Important Note:  Since this material was prepared and posted, SBIG has come out with a modification to the ST-7/8 that provides additional cooling.  The modification adds a second stage Thermoelectric Cooler with a larger fan and additional heat sinking.  Additionally, there is provision for liquid heat transfer cooling.  This modification, available starting in January 1999, is an option for new imagers and is available as an upgrade for older imagers.  It should provide all the cooling capability that astrophotographers might require in these imagers.
 

Cooling is always desirable and is used on all astronomical imagers.  Cooling reduces dark current noise by twice per 8 degrees of cooling.  In some applications cooling much below 0 C is not worth while because of sky pollution.  But for faint deep space objects, where narrow band filters are used, for high f number telescopes and any cases where long exposures are needed more cooling is generally better.  Many professional imagers cool the chip to liquid nitrogen temperatures (-180 C) or with dry ice (-80 C).  Such extreme tactics will not be considered.  But methods for cooling the imaging chip to -30 C using thermoelectric coolers (TEC) and/or liquid heat exchangers are quite reasonable.  Not only the depth of cooling but the accurate control of the chip temperature are important considerations.

While the following comments are mainly speculation about cooling of the  SBIG  ST 7/8 imagers, other imagers are mentioned as well.  These were private communications among several concerned  users.  They have been edited, updated and organized by Doc G.  The occasional I, we, our, my and so forth may be attributed to one or another of  several persons.  The comments should be looked on as experiences and speculation on the state of the art with discussion of concerns and possible solutions to the cooling problem.

When I gave up on the 416XT and decided to get the ST-7, one of my major concerns was the single stage cooling provided by the ST imagers for the same chips which are two stage cooled in the XT imagers.  I almost did not get the unit because of this factor, since I consider cooling very important.  But I was at that time desperate for some kind of working CCD imager.  The XT software was not working at that time, so I got the ST-7 despite its added cost and single stage cooling.

The ST-7 worked immediately out of the box and I have gotten a few satisfactory images with it.  So now I use the ST-7 with what I consider temperamental and sometimes inadequate cooling.   The XT imagers have better cooling and cooling control but have other problems.   I was about to get an ST-8 but believe the cooling problems must be addressed first.  This big chip deserves better better cooling and better freedom from condensation in my opinion.  The design of a -30 to -50 C cooler and the required thermal shielding, etc is not trivial as the following discussion shows.

I have been studying applications of  TECs (thermoelectric coolers), their capabilities and circuits for some time so as to evaluate the situation regarding these coolers and have studied the Melcor devices and their engineering material at some length. Melcor is the major manufacturer of TEC and other cooling devices.  They have excellent engineering data on their fine web site at  www.melcor.com.

After looking at several designs, it is apparent that a two stage TEC is a possible choice and can be made to give differential temperatures of 40 C.  This is about what the XT imagers get.  The problem with TECs is that each stage is quite inefficient.  The input power has to be something like 3 to 5 times the pumped power. The second stage has to pump the original power dissipated in the chip plus the input power to the first stage.  Thus the total power that needs to be dissipated in the final heat sink is 10 to 25 times the initial chip power.  Thus, there are other ways, not using a two stage cooler that should be seriously considered.  (these are complex and cumbersome as described later in these discussions)

The design, to get large differential temperatures, is very tough.  If the chip power, consisting of electrical power dissipated in the chip, convection power through the air, conduction power through connections and power radiated to the chip from all surrounding surface, including the hot electronics, is as little as 0.5 watt then the final power to dissipate to the ambient sink needs to be as high as 12 watts.  (this just a possible example)  To dissipate 12 watts to the atmosphere is not a trivial problem.  A good, large heat sink operating through convection only will have a rise in temperature of 3 C to 5 C per watt. (This makes the above design example impossible unless the initial power to the chip is reduced by a great deal.)  A fan cooled heat sink might have 0.2 to 0.5 degrees rise per watt.  This immediately makes the design at least possible though still difficult.

At the present time, I think the XT design is very, very good.  It seems to get good cooling with a convection only cooled heat sink.  Think of what could be done with a fan.  (an example of what can be done with a fan follows later)  It is apparent that the initial dissipation in the camera head has been kept to a very low level.  Very little of the electronics is in the head which reduces heat dissipation.  The major electronic processing is apparently in the control box (that strange and mysterious thing) which does the processing and signal transfer work.  I think this is basically a good design decision.  On the other hand I think the ST imager has too much electronics in the head and there is too much heating from all this electronics.  This is, I think, a basic design flaw.  There has always been an indication that the electronics heats the chip by radiation in a non-uniform way.  All heating on the cold side of the TEC should (must) be minimized in every way possible.

