Jump to content

Banner.jpg.b89429c566825f6ab32bcafbada449c9.jpg

SXVF-H9C - Low Sensitivity


Recommended Posts

Ever since I upgraded from an MX7C to a SXVF-H9C, I've always thought the H9C to lack sensitivity, compared the MX7C.

As time has gone by, the evidence just keeps stacking up, to support the view my particular H9C is way below par, when it comes to sensitivity.

Not having another H9C to compare it with, I can only judge by the results I get, and compare them with images taken with other H9Cs, using similar scopes and exposure times. In fact my H9C has a 'head start' in that from my location, I enjoy relatively dark skies.

Other makes of camera, employing the same 285 CCD, also seem to produce far less noisy results.

So, my question is, is there anyway that I can measure the sensitivity/gain of the camera, against some sort of specification 'yard stick' ?. I do have a good assortment of electronic test gear.

Dave

Link to comment
Share on other sites

  • Replies 42
  • Created
  • Last Reply

I've just received a response from Terry Platt, suggesting a couple of quick assessment tests, which I going to try.

However, if we have any 285 based CCD camera technical exeperts out there, your input would also be most welcome too.

Dave

Link to comment
Share on other sites

the quantity you are wanting to know is quantum efficiency. You can measure gain, but that doesnt really tell you how sensitive it is.

To accurately test the QE you need a calibrated Silicon photodiode and monochromator.

It is unlikely you have one, but a university might

or...

find someone near you with an SXVF H9C camera and ask them to bring it over.

Without a lens..ie chip exposed to outside environment, point both cameras at the same part of a even illuminated surface.....like taking a flat, but with no optics.

Take a series of exposures from very dark virtually no signal to overexposed (pure white) with both cameras pointing at a flat illumiation.

Take around 20bias frames for each camera and subtract this frame from all the light frames.

Check that the frames are virtually flat...no obvious gradients etc.

Crop all images from both cameras to about 40x40 pixels roughly in the centre.

So you know have a series of flat images from black to white, cropped to 40x40 from both cameras.

Using an image processing package measure the average intensity of each box, and measure the standard deviation. Most programmes measure the average and std dev using the histogram option.

Measure the average and std for all images.

plot std dev on the y axis and average on the x axis.

Use log scales for x and y.

You have made a PTC (photon transfer curve), which will give you gain and pixel response non uniformity info.

On the same graph add a line with equation y=a*sqrt(x).....this gives a straight line.

vary the value of a until it matches well the area of the graph around several thousand signal...this is the shot noise area.

equivalently, find all points that behave according to the equation above. Using only the points that lie near this line, use the equation K= S/ N^2...where S is the signal of the point and N is the noise.

when the line overlaps some points, it will be evident which points obey the shot noise...slope 1/2 or read noise flat slope or fxed pattern noise slope 1

K is the gain.....it converts intensity to electrons. This will vary from camera to camera as you know

using a similar technique, fit a line with equation y= b*x.....this will be found with points with high signal, like >10000 DN until full well.

The value of b is called the PRNU...pixel response non uniformity. In this region noise scales linearly with signal, and so incresasing signal gives no more SNR increase leading to a poor image.

infact SNR max = 1/PRNU...a prnu of 0.07 (7%) gives SNR max of just 14. Flat fielding your camera removes PRNU almost completely and so high SNR is possible.

You know have the gain and prnu of your camera.

Go back to the cropped images and multiply all of them by the gain...the image now is in terms of electrons.

Do the same technique for he other camera. You have made 2 PTC and have a value of gain and PRNU for each camera. You have now multiplied your images by the gain of your camera and similarly for the other camera.

Now plot the average signal vs exposure length for each camera.

This should be a straight line. Fit this line in excel. The gradient is important. It is electrons per second. If the same camera is used with the same exposre the same illumination, then everything should be the same. Compare the gradients. The lower the gradient the less sensitive your camera.

If your gradient is lower and the PRNU higher than your friends, then there is good reason to believe you camera is a bit dodgy.

That would be the most accurate, but also quite diffiecult way of doing things

Good luck

Paul

PS simply applying good flats to your images might make the difference.

