SN65DSI85: PLL_UNLOCK and CHA_SOT_BIT_ERR

Eric Steimle

Part Number: SN65DSI85
Other Parts Discussed in Thread: DSI-TUNER

We have about ten prototype systems with the SN65DSI85 running. Occasionally on a few of them we lose displays. A power cycle will bring it back. Today we were able to catch a failure while we were all instrumented up and we saw that the error register 0xE5 was 0x35. Which says, PLL_UNLOCK, CHA_SOT_BIT_ERR (leader sequence bit error), and that it detected an uncorrectable and correctable error.

I was surprised to see PLL_UNLOCK because we are using a local oscillator at 74.25MHz. Is there some kind of DSI error that would also knock the PLL out of lock? We were able to get the part working again by just re-writing the configuration over I2C, but I'd like to get to the root of the problem too. Also the distance between our DSI source and the TI chip is 2-2.5". So it's a pretty short run, all matched impedance, matched length etc.

FInal question. Can we recover from this error or do we have to reset the TI part everytime?

Thank you,

-Eric

over 6 years ago

0 Ikechukwu Anyiam over 6 years ago

TI__Mastermind 35130 points

Hi Eric,

Are you following the initialization sequence in the datasheet?

Regards,
I.K.

0 Eric Steimle over 6 years ago in reply to Ikechukwu Anyiam

Expert 2290 points

You won me a dollar bet, because I told the rest of my crew this was the exact question you would ask me. Also high on the list was did you use our initialization tool? But I felt good about this one.

I have asked firmware to look into it. This particular unit was running for 45ish minutes before failure.

0 Ikechukwu Anyiam over 6 years ago in reply to Eric Steimle

TI__Mastermind 35130 points

Hi Eric,

Happy to hear I earned you some money.

Although, it's a bit strange that it's only failing after 45 minutes. After you've confirmed the initialization sequence, if you see the error again can you go ahead and read register 0xE5 and 0xE6 again? Please clear these registers by writing 0xFF to them before reading them though, as some of these bits may be erroneously set during initialization.

Regards,
I.K.

0 Ikechukwu Anyiam over 6 years ago in reply to Ikechukwu Anyiam

TI__Mastermind 35130 points

Hi Eric,

Any updates on this issue?

Regards,
I.K.

0 Eric Steimle over 6 years ago in reply to Ikechukwu Anyiam

Expert 2290 points

We're still actively investigating it. We had firmware make a build that will log these errors for us and we're going to see if they coincided with the displays going blank. We're also building up some probe cards, and getting some new probes to look at the SI side of things.

Long story short it's still a big issue that we want to understand better.

0 Eric Steimle over 6 years ago in reply to Ikechukwu Anyiam

Expert 2290 points

Two other pieces of information. First we don't understand why the PLL unlock bit is periodically being set. It doesn't seem to stop the part from functioning but we see it so often that no one trusts it (or the PLL is really losing and gaining lock). Second we believe the problem is worse at 1080p than 720p when we are running the DSI interface faster.

0 Ikechukwu Anyiam over 6 years ago in reply to Eric Steimle

TI__Mastermind 35130 points

Hi Eric,

This may be due to a line time mismatch between the DSI side and the LVDS side. Can you share the settings you used for the DSI85 and processor configuration? You can do so with the DSI-Tuner tool for ease of reading (either share screenshots of the tabs or attach the .dsi file). Can you also share your display datasheet? You can email these to me at i-anyiam@ti.com if you do not want to share them here.

Also have you confirmed that the initialization sequence is being implemented correctly?

Regards,
I.K.

0 Ikechukwu Anyiam over 6 years ago in reply to Ikechukwu Anyiam

TI__Mastermind 35130 points

Hi Eric,

Any updates on this issue? Please feel free to continue replying via email, I just need to keep this thread updated.

Regards,
I.K.

0 Eric Steimle over 6 years ago in reply to Ikechukwu Anyiam

Expert 2290 points

We found one instance where the firmware team wasn't initializing properly in one wake up case. We've also put an interrupt handler in to catch this error and re-init the chip. It needs more testing to say if we got it or not though.

0 Ikechukwu Anyiam over 6 years ago in reply to Eric Steimle

TI__Mastermind 35130 points

Hi Eric,

Understood, please let me know if this resolves the issue or if you need further support.

Regards,
I.K.

Interface

Interface forum

SN65DSI85: PLL_UNLOCK and CHA_SOT_BIT_ERR