How long between exposures ?

[johnb]

Hi

Using the kit in my signature I typically take 300 Sec exposures at ISO 800 - Ideally I should build a gap between shots to allow the sensor to cool ?, is there an optimum time period ?

Regards

John B

[nmoushon]

I use a 450D (unmodded though-not sure if that makes a difference or not) and typically do 15-30sec in between exposures. The warmer the night the longer in between. But I dont do more than that as I don't think that amount of time will make a difference worth the time lost. Even 30sec adds up real fast...thats six 5min subs EVERY HOUR. So by the end of the night there could be several dozen 5min subs that I lost out on and I think that those subs will off set any extra heat built up over the night. On top of the normal heat build up that is. So I typically tend to stay near 15sec during the summer and drop to 7 secs during the winter. If its warm enough for me to want to use 30sec break then I usually don't even bother imaging unless its something special going on or I'm getting short subs to help with M31s core or somthing like that. The time between is more to let shutter reset and image to download and for me to take a quick peek at the screen to see if I can see anything major going on thats not suppose to be going on.

[ollypenrice]

Have you thought of physio gel packs rotated in the freeze and placed around the camera?

Olly

[ian_bird]

Olly

I noticed my H-18 heats up significantly during download. For example - I was set at -20, taking flats - around 2 seconds. By the time the download ended I would be up to -12. So I was having to build in a delay of 30 seconds to allow the camera to get back down to around -20.

Is that normal - or should I add it to the list of problems with this camera?

Cheers

Ian

[nmoushon]

Have you thought of physio gel packs rotated in the freeze and placed around the camera?

Olly

I thought about doing that for mine but worried too much about condinsation so decided against it. I've started looking into a cold box for that reason. The extra weight sucks though.

[JKB]

I have tried various lengths from 10s to 5mins and found that on an average mild evening around 1min is sufficient. During the cold winter skies with ice forming on the scope around 15s. This is on my 30D.

HTH

Cheers

Jamie

[Daddystu]

I thought about doing that for mine but worried too much about condinsation so decided against it. I've started looking into a cold box for that reason. The extra weight sucks though.

I surrounded mine with ice blocks (in freezer bags) for 20 minutes before going out tonight. Had a touch of condensation but I think I knocked a few degrees off. A 7 minute sub at 1600 ISO wasn't 'too' noisy.

[davew]

Olly

I noticed my H-18 heats up significantly during download. For example - I was set at -20, taking flats - around 2 seconds. By the time the download ended I would be up to -12. So I was having to build in a delay of 30 seconds to allow the camera to get back down to around -20.

Is that normal - or should I add it to the list of problems with this camera?

Cheers

Ian

Add it to the list Mine drops 0.1 degree. It's not an H18 though.

Re DSLR rest. Someone did a heat test some time ago. Perhaps a site search is in order if they don't see this post.

Dave.

[ian_bird]

Dave

"Mine drops 0.1 degree."

Mine drops 8.0 degrees! 0.1 degrees is noise level to me!

What camera do you have?

Cheers

Ian

[davew]

A Moravian. I'm imaging now but it just came to me that you may be running flat out ? If so your camera may not be able to keep up.

Mine's running at just over 75% at -25* right now.

Dave.

Were you doing flats indoors ?

[IanL]

First the theory, and at the end what I do in practice (which isn't the same as the theory!)

The EXIF temperature is measured at the start of the exposure on Canons and that doesn't tell you:

- The average temperature for the whole exposure.

- The rate of increase during an exposure.

- The rate of decrease during a pause.

- Whether the sensor reaches thermal equilibrium during an exposure (and if it does, at what temperature and after how long will vary with environmental conditions).

The first thing we care about is reducing the average temperature of each exposure, which reduces dark current and thus more importantly reduces dark current noise which increases as the square root of dark current. If the camera heats up rapidly at the start of each exposure, a long pause between exposures is just wasted imaging time since the average temperature of the exposure would be similar same regardless of the length of the pause. If it heats up slowly throughout the exposure, then it would be worth leaving a cooling pause between exposures as the average for the exposure would be lower.

