Dust on the Mirror: Is Cleaning Worth the Risk?

[sophiecentaur]

I keep reading the general opinion that the improvement is just not worth the risk if you attempt to clean the mirror on a Newtonian.
Clearly, there must be a point at which it really is an important thing to do. I thought I would try a back of fag packet calculation and see where it took me. This is my argument and I would be grateful if someone would cast an eye and comment whether it's conclusion is realistic.
I'll take the example of a 200p (mine). Diameter 200mm.
Main mirror area : A₁ = πR₁²
Distance D between the two mirrors is about 1m
Secondary mirror area: about A_{2 =} πR₂²
Assume that the surface of the main mirror is covered by 1% with dust and that the dust has reflectivity of 20%. (I had to start somewhere). Pointing the scope at a single star, there will be E Watts of power entering the tube and 0.002E scattered randomly from the dust. The reduction in image brightness is negligible but a proportion of the light scattered from the dust will find its way to the secondary mirror and will uniformly (?) illuminate the image that the eyepiece sees and hiding any object of similar brightness. (i.e contrast is affected)
The proportion of scattered light hitting the secondary mirror will be (assuming a point source at the main reflector, spread over a hemisphere) approximately equal to
A₂/2πD² = 10^-4π/2π = 0.5X 10^-4 π.
The energy will therefore be 0.002E X 0.5 X 10^-4 π = about 3x10^-7 E
That ratio looks to me to be a relative magnitude of 15! Can that be right?
But that's for just one star. A hundred, of varying magnitudes could make a difference - as could the Moon tonight,
Can someone find the flaw in my argument please? It seems to imply that a really grotty mirror cannot hurt.
But, otoh, regular photographers spend all their time keeping their lenses spotlessly clean - and they don't even suffer from the back scattering that my calculation deals with.

 

[Chronos]

Camera users are bit anal about their lenses. It is true that a dirty lens can affect image quality in bright daylight [makes the picture look a bit foggy] but under low light conditions it would not even be noticeable. A telescope qualifies as a low light condition instrument. It's rarely going to make any difference. Scratching your lens coating will make a difference so be smart and resist the urge to clean you telescope objective, or any other precision optical surface. If you just can't help yourself a lens brush like they sell at camera shops is usually sufficient to make the optical surface eye pleasing.

 

[sophiecentaur]

Low light yes but high contrast. Your sentiments are the same as most people's. But I would ask why, if a tiny scratch would be so deadly, why is a lot of rubbish on the surface of no consequence? You could say that it's coherent but the total area of a scratch and hence the amount of light involved would be nowhere near the same.
This is why I was trying to quantify things. Do you have any comments on my sums? If there's an error then that would be more interesting than just repeating the gypsy's warning.

 

[sophiecentaur]

Actually not even always low light conditions Photographing the Moon, you have a subject in full sunlight. 1/125 @ f8 sort of exposure.

 

[Chronos]

The truth is a small scratch is more psychologically damaging than a performance issue. It affects the sales value much more than image quality. The thing with scratches is they are basically forever. There is no practical and affordable way to correct a physical defect [scratch], whereas you can usually remove surface contaminats [dust and grime] rather easily and cheaply. The trouble with physical defects on an optical surface is they more strongly reduce contrast and encourage scattering [flare] than surface contaminants. ATM'rs learned long ago that you can simply black out a surface defect, like a chip or scratch, with hardly any affect on image quality. All it does is slightly reduce the light gathering ability of the element which has virtually no effect on optical performance in most cases. For discussion, this might be of interest http://www.bobatkins.com/photography/technical/lens_condition.html. I side with the no cleaning crowd because if you do scratch an optical surface, you know it's there and it looks as big as an elephant to you.

 

[sophiecentaur]

The truth is a small scratch is more psychologically damaging than a performance issue.

Very likely but, on PF, we would surely expect some numbers to support the argument, either way. The only numbers that seem to be on offer are my possible ham fisted estimations and calculations. Modulation Transfer Function and Airy Disk and other terms are liberally scattered about discussions on optics. There surely must be something quantitative about dusty and cloudy lenses and mirrors. If only I could find it. I usually find I can put my trust in numbers rather than 'verbals'.

