Notes on PC194C and 1004x ccd camera's

Notes(2) on PC164C and 1004X CCD Low Light camera's

From: "Andrew Schott" Date: Sat Feb 23, 2002 8:26 am Subject: Re: [QCUIAG] Re: Finally, the ExView mod revealed! There's no real "back" to the PC board in the PC164C. There are four small boards with three of them held together in a small 'H' shaped assembly with input (CCD) on one end and output on the other. I've looked at your 1004X mod with the hopes that the conversion of the PC164C would be very similar but I find the parts count of the PC164C seems to be much higher. There's lots more resistors and caps but that might not be a problem. I've taken some half ased shots of all the guts and attached them. I hope you'll find them informative? Helpful? Blurry? Since the boards aren't 'flat' I couldn't get any decent images on the scanner. These were shot with a PC23C and I think it was freaking out. It probably thought it was witness to a vivisection. Disassembly instructions: With proper sized jeweler's style screwdriver, back out the four small setscrews visible in the flats machined in the camera housing. Screw them out far enough to disengage the front thread ring and rear connector assembly. The rear connector assembly is simply extracted by carefully pulling the cap off the body, allowing the four connectors inside to disengage. The front thread ring is worked out of the threads as best as you can due to the setscrews that have crunched up the threads somewhat. It might not even be necessary to remove it if you can reach the four screws surrounding the CCD window. My smallest Phillips style screwdriver had a shoulder on it that wouldn't allow me to reach these screws directly so I just pulled the ring and had plenty of room. After you remove the four screws the inner module will slide out under it's own weight. If you remove the four screws first, the module will come out with the rear cap when it is extracted from the housing. Remember, it's best to work over a static free or static grounding padded worksurface to avoid damage to the exposed components. (I know you know but what about the other guy:-) First the input end.... The chip sits up on top of a 2mm spacer of some kind. Maybe a thermal shield so the middle board won't heat print across the center of the chip? Arrow silkscreen in upper left points to top of unit. Top of middle board. White blurry thing is crystal reflecting IR from light in kitchen. CCD is to the left. Bottom of middle board. Note white "insulator" under CCD on left. Colors in image caused by aliasing in Matrox Marvel capture function. Rear connector/output board. SMC on both sides, mostly small stuff. Front of rear board... You know, the side opposite from above... as seen from below... awww... more parts. Top of the front of the rear board... or other half of above.... Finally, the output socket board/cable/iris connector assembly. ----- Original Message ----- From: Jon Grove To: 'QCUIAG@yahoogroups.com' Sent: Thursday, February 21, 2002 10:30 AM Subject: RE: [QCUIAG] Re: Finally, the ExView mod revealed! Thanks - I think this group is all about sharing experiences and specialised knowledge so that more people can do astrophotography 'on the cheap'. Hopefully this mod will be of use to some folks! I would guess the PC164C is similar to the 1004X in terms of circuitry, but the layout will be different. It may simply be a case of identifying where to attach wires etc and then the rest of the mod would be identical. Does anyone have a (fairly detailed) image of the back of the PCB that's inside a PC164C? Attachment http://www.nitehawk.com/rasmit/pc164/whole_works.jpg http://www.nitehawk.com/rasmit/pc164/business_end.jpg http://www.nitehawk.com/rasmit/pc164/mid_board_side_b2.jpg http://www.nitehawk.com/rasmit/pc164/midboard_side_a2.jpg http://www.nitehawk.com/rasmit/pc164/rear_bulkhead.jpg http://www.nitehawk.com/rasmit/pc164/front_bottom_of_rear.jpg http://www.nitehawk.com/rasmit/pc164/front_top_of_rear.jpg http://www.nitehawk.com/rasmit/pc164/rear_connector_board.jpg

From: "ssb73q" Date: Sat Feb 23, 2002 11:07 am Subject: Re: Finally, the ExView mod revealed! Andrew, you sure did a lot of good work there, thank you. I was very reluctant to disassemble my PC164C, you increased my confidence to proceed. I hope the PC164C is amenable to modification. Regards, Richard