Link to comment
Share on other sites

Hi Paul, thank you for that very in depth explanation, which I will take the time to digest, it is much appreciated.

The main 'fly in the ointment' is that I'm not aware of anybody with an H9C, that lives even remotely close to me, and you are correct, I don't have the necessary test kit to measure the QE :) . Oscilloscope, DVM , RF Generator, AF waveform generator, digital frequency meter, etc, etc, I have, but not the Photodiode or Monochromator :(

However, following the response from Terry Platt (he designs the SX cameras) and having just carried out the tests he suggested, it has perhaps revealed some shortcomings.

It would seem that the brightest stars in my images are only 'hitting' around 54000 ADU, whereas they should 'hit' 65535 ADU.

Also the Standard Deviation (in ADU) of the noise in the bias frame, is showing at 26.5, whereas according to Terry, it should be at around 19 or 20.

However, I think the fact that the camera is only reaching reaching 65535 ADU, is the root of the problem. This means that its falling short of its full dynamic range.

I've reported this back to Terry, and will await his response.

Dave

Link to comment
Share on other sites

the fact that you are not reaching full range of the ADC is becuase you are hitting full well before ADC saturation.

you need more Dn/electron, which corresponds to a lower value of gain. You gain is too high.

you are right, the gain isnt optimised for high dynamic range.

thats my thoughts.

Link to comment
Share on other sites

As you might expect, if you have ever had dealings with Terry Platt of Starlight Express, his repsonse was swift and helpful.

I have just sent Terry a couple of subs from the camera, and if necessary, Terry has offerred to take a look at camera next week.

You can't ask for better service than that.

Buy British!!!

Dave

Link to comment
Share on other sites

Same here!!

I managed to get an adapter stuck fast on the front of the camera. No amount of pressure would relieve it - but the guys at SX did, within a couple of days as well!!

Try getting the same service from Hewlett Packard! Pah!!!

Link to comment
Share on other sites

Ok, the latest news then :(

Terry has had a look at the files I sent him last night, and has e-mailed me back this morning , quick eh!!.

The readout noise in the bias frame is maginally higher than it should be (26 ADU instead of 20), but apparently, really not enough to affect the image, and certainly not the effects that I'm seeing. In fact, it means that the gain of the camera is set very slightly on the high side, which is contrary to my feeling that the camera is down on sensitivity.

However, the dynamic range of the camera is falling some 11000 ADUs short of the full dynamic range.

Bright stars should peak at 65535 ADU (peak white), and anything above that is just clipped, which is of no real consequence, as we can't see further than white anyway.

So, Terry suggests that the anti-blooming gate adjustment is set too low, and the bright stars are certainly not saturating the A-D converter.

So, as I made up a little 'test jig' (a box with a fibre optic light source at one end) to set the anti blooming gate level, when I converted my old MX7C to mono, I'm going have a 'play' with this setting on the H9, and see if I can raise the ADU level to 65535, without introducing blooming.

This will hopefully increase the dynamic range of the camera to where it should be. If all fails, then I'll send the camera to Terry, and he'll have a look at it.

Watch this space :)

Dave

Link to comment
Share on other sites

Well, it seems that adjusting the ABG 'pot', 180 degrees in either direction, is having no effect at all, so a fault condition lays beneath it would seem.

I've just sent the FIT files off to Terry, for him to have look at.

When I did this adjustment on the MX7, as you say Steve, it was quite critical, but not difficult to do. Moving the ABG 'pot' on the MX7, made significant changes.

Further instalments in due course. :(

Dave

Link to comment
Share on other sites

Ok, futher experimentation with the camera attached to my 'test jig', and tossing thoughts around in my head :? , has maybe got me thinking along he right lines, as to the cause of the problem.

Firstly, when I inceased the exposure time up to 3.5 secs (the MX7C only needed 1 sec to 'hit' 65535 ADU), I managed to get the H9C to 'hit' 65535 ADU, when the ABG setting 'pot' was adjusted (couldn't get any sense out of this earlier, as I hadn't realised it would need so much more exposure time).

However, there should be a point when 'blooming' starts to appear, and it doesn't, irrespective of how far the 'pot' is turned up.

