Monday, March 31, 2014

RA drift might be flexure after all

Measuring Flex

There was a thread on Cloudynights recently that got my attention:

In the discussion, a user talks about a slow creep in RA (with some variance in DEC).  Sounds pretty familiar!

In the thread, a person linked to a review done by Craig Stark (of PHD guiding fame) where he looks at the Borg 50mm guidescope using the XY stage.  In it, he describes a test to determine differential flexure where he images a distant daytime object at one extreme of the scope position, then slews to the other side and images again.  The idea is that the differential flexure between the two imaging cameras would create a double image. 

Friday, March 28, 2014

Battery Pack for the Pentax 67

This pack came with a kit of Pentax gear some summers ago.  I finally put it to use last year at Calstar 2013.

Nightfly asked to see it in more detail and so here are the pictures:

The T-shaped device is the Pentax branded plate that allows the power cord to exit from the battery compartment.  This was important when using the camera on a dovetail where the whole camera base was flat against a metal plate.

Wednesday, March 26, 2014

Flare Testing

How well did the flocking work?

Last week during the clear nights, I was able to get some images of the night sky from the back yard.  I was taking some pictures near M53.  There is a somewhat bright star nearby, Diadem, aka HIP 64241 or 42-Alpha Comae Berenices.  This star is a magnitude 4 star, somewhat less bright than other stars that were giving me issues.

Here is an image of the stack with a simple stretch in PixInsight.  Note the diffuse, linear nature of the flare. 

Here's a view of how it resides outside of the field of view of the frame.  Note that image has been linked in TheSkyX and the white line shows the frame of the platesolve.  Also note the distance tool showing a relative distance from the star to the brightest portion of the flare.  The value is about 7' 57''.

Lastly, I have a screen capture of the CCDWare website tool that shows a calculation for dust donuts.  I've added the calculation of what I see as the brightest part of the flare.  Numbers for the Pentax K10D are 6um pixel size, the SV4 scope is f6.4, and the distance of 8 arc minutes is about 250 pixels at the resolution of 1.9 arc seconds/pixel of this setup.

Interesting to note that the CCDWare tool suggests that the flare may be originating at 9.6 mm in front of the sensor - or at 19mm if I recalculate assuming that the 8 arc minutes is actually the radius and not the diameter.  As the Pentax camera has a flange distance of 46 mm, this puts the reflection within the mirror box of the DSLR.  I have looked at the camera in the past and wondered if glinting could occur off some of the edges, specifically around the light meter.  The box sides are baffled with ridges but there are some metal panels that extend into the light path a bit.  I wonder if the edges of these are fully blackened.  I will try to cut some pieces of flocking paper and place them along the sides of the box. 

I would like to check the trace to ensure that the calculations above are correct.  I'll have to remember my trigonomy.

It's important to note that for this stack, I did NOT use the extended hood.  Rather, any changes in flare control came from flocking inside the flattener. I did use a longer hood on other image sequences and felt that they were causing issues with star shapes and diffraction spikes so I abandoned them in favor of running the dew heater more aggressively.

EDIT: The more I think about this flare, the less I believe that it's coming from reflection and more likely from either dirt or condensation on the optics.  Looking at TheSkyX, the star is just off of the sensor.  It's possible that the relctions are coming from components on the board nearby as well.  It's just too close to block even with a longer hood.

Monday, March 24, 2014

Backyard Learning

Over the weekend, I set up the equipment in the backyard and did some testing of the camera, power, and Gemini.  Here's what I've learned.

  • Power: I made the effort to mimic the power setup that I'd use in the field.  13 V supplied at the car with the telescope and other equipment some distance away.  In the new setup, I put a powerpole splitter near the supply source and ran two extension cords to the scope site.  Both cords were not fully extended.  The heavier cord was used to drive the mount, camera, computer, and dew heaters.  The lightweight cord was was used to drive the peltier cooler and USB power source.  Over the course of several nights I did not have any issues with running the system.  The mount never saw a power drop below 12V based on the LED on the Kendrick distribution box.  Only once was there an issue and that was when I plugged in the Peltier device when the mount was on.  At that time, there was enough of a drop to cause a loss of power.  I think that a capacitor would have solved that issue.

