Wheezy Beta won't boot with HDMI connected

Here's a weird on for you As the titles says the Wheezy Beta won't boot for me with the HDMI lead connected, the power light comes on and OK flashes a couple of times and then it just sits there. If I power it up without the HDMI lead it boots fine, if I then plug the lead in, ssh in and reboot it reboots quite happily. This is repeatable, it won't boot with the lead but will without.

Debian Squeeze and raspbmc both work fine, so I'm happy it's not a power or SD card issue.
I've tried reflashing the SD card, but the problem remains. I will try another SD card to make sure the problem's not there, but the card was working fine with Sqeeze on it.

I am using one of these HDMI > VGA adaptors http://www.amazon.co.uk/gp/product/B007KEIRNG, so that might have something to do with it, but the problem only affects Wheezy.

Anyone got any ideas?

Still no joy with this, I tried a different card with the same results.
The 10/06/2012 alpha works fine.

[dom]

Can you try a rpi-update:
http://elinux.org/R-Pi_Troubleshooting# ... g_firmware

There's been a couple of changes that could just have an effect. Basically pullups have been applied to sdcard and UART RX lines.
It is possible the additional grounding from HDMI is just enough to make a difference to these undefined signals.

(this is a bit of a long shot)

Thanks dom, it might have been a long shot but it worked!

I should have tried rpi-udate before posting.

[ukscone]

i had the reverse. the raspi wouldn't boot WITHOUT the hdmi cable being plugged in but the 24th june firmware via rpi-update fixed it.


pi@raspberrypi /opt/vc/bin $ ./vcgencmd version
Jun 24 2012 19:34:48
Copyright (c) 2012 Broadcom
version 321373 (release)

Can you try a rpi-update:
http://elinux.org/R-Pi_Troubleshooting# ... g_firmware

There's been a couple of changes that could just have an effect. Basically pullups have been applied to sdcard and UART RX lines.
It is possible the additional grounding from HDMI is just enough to make a difference to these undefined signals.

(this is a bit of a long shot)

I don't think your grounding suggestion is a long shot at all, I think it is likely to make a difference. Given the nature of many of the SD cards, in that they don't appear to meet the timeout specs, then the situation with some cards suitability is probably marginal and small changes in capacitive coupling to the ground plane could just change the waveform enough to tip it over the works/doesn't work boundary.

I have no problems with my class 4 cards but running headless I note that the ground plane is floating. If I plug a composite video lead into the TV, then I have about 2 ohms between the outer and the copper pipe on my radiator. I guess HDMI might have a chassis ground too, hence the difference in boot behaviour.

It might be informative if someone with a 'scope could look at the data lines under these two different conditions?