The HDMI 1.3a spec section 4.2.8 states:
"The Display Data Channel (DDC) I/Os and wires (SDA, SCL, DDC/CEC Ground), shall meet the requirements specified in the I2C-bus Specification, version 2.1, Section 15 for Standard-Mode devices."
The I2C 2.1 spec Section 15.2 table 6 states:
"VIH max = VDD + 0.5"
That would be +5.5V, but lets be generous, the HDMI spec has +5.3V max spec'd everywhere else. My monitor is putting out +5.25V on both SCL & SDA. Therefore, it is idling within spec.
A lower value SCL pullup would keep the SCL line elevated, but it's elevated to within specification, so that point is moot. I am using the type C power supply with the CM5IO, but I'm also having the same problem on my custom board where I can power it with a lab bench supply. I'm only using the CM5IO as an example because you guys should have access to it and you should be able to replicate my findings.
Just to be clear, I'm connecting to what I believed to be a HDMI compliant monitor (albeit on the fringe of the spec) to a CM5IO and it will not boot. That would make the CM5 the HDMI non-compliant device.
I believe when the DDC lines backfeed the CM5 with a voltage higher than it's supplied with, it interferes with the unpublished analog boot logic of the CM5. I would argue, that the CM5 requires backfeed protection to handle the full range of the HDMI spec.
I'm not expecting you guys to change the CM5, maybe an errata, or maybe just this post is good enough to warn others when they can't get their CM5's to boot after plugging in a monitor. (viewtopic.php?t=385371)
"The Display Data Channel (DDC) I/Os and wires (SDA, SCL, DDC/CEC Ground), shall meet the requirements specified in the I2C-bus Specification, version 2.1, Section 15 for Standard-Mode devices."
The I2C 2.1 spec Section 15.2 table 6 states:
"VIH max = VDD + 0.5"
That would be +5.5V, but lets be generous, the HDMI spec has +5.3V max spec'd everywhere else. My monitor is putting out +5.25V on both SCL & SDA. Therefore, it is idling within spec.
A lower value SCL pullup would keep the SCL line elevated, but it's elevated to within specification, so that point is moot. I am using the type C power supply with the CM5IO, but I'm also having the same problem on my custom board where I can power it with a lab bench supply. I'm only using the CM5IO as an example because you guys should have access to it and you should be able to replicate my findings.
Just to be clear, I'm connecting to what I believed to be a HDMI compliant monitor (albeit on the fringe of the spec) to a CM5IO and it will not boot. That would make the CM5 the HDMI non-compliant device.
I believe when the DDC lines backfeed the CM5 with a voltage higher than it's supplied with, it interferes with the unpublished analog boot logic of the CM5. I would argue, that the CM5 requires backfeed protection to handle the full range of the HDMI spec.
I'm not expecting you guys to change the CM5, maybe an errata, or maybe just this post is good enough to warn others when they can't get their CM5's to boot after plugging in a monitor. (viewtopic.php?t=385371)
Statistics: Posted by sel_lab — Mon Feb 02, 2026 10:48 pm