From:"wlbehrens" Date: Sat Feb 23, 2002 10:54 pm Subject: Re: Finally, the ExView mod revealed! Yes,I believe it is a good canadate. First it uses the 1310AQ for the AFE and uses the CDX2463 for timing and sync generator. These are the important chips for the mod. The supporting "extra parts" are for auto iris and the external controls for other camera functions. I assume the camera has external controls for sync, gain, shutter ect,ect. Now the bad news is you would more than likely have to remount the "guts" in another enclosure for the daughter board and cooling to fit. You could salvage the end pieces and just attach them to a cast aluminum project box. William

From: "ssb73q" Date: Sun Feb 24, 2002 4:56 am Subject: Re: Finally, the ExView mod revealed! William, <> Does that camera need to be cooled? A 1.7 second shutter speed would give 100 times the sensitivity, 8.3 seconds 500 times the current sensitivity. Do those shutter times need cooling? If the other boards in the PC164C are just for the auto iris, et al., can't removing them provide the room for a daughter board??? Regards, Richard

Richard, The daughter board could be cramed into the camera I suspect with some work. When I think of making a camera mod I always gravitate (no pun intended) to a cooled unit with long exposure capability. With integration software like AstroVideo you wouldn't need to cool the camera, but Jon Grove did show that the hot pixels could be reduced a great deal by cooling it. By the way do you know what version of the CCD the PC164 has in it? There are two 1/3" models the low res and high res. William

From: "ssb73q" Date: Sun Feb 24, 2002 4:22 pm Subject: Re: Finally, the ExView mod revealed! William, <> The spec sheet say "Sony 1/3" Super HAD ExView CCD, NTSC 510(H) x 429 (V). BTW, see Andrew's post on the temperature rise of his PC23C camera. Regards, Richard

From: "Andrew Schott" Date: Sat Feb 23, 2002 10:55 am Subject: Cooling the PC164C. Now that I've gutted mine and reassembled it to functioning condition, I feel a little more inclined to modify it. Whil'st I wait the verdict from Jon on the electronical mods, I'm brainstorming the possibility of cooling this little critter to give it better S/N and in general, more stability. None of the internal components are in thermal contact with the outer housing so simply chilling the aluminum tube won't be enough to get at the real heat sources inside. Simple fan cooling is tough to accomodate due to the zero clearance design and the small package. Small volumes of air could be forced through the unit with a tube/pump but such arrangement might not provide enough cooling to make much difference. I remember once I saw a cooling arrangement for an experimental computer that used air forced through a copper coil immersed in ice to augment the cooling properties. I guess one could use something similar in this device. A small squirrel cage blower, like a dust buster, forcing filtered air into a flex hose leading to a big roll of copper tubing in a 5 gallon bucket or a large cooler filled with ice. The cold air would then be fed thru another flex hose to the camera housing. One expedient would be to remove the auto-iris jack and replace it with a nipple. Since Jon will be using this connector for the long exposure/AGC mod I'll probably use the unused tripod mount hole, and run the cooled air into the middle of the camera. I would situate the exhaust so the cooled air would be forced out around the CCD escutcheon, cooling the main chip on its way out of the camera. When used with screw in C mount lenses, the C/CS adapter could be ventilated to allow the exhaust out, and when used in a scope, the air would simply flow down the focus tube and into the scope. It remains to be seen whether or not this would effect seeing. I would bet it would so my first workaround would be to use an IR filter to block the flow and divert the exhaust elsewhere. The steady outflow of air around the CCD chip should prevent any dew formation, even here in deep in the heart of darkest Florida. The incoming air itself may actually be dried somewhat by the heat exchanger. Hmmmm, this superfluous condensate could cause problems if it should gurgle up the pipe and get into the camera. I guess I'll have to rig a "spit valve" to dump moisture occasionally. Not as complex as the Shuttle yet but we're getting there. These little cameras do get hot. Mine climbs to about 105° in less than 30 minutes of operation @70° ambient. That's a 35° heat rise! On our 95° summer nights here in Florida, this thing could get hot enough to melt cheaper barlows! All that heat can't be doing the images any good.