Since all we have is the start temperature of each exposure, using the EXIF to determine whether pauses are effective at reducing the average temperature during the whole exposure is pretty hard to do. There are three ways to test it:

- First you could take a long exposure, followed immediately by the shortest exposure possible, then a pause, etc. The second exposure should give you a reasonable estimate of the temperature at the end of the long exposure since the camera will not have had time to cool much. You can then use the two values to figure out the average temperature over the whole exposure. You'd typically take a series of exposures and graph the average temperature over time. You would need to keep going until the temperature stabilises as it will rise between exposures at first due to the sensor and the camera heating up. Try different pause lengths for each series and do them in conditions similar to those in which you image to find out what works best at keeping average temperature down. Bear in mind that taking two exposures between each pause (long and short) you will not get an exact match to real life but it should be close.

- Second you could take a series of dark frames in the same conditions as you image. Again try different pause lengths. This time you simply measure the mean ADU of the entire image (or maybe median if you have lots of outliers). Most of the processing packages can do this or try ImageJ (http://rsbweb.nih.gov/ij/index.html) if you don't have one. Plotting the mean of each image over time should in theory give you the same information as temperatures above. You're first looking for the means to stabilise in each run, and then trying runs with different pause lengths to find one that gives you the lowest means (which results in the least dark noise). The problem with Canon cameras is that they reduce the apparent dark current during on-camera processing (unlike a CCD camera), so the results are not easy to interpret and may be useless.

- The third method is to take the dark frames as above, but measure the noise instead. Take the standard deviation of the whole dark frame and that will give you a reasonable estimate of the noise. (It will be the read-out noise plus the dark current noise, but if you keep all other things equal, the read-out noise should be fairly constant between frames and only the dark current noise will vary. If you want, you could subtract a master bias from each dark first to get a better estimate of the dark current noise only, but I wouldn't bother). The Canon preprocessing does not reduce the dark current noise so plotting this against time will give you a direct measure of how effective your pauses are at reducing it.

The second thing we care about is maintaining as constant an average temperature as possible for each exposure so that we can create dark frames with matching temperatures and thus subtract the unwanted dark current. If you have a strategy to create frames with the same average temperature (or mean or standard deviation) then it should be reasonable to assume that darks and lights with the same average temp can be used together.

The problem is that due to the Canon pre-processing it may not be that easy to successfully subtract your darks. Even though you have matched average temperatures it isn't clear if the pre-processing treats dark frames and lights differently. It may not be a major issue as astronomical light frames are virtually unexposed by daylight photography standards and the camera may treat them the same, but worth bearing in mind. There is a lot more on the subject here: http://www.cloudynig...hp?item_id=2786

So much for the theory, what do I do in practice:

My current strategy for darks is to use a mini-fridge and ice packs with the camera inside. I pre-cool for a couple of hours and then take a ten minute dark, followed by a much longer pause (20-30 minutes). By varying the pause length I can get starting EXIF temps of 9, 11, 12 or 13C for a period of 12 hours or more (never had a single 10C though for some reason). I just set it up during the day and leave it running under the control of APT. Over a few weeks I have built up a library of different temperature darks from which I make a master dark. I'm pretty confident that the average temperatures are the same for darks with matching EXIFS as I measure the standard deviations and they line up nicely.

I don't have any control over the camera temperature for lights though. My experience with the mini-fridge and ice-packs is that any external cooling system is not that effective in practice (i.e. a peltier cooler box or some kind of ice pack arrangement). I doubt I could come up with something more effective than my dark-fridge that would hang off the end of a scope, and the dark-fridge requires pauses that are at least twice as long as the exposures to achieve a usefully low EXIF temperature and keep it constant. If I take darks in the mini-fridge without pauses or with pauses much less than 20 minutes the camera will heat up to 30C or above fairly quickly, so 2/3rds of the night would be wasted, even if I could make a smaller lighter version of the thing.

My current strategy is to image during the colder months and I don't use any pauses (other than for mirror lockup/vibration). I can get results with EXIFs of somewhere between 10 and 20C on nice cold winter nights. Bear in mind that radiative cooling to the night sky in winter is very effective and it's not just a question of how cold the air temperature is at ground level. Maybe a metal cooling box with the peltier directly attached be more effective than my mini-fridge. In the summer I don't bother imaging much as I just end up with 30C+ subs and a nice snowstorm effect. Planetary/lunar imaging is the way to go when it is warm outside!