    Power readout at the supply was usually over 13V and a max draw of 8 Amps.

    Will use this setup in the field for now on.  Hopefully the Prius won't run too often with this higher drain.

  • Camera: Continued to wrap the body of the camera in the sock. The camera EXIF temps are stable.  They get to a point and then don't change over the course of the night.  I will install the sensor when they arrive and put the PWM controller on the system.  In the meantime, the camera is cold and the system does work without overheating.  Haven't stacked the images yet to see how much latent noise is still in the system.

  • Gemini: The whole system still has the slow RA issue.  On the last night, I swapped out the G2 for the G1 setup.  Running the G1 on 17V gave much better response to commands than I'd experienced at Calstar.  I understand more about how it works.  Was able to get TheSkyX and PHD to work.  Was not able to get Maxim to work.  Did have some issues with gotos, but that might have been due to bad models.

    I did some PEC training with the G2 with the new firmware.  Here's what that graph looks like:

Look at how the slope of the curve goes downward.  Not good.  There's something amiss.

Happened across a comment online in an old archive that the Ovision worm uses a period of 240 vs the Losmandy standard 239.  This one difference may have been enough to cause the issue of running slow.

Here's what PHD2 looks like when guiding using the mount using the G1:

Note how the scatterplot is skewed to the left.  This is after I've replaced the counterweight from an 11 to a 7 pound version.  There was not much east bias on the mount.  The graph prior to the replacement was the same.  In this case, the RA aggressiveness was set to 75%.

I checked the Ovision spreadsheet that was provided and there's something interesting here:

Looks like there is a 240 second period?  Of course, this could just be rounding.  

Based on the math in the Gemini guides, I should be using a custom Divisor of:
For G1, the RA divisor should be: 56250
For G2, the RA divisor should be: 450000

I don't have the chance to test it tonight as there will be rain coming.  When I can get the gear setup again I'll try the G2 again as the new firmware allows custom divisors to be used.  It also has the ability to train custom rates.  I'll also want to do PemPro with the updated worm period.

Saturday, March 22, 2014

Data from Cooled Camera

I think the current issue with the stacked Peltier devices is that once they cool, they start generating too much of their own heat.  This becomes a problem and it keeps the camera from cooling further.  Also, the hot air from the heatsink fan blows onto the camera body. 

To be tested:
1. Put a physical barrier between the camera body and the airflow.  I'm wrapping the camera in an old sock to act as a thermal blanket.  If this works, then I can use adhesive material and wrap the closed-cell foam tape I have around the critical bits.  Once stabilized, I'm seeing that the bottom of the camera is 55.4F as measured under the thermal layer.  I'll check it periodically and see how it changes.

55.4 at 9:00     (EXIF 17C)
57.8 at 9:55     (EXIF 21C)
59.5 at 10:48    (EXIF 22C)
60.8 at 11:34    (EXIF 23C)
61.4 at 11:53    (EXIF 24C)
62.1 at 12:13    (EXIF 24C)
61.7 at 12:40    (EXIF 24C)
62.2 at 12:57    (EXIF 24C)    TEC powered off to replace cord
62.8 at 14:06    (EXIF 24C)    Ambient is 74.6

2. I've ordered replacement thermistors for the controller.  Once I have these in hand, I'll put the sensor into the camera at the place where the finger touches the chassis.  That will give some information about the performance of the cooling and may be able to throttle the devices, letting them work less hard and thus not heat up as much.  I would have preferred to find the device locally, but I was not able to find it online at Frys or Halted. 

3. Still bench testing to see if the camera holds the warm temperature or if it doesn't change.  The new dark data is useful as there are hot pixels that have developed over time.  Over the coming months I'll be adding more darks to the library. 

I've looked at the data from the darks of the same EXIF temperature measured last August vs yesterday.  Aside from the hot pixel outliers, the range is still very similar across the whole frame.  In other words, much of this effort hasn't changed things. 

Friday, March 21, 2014

Stacked Peltier Devices

I stacked the Tellurex C2-30-1503 and Tellurex C2-30-1505 devices as noted.  Just needed to pull out the old pad of paper that was supporting the copper plate and add some extra foam insulation to properly close the area around the parts. Ideally, I should be using some chemical foam, but that's pretty messy stuff and I'm not finished yet.  No pictures of the assembly this time, I was working quickly. 