Study of the Melcor Engineering data shows that a two stage cooler is relatively easy to design and build and is about the optimum TEC combination.  (more stages get too inefficient and expensive)  But any design  requires designing a very minimum power dissipation in the chip and its immediate electronics.  And that must include all sources of heat.  Electrical, radiation, convection and conduction.   I am generally impressed with the XT (Gordion) design and not impressed with the SBIG design in this respect.  When the XT works it is a very good camera.  (It is just that software that is a sore point.)

WHAT TO DO.  That is the question!!

So we have a possible situation where the XT seems to have the best thermal design and the ST the best data transfer and software.  (At the moment and in some users opinion, subject to change.)  Thus. some speculation about the two imagers and what to do to fix either or both is interesting and follows.

The two photos show that it is a tight fit, but it is possible to install a second stage cooler and a heat exchanger on the ST imager.  The photograph on the left shows an ST imager from the topside.  As can be seen, the imaging chip is in the socket at the top with the TEC below it.  The TEC is under the electronic circuit board, the cooled side to the top and the hot side to the bottom.  Note that the electronics is largely discrete components.  This is a bit surprising for a 1990s design.  On the right is shown an ST imager with a secondary TEC cooler added and a liquid heat exchanger.  This is a very neat installation.  This image appeared on the web some time ago.  It is of an imager owned by Benoit Schillings.  The modification was made for him by Sigfried Salat.  This is avery neat modification.  The second TEC and heat exchanger go in place directly on the rear surface of the original TEC.
 

It is very difficult to modify the ST imager as the photos show.  There is little room inside the imager.  There is room for a heat sink, shown above left, that  is very small and has only a few shallow fins.  Even with the fan blowing directly on it, the rate of heat exchange is only adequate.  With a second stage of TEC cooling, a massive heat sink to the air would have to be added to the second stage cooler. (and a fan added as well)   All of these seem to require a significant rebuild of the imager.  It should also be realized that cooling has a price in that the entire imager runs very cold and dew forms on everything.  This can fog the imager window and in the case of the AO unit the deflection mirror. In the installation shown above right the comment was made that the entire unit dripped with dew on some occasions.  This is because the entire unit cools well below the dew point and dew condenses on all surfaces including the circuit board.  In the long run this wetness will likely be deleterious to the entire electronic system.  There is also an increase of the number of wires and tubes that go to the imager unit which makes some of these techniques cumbersome.

Any cooling method requires providing cooling hot side of the original single stage TEC.   This might be done with a second TEC cooler that is conductively connected to the hot side of the current TEC.  The hot side of the current cooler is directly available for a second TEC attachment which could be attached in place of the small finned sink. The addition of a second cooler at this plate would make sense.  A TEC with sufficient capacity is available as a standard Melcor part.  Of course a much larger set of final cooling fins would be needed to dissipate the extra heat which would be 3 to 5 times greater than currently.   A much larger set of fins of the proper design and with forced air cooling might be acceptable for a custom design. I think an additional 20 C of cooling could be obtained this way.  Still this would make a larger heavier and rather cumbersome looking imager.  The problem with the use of the second TEC that the cooler has not only to cool the chip but the electronics which is inside the inner chamber of the imager.   (More comments about this electronics later.)

Another, and probably better, way to provide the massive heat exchange necessary, which would be very efficient, would be to fasten a liquid cooled plate to the backplate and cool it with an auxiliary cooler plant. One such design,  shown in the top right photo,  (this is a photo of the Benoit Schillings modified ST imager) uses a second TEC and a liquid cooler to cool the second TEC.  This makes possible massive cooling with a liquid temperature of only 0 C or slightly higher.  Liquid heat exchangers remove a lot of heat because of the high specific heat of a liquid compared to air.  With this unit, it seems to be possible to get a full 40 to 50 C differential cooling.  That is, in this case,  -50 C.  More than cool enough for any amateur application.  This seems like a fine solution to the cooling problem.

A very common cooling method used in the laboratory is a commercial cooler using a circulating liquid.  Coolers of almost any capacity and temperature range are available.  They work by using a standard compressor to cool a liquid to  0 C, -20 C or even lower pumping the liquid through tubes to the cooler element.  Such a unit could be put together for well under $1000.  Melcor makes standard parts for such a setup.  With a super cold fluid heat exchanger coupled directly to the original TEC, the second TEC would not be required at all.  The liquid cooler could be set to say -20 C and the original TEC drop the temperature another 20 C to give -40 C chip temperature.  This would be in some ways a more complex system which required a commercial liquid cooler system but on the other hand simpler since it would not require the second TEC.

