AnsweredAssumed Answered

i.MX6 MIPI Sensor debug

Question asked by Raman Subramanian on Apr 20, 2016
Latest reply on Apr 20, 2016 by igorpadykov

This customer is trying to bring up camera sensor interfacing to i.MX6. It is not clear from the description below whether the issue is with the camera driver. Maybe, someone more knowledgeable than I am can provide some suggestions.

 

?  I'm working on
our Linux camera driver and I need some clarification on the behavior of the
system.

>

> Because we don't have the equipment to analyze the
MIPI signal (which is currently set to 650Mhz), I'm relying on the i.MX6 MIPI
PHY status registers to tell me whether there is a good signal coming in, but
the things I'm observing elsewhere in the system and some comments I've seen on
the message boards indicate that I may be wrong about how I'm interpreting
them.

>

> When I start capture, MIPI_CSI2_PHY_STATE goes to
the value 0x00000330 and alternates between 0x00000330 and 0x00000300 as I
sample it while the camera streams.  I
interpret this to mean that the PHY is seeing readable data at the correct
clock frequency from the sensor, but I'm seeing messages from people in the
forums suggesting that I can get these readings even if there's no stream.  The error registers are both zero, so I'm
trying to determine whether I can assume the analog stream is good and I should
look for problems with frame interpretation or if there's still a chance that
the PHY isn't getting a meaningful stream.

>

> I've had to do more work than the driver porting
documents describe for this sensor because it is a raw bayer image that needs
to be interpreted as grayscale and not transformed by the image processing unit
and the support for that in the kernel is limited/missing, so in order to
isolate the problem I'd like to be able to at least rule out the analog front-end.

 

Raman

Outcomes