EMCCD vs BT sCMOS dragonfly dual camera

Hello, we’re looking into a dragonfly 2camera system with SRRF and TIRF.

Typical config is Ixion 888 EMCCD + Zyla sCMOS. But with the Sona BT 95% QE it seems this could make the ixion 888 defunct and not worth having on there and that 2x SONA is the way to go.
The SONA also has SRRF built in as a bonus.

Am I unfairly discounting the EMCCD and assuming it will be defunct in 5-10 yrs time?

We typically image dim proteins in fungi and bright dyes in host/pathogen for 4D capture.

Many thanks for any input!


Darren Thomson

There is QE and there is shot noise. EMCCDs amplify before A-to-D readout so will have a lower floor of detection. But they come with the risk of burn damage if you are also doing bright TIRF studies. I have owned three EMCCD cameras and all were problematic. So if you can see what you need to, the sCMOS will probably be much more reliable. You may have too high a background from the spinning disc to make good use of the EMCCD low end anyway, which was true on CARV systems. Yokogawas see some benefit from electron multiplication, if you have the funds to repair, or only have very careful users.

Hi Darren - I would perform the direct comparison with the cameras. I have not seen the Sona, but we have always found EMCCDs better for some samples especially dim signal or when you need to collect images faster. Granted the pixel size is smaller for sCMOS if you need the resolution, but that also means the pixel well capacity is less (a quick check of the spec online shows Sona 4.2B6 is half the 888 EMCCD. The EMCCD read noise is also less. We have a Zyla 4.2, as the other models were not out yet, but I still want an EMCCD for at least 5 more years. But, again, the best way to go is to compare directly with the samples you will most often image, if you can.
Lisa Cameron
Duke Univ LMCF

Hi Darren,
We have a Dragonfly 301 CF40 with the iXon Du-897 and currently demoing the Sona 4.2B6. The main reason is getting better resolution (Sona4.2.B6 matches), bigger field of view and higher QE. The samples people are using do not require an EMCCD, so we would theoretically benefit from the Sona, we thought. In the end we had to go through the tests with all our users, as the first results were disappointing and the main/heavier users were not satisfied, mainly due to the fact that the pixel well depth is 42k e-, but also because the people were told to use the EMCCD with a 300 GAIN as default!!! This is not Andor’s fault, more like the third-party company that is representing Andor in Spain… and thus users had to vamp up the laser power and completely reset their imaging settings, but it is difficult when they had optimized their samples (concentrations etc) with the 300 gain and had been working like this for years. In the end, for those affected, we tried the airyscan and it worked well with them. For others, they are just carving back the EMCCD and don’t care that the Nyquist doesn’t match, they want the signal… On the Sona their samples bleach or cells dye becasue of the high power/long exposure before they get any meaningful.

At this point, I am not sure what’s next for this system.

I suggest you demo the camera/system before you purchase. It’s really down to the sample, and not possible to know until testing.

Hi @nhalidi thanks for sharing your experience. Interesting to hear the SONA just could not reach the performance of the Ixon897.

It sounds like the Ixon888 ultra would outperform further.

I think noise suppression can be overlooked for sensitivity (e.g 10% increase in QE). If you half your noise it doesn’t sound very much, but when you express as Signal:noise it’s much more impactful!!

We will demo both cameras next week and will update what we find.
The other further consideration is the SRRF on a smaller pixel sCMOS format (but may need longer exposures; possibly optimal for fixed SRRF).

Very exciting technology!