The liquid cooled concept is attractive since the liquid can be kept close to the desired chip temperature and the TEC will have a modest and almost constant temperature drop across it.  If  the liquid were at say -20 C the chip could easily run at  -40 C with just the single stage TEC.  The problem then would be to keep everything else in the imager from frosting up.  There are significant problems in running a chip well below the temperature for which the imager was originally designed.  (this issue discussed below)  A flow of dry Nitrogen or some other tactic might be necessary.

Since I do nothing but observatory work, (I no longer do portable work at all) I have thought seriously about a liquid cooling system.  Standard parts are in fact available to do this.  Only the small cooler heat exchanger would have to be inserted into the camera.  I believe that the ST temperature control circuits could be adjusted to work at a lower temperature like -40 or -50 C as Wallis claims to use all the time.  There are tubes to the camera to consider, but they can be small and flexible with a suitable liquid.  Standard low temperature tubing which stays flexible is available.

An alternate scheme is to use the camera as it is but add a cooled air manifold to it.  This would be a light weight hose, like a vacuum cleaner hose, but lighter weight since it has a slight internal pressure instead of a vacuum and does not have to be strong.  One would connect this hose to the air input at the fan and add forced cooled air from a suitable air conditioner unit.  Such equipment is also available from laboratory supply houses.  Air at near freezing temperatures (say 2 C) is available from such units.

I have been personally interested in both the ST and XT imagers cooling for some time.  There is a lot of unrealized performance potential in the ST-7/8 as a result of the limited temperature control precision and cooling depth.  The XT can achieve a cooling differential of 40 C with some care.  At this time there seems to be no interest at SBIG for a factory solution to additional  ST-7/8 cooling and temperature regulation.  Perhaps with revised ST-5 and the AO-7 complete there will be time to address this important issue. Your observations closely reflect my own.  I have the Melcor catalog and have a few TEC's from them which essentially confirms what we know about these devices.  I believe less waste heat could be realized by a cascaded 2 stage TEC whose first stage is smaller and  a larger 2nd stage to handle the power dissipation.  The Melcor catalog also contains detailed information about their AZTEC thermal design software and liquid heat exchangers.  The current ST TEC is rectangular with dimensions 1.18" X 0.55" X 0.14" thick.

The ST-7/8 use the internal back plate for heat dissipation and fins are located directly under the intake fan. (See photographs above)  The back plate is machine screwed to the outside housing (and fins).  These fins are few and are not deep, unlike the 416 which has many thin fins that are quite deep.  There is just enough room for a 2nd cascaded stage if the inner back plate is cut out.  See the photograph with the inner backplate electronics removed  the TEC.  The desiccant container on the left.  Any added plates to slightly extend the back plate and cutting out the existing back plate would need to be of sufficient strength and secured to assure the camera doesn't leak or flex in this area.

A home made liquid heat exchanger is shown in the photograph lower right above.  It replaces the finned heat sink and is routed to a couple of small copper tubes that extend through the base of the imager.  In the front of the imager is a fitting for a nitrogen hose for supplying dry nitrogen to purge the imager and prevent condensation on the very cold imager window.   I too have noticed that the ST does not get as cool as one  might like.  When questioned about this in some detail before I bought  the ST-7, they assured me that one stage of cooling was enough.  This may be true for most users.  But with much discussion about this issue on SBIGUSER there is reason to doubt that this is the case for more advanced applications.  Clearly you can get better cooling by supplying a  cooler environment.  Winter  in  Wisconsin should do the trick..   But I would like to use the imager on hot summer days as well.

Meanwhile, I do not know how one gets a lower ambient except by  delivering refrigerated air to the cooler fan opening.  This sounds  difficult at best.  I wish someone would have a scheme to do this.  One is alluded to in the above discussion but has not been tried.  While I agree the low dark current in the Kodak 0400/1600 chips should make a 30 degree delta T. adequate, I found in practice, a lower chip temperature compensates for ST-7 temperature regulation anomalies and dark frames taken under different ambient conditions.  Basically said, cooler is better.

In the case of the XT,  a simple muffin run at reduced speed allows continuous operation at a differential temperature of  40  C.  It just takes a small amount of forced air to make the heat sink more effective.  This is well knows from standard heat sink theory.  Forced convection will reduce the heat sink temperature to be very nearly the ambient temperature rather than an increased temperature allow by normal convection only. . The XT software indicates temperature regulation to within 0.1 degrees C. when calibrated with the hardware.  The calibrated images indicate regulation to be very good.  Still, the chip temperature attainable is dependent upon the ambient temperature.

