DS90UH929-Q1: How to measure the eye diagram of FPD-Link III output

eyemask_estimate_draft.xlsxHello,

The DS90UH929 is an FPD Link3 serializer normally used with STP differential output DOUT+, DOUT-.

Using the 35 bits per symbol, and max clock input of 96Mhz, has a max serdes bit frequency of (35*96) 3.36Gbps.

An external Clock and external Sync can be used with the Pattern Generator, to generate a video data pattern.

8.3.18 Datasheet.  Also Application Note AN-2198.

The reference I used is -

In the datasheet, we don't have an eye mask, we have FPD 3 parameters in electrical characteristics.  see section 7.5, 7.6

I have created a draft EXCEL, please look at it, and send questions.

Regards,

Joe Quintal

Hi, Joe

Thank you for your update.
I have 3 more questions about measurement of eye diagram.

1. measurement of the forward channel signal.
Sorry for confusing you.
I want to know the effects that is caused to forward channel signal by back channel signal.

> The FPD-Link output for back channel is the Deserializer send back through the STP to the Serializer.

Is my understating correct, if we measure the the forward signal by operating the UH929 standalone,
it will not be affected by the back channel signal?

2. eye diagram excel
I checked your excel file. I can't find these value in datasheet.
Where can I find this value?
Is this the recommend value of TI for FPD-Link signal? (The customer does not need to change it?)



3. setting of measurement instruments (Tektronix)
Our customer and Tektronix engineer  are measuring the UH929 signal the configuration of the attached file.
Could you please confirm this setting is right way for measurement of eye diagram?

FPD-LINKⅢ(Tek Oscillo DPOJET)setting.zip

Regards,
Nao

Hello

The additional 3 questions.

1) the eye diagram is looking for data transitions at the sample rate.   This is used just for the Forward Channel, or just for the Back Channel, not both at the same time.    Most Deserializers have an I2C register to turn off the back channel, or you can use the Deserializer CML Out (which cancels the Back channel, and adds the AEQ gain).

Yes if you terminate the FPDLink to a passive board, and attach the oscillscope to just the Serializer, there will be no backchannel.

2)  pg 10 of the 929 datasheet, under FPD Link III Differential Driver

VODp-p - 900mv

Delta VOD - 50mv

Delta VOS - 50mv

3.   You need to have a low jitter input clock, to make a low jitter FPDLink measurement.  pg 11  .3*UI (TMDS)

You can substitute an LVDS differential signal generator. 

J10-10, INCLK+, J10 - 12 INCLK-, pg9 datasheet  200mvp-p

"  www.tek.com/.../dpojet  "

There is a Tektronix reference for the DPOJET software.

on page 39 of the EVM User Guide, if you measure DOUT0P and DOUT0N together check the part installation, or for SMA measurement you need to add resistors.  You need to make sure each line is terminated with 50 ohms to GND.

Note: if you make SMA measurements, the eye height is 1/2 the total.

The average cycle to cycle jitter, is needed. 

In the case of the forward channel (Serializer Output - not connected to Deserializer, terminated at oscilloscope or termination board)

see Serializer_FwdChannel_Example_EyeDiagram85Mhz_Clock.jpeg

In the case of reverse channel, I do not have an example picture, different oscilloscope metrics were used.

See Deserializer_BackChannel_Example_OscMeasurement.10Mhzjpeg.

  Regards,

Joe Quintal

Hi, Joe

Which bit?



Regards,
Nao

Hi, Joe

I have a question about "eyemask_estimate_draft.xlsx".
What does this parameter come from?

Is this value "Rx In" or "CMLOut", if min loss is total loss to Rx of des device from Tx of ser device?

Regards,
Nao

Hi, Joe

Thank you for your quick update.
Is this measurement done on the DS90UH929 EVM?

Regards,
Nao

Hi, Joe

Thank you for your update.
Do our customer need to care this parameter (Min loss) in their system board?
Do they need to measure it on their board when they measure the eye pattern to make Eye mask?

I have one more question.
I checked the eye pattern in eyemask_estimate.xlsx
Is this setting correct?

Regards,
Nao

Hello Nao,

The spreadsheet calculation is based on the user input.  While the fraction of UI will not change, the value is seconds will change by entering the customer PCLK value (for eye width)

The spreadsheet value for eyeheight can be based on the spreadsheet method where we estimate the loss from the device to the output connector

it has a user value for reflective and insertion loss of 2db.  This is a starting value for eye width.

Note: If the customer FPDLink deserializer has a CMLOut, you can measure the receiver Forward Channel after AEQ is applied.  you want an eye height >= 100mv.

Other than those comments its OK.

Regards,

Joe Quintal 

Hi, Joe

May I confirm your 2 comments?

> We estimate the loss from the device to the output connector it has a user value for reflective and insertion loss of 2db.

Is this here?

> This is a starting value for eye width.

Is this setting correct? PCLK is 96MHz which is orignal value of the calculat excel file.

Regards,

Nao

Hi, Joe

I understand that how to measure min loss value
Thank you.

Sorry for confused you about Eye mask.
I would like to know which value of Eye mask the value calculated by Excel.
Could you draw a Fig, if my understanding is wrong?



Regards,
Nao

Hi, Joe

Thank you for your quick reply.
Is this setting correct?



Regards,
Nao