REVISED March 7th, 2021- See the bottom of this post for more info…
The QHY294C OSC camera has a LGC/HGC mode that switches to it’s low noise High Gain Control[HGC] mode when the GAIN equals 1600 or higher. Normally, a user would not expect that the camera to be limited by it’s fullwell. The QHY294C has an impressive fullwell of ~65,000e-. QHY’s current implementation of the HGC mode causes the fullwell to dramatically drop to ~10,000e-. These cameras have lots of sophisticated hardware on the imaging chip so it’s difficult to understand why the fullwell would decrease so dramatically.
This is the fullwell graph for the QHY294C. These curves don’t actually show you the fullwell capacity of the pixels. It’s a fixed value that, for the QHY294C, should be 65,000e-. The graph shown above is the effective fullwell which is the maximum number of electrons that can be stored in a pixel before the data value reaches the maximum 65,535 counts. 65,535 is the maximum value available with 16 bit unsigned integers which is the data type returned by this camera.
The GAIN=1600 setting corresponds to an effective fullwell limit of ~14,309e-. All this means is that a pixel value of 65,535 is equivalent to ~14,309 electrons stored in the pixel.
This is the histogram of an overexposed image I took with the QHY294C at GAIN=1600. In most cameras this histogram would be a vertical line at an X-axis pixel value of 65,535 simply because all of the pixels would be saturated. Instead we have a histogram that looks like a bell curve centred at 45,900. After posting to QHY on their forum and reading all the related posts I believe that this histogram is what the pixel values will look like if they were allowed to reach their fullwell limit. All of the pixels, red, green and blue, were overexposed and all of them reached their fullwell limit.
A pixel’s fullwell capacity is the number of electrons that can be stored in a pixel well. I think it makes sense that, due to minute differences in each pixel, that the fullwell capacity is actually a range of values that can be visualized as a bell curve of possible values
The mean value of the histogram is roughly 45,900 counts which is equivalent to ~10,000e-. What this means for astrophotography is, at GAIN=1600, pixel values higher than ~30,000 counts can potentially be limited by this issue. If you study the histogram you will notice that some pixels were being restricted to 32,000 or less when they reached their fullwell limit. Bright stars will be affected by this limitation. Also care has to be taken with generating FLAT frames so that pixel values do not get any higher than 30,000.
Luckily this limitation moves to the right as you increase the camera GAIN. The above graph shows how this fullwell effect has moved to the right when the GAIN is increased to 2000. At GAIN=2500 this fullwell limitation issue is centred at the 65,535 value.
At GAIN=1600 this fullwell limit will distort the star profile of bright stars. These stars will look bloated with a mottled surface because of the variability of the fullwell limit. The QHY294M Pro camera originally suffered from this same issue at GAIN=1600. QHY claims to have fixed this issue with their latest drivers and SDK. The charts that I saw indicate that they were successful in increasing the fullwell limitation so that the mean value was at 65,535 or ~14,309e-.
QHY has told me that they are working on fixing this same problem in the QHY294C. When they do release a software fix I will update this post with my findings.
Peter
Feb 24th Update: QHY has let me know that the next SDK and driver update on their site will include the fixes for this issue. They said that the logic for the QHY294C and the QHY294C Pro will be identical except that the QHY294C will not support the 47Mp mode.
March 7th, 2021 Update:QHY has now released version 21.02.26 of their All-In-One installation package for their drivers. The release note indicate that it will update the firmware and SDK for the QHY294C. Yesterday I had a chance to install and test this update.
I can now confirm that the fullwell limitation issue has now been corrected. At GAIN=1600 then camera will saturate at 65,535 ADU when the image is overexposed. I no longer need to be cautious when taking FLATs. I haven’t tested the new software under the stars but I have taken several images with this set-up.
A GAIN=1600 BIAS frame has the same mean and standard deviation values as before which means my old masterBIAS should still be valid. I will be regenerating my masterDARKs and BIAS for both 0C and -10C because it is just good practise to re-take your masterDARKs every year.
I also used the sensor analysis feature in Sharpcap to verify that the camera gain and read noise characteristic have not changed. One less thing to worry about for 2021.
47 Megapixel mode
I E-mailed Cha at QHYCCD about the 47M mode available in the QHY294PROC version of this camera. I can update my QHY294C to become a QHY294PROC by sending the camera back to QHY or a local rep. where they will update the firmware in a manner that is not available via a download.
The QHY294C is an 11 Megapixel camera with a very unique pixel arrangement. Every single pixel is actually a cluster of 4 mini-pixels. There are some interesting features of this camera that need the camera to be “un-locked” via the factory firmware update. Once “un-locked” the camera is able to operate in either 11Megapixel or 47 Megapixel mode.
As it stands right now, QHY can “unlock” my QHY294C but the 47M images are not usable. The very custom bayer matrix presented by an “un-locked” QHY294C is not currently supported by QHY. QHY says that they are working on developing software that does support the custom bayer matrix. It will be very interesting to see how they plan to achieve this.
Right now, a 47M image from the QHY294PROC looks like a monochrome image. Unfortunately the color pixels prevent this arrangement from yielding images that would be useful for developing into a quality monochrome astrophoto. I would like to get my hands on a 47M image from a QHY294PROC just to play with.
The imaging chip in the QHY294C [IMX294] has a High Dynamic Range mode which requires the camera to be “un-locked”. The cluster of 4 mini-pixels mentioned above can be assigned into two clusters of two mini-pixels. Each cluster of two mini-pixels acts as a single pixel. This concept is expanded to the point where the camera can operate as two virtual cameras where each virtual camera is assigned to one of the two new mini-pixel clusters. This is clever enough but it turns out that each virtual camera can have their own exposure times.
This results in an 11 Megapixel camera that can take two pictures at once. One picture is a long exposure and then other would typically be a shorter exposure. The two images can then be combined in a 3rd party program to yield a High Dynamic Range image. This could mean that the standard 14b image could be eclipsed by a HDR image that has the equivalent of 16b or more of dynamic range.
Lots to look forward to for 2021
Peter