Several ST-7 imagers checked varied somewhat in temperature capability.  Average differential temperature is 30-32  C. with an apparent 0.5 degree C. regulation precision.  Apparent because the indicated chip temperature moves back and forth between the 0.5 degree temperature sensor setpoints in the ST-7/8.  This can be seen by observing the software readout that indicates a chip temperature from +0.19  C to -0.23  C. when set for 0 degrees C.  The indicated temperature may change back and forth in less than a second.  Of coarse the "actual" chip temperature isn't changing that fast because of thermal inertia.  So we do not know true chip temperature or regulation precision.

Here is how  SBIG explains the ST-7 anomaly described:  " The ST-7 is acting normally by dithering between +0.19 and -0.23 degrees.   The temperature sensor in the camera has approximately 0.5 degree setpoints.   The way it regulates the temperature is by forcing the temperature to the  boundary between these two temperatures.  While the CCD's temperature is changing a small amount (typically 0.01 degrees) the reading changes by 0.42  degrees because it is right on the boundary.  Do you understand what I'm saying?  It regulates the temperature to be between the two positions and is  constantly raising and lowering the temperature by small amounts to achieve  that temperature."  (This basic operation is a description of a small amplitude, closed loop limit cycle which is not uncommon in temperature servo control systems.)

As long as a dark frame is taken at nearly the same time as light frame,  good results with the calibrated image can be obtained.  But, if library dark frames are used, results are not nearly as good as many hot and cold pixels are left in the calibrated image.  While CCDSoft will "remove" these artifacts by averaging the hot or cold pixels with surrounding pixels, the data integrity is compromised.  For those using their camera for illustrative purposes, this may be of little concern.  I tried taking several same length light frame exposures using one dark frame taken that night but I was not satisfied with the results and certainly didn't like the constraints. The Pictor 416 is better in this respect.  For every  8 degrees C reduction in chip temperature, the dark current is cut in half.  I can't control temperature regulation and find it difficult to reproduce ambient conditions.  The only practical way to minimize these hot and cold pixels when using library dark frames is to take images at a lower chip temperature.

The lower differential temperature of 30-32  C  is not adequate for precision work.  A second TEC added to the ST-7 with disappointing results as the amount of surface area needed in the heat sink cooled by a much larger fan has to be very large due to the poor thermal capabilities of air.  However an internal heat exchanger added to the ST-7 with very small hoses would easily achieve a temperature differential of 55  C.  Since the added cooling tends to cool the entire imager body,  the dew point outside the camera becomes important and the user risks condensation on the imaging window exterior and everything else.  So added heat or inert gas may be needed near the imaging window. This is accomplished with a nitrogen purge in one of the modifications shown.

Adding secondary cooling to the ST-7 is not practical for the average ST-7 user considering the added power used and the skill required for the retrofit, not to mention still more wires or hoses coming out of the camera head.  All of the extras are painful to utilize even in a permanent pier setup and damage do to leaking liquids and dewing are added disincentives.  Ice bags or cold packs help, but add weight and require frequent changing.  A better solution would be for SBIG to improve chip temperature regulation along with a modifications to the TEC cooling system, perhaps a second stage.  Experiments also show that a temperature differential of 40 to 50 C would also require changes to the camera case and cooling fan air volume.  This all sounds like a fresh start is the required design path.

While it is true that a lower chip temperatures can be reached in winter, actually controlling ambient temperatures might be more useful even then.  The  XT displays head temperature, which is very useful to match a given head temperature to a chip temperature to determine what dark frame to use.  For example,  416XT dark frames shot at 25 degrees C. head temperature and -10 degrees C. chip temperature and dark frames shot at 20, 15 and 10 degrees C. head temperatures at the same -10 degrees C. chip temperature are useful.  Then  a head temperature (reflecting ambient) can be related to chip temperature.  Coupled with the temperature differential of 40 degree C obtained by very slight forced air movement from a small fan on the imager and the good temperature regulation of the imager results in excellent calibrated images that are difficult to tell (if at all) from those shot right before or after the light frame.

The ST-7 requires a little more work to obtain similar results as it lacks the head temperature readout, regulation precision and a large differential temperature of the 416XT.  The heat exchanger is used all of the time, Winter and Summer, even if the temperatures get below 0 C.  Then, the reason isn't so much to lower the chip temperature, but to provide an ambient that will minimize the coupling effect the temperature sensor has with ambient temperature of the imager body (nominally the outside temperature).  The idea is to maintain a set temperature for the ambient of say 5 C. with the heat exchanger to match dark frame libraries made at the 5 degree C. ambient.  The heat exchanger increases the efficiency of the TEC that results in a temperature differential of 40 degree C, but also warms the camera when outside temperatures reach well below 0 C.  All of this is to stabilize the conditions that the chip sees.

