EMMC_EN_N_1V8 troubles

[__deets__]

Hi there,

we are using a CM3 lite module in a custom hardware design. We faithfully replicated the CMIO V3.0 circuitry around the signals EMMC_EN_N_1V8 and EMMC_DISABLE_N with the exception of the jumper J4 being replaced by a hall-sensor plus inverted schmitt-trigger to allow to affect the USB boot behaviour from outside the case through a strong magnet. We can successfully boot into USB mode when using the hall-sensor. We can boot into our system when not using it.

But we see a big (and very relevant for us) difference in behaviour that so far I have not been able to explain to myself: the EMMC_EN_N_1V8 that is supposed to go low for actual booting stays high for ~10-12 seconds. Only then it falls back to low, and our system comes up.

On the CMIO V3.0 that sits right besides the hardware on my table the sequence is

POWER ON
EMMC_EN_N_1V8 HIGH (~500ms)
EMMC_EN_N_1V8 LOW (~250ms)
EMMC_EN_N_1V8 HIGH (~800ms)
EMMC_EN_N_1V8 LOW (system boots)

On our own design the sequence is

POWER ON
EMMC_EN_N_1V8 HIGH (~500ms)
EMMC_EN_N_1V8 LOW (~250ms)
EMMC_EN_N_1V8 HIGH (~12 seconds!!!)
EMMC_EN_N_1V8 LOW (system boots)

I'm at a total loss to where this behaviour could come from. So any insights are more than welcome.

Diez

[6by9]

Have you got pull ups on the HDMI I2C lines? If not then the firmware ends up clock stretching and timing out whilst trying to read the EDID.

[__deets__]

Nope. That's a difference to our design. I will try this tomorrow & report back.

[__deets__]

Have you got pull ups on the HDMI I2C lines? If not then the firmware ends up clock stretching and timing out whilst trying to read the EDID.

For those digging up this nugget of information in a future yet to come: this was in fact the issue. Once pulling the signals up the boot process was normal. Thanks a million 6by9.

A quick follow-up: is it possible to deactivate this HDMI interrogation process to reduce boot-time even more? E.g. using config.txt?