From: Jon Grove Date: Mon Feb 25, 2002 2:39 am Subject: RE: [QCUIAG] Re: Finally, the ExView mod revealed! Thanks Andrew, lots of useful pictures! Looking at the 'bottom of middle board' picture, there isn't an obvious component connected to pin 5 of the 2463R. It's possible that with a good lens you could peer underneath the chip and see where the track leads from this pin (especially if you've already lifted the pin) - alternatively a multimeter (set to resistance measurement) should help to locate a component connected to Pin5 (prior to lifting it). I would guess it would be connected to a capacitor or a voltage regulator somewhere. Failing that, it may be possible to solder a wire to the pad beneath pin 5 once it's lifted, but that would be very fiddly and prone to shorting onto other things. If that stage of the mod can be done, then the rest ought to be pretty straightforward. You'd need to locate a +5V supply somewhere on the PCB, and take a ground and +12V from the input connectors. Pin 2 on the CCD looks easily accessible, as is Pin 9 if you choose to do the amp-off mod. I'm pretty sure the mod is achievable, with a bit of investigation! The daughterboard probably won't fit inside the case as it looks pretty cramped. But I don't think it'd hurt to mount it outside on a short cable, if a suitable access hole could be found or made. There'd be about 10 wires needing to go through it. Jon.

From: Jon Grove Date: Mon Feb 25, 2002 5:29 am Subject: RE: [QCUIAG] Re: Finally, the ExView mod revealed! No, I've not done any current or temperature measurements (other than feeling the chips and thinking "that's warm!"). I wouldn't have thought the basic long exposure would affect the CCD temperature, since all it does is disable the charge transfer pulse which is only a few microseconds per frame. The amplifier-off mod may have an effect though, since it reduces the voltage supplied to the CCD. The specification says the CCD typically draws 3mA at 15V, which crudely works out at a resistance of 5K. The amp-off mod puts another 10K in series with it, which will drop the CCD voltage to 5V, and so the heat generated in the CCD will drop to 1/9 the 'normal' level. So in principal the amp mod could help the CCD run cooler. In practice I think it will be warmed up by all the other components around it. It would be interesting to measure the current vs supply voltage characteristics of the camera as a whole though, since it may help to reduce the temperature by dropping the supply a bit. When I get a chance I'll give it a go. Jon.

From: Jon Grove Date: Wed Feb 27, 2002 1:42 am Subject: ExView dark/cooling test Last night I tried some experiments with the ExView camera to see how cooling it affected the dark current/noise characteristics. The results are attached - each swatch is a 100x100 section from the top left of the full 640x480 frame. I switched the camera on, uncooled, and took a sequence of three 4-minute exposures with the 'lens cap' on. (The front of the camera is a sawn-off film cannister, the lens cap is the lid from one but it fits onto the sawn-off end so isn't very positive). As you can see, the noise levels gradually increased as the camera warmed up, although the hot pixels remained largely constant. The background grey level is because the lens cap wasn't very light-tight. I then switched on the Peltier and took another three 4-minute exposures while the camera cooled down. as expected, the noise level decreased until it was better than the first uncooled frame. Also the hot pixels seemed to improve. I then realised that light was leaking in to the camera so I turned off the room lights and put a cloth over the camera, and took another 4-minute exposure. This looked pretty much like the previous one only darker, as expected. So I tried a 10 minute exposure and whilst there was some noise visible I was quite surprised at how dark the frame was. Finally I turned off the Peltier (which would allow the CCD to warm up, possibly to a higher temperature than it would normally reach because of heat conducted from the hot side of the Peltier) and took another 10 minute exposure. This was very noisy as you can see. So it appears that it is worth cooling these devices for long exposures - even if only by using ambient air. It's interesting to note that the noise has a very similar pattern in each of the images, which indicates that it might be largely treatable by darkframe subtraction. I may need to do some more experiments to determine how much of the 'noise' is random (thermal) and how much is fixed (dark current). Food for thought, anyway. Jon Grove.

