MT9M001 is a CMOS image sensor. As its output it provides FRAME_VALID, LINE_VALID and DATA. The output signals are synchronized (edge-aligned) by PIXCLK, which is generated by the sensor. The datasheet is for example athttp://www.kynix.com/Parts/4016034/MT9M001C12STM-DP.html
I read the senosr output using FPGA, it somehow works, but I have a hard time understanding timing of LINE_VALID. Since this is the most critical signal for the image shape, I cannot ignore these problems anymore.
The datasheet claims that the maximum frequency of the camera is 48MHz. This is the frequency I use, the period is 20.833. I am supposed to read at falling edge, which means at the 10.416 mark. This is a diagram from datahseet:
To setup valid timing constrains, I have to focus on t_PLH and t_PLL. Let's see how they are defined (min, typical, max values):
Acording to these data, LINE_VALID goes from low to high after up to 7 ns after rising edge of PIXCLK, which is at least 3.4 ns before falling edge (at 48MHz). This means t_LVS min value should be 3.4 ns, not 2 ns ...?
But never mind, let's see t_PLL. Maximum value is 13 ns, which means LINE_VALID goes from high to low no later than 13 ns after PIXCLK rising edge. But PIXCLK falling edge happens 10.4 ns after PIXCLK rising edge, so LINE_VALID falling edge arrives later than PIXCLK falling edge. But only sometimes, because there is no typical or minimum value. Furthemore, if t_LVS is 2 ns, t_PLL would have to be lower or equal to 8 ns.
How to handle this? For me it's a real problem, as my line lengths get messed up sometimes (especially when I overilluminate the camera).
Based on t_OS and t_OH my data signal constraints are:
Question
ailee
MT9M001 is a CMOS image sensor. As its output it provides FRAME_VALID, LINE_VALID and DATA. The output signals are synchronized (edge-aligned) by PIXCLK, which is generated by the sensor. The datasheet is for example at http://www.kynix.com/Parts/4016034/MT9M001C12STM-DP.html
I read the senosr output using FPGA, it somehow works, but I have a hard time understanding timing of LINE_VALID. Since this is the most critical signal for the image shape, I cannot ignore these problems anymore.
The datasheet claims that the maximum frequency of the camera is 48MHz. This is the frequency I use, the period is 20.833. I am supposed to read at falling edge, which means at the 10.416 mark. This is a diagram from datahseet:
To setup valid timing constrains, I have to focus on t_PLH and t_PLL. Let's see how they are defined (min, typical, max values):
Acording to these data, LINE_VALID goes from low to high after up to 7 ns after rising edge of PIXCLK, which is at least 3.4 ns before falling edge (at 48MHz). This means t_LVS min value should be 3.4 ns, not 2 ns ...?
But never mind, let's see t_PLL. Maximum value is 13 ns, which means LINE_VALID goes from high to low no later than 13 ns after PIXCLK rising edge. But PIXCLK falling edge happens 10.4 ns after PIXCLK rising edge, so LINE_VALID falling edge arrives later than PIXCLK falling edge. But only sometimes, because there is no typical or minimum value. Furthemore, if t_LVS is 2 ns, t_PLL would have to be lower or equal to 8 ns.
How to handle this? For me it's a real problem, as my line lengths get messed up sometimes (especially when I overilluminate the camera).
Based on t_OS and t_OH my data signal constraints are:
But how to continue with LINE_VALID?
Link to comment
Share on other sites
1 answer to this question
Recommended Posts
Archived
This topic is now archived and is closed to further replies.