Instead of trying to temperature match the darks I use PixInsight for the pre-processing. This has a method of scaling darks by measuring noise statistics, and in theory you don't have to have a temperature matched master dark for the light. PI will try different scaling factors for the dark until you get the best noise results in the calibrated image. I did a lot of experimenting earlier in the summer to see what worked and I found that making a single master dark from my biggest set of darks of one temperature/exposure length/ISO was the way to go. This produced less noise in the stacked image than using (smaller) sets of darks with temperatures that matched each light frame, so it does appear to work. I plan to spend some time making more darks of the same temperature as this should further improve the results.

In theory the same scaling process should work for scaling one master dark to different exposure lengths for the lights, but I haven't tried that yet. One thing that really helps is to make a really big stack of 'bias' frames to remove fixed pattern noise from the images. Taking 200+ bias frames doesn't take long (though stacking them in to a master bias was a challenge). There were measurable, if small, improvements in the noise over smaller stacks of 30, 50 or 100 bias frames.

[dph1nm]

I seem to remember that a few years ago when the 1000D was new, someone (probably on Cloudy Nights) pointed out that the temperature sensor was nowhere near the CMOS chip, so what it was measuring the temperature of was rather a moot point!

NigelM

[Prador]

Since people on here seem to love tearing 1100d's to bits does anybody know where the internal temp sensor is?

Magic Lantern gives a figure for internal temperature but I have no idea where it takes the reading from. If it is very near the sensor at least we would have an idea how much heat is generated on longer subs and wait for it to cool between exposures.

Dave

[ian_bird]

A Moravian. I'm imaging now but it just came to me that you may be running flat out ? If so your camera may not be able to keep up.

Mine's running at just over 75% at -25* right now.

Dave.

Were you doing flats indoors ?

Hi Dave

Outside - during the day.

Will try again tonight.

Cheers

Ian

[davew]

Hi Dave

Outside - during the day.

Will try again tonight.

Cheers

Ian

Sounds like it may have been running on the ragged edge. They do -40 from ambient I think. They also do a fan to cool the body. If I spoke too soon I apologise !

Dave.

[ollypenrice]

Readout does cause heat build up. This is very prevalent when shooting flats whatever make of camera you have so for rapid successive subs I'd build in a delay.

Olly

[IanL]

I seem to remember that a few years ago when the 1000D was new, someone (probably on Cloudy Nights) pointed out that the temperature sensor was nowhere near the CMOS chip, so what it was measuring the temperature of was rather a moot point!

Yes it is definitely another problem to factor in, since the location of the sensor (and construction of the camera) will determine how representative the measured temperature is of the actual sensor temperature. If it is floating around in the air a long way from the sensor, it could be relatively well insulated from the real sensor temperature and under read, or it might be stuck on something with a significant thermal mass and provide a reading that is much higher. About the best you can do is have a sensor on the PCB directly under (if not touching) the sensor chip package and even that isn't perfect.

The gold standard (for Canons) would have to be the standard deviation of the darks. How I dream of set point cooling, may have to put one of the kids up for sale in the classifieds section.

[davew]

Readout does cause heat build up. This is very prevalent when shooting flats whatever make of camera you have so for rapid successive subs I'd build in a delay.

Olly

If you are using a camera within it's design parameters and the cooler actually works then no, you shouldn't have to leave a delay.

As one measurement is worth more than a thousand guesses I tested my camera. I setup indoors so the wind didn't play a factor. The ambient, according to a digital thermometer, was 21.8*. I set the camera to -20* to replicate Ian's temperatures and made a series of flats with no delays using the flats plug in inside Maxim. The start temperature shown on the fits header was -19.9* which then went to four readings of -20*. The next two were -20.1* and then shock horror the next two were -19.8*. No problem, the next was -20* and that's how it continued with very slight fluctuations for over forty flats.

As anyone who looks at their Maxim fits files will know, the temperatures are recorded to fifteen decimal places !! I rounded up and down for my figures.

So, in conclusion, I think that an increase of 8* degrees is quite wild. Either the ambient was too high or the camera is very inefficient or defective. This could be the reason SX make a fan add on. It may be best for you to increase flats rest periods but not all camera makes require it.

Dave.