My math was off with figuring out what the amp draw of the two devices would be stacked.  With 12V applied, the draw is 5.5 to 5.7 A.  This is only about 1 A more than the 1505 by itself.  I believe that the difference is because the pull and temperature is derating the resistance of the devices. Handling 6A is not a small amount of power for use in the field.  I'm considering what it will take to provide this much juice to the scope. I'll play with putting a splitter closer to the battery/car vs putting it at the Kendrick power distribution box.  Running two cords should give the effect of less cable resistance.

The only part that really complains about low voltage are the dew heaters.  If necessary, I'll put a voltage regulator on it to help support them  They draw 1.1 A under full load so it shouldn't be much of an issue.

So far, the air based heat sink seems to be keeping up with the upgraded power requirements.  I'm testing it on the desktop and I can feel that the tabletop is warm.  That shows me that there's quite a bit of heat being moved.  I looked into some closed-circuit water cooling devices to see if they would be good for this setup.  While the concept is good, the hoses are too short for my use.  Most are one foot to 18 inches - great for use inside a computer case, horrible for trying to route hoses from the back of a camera on a telescope. If I need to go this way, I'll roll my own system as I'll either need 2 feet of hoses to mount the system on top of the scope (piggyback style), 6 feet to mount it at the tripod head, and 10 feet to go all the way to the ground. 

After all the testing on the desktop and again in the refrigerator, I'm disappointed that the EXIF temperatures did not drop below 17C.  I believe that either I've misconfigured the TEC devices OR there's not enough watts pulled out of the camera OR that the EXIF sensor is no where near the cold finger.  Best I can do at this time is continue to take darks and compare with earlier configurations.

As another followup, I see the reason why the PWM controller stopped working.  Apparently the thermistor has gone open.  I'll get another to replace it and when I can, I'll put it into the system.  May be able to wire up the assembly with the new stacked devices to better control the power needs.

Thursday, March 20, 2014

Followup on hot camera

Did some work on the camera without spending money.  As noted earlier, I had some Arctic Silver compound in my kit from other projects.  I removed all of the old compound, including the most recently applied material and used the Arctic Silver in its place.  This material is supposed to be the bee's knees, even though it has a long set time (200 hours).

I also put a terminal block on the heatsink baseplate to facilitate changing parts. 

I put the Tellurex C2-30-1503 in place of the larger wattage device.  The drop of 10 watts should help control the generation of unwanted heat.  Since I'm driving the device at full Voltage with no modulation, there seems to be a lot of waste heat with a larger device.  A nice advantage is that the current use drops by a full amp.

While I had the camera apart for the Arctic Silver exercise, I looked deep inside the guts.  I could see where the sensor assembly floats and moves.  With looking at the way stuff fits inside the path, I cut a small pad of the thermal gel and slipped it inside the cavity.  I did not remove the cling plastic as I wanted to be able to have it slip inside the camera without snags.  Learned a few things about the way the sensor plate moves when the camera takes a picture.  I'd assumed that the resting point for the sensor plate was "neutral" and that when the camera takes a picture with the shake reduction turned off, it would not need to float.  After I put the thermal pad in place, I noticed the way the plate moves.  The sensor floats free when unpowered and then as the camera takes the picture, the sensor "unparks" to approach the proper distance and location.  Even with the shake reduction turned off, the sensor must move.  Very interesting!

With testing after this addition of the thermal pad covering part of the back of the sensor, I checked some darks.  I was pleased to see that the relative darkness of these new images is notable.  Yes, amp glow still reigns in the same place.  However, the standard deviation of noise in the central section drops from 13 to 12 when at 19C.  I'll check other temperatures as I can.

After a few hours of running the camera on the desktop, I'm a little annoyed that the EXIF temps hold at 19-20C.  There's no real movement to a lower setpoint.  The standard deviation on the dark frame shows the same range - 13 to 15.  Interesting to note that the hot pixel library in Maxim is still leaving a few stragglers, may be what's causing the high numbers relative to the old examples.  Std Dev of the bias signal is 1.096.  Tomorrow I'll put the camera in the fridge and see what happens vs the desktop.  If the results don't show much additional promise, then I think I'll need to look at stacking devices for a trial.  I have several to try.  I can rig in parallel to see what happens. 

