Weird HDMI issue on TV

[TheMG]

Background: Panasonic TC-P65S1, had problem with image corruption and intermittent signal loss on HDMI inputs only at higher resolutions and refresh rates. Replaced HDMI switch IC (Sony CXB1444R), no improvement. Next thing down the line... the main SoC IC, a Toshiba TC90413XBG. Replaced it, with one taken from a donor board (same physical board but for different TV model). Now TV works 1920x1080 60Hz no problem, no more corruption.

A new problem has now appeared: the TV only works with PCs but shows no image at all (or sound, and does not even acknowledge there's something connected) when connected to an A/V receiver, a Blu-Ray player, or Steam Link. Note that these devices used to work perfectly fine (at 24Hz) prior to swapping out the SoC.

This one has me scratching my head. Why would a computer connected to the TV work just fine, but nothing else does? Anyone more knowledgeable about HDMI that can shed some light on this?

[Rasz]

what is it advertising in its edid? might be missing audio and YCrCb  modes now, did anything you replaced needed reprogramming?

[macboy]

Check if HDCP is working. The PC's video card should be able to report this somehow (e.g. through AMD Radeon software).

[TheMG]

Well I finally got a bit of spare time to dig a little deeper into the problem.

I think you may be up to something regarding the EDID data. On the laptop that I had been connecting to this TV all along, the EDID data is all there. However, I think it just remembers the EDID from before, because when I connected a different PC to the TV, it was not detected.

Now here's the interesting bit... when I power off the TV, the PC immediately detects it and gets the EDID data!

Looking into the schematic, it looks like each HDMI port has its own I2C EEPROM, presumably to store the EDID. These EEPROMs appear to be powered not from the TV itself, but from the HDMI port. This explains why the TV is detected properly when turned off.

So something is happening when the TV is turned on that is preventing the EDID from being sent. The I2C lines, in addition to going from the EEPROMs to the HDMI ports, also go to the HDMI switch IC and then there is also I2C going from the HDMI switch IC to the main SoC.

So perhaps when the TV is powered on, the SoC takes over the job of providing the EDID data (I assume the HDCP stuff also uses these same I2C lines?). Anyways, somehow this isn't happening as it should.

Time to bust out the oscilloscope again. Hopefully something simple like a pull-up resistor got blasted off the board, rather than an open or short on one of the BGA pads...

[m k]

Maybe it's something with the standby voltage.
Like what power is powering the connection and when.

Maybe port selection is messing it and powering off clears it.
Like no connection, no power.