[ian_bird]

Hi Dave

Very useful info. Thanks for that. So which Moravian are you using? I am guessing you are using their own software to control.

Appreciate it.

Regards

Ian

[davew]

Hi Ian,

I use the G2 8300 with internal filter wheel ( They also do a 12 position external bolt on wheel ) It's the same one that Nik Szymanek tested recently. I like him use Maxim. The only reason in my case is that I know the programme and feel quite at home with it. If I used the very similar Moravian programme I would need to re learn a few things I'm sure.

Their software does have a few good points about it. One is that just about every parameter you can think of can be followed including the input voltage of a battery. Great if you want to watch your Leisure battery's progress as the night wears on ! Some are quite frankly over the top. Who, other than me, would want to monitor the chip temperature ? The drivers have a built in cool down and warm up routine so as not to shock the chip.

This camera runs at -50* below ambient ( They suggest sticking to -45* ) so even in summer the Kodak chip runs cold. If the outside temperature really is below freezing then yes, it will run at over -40* easily. Not that I would but it's fun testing it.

At this point I'll depart from my usual test it and see approach and give an opinion that I only believe is true. One of the reasons larger bodied camera like the Moravian, QSI and FLI etc can cool so well is that it can actually cool properly. The camera isn't hampered by trying to appeal to a few imagers that want to use it on the front of an SCT. The Atik 383 uses the larger form factor but still won't cool as well as the others mentioned. This I only BELIEVE to be true and you won't read it on Wikipedia !

There are downsides however. You won't find shelves full of the things in the local Tesco. They are usually built to order so in effect you'll get a personal hand built camera with it's own identification number in the software / firmware. This does mean you can attach as many cameras to your computer as you like but I think you can only run two at a time unless I've miss read the instructions.

There is only one retailer, though another keeps indicating he'll be selling them.

Only three Moravian imagers post here so there won't be a collective back up.

If you need any specific information then please PM me. ( That isn't necessarily a downside )

Dave.

[ian_bird]

Hi Dave

"Who, other than me, would want to monitor the chip temperature ?"

Well - me for one!

365astronomy.com ?

I use Maxim too. Would seriously consider the Moravian - but I use the SX AO unit - which only works with SX CCds. It would be a bit hard to give that up!

But you are giving me serious thought to consider it!

Cheers for all the info. Really, really useful!! Thanks a lot!

Cheers

Ian

[ollypenrice]

If you are using a camera within it's design parameters and the cooler actually works then no, you shouldn't have to leave a delay.

As one measurement is worth more than a thousand guesses I tested my camera. I setup indoors so the wind didn't play a factor. The ambient, according to a digital thermometer, was 21.8*. I set the camera to -20* to replicate Ian's temperatures and made a series of flats with no delays using the flats plug in inside Maxim. The start temperature shown on the fits header was -19.9* which then went to four readings of -20*. The next two were -20.1* and then shock horror the next two were -19.8*. No problem, the next was -20* and that's how it continued with very slight fluctuations for over forty flats.

As anyone who looks at their Maxim fits files will know, the temperatures are recorded to fifteen decimal places !! I rounded up and down for my figures.

So, in conclusion, I think that an increase of 8* degrees is quite wild. Either the ambient was too high or the camera is very inefficient or defective. This could be the reason SX make a fan add on. It may be best for you to increase flats rest periods but not all camera makes require it.

Dave.

OK but on the two makes I use regularly I do have to build in a delay when shooting flats, partly because it is a daytime activity in the observatories and it tends to put the coolers on their limits. I don't temp match my flats but I do try to shoot them as cool as possibe. I'd just advise people to check that rapid repeating flats don't bump up temperature because it is likely to happen. The SX add on fan still doesn't match the standard Atiks for cooling. The difference varies but can be in the order of 5 degrees on hot nights. I feel that this is a lot.

Olly

[Luke]

With my Canon 450D I used to leave as long a gap as the length of exposure. So, 5 minute gap between each 5 minute sub.

I did this because I thought I'd read somewhere that's what you should do. Perhaps it was for cameras with amp glow problems though?

I later found out others were using very short gaps and swtiched to something like 10 or 20 second gaps.

I can't say I noticed any difference in the individual frames apart from I had almost twice as much data.

That's just my personal experience, no carefully controlled testing.