 

[Chronos]

One of my most trusted sources for information like this is Cloudy Nights. This is what they had to say http://www.cloudynights.com/topic/87358-why-not-clean-your-telescopes-objective-lens/page-2. It's not quite as quantitative as you may desire, but, not bad. Here is another discussion https://books.google.com/books?id=O...tches on modulation transfer function&f=false

 

[sophiecentaur]

One of my most trusted sources for information like this is Cloudy Nights. This is what they had to say http://www.cloudynights.com/topic/87358-why-not-clean-your-telescopes-objective-lens/page-2. It's not quite as quantitative as you may desire, but, not bad. Here is another discussion https://books.google.com/books?id=OXWGvFMHUJUC&pg=PA50&lpg=PA50&dq=effect+of+scratches+on+modulation+transfer+function&source=bl&ots=z98eyMgTtB&sig=j8nem_8lySJUkMrUjpzfW2D8sQY&hl=en&sa=X&ved=0ahUKEwij6ZaUpajQAhUBWCYKHSfrAZ4Q6AEIMTAD#v=onepage&q=effect of scratches on modulation transfer function&f=false

Interesting 'opinions' and the longer reference looks at various forms of imperfection. I am still disappointed in the lack of actual measurements. If you take another example of a rather anal hobby, HiFi has a mixture of subjective and objective data available. There rubbish about gold connecting leads etc. is not backed up by measurement but amplifier linearity and SNR are both measured and the results are quoted. Usually we find that enthusiasts love figures - why not Astronomy Enthusiasts? It really beats me.
I live in hope that someone will come up with a source of hard figures before the thread plummets to the bottom of the pile.

 

[Chronos]

The most commonly used optical standards originate from 3 basic sources; MIL specs, ANSI and ISO. These are fundamentally cosmetic in nature as opposed to quantitative performance standards. It is up to the customer to provide any desired performance specifications. These are typically unique to the customer and often only apply to a specific order or orders. It is a rather confusing and inconsistent situation for most optical suppliers [which helps explain why optical elements are often outrageously expensive]. For discussion see; http://www.photonics.com/EDU/Handbook.aspx?AID=57165.

 

[sophiecentaur]

It is up to the customer to provide any desired performance specifications.

which helps explain why optical elements are often outrageously expensive]

Yeah. Well put. Specialist equipment makers tend to have their customers over a barrel. Mostly the customer doesn't actually have a way of measuring performance because it often requires further special equipment.
Dust is an easy thing (relatively) to avoid and it won't be the manufacturer's fault when it arrives on the mirror. So you can't expect help from that direction. But it wouldn't be beyond an amateur to do some measurements with an artificial pinhole light source and a camera. I'm really surprised that no one has such results available.
As you have been prepared to dig out those useful references, perhaps you could look at my figures and see if they make any sense?
I have a very strong urge to clean the mirror in a recently acquired (very cheap) 10" Newtonian and I propose to find some way of characterising the existing dust situation. It looks much more dusty than my existing 8" Dobs and I can do a before and after measurement of flare with the DSLR (8bit resolution of light level) The problem may well be with measuring the dust coverage. Perhaps an off-axis picture of the mirror would give a satisfactory estimate of density X reflectance. I can also try a close-up shot of the surface for an idea of particle size and number per cm².

 

[andrew s 1905]

Harold R Suiter covers this in "Star Testing Astronomical Optics". I only have the first addition where the discussion is quite brief but he discusses the impact on the MTF and he concludes the optics should cover less than 1/1000 of the surface area.

Regards Andrew

 

[sophiecentaur]

Harold R Suiter covers this in "Star Testing Astronomical Optics". I only have the first addition where the discussion is quite brief but he discusses the impact on the MTF and he concludes the optics should cover less than 1/1000 of the surface area.

Regards Andrew

Interesting. I wonder what that looks like at the far end of a Newtonian Tube with an LED shining down it.

 

[Andy Resnick]

I keep reading the general opinion that the improvement is just not worth the risk if you attempt to clean the mirror on a Newtonian. Clearly, there must be a point at which it really is an important thing to do.

Jumping in late...

One thing to remember is that it matters where the dust is: if it close to a pupil plane, the effect on image quality is (usually) small. If it's near an image plane, the effect is large. Qualitatively, I expect the effect of dust to be approximately the same as a scratch/dig specification (mentioned in post #9), and as can be seen from the link, there is no clear quantitative metric correlating scratch/dig with image quality. FWIW, a good surface has a scratch/dig of 20/10, while an 'average' surface is closer to 60/40, but I'm not sure how those specs relate to dust contamination.

Somewhat related- I've used microscope objectives that are hideously fouled, and the accutance is definitely lower than when using clean lenses. Your rough calculation is reasonable, and since the primary mirror is essentially the entrance pupil, I'd expect dust and scratches to weakly impact image performance. As for cleaning, a lot depends on your experience and ability- for example, can you disassemble the mirror and properly reassemble it? Do you know how to clean a metalized glass surface without flaking off the metal?

It's probably ok to perform a light cleaning- blowing some compressed air to remove the loose stuff- but not worth extensive cleaning, because if you try and clean the mirror and accidentally leave behind a bunch of parallel scratches/grooves, you have just created a diffraction grating in the worst possible location. Besides, unless your primary is sealed behind a window, dust is going to start accumulating immediately.