Now when I bought the camera new, it had a 'steaking' problem (blooming) on bright stars, and Terry Platt advised me to add a 4.7nF tantalum capacitor, across a smaller value surface mount capacitor, to increase its value.

This was an easier option than removing and replacing the surface mount cap. BTW, I should add that Terry was more than happy for me to return the camera for him to do the job, but was happy for me to do it if felt confident to do so.

Now, following on from an exchange of PMs with Steve (steppenwolf), who has both an H9C and a 25C, he tells me that he also had this 'streaking' problem, but this was resolved by adjustment of the ABG setting.

So, my poor old brain 'kicked into gear', and started thinking :scratch: , is this additional capacitor 'padding' the ABG circuit too much, as indicated by the fact that turning up the 'wick' on the ABG, doesn't push the camera into 'blooming', which it should do.

The 'bottom line' is, did my camera leave the SX factory, with full sensitivity, but also with the ABG level set just too high, and then my adding the recommended capacitor, knocked the sensitivity back to where its been since that time ? :scratch: .

Without doubt, ADU measurements taken from the big stars around IC434, that I imaged a couple of nights ago, were only showing 54000 ADU after a 600 sec exposure, when in reality they should have been 'topping out' at 65535 ADU, or may be even 'clipped' at that point.

So, I've just e-mailed my thoughts to Terry Platt, to see if my logic makes sense, or whether I'm talking rubbish.

Dave

Link to comment
Share on other sites

Dave,

If you are correct, this will be the most complicated "user error" I have ever encountered. :(

Assuming you are correct, presumably the cutters will be out, you'll have a spare capacitor and a more sensitive camera?

Mike

Link to comment
Share on other sites

Well it will be nice if it does turn out to be as simple as removing, or changing its value.

Albeit, it will mean that I've struggled with this camera almost from day one, as a result of installing an SX recommended modification, which cured one problem but created another. :(

However, I'll wait and see what Terry has to say, before proceeding.

Dave

Link to comment
Share on other sites

Dave, having moved from a b/w camera years ago to the SXV-H9C, I noticed an immediate difference between the two.

The main thing I had to come to terms was, in order to shoot colour images I would sacrifice sensitivity due to the bayer matrix BUT given the fact that I would have about 2 - 3 hrs MAX per night meant I could get ALL the colour data for an object that night.

Resolution is also compromised as LRGB typically gets all the resolution from the L channel & the RGB can then be binned to get more sensitivity for less time...

All in all, the one shot colour will not stand up to (L)RGB EVER in a 1 - 1 comparison - the benefits are elsewhere...

I am not sure if this rambling is of any help but - the main thing I noticed:

Need more exposure time compared with b/w cameras doing the equivalent (L)RGB

I have MANY images for comparison - if you think it would help, I can provide any number of images to see what you did at that time.

Phil.

Link to comment
Share on other sites

Hi Phil

I've gone from OSC imaging, to mono + LRGB, and back to OSC again, so have a very good feel for the difference in sensitivity.

Compared to my MX7C, the sensitivity of this particular H9C, is way down, and I've thought this right from the start, but without another H9C to compare it with with, I've never been able to prove it.

As Terry Platt points out, the MX7C uses a CMY filter, which passes more light than does RGB, and it does have larger pixels. However, IMO, that still doesn't account for the degree of sensitivity loss than I'm seeing.

I'm convinced that this H9C is way down on what it should be.

Dave

Link to comment
Share on other sites

Well, well, well, :( , I lifted one leg of the SX recommended additional capacitor, and the camera now 'hits' 65535 ADU at around 0.5 secs, as compared to 3.5 secs that it took before.

So there is non doubt that adding to the additional capacitance, to resolve the 'streaking, (blooming) problem, as advised by SX, had as I suspected, reduced the dynamic range of the camera significantly. Albeit, it did resolve the 'blooming' problem, that the camera had when bought new.

Of course what I don't know yet, is if the camera will once again 'bloom' on bright stars, and if so, can I then adjust the ABG 'pot' to set the threshold at the right point.

