DS90UB928Q-Q1: DS90UB928Q-Q1/927 LVDS LK Signal attenuation issues

Shuming Zhang

Part Number: DS90UB928Q-Q1
Other Parts Discussed in Thread: ALP,

Problem phenomenon:

When the SOC (927 LVDS input) is set to 54MHz, the CLK amplitude is about 48mV, but the CLK amplitude of the 928 LVDS output is attenuated by about 24mV, which is inconsistent with the 927 input signal, when we reduce the CLK frequency to 20MHz, the 927 input and 928 output do not attenuate.

Needs:

The LVDS CLK transmits a consistent signal with no attenuation!

Drawing of the 927 LVDS input：

Drawing of the 928 LVDS output:

11 months ago

0 Jack Scherlag 11 months ago

TI__Mastermind 23665 points

Hi Shuming,

Thank you for attaching the drawings of the 927 and 928 signals. How are you measuring the signals using the oscilloscope? Is the OLDI clock trace terminated with a 100Ω resistor? The amplitude of the 928 signal looks too low.

The above figure is from the data sheet. The differential output voltage is typically 400mV with the single-ended voltage swing typically being 200mV.

Best,

Jack

0 Shuming Zhang 11 months ago in reply to Jack Scherlag

Prodigy 10 points

When we tested it, the OLDI clock trace was already connected to a 100Ω resistor

0 Jack Scherlag 10 months ago in reply to Shuming Zhang

TI__Mastermind 23665 points

Hi Shuming,

Thank you for confirming. Is a differential trace used to measure the signal? What's the bandwidth of the equipment you are using?

Best,

Jack

0 Shuming Zhang 10 months ago in reply to Jack Scherlag

Prodigy 10 points

Hi Jack，

Thank you.Our bandwidth is 100M.

Best，

Shuming

0 Jack Scherlag 10 months ago in reply to Shuming Zhang

TI__Mastermind 23665 points

Hi Shuming,

A rule of thumb for measuring digital signals is that the bandwidth of the oscilloscope should be at least 5x the signal frequency. Although the 927 and 928 measurements do have different amplitudes for similar frequencies, I recommend remeasuring with a higher bandwidth oscilloscope (>1GHz BW would be best).

This app note is helpful for rules of thumb and theory regarding digital signal measurement: Link

In the meantime, a quick way to increase the VOD of the 928 output is to change the LVDS setting (0x4B) register. In ALP or over I2C this would be:

board.WriteI2C(928_address,0x4B,0x01)

Best,

Jack

0 Shuming Zhang 10 months ago in reply to Jack Scherlag

Prodigy 10 points

Hi Jack，

I've tried this before, but it didn't work.

best，

Shuming

0 Jack Scherlag 10 months ago in reply to Shuming Zhang

TI__Mastermind 23665 points

Hi Shuming,

Modifying the VOD control will change the output behavior of the 928 LVDS output. I suspect there is something downstream causing issues.

Are the OLDI traces connected to anything else besides the 100Ω termination resistor? Are these measurements done when a display is connected to the 928 OLDI output? Some displays have internal 100Ω terminations for LVDS inputs.

Best,

Jack

0 Shuming Zhang 10 months ago in reply to Jack Scherlag

Prodigy 10 points

Hi Jack，

The 928 output is connected to a common-mode inductor, then to 100Ω, and there are no other devices.

Best,

Shuming

0 Jack Scherlag 10 months ago in reply to Shuming Zhang

TI__Mastermind 23665 points

Hi Shuming,

Given that these measurements are not done with a high bandwidth scope and there is no underlying issue with the system. I cannot conclude there is an issue with the 928 OLDI output. If an LVDS/OLDI display is not able to correctly interpret the data from the 928, then we can return to this topic.

The TIA/EIA-644-A standard for LVDS recommends using a 5GHz scope and a scope probe of at least 100kΩ differential resistance with no more than 1pF differential capacitance.

Best,

Jack

0 Shuming Zhang 10 months ago in reply to Jack Scherlag

Prodigy 10 points

Hi，Jack，

there are still some questions that I need to ask you.

