We have developed our own PCB with an i.MX28, using type MCIMX287CVM4B. There are 128MB DDR2 RAM are installed, which are supplied by an external dc-dc converter, as well as NAND Flash, ethernet phy, 4.3" TFT display and other periphery, also supplied external. So, the internal dc-dc converters are used for cpu interfaces, JTAG and boot mode selection only. As OS we are running Windows CE 6.0.
When we put the first 24 boards into operation, we observed a strange behaviour of the internal DC-DC converter on 4 of these boards:
Case 1: The device is booting properly, but when Windows CE 6.0 switches the display mode to show the desktop, a reboot is performed, caused by internal dc-dc converter that stops switching for ~ 1ms. In consequence, VDDD drops from 1.55V to 1.15V, VDDA drops from 1.8V to 1.6V. After this pause the converter starts switching again, but all voltages rising too high (VDDD (cyan)=2.3V, VDDA (pink)=2.7V, VDDIO (green)=5.1V), while VDD4P2 (yellow) keeps 4.2V continously. It takes nearly 1 ms until all voltage rails have returned to its nominal values.
In picture 2 the pink graph shows the switching node at the external inductor, pin A16, instead of VDDA.
Case 2: The device is not booting at all, VDDD swings from 1.0V to 2.4V, VDDA swings from 1.8V to 2.9V and VDDIO swings from 3.3V to 5.3V.
In both cases we replaced the cpu with new parts and all boards were running properly. So, obvious there was no problem with our board but with some CPU's.
Did anyone of you observed the same behaviour? What could have caused these defects? A failure rate of 17% is too high to enter series production of our board with good feeling.
Although we have put another 15 boards into operation in the meantime without failure, we are still wondering about these 4 defects.
Best regards,
Maik
Message was edited by: Grant Whitacre Maik, your images did not come through so I reproduced your discussion entry with images intact. Regards, Grant
Hi Maik
i.MX28 VDDIO can not tolerate 5.3V according to
Table 6 "DC Absolute Maximum Ratings" datasheet
chip may be damaged.
Probably some peripheral (may be LCD ?) backs power
processor.
Also it may be recommended to add patch below (seems it as not
implemented in Windows CE 6.0 package)
L2.6.35_10.12_5V_SUPPLY_PATCH : Linux patch for i.MX28 SDK 2010.12 to
add the most robust support possible for a VDD5V or DCDC_BATT only configuration.
For testing one can test board with minimum soldered components,
then add them one by one.
It is highly recommended to test with OBDS bare metal tests
IMX_OBDS : On-Board Diagnostic Suit for the i.MX28.
Best regards
chip
-----------------------------------------------------------------------------------------------------------------------
Note: If this post answers your question, please click the Correct Answer button. Thank you!
-----------------------------------------------------------------------------------------------------------------------
Hi chip,
Thanks a lot for your fast reply!
We downloaded the patch and are analyzing the source code at the moment. Obviously, this is a patch for linux. Is there also a version for Windows CE available? Or is the patch used for eboot?
By the way, we have another board found yet, that shows the same behaviour. It does not matter if the LCD is connected or not. So, we now have a hardware for approving effectiveness of this patch.
We will also have a look at the OBDS tool.
I will come back with our results.
Best regards
Maik
Hi Maik
I think there is no such patch for Windows CE.
Best regards
chip
Hi chip,
when we analyzed the patch for linux we found that we forgot to disable the battery as possible voltage source in our software. Because we have an external 5V supply as only source, sometimes there were a false detection of the connected power supply.
Now all boards are working proberly.
Thank you very much for your support!
Best regards
Maik