    The DS measurements are a bit imprecise due to human error in the measurements. 

    What I recommend is just use the IBIS model to estimate VOD for your board stack up. 
    This is a much better approach.


    I quick check on IBIS model gives me a VOD a bit over 800mV @2.5GHz



    Best regards, 

    Vicente 

  • 0 in reply to vicente flores prado
    Prodigy 120 points

    For AC-Coupled......the Data Sheet is showing 540mV @ 2.5GHz. That's a "Pretty" big difference from 800mV via your IBIS Model Results......wouldn't you agree? I'm thinking the Data Sheet "Typicals" are Way Off?

    From an IBIS Model Perspective, the model would be (1) Shunt 120Ohm to Ground, (1) Series Capacitor {AC-Coupled as per LMK Data Sheet} and (1) 100Ohm Shunt internal to the DS250DF410

    Could you adjust your IBIS Model to "Match" what I'm showing above {It would be "Great" if you use the DS250F410 IBIS Model}!

    Do you have an IBIS Module for the LMK04832 and for the DS250DF410?

    If so, please send me a "Link" where I can find them.

  • 0 in reply to Dallas Pace
    TI__Expert 8830 points

    Hi Dallas,

    Just to piggyback off of what Vicente said, in your calculations at 2.5 GHz, you characterized the output pin at 50 Ohms.

    However, in our datasheet, we state that at high frequencies (above 1.8 GHz), the pin impedance is low, and typically requires a resistive pad to bring the impedance up to 50 Ohms. 

    Thanks,

    Michael

  • 0 in reply to Michael Srinivasan
    Prodigy 120 points

    Where is there a reference to Output Pin Impedance being low......requiring a Pad?

    The Only reference I see in the Data Sheet to Resistive Pads being required is on the Fin Pins:

    Another thing I'm not sure of is whether not my Model is correct.

    I did "Not" take into account of the 120Ohm Shunt.

    This 120Ohm Shunt, is in "Parallel" with the Series Cap + 100/2 Termination.

    So, is your recommendation to "Ignore" the Values listed in the Data Sheet when AC-Coupling the LVPECL 1600 Outputs using 120Ohm Shunt, AC-Coupled to 100Ohm Termination?

    Thanks!

  • 0 in reply to Dallas Pace
    TI__Expert 8830 points

    Hi Dallas,

    I misread your list of calculations - the input and output pads share the same impedance characteristics, and as a result, at high frequencies, the input and output impedances are no longer 50 Ohms. However, the output impedance is not of any significance to your calculations.

    Accounting for the 120 Ohm emitter resistor as being in shunt, followed by a series cap, followed by a shunt 50 Ohm resistor to GND is the most appropriate way to model.

    Thanks,

    Michael

  • 0 in reply to Michael Srinivasan
    Prodigy 120 points

    Agreed!

    Last Question: Do you have IBIS Models for the LMK04832 and for the DS250F410?

  • +1 in reply to Dallas Pace
    TI__Expert 8830 points

    Hi Dallas,

    I am not certain that an IBIS model exists for the DS250F410, but the IBIS model for the LMK04832 is on the product page for the part, under design tools and simulation (https://www.ti.com/product/LMK04832#design-tools-simulation). Any similar offerings for the DS250F410 would be on the corresponding product page.

    Thanks,

    Michael

  • 0 in reply to Dallas Pace
    TI__Mastermind 29172 points

    Dallas, 
    The IBIS model for LMK04832 can be found on the device homepage under design tools and simulations. 
    For the retimer the IBIS model should be in the same place and if it's not it's most likely in a mysecure folder. Some products lines require users to request access to their IBIS-AMI models.
    For that you need to make a new E2e request for asking for access to the IBIS-AMI model and the retimer team will grant you access 
    As mentioned, the DS values are only typical values and provide no guarantees. 

    Best regards, 

    Vicente