Bottom line- there's no 'theoretical' way to decide that your mirror (or optical system) is 'too dirty'- you decide based on the performance.

 

[Chronos]

Two of the most important factors in the quality of an optical surface are reflectance/reflectivity and scattering. There are some very good and reasonably practical ways to quantify the reflectance/reflectivity of an optical surface. See http://ricmorte.com/index.php/light-a-colour/optics/reflectance-a-reflectivity for discussion. The more critical factor in image quality is scattering. This is a largely a function of surface roughness which is obviously affected by surface contaminants and finish. This is termed total integrated scatter [TIS] and is discussed here http://photonics.intec.ugent.be/education/ivpv/res_handbook/v2ch26.pdf. There is a device called a scatterometer or scatteroscope that can quantify this measurement. It is, of course, expensive and requires an [also expensive] expert operator and controlled environment to acquire reliable measurements. See here for further information https://www.researchgate.net/publication/252182004_Scatterometer_and_scatteroscope_for_testing_optical_surfaces .

 

[sophiecentaur]

I'd expect dust and scratches to weakly impact image performance.

I does seem to me that the effect of dust on a mirror surface is different from the effect on a lens because of the back scatter. A proportion of the incoming light will end up spread all over the image plane. Does this make sense? Will this not have the effect of raising the mean black level in a different way from what would happen when transmission is reduced - as in a lens? I wasn't looking for distortion - just a reduction in contrast.
But, as I always say to people with dodgy ideas "the numbers count" and I will need to measure something to back up my suggested figure for non-specular reflected light at the mirror surface. I will have to photograph the mirror surface and 'count the dust specks', for a start.

 

[Drakkith]

I might be taking out my backside, but here are my thoughts:

I does seem to me that the effect of dust on a mirror surface is different from the effect on a lens because of the back scatter. A proportion of the incoming light will end up spread all over the image plane. Does this make sense?

Of course. But I doubt it's as bad as you think it is. The cone of scattered light from any particle of dust expands as it travels. Since the primary mirror does most or all of the focusing (and can't bring the light scattered towards the mirror's surface to a focus since the dust is so close), this cone continues to expand on its way to the secondary and then to the sensory/eyepiece. And if you have a cassegrain or other type of telescope with a negative-power secondary, the light reaching the focal plane is reduced even further.

But, otoh, regular photographers spend all their time keeping their lenses spotlessly clean - and they don't even suffer from the back scattering that my calculation deals with.

Perhaps, but I suspect that daytime imaging is different from astrophotography.

Most astrophotography images are of very, very dark backgrounds peppered with bright points and possibly some diffuse objects. In this situation a small portion of scattered light might add, what, one or two dozen photon counts to each pixel in the sensor array? Hardly detrimental to image quality when the difference between the background and all but the faintest objects is hundreds or thousands of counts, and the background itself is already several hundred to several thousand counts from dark current and other factors (upwards of ten or twenty thousand counts from light pollution if you're anywhere near a city. Which, now that I think about it, really doesn't factor into this. If you're imaging near the city you're probably more worried about how to get rid of all that darn light than you are about a bit of dust!). Very faint objects are already in danger of being buried in noise and usually can't be seen anyways. Plus, if you're really trying to pull in those faint details you can always take and integrate more images to make up for slightly dirty optics. Resolution and image sharpness are almost certainly limited by seeing and not surface irregularities and galaxies and nebulas are already "fuzzy" to begin with.

Visually, these objects are already so close to the background brightness that other factors such as seeing and light pollution are much more dominant in determining what you can and can't see. When I lived in Shreveport, I would often take my equipment about 20 minutes south of the city to the local club's observatory. The "clarity" of the atmosphere was noticeably different from day to day, especially after it had just rained and all the gunk in the air had been cleaned out. Seriously, the light dome from the city seemed to shrink in half after a good downpour!

Daytime photography typically involves exposure times of less than a quarter-second or so. The background noise is dominated by sensor bias and the dark current is negligible, unlike in astrophotography. The dominant source of noise in these photographs is shot noise from the very, very bright scene. While the contrast between a dark and light portion of a scene can still be extremely high, a much larger percentage of the scene is "bright" compared to that of an image of the night sky where there are, at most, a few hundred to a few thousand bright spots in the form of stars. So the amount of light coming into the camera lens is probably much higher than that of a telescope, even taking into account all the light you had to bring in and integrate to get a decent astrophotography image.