I did some math on what to expect with the stacked peltier/TEC devices.  I have decided to stack the C2-30-1505 on top of the 1503 device.  Based on the numbers that I have for the current use and Qc at this current, it should work.  The question becomes whether the heat sink will be up to the task of removing the waste heat.  If it's not, then the experiment moves towards water cooling.

At this time, the old thermocouple sensor wires are still hanging out of the camera.  I'd like to find a way to use this available data.  I have found the resistance to C chart for the probe.  I could just use the multimeter to see how it performs.  Unfortunately, I did not pay attention to where the probe ended up during the last assembly.  It may be loose and not contacting anything of interest.

A final step that I did to combat the rotation problem at the TMount, I put a few dabs of hot melt glue in the seam and gap from where the locking pin goes.  I can remove this glue with just picking at it with my fingernail if I must change it.  Otherwise, it seems to hold the bayonet adapter nice and firm.  Should have done this long ago!

Tuesday, March 18, 2014

Summer is Coming

And my camera is hot.

I don't want it to be hot, really.  I've been using a peltier device to try to cool the camera with fair success.  Recently, I pulled apart the bottom of the camera to re-paste the finger to better transfer heat from the chassis.

On testing in the thermal extremes of the shed, I had some challenges.
  • First, the extension cord I used initially had too much voltage drop and was appearing to cause problems. To solve this, I've put the power supply at the source of AC and use my heavy extension cord to provide 12V.  This method is similar to the way the system is used in the field.  
  • Secondly, there has been a failure in the PWM controller.  It appears to no longer work, not powering the TEC/Peltier at all.  The failure could have been from a short across components. Thus, I removed the controller from the system and hard wired the fan and TEC together.  
  • Third, even with the thermal control at full power, the camera continued to heat up.  With strategic temperature probing, I was able to see that the cold finger wasn't very cold.  To solve this problem, I pulled the heatsink again and replaced the thermal compound on the cold side of the TEC with the silicon mat that I've used.  I think the thermal compound that I have is not very good at cold temperatures.  I think it needs heat to "set."  After I made this change, the cold finger worked much better.  However, the camera still didn't get cooler.
  • Fourth, while checking, I did see/feel that the base of the camera gets too cold.  The dovetail mounting bracket acts as a place for heat intrusion.  This is a massive leak.  I've removed the bracket and will have to try to protect the area.  Right now I'm using the old sock method.  Because I've removed this firm mounting point, I'll have to re-solve the issues with flexure, specifically rotation as the T ring has a lot of slop on the mounting pin.

I think the remaining issue with ut is that the air cooling just can't dump heat fast enough and the whole setup builds heat.  I'll need to look at optimizing things.  I've left it running overnight and I'm hoping that the cool air of the evening will show better results.  There might be an ambient temperature where it starts working.

All this has me thinking again about water cooling.  Nothing else I know can handle 100 Watts of waste heat in a light form factor.  I do have an old water block that I could press into service.  Would need a radiator with fans and a 12V pump.  I did a little searching online for parts and have some ideas.  The idea of water cooling makes sense to me since I use the method for my desktop computer.

While the idea of having a system that allows the introduction of ice to speed things up, I recognize that this setup is not very portable.  I think that using just a radiator, pump, hose, and water block will be all I can do.  I may be able to mount it on a dovetail clamp for riding piggyback.

EDIT: As a followup with more thinking, I believe that I may be using the wrong thermal compound.  I'll open it all back up and recompound with some better stuff I already have (Arctic Silver).  Will also try a smaller TEC since the system is running flat out with no PWM.  To facilitate the ease of changing components, I'll put a terminal strip on the board. 

Sunday, March 16, 2014



Left the car unlocked and my toolbox with scopestuff was taken overnight.

They made away with a 41mm Panoptic, 40mm Scopetronix Maxview (2inch version), 3-6mm Nagler Zoom, a spare Losmandy Gemini motor, IDAS HEUIB II 52mm filter, IDAS LPS P2 48mm filter, metric hex set, a baggie of half-height 6mm nuts, plus my supply of 6mm and 1/4-20 socket head screws.  I had a backup stash of lithium coin batteries plus a bunch of dead 9V.  A wood-inlaid compass was also in the box as well as the tiny set of hex wrenches for aligning the reticle for the Losmandy polar scope. 