Controlling ambient temperature to (+ or -) 3 degrees C. is adequate to make the coupling effect negligible.  The liquid cooler system shown in the photographs uses a precision Gorman-Rupp positive displacement pump coupled to a gear motor for around 225 ml/min.  The rubber hoses use low water vapor transmission type insulation with a maximum absorption of 5% and covered by silicone to minimize losses and condensation as well to provide flexibility and durability. Special plastic hoses designed for low temperature use are also available. The heat exchanger is mounted inside the ST-7 on the back of the TEC.  It also serves to lower the camera head temperature, which is arguably more important.  This cannot be done with a second stage TEC as the surface area required to dissipate the heat removed (and heat generated by the TEC stage itself) requires more fin surface area than provided by the current ST-7 case.  What happens is the case temperature rises, couples to the temperature sensor and effects the resulting calibrated image taken at a given chip temperature setpoint.  This result has been verified with experiments with a second TEC operating at different capacities.  A commercial 1/5 HP chiller/heater was used to simplify maintaining the 5 C. temperature in the liquid cooler loop.

Of course, liquid nitrogen used on the cold finger of scientific imagers would do the trick, but I don't think the Kodak chips can handle the extremes, although I am highly skeptical that the SBIG cooling is too fast for these chips as I have never heard of a damaged chip due to fast cooling of SBIG, Axiom, Meade, ISI, Micro Photonics, Photometrics, or SpectraSource.
I am putting inquiries in to get a Kodak engineer with a name that will go on record to advise me on this. (no luck to date).  Lowering the CCD chip temperature, controlling ambient temperature and temperature regulation precision have synergistic effects that can improve results of calibrated images that will allow the use of dark frame libraries minimizing the need to shoot dark frames during observatory imaging time.

I have to believe that almost all that is possible to improve the cooling of the ST-7/8 package has been considered or speculated upon above.  I assume Wallis and possible others have done some of the same things since he talks about running at -40C and using a liquid cooler with a bucket of ice.  From the photographs it looks like the camera structure is simply not designed to go to and hold temperature differentials much more that 40C. Just cooling the chip is not enough.  One has to also provide vacuum for really good thermal insulation.  There also has to be sophisticated thermal shielding to prevent radiative heating of the chip and the consequent heat load.  Then there is the window which has to be kept free of condensation.

It looks to me like SBIG has made a nice camera, but it is destined to be just that.  A nice amateur camera which will take reasonable good (even wonderful) pretty images.  Nice fun stuff but not a scientific tool.  Their electronics is a dead give away.  They have an inner chamber full of discrete components which generate heat in that vital region and outgasses who knows what else.  I am quite appalled by the primitive looks of the electronics.  There should be minimum electronics near the chip and it should be very low power and  in sealed modules.  The electronics looks like about 20 year old design.  It should be all microcircuits.  I have always suspected that one corner of the chip is seeing heat.  Now I believe it is from the electronics. Additional shielding around the imaging chip might be required in a new design.

At this point, I can see no reason to try to do much with this camera design except for a bit more cooling.  It is a nice amateur imager and has to be treated as such.  It seems clear that SBIG has a certain market in mind.  This is the amateur astronomer.  The Meade camera, with its two stage cooler, is basically better design.  It is too bad that both the electronics in the Box and the software are so poor.  I do believe that one of the smart design decisions that was made in the XT design was to put a minimum of electronics in the camera head, thus reducing heating, and using a high speed serial link to the box. (A company called Gordion Engineering actually designed the camera.)   The box is a bit of an aborted effort with the holes and slots for non-existent accessories. And, the one that makes me just plain mad, buttons that do little if anything with total lack of instructions.

But, I digress.  The above picture is not a happy one. Both Meade and SBIG really fail in different ways to meet the needs of amateurs with growing aspirations.  There are some things we amateurs can fix or work around and some we can't.  It may be that the Meade and SBIG cameras can not be modified very much by the user and have to used as best as possible.

Better imagers, like better telescopes are available at the next semi-professional level but also at the next level of cost.  If we were professionals, the government would pay for the equipment and it would not be a hobby any longer.  Well, I have gotten a bit off topic but I hope you understand my viewpoint.

Go to Beginning

Go to Home Index for Doc G's Info Site