We want to measure the jitter，how do I set the parameters for the Ti 927 IC?

Best，

Shuming

0 Fadi Abdulhameed 9 months ago in reply to Shuming Zhang

TI__Mastermind 22830 points

Hey Shuming,

I will review your request and get back to you tomorrow.

Regards,
Fadi A.

0 Fadi Abdulhameed 9 months ago in reply to Fadi Abdulhameed

TI__Mastermind 22830 points

Hey Shuming,

Here are the settings needed to test input jitter.

The UI for the jitter analysis should be the UI of the serial data 1UI = 1/(35*PCLK)

- Method = PLL Custom BW

- PLL Type = Type II

- Loop BW = f/40

- Low pass filter = f/20

- Damping = 2

- Target BER = 1e-10

- High pass filter: None

Regards,
Fadi A.

0 Shuming Zhang 9 months ago in reply to Fadi Abdulhameed

Prodigy 10 points

Hi Fadi A,

Thank you so much！

Regards，

Shuming

0 PO-HSIANG WU 9 months ago in reply to Fadi Abdulhameed

Prodigy 20 points

Hi Fadi,
I also have some more questions for you.
a. How to set the parameters of Ti 947 IC?
b. Where to get these parameter information, e.g., Damping, BER and PLL Model Type.

Best Regards,
Dennis

0 Jack Scherlag 9 months ago in reply to PO-HSIANG WU

TI__Mastermind 23665 points

Hi Dennis,

For the input jitter measurement, the parameters that Fadi mentioned are for the oscilloscope. DPOJET or EZJIT are common programs for measuring jitter on high-speed links.

The parameters used to program the scope to measure jitter are chosen to match the performance of the PLL inside the 927.

Best,

Jack

0 PO-HSIANG WU 9 months ago in reply to Jack Scherlag

Prodigy 20 points

Hi Jack,

I need to confirm the TI 947 parameters (not TI 927) to set in DPOJET for measurement.
I can't get the parameters from the datasheet, so I need to know where to find the parameters.

Best Regards,
Dennis

0 Shuming Zhang 9 months ago in reply to Fadi Abdulhameed

Prodigy 10 points

Hi,Fadi A,

The parameters are not very specific, what are the units behind the parameters?

For example，Loop BW，f/40？Hz or KHz？Please mark it out, thanks

Best Regards，

Shuming

0 Fadi Abdulhameed 9 months ago in reply to Shuming Zhang

TI__Mastermind 22830 points

Hi Shuming/Dannis,

PO-HSIANG WU said:
I need to confirm the TI 947 parameters (not TI 927) to set in DPOJET for measurement.

Is this request for the 927 or 947? Those 2 devices don't use the same PLL parameters. This e2e is titled "DS90UB928Q-Q1/927 LVDS LK Signal attenuation issues"

To get these settings I need to sync-up with design team to understand to how configure the scope setting to mimic the device PLL. So please specify which Ser and Des is used here.

Shuming Zhang said:
For example，Loop BW，f/40？Hz or KHz？Please mark it out, thanks

f is your PCLK frequency which is in MHz.

Regards,
Fadi A.

0 Shuming Zhang 9 months ago in reply to Fadi Abdulhameed

Prodigy 10 points

Dear Fadi,

If we use f=54MHz(Loop BW=f/40=1.35MHz, LPF=f/20=2.7MHz), are the parameters in the table correct?

[note]

TJ@BER parameters setting: Target BER=10^-10、PLL Model=Type II、Loop BW=1.35MHz、Damping=2V、LPF=2.7MHz、HPF=none

Oscilloscope: MSO64B (4GHz、50GS/s)、Differential probe: TAP4000 (4GHz)

Best Regards，

Shuming

0 Jack Scherlag 9 months ago in reply to Shuming Zhang

TI__Mastermind 23665 points

Hi Shuming,

The parameters in the table are correct.

Best,

Jack

Interface

Interface forum

DS90UB928Q-Q1: DS90UB928Q-Q1/927 LVDS LK Signal attenuation issues