And my nice LED red flashlight.


Surprisingly, they didn't take the 50' extension cord nor the two sets of binoculars they had to move to take the toolbox.

Likely they'd get just a few bucks from the fence for the whole kit.  It's a shame, really.  I'll be looking into what coverage the insurances might have.

Rather annoying as I had been considering selling the eyepieces as I don't do visual except on outreach nights.  Also, I was hoping to do some maintenance on the scope this week as a followup to the work I'd done last week.


Thursday, March 13, 2014

What I'm Doing About Flare

As I've noted in earlier posts, off axis flare is a real thing that I've started to address.  I believe that the reason why flare wasn't noticed before was that my efforts with picture taking was mostly on brighter objects and my poor methods of calibration wasn't pushing the exposures enough to see the flare.

Now that I'm aware of it, I can see it more readily.


More Flocking

I've been marginally successful with controlling it with some flocking efforts, mostly concentrating on the bright anodized components in the feathertouch focuser.

At the latest project, I noticed that the flare was coming from an area about 1.3 degrees from the edge of the frame.  This got me to think about how to better flock the components.

I added some rings of felting in the flattener.  See these photos for a view of what I did.  Note how I've added flocking to the T42/Kmount ring.  I don't expect there to be much possibility of glinting on these elements, but it doesn't hurt to control it.

I'm expecting a certain amount of fiber shedding.  I made the effort to drag some tape on the felting to pull off loose threads.  Hopefully this will keep the shedding to a minimum.

As the photos show, there's some exposed metal between the elements of the flattener.  These may reflect light if they are exposed at the proper angle.  I checked with Vic of Stellarvue and his comment was that the retaining ring should be just finger tight.  Well, mine is gummed up from previous experiments with adhesive flocking.  Not going to be able to get into this spot.


Longer Hood

Instead, I'm going after the idea of using a longer hood to control off-axis light.  The idea is that by keeping the sensor from seeing the bright stars, it's unlikely to have as much of an effect.

I was looking at the potential of a custom lens hood like the ones on this site.

Based on some math provided by the HOODCALC.xls spreadsheet that I found on the Pentax Forum, I've found that the hood length for the APS-C sized camera should be about 450-500mm long.  That's pretty big.  Up to now, I've been using a foam paper hood rolled around the existing dew shield to extend the structure.  Based on the math, I'd need to add another 6 inches or more to the end of this structure.

I played with adding different shapes to the foam paper recently.  I tried rolling it up inside the dew shield.  This works well but it does raise the challenge of repeatability.  Plus, it shows the fact that a long extension of material will sag under its own weight. For anything that's going to be permanent, I'll need something that's light, won't sag, and fairly durable.  I've been thinking about rolled aluminum.  Advantage is that I could get it made from something fairly sturdy and not worry about crushing it.  Disadvantage is cost.

What remains to be seen is how this extended tube will impact piggyback shots.  I ruined a few film shots back at Calstar because the hood extended into the frame.  I'd like to avoid this happening in the future, too.

For now, I can use the newly rolled up foam paper that has some carefully placed velcro tabs.  It may serve the purpose well.

A possible concept that I may return to try would be to cut a rectangular mask out of some cardstock and insert this into the dewshield.  If I put it right at the objective, it could mask the offending areas pretty easily.  This would allow a shorter hood to be used with minimal changes.  I'll do some math and cut one as I have the chance.


 Camera maintenance

Another exercise that I completed today was getting back into the camera.  I'd been suspecting that the cold finger wasn't making good contact with the chassis of the camera.  Finally tore it apart and checked.

Sure enough, the old thermal paste tells an obvious tale.  The old material was barely touching.  To remedy this situation, I bent the copper to fit a bit tighter.  I also inserted a slip of foam paper to apply some slight pressure as well as thermally insulate some of these parts.  Lastly, I moved the thermal probe to a better place inside the assembly where it can make contact between the finger and the chassis.

