TDP158: TDP158 failed on HDMI1.4

Hi,

Can you please share customer's schematic, layout, and I2C register dump?

Thanks

David

Hi David,

See the sch pcb and the test result in the link below, in TIDrive. EQ they used is 15.5dB for data and 4.5dB for clock, they used this value because they tried all the settings found the difference is very small, and they think it is better. It may due to their on board trace is not very long, and the mean problem is on the output. 

https://tidrive.ext.ti.com/u/gZfoBpaX6BWTl-gB/4cf66cac-6a0d-45ba-bfaa-8838cf083356?l

https://tidrive.ext.ti.com/u/l8lAwYDJQNOFyUDG/e86df2f9-fe34-476a-9509-6fee4508a292?l

The first eyediagram is what they got in their lab, with 15.5/4.5dB EQ, 3.5dB emphassis. The second is from the 3party certification, the return lost is harder. Any suggestion? Customer is asking if they can have a concall with you, any time proposal? 

Hi,

What is the code to access the TI Drive content? 

I can support a call, but would like to look at their design first before the call.

Thanks

David

https://tidrive.ext.ti.com/a/OA4cmy8OHIsqDhyc/e86df2f9-fe34-476a-9509-6fee4508a292?l

https://tidrive.ext.ti.com/a/OA4cmy8OHIsqDhyc/e86df2f9-fe34-476a-9509-6fee4508a292?l

Can you try this?

Hi,

1. For trace length of 3.682in at 3Gbps, I would not set the EQ to maximum. I would set the EQ for data at ~2dB, and EQ for CLK at 0dB. 

2. Change the TERM bit to 0x00 (0x0B = 0x40) for HDMI1.4, does it improve the eye?

3. They have a 3.3ohm resistor at the output, can we reduce it? 

4. Have you tried to increase the SWING or change the VSADJ?

5. Please try to remove the choke and see if it helps? The choke datasheet looks to be having sufficient bandwidth but I want to double check

6. Please check the TDP158 thermal pad soldering, do we have enough solder paste coverage?

I can support a call at 8PM Wednesday (Dallas Time), 10AM Thursday (China Time) if needed.

Thanks

David

Hi David,

1. As I mentioned they found the difference on EQ does not effect too much on the eye, so they keeped high EQ, but they said yes, they can change it to low later;

2. They tried change the TERM bit to 00, 150~300ohm, the eye looks better, but the VL- failed. They can show you on the meeting.

3. They tried to replace it with 0ohms, said no big difference;

4. Yes, both can make the swing higher, but they are worrying about the EMI, so they prefer low;

5. The chock is 0 ohms resistor for now;

6. They confirmed it solder well.

As I mentioned in the email, they hoping us to move the meeting to Friday Morning China time. 10AM.

Hi,

I can support a call tonight, let's talk.

Thanks

David

Hi David,

See the test result from them. https://tidrive.ext.ti.com/u/11vVcMBb1W7FW9V3/7c73c9cc-be7c-4dee-9cc3-b93ddbaad3f2?l password g9M3m9`d

They retest VL with 300ohms termination, still not equal, and all the other test with different termination, which was done at the same time, all the VL are OK. so it maynot be the test issue. 

And from the test, seems the high EQ helps on the eye, and the slew rate does not help too much.

The each file name stands for 0x0a 0x0b 0x0c 0x0d's value. Please help to advise the next step try.