Using my 'homebrew test jig' which uses an optical fibre as a point light source, and a pin hole in a lens cap, on the front of the H9C, I have no idea how this compares to the light from a bright star. So its difficult to know, just how much to turn the 'wick' up or down. I do know that there's a very sharp cut off, between a very noisy signal, and good one that 'hits' 65535 ADU, and not a gradual transition, as I would expect. This change change/flip point represents a VERY tiny movement of the 'pot'. Which means that if the camera 'blooms', when set just on the good signal side of this 'break point', it is going to be impossible to set effectively. On the other hand, if the capiacitor goes back in, the 'blooming' will go, and along with it the dynamic range of the camera. Catch 22 ?.

So, although I am a very patient bloke, and a great fan of Starlight Express products, This SXVF-H9C has really put my patience, and loyalty, to the test.

First the 'blooming' problem with the brand new camera, then I discover it leaks light into the CCD chamber, via the rear panel, and worst of all, I've struggled with this lack of sensitivity issue for two years now, and in doing so, wasted two years of imaging time, cloud permitting that is.

This does not detract from the help/advice that Terry Platt has provided, which he does so willingly, but although I'm relucant to say it, I am disappointed, and that's being polite.

To keep this in context, I am not denegrating the SXVF-H9C, its a very good camera as shown by the many excellent images taken with them, and posted/published.

Its just that I seem to have paid £1500 for a 'friday afternoon' camera, and I guess the answer is YES, I am now a bit P***** off!! :grin:

There, I feel better now, and if the stars show tonight, I'll struggle on, take the camera out, minus its case, attach it to a scope, and fiddle some more. I must be a saint :) .

Dave

Link to comment
Share on other sites

Hi Dave,

I'm really glad you got to the root of the problem.

I'm sure that if you 'copy and paste' that last entry of yours to Terry Platt, he will be more than sympathetic and would probably replace the camera as a whole unit, after what you've been through.

I would hope that to be the case anyway.

Barry.

Link to comment
Share on other sites

Well, Dave at least you know the root cause but it remains to be seen if the adjustment to the 'pot' resolves the problem.

As you say, the results obtained by many imagers from the H9C testifies to its quality but you do seem to have been very unlucky with this one and I have wondered for some time why you weren't singing its praises but had put this down to the fact that you have also owned a mono version - an advantage I have not had so I have nothing to judge mine by but I would say that mine has met my expectations completely. As a matter of interest, I did have to send mine back to SX for a capacitor mod. as after several months of ownership, I noticed that I had a small black 'cutout' to the right of all my really bright stars - the capacitor change resolved this to my complete satisfaction.

With regard to the M25C, I had the opposite problem - a bright streak to the right of really bright stars - but this was resolved by adjustment of the 'pot'. In fairness to the company, there is a fine line between 'bloom and gloom' as I guess these types of camera in this environment are operating on the edge of the sensor's performance envelope?

I really hope that the problem is finally put to be so that you can get on and enjoy what is genuinely a really great camera.

I know you'll keep us posted.

Link to comment
Share on other sites

Hi Steve

I mentioned to Terry about your 'streaking' problem being resolved by adjustment of the ABG setting, and he said that there are a number of things that can cause 'streaking' in fast downolad cameras, and that yours was most likely the result of something different to mine.

I have a great respect for Terry's knowledge, and so I'm quite happy to accept what he says.

I had another reply from Terry this evening, and he says that the capacitor should be left in place, and that I should try and make adjustments, in a telescope and star situation. Albeit this is not so easy as making finite adjustments on the bench.

I have said that I will do this, although looking outside, goodness knows when :( .

I have also relayed my wholesale disappointment and frustration, with this particular H9C, and annoynance at having wasted so many imaging hours, over a two year period, without getting the results I should.

As I said before, I have a lot of respect for Starlight Express products, and they are small BRITISH company that deserves our support. However, in fairness to me as a customer, the camera should not have left the factory with the 'streaking' problem in the first place, and I shouldn't have needed to make a modification to the camera to stop light leaking in.

I guess its a matter of all these things adding up, and the poor performance of this particular camera (and I stress, this particular H9C), that have eventually worn me down. :)

Dave

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue. By using this site, you agree to our Terms of Use.