In my testing afterwards, the thermistor controller showed no drop in temperature below 10C. I base this measurement by turning the power controller down until it stops pulling a high amp load.  Ambient temperatures were 22C.  EXIF temperature was 17C.   I also measured the temperatures on the camera body at a few different places.  This tells me that wrapping the camera in a thermal blanket is important.  I'd already inserted a piece of foam paper between the dovetail and the camera body.  This should help control unwanted heat entry to the camera.

On further testing, I mounted the camera to the telescope, wrapped the camera body in thermal layers, and set it up to do a run of darks while in the shed.  The shed temps are approaching 28C and surprisingly, the thermistor controller is suggesting the sensor is -15C.  I'll need to check later to see what the EXIF temperatures are.  If this works well, then I can expect very low noise in the field going forward!

Returning to this post after some time.  The shed experiment was inconclusive.  EXIF temperatures started at 18C and went up to 25C before the voltage drop from the outside power to the 12V converter was too much.  Will have to try again and see what I get using a different cabling situation, more like the way the car setup where the 12V is provided at the start of the extension cord.

Thursday, March 6, 2014

Clear Skies Returning?

Just in time for the full moon...

An update from Weather West suggests that the rain-blocking ridge may return.

While I'm not terribly happy with the prospect of dry conditions being that we're in a drought, I'm bummed that the clear skies comes right when the moon is moving into full.

Maybe it will give me the chance to check the system for flare control.

Ghost of Cassiopeia (IC 59 and IC 63) at Calstar 2013

Via Flickr:
This area is a complex of dust near the bright star Navi in Cassiopeia. The nebula glows with a combination of emission and reflection, giving a reddish and blueish white color. The whole complex of IC 59 and IC 63 also are in the Sharpless Catalog as Sh2-185.

Taken over two days while at Calstar 2013 in October. Most data was collected on the first night after I'd done polar alignment and while I was struggling with polar alignment and guiding of the GM8 for film work. The rest of the data was gathered piece-meal as I had shut down the GM8 and went piggyback for the film camera.

Standard setup for my digital work:
Stellarvue SV4 telescope using SSF6 flattener
Modified and Cooled Pentax K10D camera
No light pollution filter, just used the B+W 486 UV/IR filter. For a time I was using this filter to control NIR and UV on the full spectrum camera to control what I thought was out of focus light causing problems.

10 subexposures at 1200 seconds each giving 3 hours and 20 minutes of integration time.

Guiding and calibration with Maxim DL.
Stacked with DSS.
Processed with PixInsight: crop, DBE, masked stretch script, histogram stretch to reset black point, masked use of SCNR to control green, MT with a mask for brightest areas, finally unmasked curves to boost contrast a bit.

I had some challenges with flare on this data. Because the bright star Navi (Gamma Cassiopeiae) is near the center of the frame, it blazes at 2.5 magnitude, causing reflections to show up around the scene, specifically near HR266. The limited amount of data I had to use made it difficult to control the flare. Only by stacking with rigorous rejection routines (DSS Kappa Sigma K=1 x 49) was it possible to control the flare. Here are some notes from that effort.

I have not taken a similar image recently that features such a bright star. I'm curious what the recent centering and alignment and flocking work I've been doing will change the behavior of the system.

Here's the plate solve from PI:

Image Plate Solver script version 3.4.1
Referentiation Matrix (Gnomonic projection = Matrix * Coords[x,y]):
+6.55284e-006 -0.000530856 +0.631552
+0.000530703 +6.70795e-006 -0.947467
+0 +0 +1
Projection origin.. [1769.992960 1211.534127]pix -> [RA:+00 56 47.60 Dec:+61 03 08.20]
Resolution ........ 1.911 arcsec/pix
Rotation .......... 90.707 deg
Focal ............. 653.03 mm
Pixel size ........ 6.05 um
Field of view ..... 1d 52' 44.8" x 1d 17' 10.2"
Image center ...... RA: 00 56 47.613 Dec: +61 03 08.21
Image bounds:
top-left ....... RA: 01 01 51.623 Dec: +60 05 55.84
top-right ...... RA: 01 02 22.028 Dec: +61 58 35.00
bottom-left .... RA: 00 51 32.283 Dec: +60 06 52.69
bottom-right ... RA: 00 51 24.872 Dec: +61 59 35.24