I think you're right - you could probably go to about 5 mins uncooled and use postprocessing to reduce the effects of dark current. You'd have a higher noise level on top of the dark current, which you couldn't get rid of (other than by stacking/summing/averaging multiple exposures). But I need to do some more experiments to work out how much random noise there is in the warm vs cold exposures. Good point about the flatframe generator - I hadn't thought of it like that! Jon. > > Jon, very interesting experiments, thank you very much. I find the > very early pixel saturation of the ExView HAD interesting, cooled or > uncooled. You experiment confirms my conjecture about an uncooled > ExView camera being useful with dark frame subtraction to ~5 minutes. > If the cooled camera only gives 2-3 times the uncooled exposure, is > the effort and complexity of the peltier cooling the camera worth it? > > Again, thank you for your time doing these experiments, it is very > useful work that allows us to make decisions based on knowledge, not > wishful thinking. > > BTW, it looks like your "light leaking cap" makes a nice flat frame > for testing exposure intensity as a function of time. That test may > be useful for estimating CCD sensitivity linearity. > > Regards, > Richard > >

From: Jon Grove Date: Wed Feb 27, 2002 5:23 am Subject: RE: [QCUIAG] Re: ExView dark/cooling test Hi Gene [and in spite of your misspelling my name, I won't comment on the sound of that greeting as I'm sure you've heard it too many times ;-) ] I don't know how cold the CCD gets yet. I bought a cheapo digital indoor/outdoor thermometer and one day will attach the sensor to the cold finger near the CCD, but I haven't tried it yet. A layer of frost forms fairly quickly most of the way up the cold finger, but I've never noticed it forming near the CCD. So I'd guess it runs at a bit above freezing, but not much. I'm not running the Peltier at full power though, since the PSU I'm using won't supply enough juice. Regarding noise levels, I just did some quick'n'dirty experiments with the dark images I took yesterday. I averaged them all (except the last one) to produce a 'master dark frame' with, I hoped, less thermal noise than any of the individual frames. I then scaled and subtracted it from image 3 (warm) and image 7 (cold), adjusting the scaling so that the result was as flat as possible. It was done by eye so was rather approximate. Naturally the warm image required a bigger multiple of the darkframe to be subtracted than the cold one did, by a factor of about 3. The residual noise had an amplitude of around 10-15 in the warm image and about 5 in the cold. So it looks as though, roughly speaking, cooling the thing cuts the repeatable dark current and the random thermal noise to about a third, maybe less if I do the analysis more accurate ly. Reducing noise levels to a third would also be achievable by stacking 9 images, so the question of whether it's worth cooling maybe boils down to whether you have the patience to take 9 times longer per imaging session. Personally, I'd rather use the cooler, since it's already built! But it's probably worth investigating less invasive methods of cooling the CCD such as sucking cold air around it, or cooling the metal lens-holder that comes with the camera, or even making a snug-fitting cooled metal clamp that fits to the ends of the chip.

From: Jon Grove Date: Fri Mar 1, 2002 2:06 am Subject: RE: [QCUIAG] Not enough images in astrostack Surely it's not possible to give a hard and fast rule of the form "Using the best x% of your frames will give the best resulting image". If you have very stable seeing then the main difference between each image will be noise. In that case, stacking a large number of images will reduce noise, and leave the image detail unchanged (since the same detail is present in the same place in all the images) If the seeing is variable, as is more often the case, then the best image might contain good detail, but as you select more of the 'next-best' images the true detail will be both blurred and displaced relative to the best image. As more of these images are stacked, noise will decrease as before, but so will image detail - because the underlying (noise-free) image is of poorer quality than the best one. Even slight movement of features caused by turbulence will have a blurring effect on the result after stacking. I feel it is a subjective thing. In a given set of raw images, you can choose how to trade off noise against detail. If the seeing was excellent you may be able to tip the balance heavily towards detail, by stacking more images. If it was variable then you'll have to put up with more noise to keep the detail, and stack fewer images. Wonder if it would be possible to have a program which, given a sorted (either manually or automatically) set of raw images would start with the best, and keep stacking images, monitoring the 'quality' of the result, until that quality started to deteriorate. The 'quality' metric would have to be supplied by the user, as a tradeoff between noise and detail. Of course it's hard for a program to tell the difference between noise and detail, so it may require human guidance. Could be an interesting project! Jon Grove.