There's also a problem with glare during the daytime. A photograph with the Sun shining off of a car window or a surface of water in the background isn't that big of a deal if the light is confined to landing on the sensor. You may saturate your pixels in that area, but so what? However, with dust and scratches on your lens, even a small portion of that light being scattered across your sensor may be enough to noticeably degrade your images. Even an image taken with "perfect" optics will have noticeable degradation if the glare is bright enough, so I assume a dirty lens just makes the problem that much worse. And I'm sure that if you ever get sunlight shining directly onto the lens surface itself, all that dust and any scratches suddenly become VERY bright and noticeable.

Anyways, that's my thoughts on it. Like I said, I could be speaking out my backside and I haven't ran any numbers, so take it for what you will.

 

[sophiecentaur]

The cone of scattered light from any particle of dust expands as it travels.

I read this comment and it made me think but . . .. I assume that the secondary is only as big as it needs to be and that it intercepts the cone between the image disc and primary. This implies that all the light landing on the secondary from the primary will hit the image. I calculate the fraction of light from the primary dust getting into the secondary with the factor:
A_s/2πd², which is the proportion of the hemisphere at radius D that hits the secondary. (There may or may not be an error because of the width of the primary but I don't think there is)
Any part of the image disc will be subject to its share of this backscattered light so it can't be eliminated by using a short EP lens, which will select only a part of the disc (i.e. magnifying the central part of the image). Any light sources that are admitted onto the primary will contribute to the background of a faint object so increasing exposure doesn't help with SNR and this ' noise' is DC, largely, and can't be reduced by long integration time / multiple images. This relates to several of your comments about the way images can be processed and it's an important factor in communications theory. You are stuck with any noise that falls in band - if you don't have the benefit of fancy coding and error correction when you can choose the signalling format. But my main interest here would be to improve the visual performance

Most astrophotography images are of very, very dark backgrounds peppered with bright points

Yes, I agree and that constitutes a high contrast scene. It's true that the proportion of bright sources, in most directions is small but, otoh, we are trying to see a huge range of brightnesses - greater than the range we would expect in a normal photograph. There's a difference between day and night vision and a 'good TV display will have a contrast ratio of around 500:1, which would correspond to a relative magnitude of around 7. Now, 500:1 is only near enough for satisfactory viewing and we can detect inadequate blacks on TV pictures under good viewing conditions. But I have to insist on the importance of actual numbers here if I am going to come up with an answer to the question "How bad does it have to be?" What I am attempting to find out is how much (numerically) does the telescope manage to reduce the effect of the dust compared with how it looks when we look down the tube.
I take all your points about the extreme requirements when attempting to take photographs with a strong sun, even when it's off axis (hence lens hoods). The demands are different.
With great respect (really!) astrophotographers are quite shameless in what they do to their images on the way to a final presentation. The results are often stunning and lovely to see BUT many of them could never be used as the basis for actual measurement. The eyeball is also guilty of cheating about what it sees but I can accept that.
So far, the 1/1000 figure from andrew s 1905 is the only actual number that's been submitted. I would like to know just what that looks like on a mirror surface - it corresponds to about 1/30 on a line. which would be a 1mm dot, every 3cm. That seems pretty fussy (the argument seems to be in favour of my attitude - for once)

 

[Andy Resnick]

But, as I always say to people with dodgy ideas "the numbers count" and I will need to measure something to back up my suggested figure for non-specular reflected light at the mirror surface.

Of course. But I doubt it's as bad as you think it is. The cone of scattered light from any particle of dust expands as it travels.

Conceptually, I think of dust and scratch/dig in terms of rough surface scattering:

http://gea.df.uba.ar/giambiagi/material/mendez_c01.pdf

 

[Drakkith]

I read this comment and it made me think but . . .. I assume that the secondary is only as big as it needs to be and that it intercepts the cone between the image disc and primary. This implies that all the light landing on the secondary from the primary will hit the image.

Hmmm. I treated a dust particle like a new point source on the surface of the primary. However, Andy's link made me realize I really don't know how light is actually scatted from dust.

But my main interest here would be to improve the visual performance

Ah, okay. I thought you were talking primarily about astrophotography.

But I have to insist on the importance of actual numbers here if I am going to come up with an answer to the question "How bad does it have to be?" What I am attempting to find out is how much (numerically) does the telescope manage to reduce the effect of the dust compared with how it looks when we look down the tube.

Well, I don't think I can help you with numbers, as I lack the knowledge and/or the equipment to do so.

 

[sophiecentaur]

Well, I don't think I can help you with numbers, as I lack the knowledge and/or the equipment to do so.

But there must be a man who can, somewhere. They don't instal super large telescopes without having a good idea about how often they need to be checked for dirt and cleaned. Apparently, when it's really important, they will often re-coat the mirror, rather than giving it a lick and a promise. Such an expensive process will only be carried out when it's really needed. That decision will be taken on the basis of performance and inspection of the mirror.