TMDS181: Signal detect register

Hi Ken,

The signal detect procedure mentioned in that other thread is the clock detect. The TMDS181 will ignore data on the other lines if it does not see a clock.

Regards,
JMMN

Here's the problem.  The TMDS181 occasionally gets in to a state where:

1.  There is valid video (including a valid clock) going in to the TMDS181 (also yes the TMDS181 is configured correctly, Hot Plug is asserted, The clock ratio has been written and the clock ratio setting is confirmed by reading TMDS181 status)

2.  There is no video coming out of the TMDS181.

3.  TMDS181 status reports no PLL lock (or enable), no rate status (value is 0xF) and no clock detect.

4.  Once it is in this state, the only way to clear it is to re-initialize the TMDS181.  Cycling Hot plug does not work.

5.  The problem is always cleared (good video and good TMDS181 status) by re-initializing the TMDS181

My problem is that it is impossible to know when to re-intialize the TMDS181 since the status it reports when it is stuck is identical to the status it reports when there really is no video present.

BTW:  Using Hot plug to detect this scenario is problematic.  Some HDMI sources leave there video off for a long time after Hot Plug is asserted back to them (this is particularly true when they are just powering up).

ALSO:  Yes, I am certain video is  present.  The video source a quantum 980.  Plug the 980 directly in to the source, you see video.  Plug the 980 in to the the TMDS181, no video output (not just bad, none).

I either need some way to reliably prevent the TMDS181 from getting in to this state (which may be difficult since I have no control over when video is asserted to the TMDS181) or I need a way to detect this condition and recover from it.

Thanks,

Ken

see below

When it gets hung in this state, can you send register dumps of Page 0 and Page 1?

A few questions:

Do you see this with a particular resolution?
Is the clock ratio being set manually or through DDC snoop?

It is very problematic to dump all the registers. Our modules only includes a tiny micro-controller to drive an LED and configure components on power up. The only way to dump registers is to perform dummy reads and then manually copy the results from a Beagle I2C trace. It would be much easier if you could tell me the specific registers you want to see.
The problem does not appear to be associated with any particular resolution.
The clock ratio is set through DDC snoop (for debug I currently check page 0, Register 0xB, Bit 1) to verify that the snoop is successful (it is).

Hello. Any feedback???

Hi Kenneth,

Can you confirm the Page 1 registers - registers 00h and 01h, and then 4Eh and 4Fh when the problem occurs?

What does "re-initialize" the TMDS181 entail? Can you clear out the issue by just setting the apply-rxtx bit?

Regards,
JMMN