Can anyone explain why the MICRO has a 10PF cap to from DCDC_LP to ground (C45)?
It is NOT on the Freescale REF design, and if I remove it I do not boot so it is needed.
This is VERY strange. Why was this added?
Thanks to all!
I'm curious about this too. I haven't found anything in any of the Freescale App Notes that suggest or explain it.
Kean
Is there anyone at Olimex who can answer this?
Thanks
this capacitor is not in the Freescale design but when we run our first lot of iMX233-OLinuXino-MICRO we found interesting issue: the DCDC didn't start-up on some of the boards, when we touch the DCDC with the oscilloscope probe the DCDC was starting successfully so we add 10pf to the next schematic revision and since we place this capacitor we never saw problems with the DCDC startup
so it's not in the datasheet but it's necessary for correct DCDC startup
Thanks Tsvetan - that was my suspicion. Hope that answers the OPs question.
Kean
We found the EXACT same issue on different HW .... It turns out it's all about layout and how close the DCDC caps are to the pins and some other 'mysterious' layout issues.. (i.e. the Freescale EVK does not do this... but it is also a 169BGA so the package may be the issue) .....
In our case in an old proto design, before we saw this cap and used the Mini, we turned off the DOUBLE_FET in the code and this did not happen. It seems that the DCDC turn on breaks the powerup routine.
When we got some MINIS we saw this and realized your HW fix !!!OR!!! our SW fix solved the issue.
The FIX IS a> Leave the cap or b> Find all instances where double_fets are turned on (Max Power) and be sure they are cleared.
THANK YOU FOR THIS INFORMATION!
Great HW fix!
So if any of you great low-level engineers are still reading this I have a follow-up question (unrelated).
Question: Has anyone ever made the IMX23 "Resume"reliable from Suspend
Has anyone done this successfully? (i.e. send a character or I2C to wake up the IMX23 from suspend)
I can post this in a new topic if this has merit.
thanks
-sUngerson
Despite the hijack of the thread, suspend does not seem to suspend much. Measure the input current while in suspend to see.
Has anyone learned more about this issue.. ..We still see unreliable Suspend and resume
Was Installing/Populatin C46 advantageous for anyone?
Thanks
SR