LMH1219: Register flips causing video dropout

Hi Dajon,

Let me look into this. Do you see the issue at both 3G and 12G data rates?

If so, is the same register change seen on both data rates?

Best,

Shane

Hey Shane,

The team said they saw the same issue with both 3G and 12G data rates. The same Register 0x7 from the CableEQ/Divers Page goes from 0x24 to 0xFF for both 3G and 12G.

3G and 12G are the only data rates they use and saw the issue on both. They didn’t try other rates.

Any ideas yet? Have any additional questions?

Hi Dajon,

Are we certain the customer is not writing 0xFF to this register when the issue happens? From what I can gather, the device should not be changing the value in this register by itself.

The fact that this is only seen on certain devices is strange as well. If the customer performs an ABA swap between a working board and failing board, does the failure follow the Device or the board?

By video drop out over time, do you mean the Video stops transmitting from the LMH1219 after a certain point? How long does it take before video drops out?

Best,

Shane

Shane Hauser said:

Are we certain the customer is not writing 0xFF to this register when the issue happens? From what I can gather, the device should not be changing the value in this register by itself.

[Alex] That’s good to know. When this occurs, software should only be reading status – not performing any config. That said, our interim field solution is to disable all device access following initialization, in which case the video dropout issue has not been observed. So, we’re open to the idea that an intended read may, somehow, become a write. Albeit, so far, we haven’t managed to capture an SPI transaction that looks like a write on oscilloscope. Below are measurements of timing parameters, with the one marked in red now addressed in software that’s pending test (doesn’t seem likely to be the issue). Looking at how the software person implemented the protocol, I’ll likely seek clarification on a couple of things that the datasheet doesn’t confirm for me, but I’m currently awaiting another unit for further probing, and to capture waveforms for context. I should note that SPI devices are daisy-chained four-deep, with the first two devices being LMH1219 and the other two being LMH1218. The issue has, so far, only been observed for the second LMH1219 device in the chain.

Shane Hauser said:

The fact that this is only seen on certain devices is strange as well. If the customer performs an ABA swap between a working board and failing board, does the failure follow the Device or the board?

[Alex] We’re currently awaiting additional units from the field and could try this once we have enough.

Shane Hauser said:

By video drop out over time, do you mean the Video stops transmitting from the LMH1219 after a certain point? How long does it take before video drops out?

[Alex] We’ve been replicating by interrupting video to the device repeatedly and it can take a few, to many interruptions. Temperature also seems to have some effect (which makes it seem like it could be a timing or amplitude margin issue, but amplitudes look good and timing is as given above). We can restore video by writing the register back to 0x24. It can also be replicated by writing the register to 0xFF followed by a video interruption.