From: Jon Grove Date: Fri Mar 8, 2002 9:18 am Subject: RE: [QCUIAG] ExView 1004X Hi Adam, How are you monitoring the output from the camera? Are you viewing it on a TV/monitor, or are you capturing it on your PC? I found that using a TV I would always see the 'bright' frame (except on very rare occasions, <1% when the timing was just wrong). Using a capture card in preview mode you may well not be seeing all the frames, and so the magic one might not be displayed. If you can capture an AVI at 25 [30 US] fps without dropping any frames then you ought to see the magic one in there somewhere! Jon. > -----Original Message----- > From: vilarestil [mailto:hodsonfamily@madasafish.com] > Sent: 08 March 2002 17:09 > To: QCUIAG@yahoogroups.com > Subject: [QCUIAG] ExView 1004X > > > Dear ExView users, > > When testing out stage 1 of the mod, I discovered that the 'magic > white frame' is not being generated every time I reclose the switch. > I tried to ascertain whether it was releated to time but it doesn't > seem to be the case. It's random whether I reclose the switch after 2 > or 10 seconds. > Have any of you experienced the same thing i.e. sometimes it > integrates, other times it doesn't? > I have checked my wiring so I don't think it's a loose connection! > > Many thanks, > > - Adam

From: Jon Grove Date: Tue Mar 12, 2002 1:44 am Subject: RE: [QCUIAG] "good frame" in 1004X mod. Sorry, been away from my PC for a few days! To capture a frame it is currently necessary that software can 'see' every frame as it is generated. It only needs to select the exposed ones to store in an AVI. I have a preliminary program that does all this work, and compiles an AVI consisting only of exposed frames. When the program has had a bit more testing, and if there is sufficient interest, I will make it generally available. Jon Grove.

From: Jon Grove Date: Mon May 13, 2002 8:06 am Subject: RE: [QCUIAG] Re: Why is on-camera integration superior to off cam era integration? From the online description of the 1268 Mintron http://www.mintron.com/HTM/News/HTM/Alpha.htm, it uses 'Star Light' technology. On a Mintron Q&A site http://www.mintron.com/HTM/Q&A/Htm/Star%20light%20mode.htm, I found the following info on StarLight mode (admittedly a different model camera): =============================================== Star light mode will make a CCD camera able to see clear color image under very low lighting situation such as 0.0002 lux illumination level. All CCD sensors were designed to work on 1/50,1/60 ~1/2000 sec. shutter speeds, hence the minimum illumination level or so called sensitivity is then limited on 3 to 6 lux with a F1.2 lens on 5600k Mintron proprietary DSP can drive CCD shutter speed to as low as 1 sec ~ 10 sec. As a fundamental of physics the longer the shutter is open, the more photons the CCD will capture, hence increased the sensitivity to 100 ~ 600 times more then a legacy camera. Mintron V series camera's are having 4 times better sensitivity on legacy mode then other camera's . Plus it will multiply light by 128X on the socalled star light mode, all these make V series camera's 512 times more sensitivities then all other camera on the market. Jaguar MTV-63W1 is also equipped with a 64X star light mode. How does the Star light mode kick in? Is it automatically ? How can I activate it? The user can turn on the Star light mode ( slow shutter frame integration mode ) On model 6318 / 6368 by the Dip switch No 8, and on the advanced models V and W series such as 63V1 and 63W1 the user can turn on the Star light mode by OSD screen and an interactive push bottom. If a camera has the Star light mode turned on as above, while the incoming light is getting weaker, the camera will automatically increase AGC gain to bring video level back to 100 IRE. After the AGC has pulled up to the maximum and the system still can not maintain 100 IRE, the frame integration feature will automatically kick in, the shutter speed will start to drop from 1/60 to 1/30, 1/15, 1/8 1/4 1/2, 1, 2 ,4 ,8 ... and all the way down to 1~10 sec. Due to the manufacturing methods of CCD sensors , most of the CCDs will have too much dark current noise on 64X multiply or 1 sec shutter speed. hence the 6318 and 6368 have max 64x integration. The new V series camera 63V1 / 63V0 use better selected CCD's and adaptive noise cancellation circuits , it's can go further up to 128X multiplication or 2 sec. slow shutter. hence it will add two times better sensitivity. In V series an even more advanced feature is added =============================================== From this description, it sounds very much as though StarLight mode actually does keep the shutter open for several frames, rather than integrating multiple 'normal' frames. In particular, the sentence about the custom DSP driving the CCD shutter speed to 1-10 seconds seems very indicative of on-chip integration. Thoughts, anyone? Jon.

