120MC-S odd recurring glitches - debayering bug?

fivey

I am getting reasonably frequent read glitches on my 120MC-S like this

image-20190917112540SMALL.png

For other examples (including the correctly-read images either side of the glitches) see my dropbox:

https://dropbox.com/sh/9mu4c5da005dn7g/AAAZ0_wu8-HNLtORfwZ9TbcYa?dl=0

that will also give you an idea of the frequency of the issue and how sometimes it is one image and sometimes a whole series. Sometimes the image is almost OK, but there is a discontinuity, like part of the image is offset (eg see image-20190917120920.png in the dropbox).

Looking at the step-stripe nature of the glitches and the patterns in the 'off' bits, I wonder if this is some sort of debayer bug? What can I do to make this go away? Even if there is no firmware fix for this, is there anything I can do to avoid triggering this bug?

I am capturing the images using Thomas Jacquin's "allsky" https://github.com/thomasjacquin/allsky

katherine.tsai

Are you using 120MC-S in Raspberry Pi? Could you turn down the USB bandwidth and try again?

fivey

Yes,, with an R-Pi. What USB bandwidth would you suggest?

katherine.tsai

I am not sure, may be you can have more try, such as 40

fivey

The Default and Minimum is 40 - which was causing the problem. I can't go lower than that. I've tried increasing to 50, but the problem still occurs.

Why would *decreasing* the USB bandwidth fix it? If it is a bandwidth problem, wouldn't that make it worse?