Linux/DS90UB954-Q1: DS90UB954-Q1 encoder errors for different wire

Hi, Team

Is there any progress?

Thanks!

Hello Lei,

Please give me some time to address your questions. For now here are answers to a few of your questions:

1. Look at the MAP tool data, the pass criteria is 3 EQ levels and 4 strobe positions. From what you have shared, the link should pass.

2. You could manually adjust the AEQ setting and override it. However, this is not recommended due to manual adjustment of AEQ is solely meant for debug purposes.

3. I will need to get back to you on what EQ Hi/Lo means.

4. S-Filter controls the min and max value allowed for clock to data sample timing. For more information, see the 954 datasheet.

5. Lock Chg Cnt is the lock change counter. It is meant to keep track of how many times lock has been lost since the system has been brought up. Typically we recommend customers to Reset Statistics once the full system is initialized, and then monitor the Lock Chg Cnt. 

Hi, Mostafa

 Thanks, I will wait for your further reply!

And add a question, what's the meaning of AEQ status of reg D3?

Hi Lei,

AEQ_STATUS (reg 0xD3) is brought out to ALP in the EQ Hi/Lo line and is simply reporting back the EQ stage that is being used by the chip's adaptive equalization. This value can be different across different cable types/lengths, as the deserializer will tune the EQ until a stable lock is achieved. You can read more about it in section 7.4.9 of the DS90UB954-Q1 datasheet.

Dear Sir,

Is there any progress about my original questions?

When I use the ALP "Current RX Port Status" to test my device, it can works well using 3.5m STP wire at about 25°C，but it will be lock changed and encoder errors at about 60°C. The thermal design seems well, Tj of 953/954 and other power ICs will be not exceed 80°C.

Hi Lei,

Can you confirm that you clear all errors once the link is establish? 

Typically, at start up there will be some errors accumulated when the devices initially link. We recommend after establishing the first link to clear all errors, and then proceed with any long term testing. This will give a better indication of whether a true error has occurred during operation.

Hi, Mostafa

Yes, I have clear all the startup errors after link status is established. And I have test more than 72 hours at about 25°C in this way, there is no any error.

I don't know if it's is right to test the system in this way, if yes, then we need check where is the issue.

Hi, Sir

Is there any progress?

Thank you very much!

Hi Lei,

At 72 hours of test @ 25C, are you getting errors?

Can you also share your test methodology?

Hi, Mostafa

At 72 hours of test @ 25C, I get no lock change, no error.

First, connect the I2C interface of USB2ANY to the I2C of 954, open the ALP software and start the device to ensure that our device can produce images normally.There will be lock change and error in the first startup, and the error will be cleared after the system is stable.Continuously observe whether lock change and error occur during the operation.In the previous test, I used 3.5m cables of three different brands, and there was no error. Of course, some manufacturers also failed the test.

However, when the high/low temperature test was carried out, none of the cables from these three manufacturers passed the test.

Hi Lei,

Thanks for sharing this bit of information. Would it be possible to share the cable type/vendor?

These devices have relatively strict requirements on the cable characteristics. That could explain why they all fail with temperature variation.

The 953 and 954 devices were designed to support the DACAR 462 cable up to ~10 to 15m.

Hi, Mostafa

One of the HSD cable we used is from LEONI's Dacar535.

I take Margin test at high and low temperature, which looks good, as shown in the figure below

So, you think my test method is correct? In your opinion, does Dacar535 wire with a length of 3.5 meters meet the attenuation demand? We have test the loss @about 25C, it is no problem.

Thank you. Later I will test with shorter cables or Dacar686 cables to make sure the problem is related to cables.

Thank you for your long time support for my problem. If there is any good news in the future, I will also synchronize it here.

Hi Lei,

Just another thought I had. Can you do the margin analysis at the 3 temperatures: low, high, and 25C? Then from there you can observe what are the optimal EQ and Strobe Position settings for each temperature scenario to manually program.

However, keep in mind, in normal operation TI does not recommend manually controlling the EQ and Strobe Position -- this is simply for characterizing the system and debug. So if the device can't automatically compensate for any temperature variation, you're still in the same situation but at least you can identify what's causing the failure a bit more.

Yes, I can do it follow you suggest to debug this issue, but the the final program will remain the default AEQ.