http://www.sel.sony.com/semi/PDF/CXA1310AQ.pdf http://www.sel.sony.com/semi/PDF/CXD2463R.pdf

From: Jon Grove Date: Wed May 15, 2002 1:14 am Subject: RE: [QCUIAG] Re: Why is on-camera integration superior to off cam era integration? Hi Rein, You are largely correct in what you say below. For the sake of clarity I'll assume we're working with a 50Hz field (25 frames per second), and I'll describe the reasoning behind the 1004x mod. I guess this is similar to the Mintron, except that they have the luxury of designing their own version of the 2463R. The SUB pulse is of duration Td < 20ms, so that the interval between the end of the SUB pulse and the CT (charge transfer) pulse determines the exposure time. The 2463 can generate exposure times at certain discrete values, between 1/10000 and 1/50 second. The shutter speed is determined by a voltage that's derived from the previous frame, by integrating the brightness in a particular area of the image (the area is selectable from various patterns in order to allow for backlight compensation, etc.). Also, the video signal is sent through an AGC which allows finer control than shutter speed alone would give. So as a video image gets brighter and brighter, the AGC pregresively attenuates it until the unattenuated signal reaches a level where the 2463R decides to switch to a shorter shutter speed - at which point the AGC will compensate by increasing the gain - and so on. The SUB pulse can be disabled by applying +5V to pins 20 and 21 of the 2463R - this makes the shutter effectively fully open all the time, which is what is required for long exposures. Then the only requirement is to disable the CT pulses. These are also generated by the 2463R, and are derived from the +15V input at pin 5. (The SUB pulse is also derived from this voltage). So disconnecting pin5 disables the CT pulses. When pin 5 is reconnected, the next field will transfer charge from all the CCD array into a single field - adjacent lines are binned into one line in order to generate a field with half the number of scanlines as are present in the sensor array. Depending on whether an odd or even field is being generated, the binning is either lines 1, 2+3, 4+5, 6+7, ... or 1+2, 3+4, 5+6, ... This allows the full area of the CCD to be used each field, whilst preserving some of the resolution of the full array. Unfortunately it's not the behaviour that we want when reading out a full resolution long exposure, since reading (and binning) the lines removes the charge and there's nothing left for the alternate field - we get a frame in which every other line is a long exposure made from two lines of the CCD sensor. The solution to that problem was to prevent one of the sets of lines from being transferred when pin 5 was re-enabled. Fortunately the CT pulses for odd and even CCD lines are presented separately so we can disable, say, the odd lines CT by shunting the pulse to ground through a diode. Then when pin 5 is enabled, the even lines will be read out (and the CCD will try to bin them with the odd lines but the odd data will not be there). Subsequently we can re-enable the odd lines and they will be read out (and binned with absent even-lines) at the next field. So the whole array can be read out at full resolution, no binning, in two separate fields. It's a little bit complicated, but it works. Hope that's clear! Jon Grove.

From: Jon Grove Date: Fri May 17, 2002 1:28 am Subject: RE: [QCUIAG] Re: Why is on-camera integration superior to off cam era integration? Hi Rein, Some of your questions have already ben answered, but I'll answer the ones about the 1004x operation. The binning of adjacent scanlines is performed inside the CCD chip. Depending on whether an odd or even video field is being generated, the CCD bins in one or other of the patterns I mentions (1, 2+3,... or 1+2, 3+4...). The only way to get unbinned data is to fool the CCD into binning data lines with blank lines, which is what the mod does. When, for example, even data lines are prevented from being transferred to the shift register, the CCD will produce a field consisting of only odd data lines. However this field may be presented as an odd OR even field. The iCatch program knows which set of data lines it has requested, and when it receives a frame in which alternate lines are bright and dark it keeps the bright lines (which can be odd or even lines) and then requests the other set of data lines in a new frame. Again it grabs the brightest set of lines from that frame, and reinterlaces the two sets to produce a full-resolution image. It's complicated to descibe, but not too hard to write the program. Are you a programmer yourself? If you can grab the two frames into an AVI then you don't have to use iCatch to re-interlace them, it can be done manually with a graphics package. But it's a bit laborious! What sort of device is your Snappy - is it a USB or a PCI (internal) capture device? Jon.