Re: First light for a new camera

Dale Ghent

I have the ASI1600 with the Panasonic chip, 294 color, and IMX455-based QHY600 which is the same sensor as the ASI6200.

The Panasonic chip that's in the ASI1600 and other cameras has been a fantastic chip over the years and, like you said, the microlensing effect on super bright stars has been the only major complaint about it. The 12bit nature of its ADCs can be compensated with by just making more exposures. For what it is/was, it has been a very good standard in the CMOS realm. It was a sensor designed in the early 2010's with tech of that era, and despite that, it has decently controlled amp glow and dark current. And as a chip of that era, it is a front-illuminated design with a single gain domain. These are outdated designs now.

The IMX294/492 is back-illuminated which means more photons actually fall into the pixels, which raises the QE. The vast majority of current era designs from Sony are back-illuminated. The other thing that many current-era chips have in their corner is dual conversion gain. Dual conversion gain is where a good bit of the magic occurs in these sensors and lets them be flexible. DCG sensors came about to address the need for a sensor that can perform well in both bright and dim settings. The Low Conversion Gain (LCG) domain grants you a large full well, with the trade-off being read noise. The High Conversion Gain (HCG) domain gives you lower read noise at the cost of a shallower well. Both have DR curves that largely mirror each other. LCG is meant for exposing in bright environments where the deep well would be best suited, where the HCG domain is for dark scenes, where sensitivity and low read noise is needed. Both the IMX294/492 and IMX455, as well as others, are dual conversion gain sensors.

You can spot these types of sensors by looking at the graphs commonly posted by ZWO, QHY, and others. They have two distinct areas, like this one for the IMX455-based ASI6200MM:

You can see the two domains clearly in the DR and Read Noise charts. LCG is on the left, starting at gain=0, and then the transition to HCG happens at what looks like gain=100. You can see that LCG and HCG start out more or less similarly, with the LCG having a larger well. Once you get to gain=100, the read noise drops, DR (mostly) recovers, and the downward progression resumes as gain increases. The difference is that at gain=0, you have a 50k e- well, and at gain=100 (the start of the HCG domain) you have what looks like a ~15k well. These charts aren't the greatest but that's my guess assuming they're on a log scale. But you can see the difference nonetheless.

So what does this mean in practice? You can shoot in two different ways:

1. At gain=0 and with long exposures
2. At gain=100 and with shorter exposures

both will yield effectively equivalent results. Which one is best to use (note: I'm purposefully avoiding the word "better" here) largely comes down to personal preferences. Personally I shoot at the start of HCG (the equivalent of gain=100 here) because that's where the sensor performs its best. The very slight hit to DR compared to gain=0 is more than made up by the larger number of exposures I can rip through. People doing photometry or planetary with these kinds of sensors might find the LCG domain a more useful place for their purposes. In this way, you can look at LCG vs. HCG more in terms of what's best for the use case.

If you have one of these kinds of dual conversion gain sensors and are not entirely sure where the LCG->HCG transition is on the gain scale, it's easy enough to figure out for yourself. Cap off the camera and run bias exposures in loop. Look at the histogram or image stats as you steadily increase the gain. As you increase the gain, you will likely also see an increase in the measured standard deviation. At some point, the stddev will suddenly drop. The gain value in which that drop in stddev happens marks the entry into the HCG domain, as stddev is a representation of the read noise.

As for IMX294/492 vs. IMX455, the 455 (and its APS-C analog, the IMX571) are hands-down the best we have on the consumer market right now. With zero amp glow, a good middling pixel size that applies to a wide range of focal lengths, QEs in the mid-70s for Ha/Sii and mid-80s for Oiii, and 16bit ADCs, they're really cement the fact that modern and highly compelling CMOS tech has arrived and is available. There's not much left to want for the amateur. These sensors are absolute monsters on fast optics as well.

On Feb 8, 2021, at 10:55, Terri Zittritsch <> wrote:

[Edited Message Follows]

Dean, thanks for posting the information.
Can you say what you think you're getting by replacing the ASI1600MM with the ASI294MM? I have the ASI1600 and many great images are taken with this camera. The biggest downside that I've noticed is the micro-lensing effect you can get on bright stars. I know the ASI294 shows deeper well depth, and better DR, but do these directly relate to better images? I know that some poo poo-ed the ASI1600 because it only has 12 bits of depth, but that didn't keep people from producing great images. Since you've owned both, what are the benefits that you can see in the images?
Maybe a tough question. I wonder if there is any micro-lensing?

I've recently purchase an ASI6200MM which is on my stowaway now.. and waiting to get a first image.


Join to automatically receive all group messages.