Hi Martin, Here are some responses. PIXPHASE: I know that there is a delay in sensor data output respect to the clock but this delay is not fixed because it keep changes and I have to adjust the PIXPHASE data everyday as shown in the images when I start using AFE? Response: If Pixel Rate INCLK is being used, the PIXPHASE should be consistent with respect to the CIS sensor output waveform. Is the AFE input clock consistently aligned with the clock signal to the sensor? These are both coming from the FPGA and if that alignment is not consistent it will cause problems.Please provide a scope photo of the clock to the sensor and the clock to the AFE and the AFEphase signal simultaneously. Capture this information for several power up and configuration cycles to see if there is any shift and what is the shift amount? Clamping Pulses: Based on the scope image you provided (Nov. 16, 10:27PM), you are getting 8 pixels clock worth of clamping pulses. You can increase this number by using Page 2, Reg. 6-7 (input clamping start / stop). Increase the end setting to get more clamp pulses. There is only one clamp pulse as shown in the image. Response: In this response above we were referring to the Auto CLPIN reference level Clamp switch pulses (highlighted below), which are used to bias the AFE side of the AC-coupling capacitors at the proper reference level, not the input signal sampling. The terminology is confusing so it is easy to miss the difference. The settings affecting the Auto CLPIN input clamping are Page 0, Register 0x03 Bit 5, Page 0, Register 0x04 Bits 3, 4 and 5 and Page 2 Registers 0x06 and 0x07. If I increase the end setting this means there are more sample point within a pixel period? Response: Increasing the end setting makes the Auto CLPIN operation occur for more pixels at the beginning of the line. VCLP: As you said to keep the VCLP value below the minimum CIS sensor output but when I do so there's still output on the ADC pins when no light is falling on the sensor but when I keep 0.9VCLP there is no output in dark condition except for some peaks. Response: When the sensor is in full darkness, the output data should be a value near the 0 codes end of the AFE output data range. If it is at 0 codes then some information may be lost. The VCLP value in combination with the Black Calibration feature can be used to set the black (no light) output level to a desired target code. Alignment: Based on your scope image (Nov. 17, 12:03AM), the delay pulses can be delayed a little bit relative to the sensor output as the actual event is a little earlier than what you see. You can use the little glitches on your sensor output (coincident with internal switches opening and closing) to see what is being sampled and to fine tune the alignment (as another technique to employ). Didn't get your point. Scope image Response: I have highlighted the small pulses I was referring to for easier understanding, below. Hope it is clear now. Regards, Hooman
↧