Author: Tom Droege Date: 960912 Revision: #0 960912 Key Words: instrumentation, CCD, telescopes
The hardware design for the Mark IV camera has now been pretty well formed. This note will outline the design that exists in my head.
The camera will be approximately a one foot cube. It will have a glass top, and will be designed to sit out in the elements with no other protection. The optics will be mounted on a polar axis. It will look straight up. It will track the sky for a period like 100 seconds, then close a shutter cog back and track the sky again. With the planned 50 mm f/1.4 optics, the frames will be 35 degrees square with 1 minute square pixels. With 100 second exposures each object will appear in 80 frames. Other optics will be possible for a narrower field of view and slightly more sensitivity. Stacked sensitivity will be similar to that of the Mark III. The unstacked frames will allow tracking fast moving objects if they are bright enough.
The electronics will be designed to accommodate a variety of CCD chips. Data push from the camera to a PC resident memory card will be used. Control will be by RS-232. It will be possible to use different computers for data acquisition and control.
The hardware will be packaged in an aluminum box about 12" x 12" x 12". Inside the box, the camera will be mounted on a shaft that is angled at 45 degrees that is to be aligned with the earth's axis. Alignment for most locations will be possible by adjusting a three point mounting for the entire box. An adjustable level will be included in the box that can be set to the desired polar angle to facilitate set up.
The usual mounting of the camera will be to point straight up. At 45 N and with 50 mm optics, the camera will cover 27-63 N. Stars will thus traverse circular paths on successive frames.
Sidereal drive will be by a stepping motor positioned 10 inches from the axis. Half step drive of a lead screw will produce motion of 0.0005" at the stepping motor, which corresponds to 10" of arc per step. This will be 6 steps per pixel for the planned optics. Longer focal length optics may require a different drive scheme.
The box will be designed with a glass top. It will be removable should it prove to have more light loss than it saves in pain of operation. The design will stand continuous exposure to the elements. We plan to put a cover on our installation when it is not operating so that birds do not nest or do other things on it.
The camera head will be sealed, but not evacuated. A desiccant will reduce condensation problems. A design is planned that is similar to the Mark III.
A model servo motor will be used to move a shutter in front of the lens. Response from full open to full shut should be about 0.2 second. It will be designed like a focal plane shutter. An aperture will be moved in a single direction to go from full closed to full open to full closed.
A second model servo will allow remote focus adjustments.
The whole box will be mounted on a three point mount. This will allow a single 45 degree design to be used at a variety of locations. The camera itself will be movable in declination, probably to a few fixed locations so that several cameras can be placed at a single location to cover more sky.
Connections to the camera enclosure will include two water cooling tubes, a DB-25 data cable, a RS-232 control cable, and a connector for the TEC power.
Significant design effort will be given to lightning protection. I expect this design will survive anything but a direct hit. I plan to have an installation that can stand even that.
The camera head will contain the CCD, protection circuits for the CCD clock lines, thermisters to measure the temperature, connections to the thermoelectric coolers, and not much else. Leads will be brought out through the not so great seal scheme of the Mark III.
The fast electronics will be mounted close to the camera head. It will be designed to function at a 10 MHz pixel rate, even though the available Ford chips may barely operate at 2 MHz. The only items where speed is critical are the clock drivers and the ADC path.
We will use a clock drive scheme similar to that used by Jim Gunn for the Sloan survey. This uses capacitors charged by DACs which are then switched by CMOS devices to the clock lines. Paralleled devices will be used on the vertical shift lines where the peak current is high.
I will design the data path using the Analog Devices AD9220. This is a low power, low cost, 5 volt, 12 bit, 10MSPS flash converter. I really expect a 16 bit version of this type of converter soon. Note that for this scheme, we will almost surely be sky brightness limited. We will be lucky to have a dynamic range of 1000 to one. There would appear to be no reason to require 16 bit conversion. Once a read out scan is triggered, the entire frame will be pushed into the PC memory module. Ready or not, here it comes.
I am designing a very general purpose scan engine generate the clock sequences. It will use two PROMS. One will generate the horizontal clock sequence and the second will generate the vertical clock sequence and thus the number of lines to be scanned. I am an optimist, so the design will accommodate anything up to about a 16k by 16k device.
There will be two built in control processors. The present plan is to use the BASIC Stamp II.
One will mind an RS-232 interface. It will receive slow commands from any serial port, and will send back slow responses. It will mind things like level adjustments to the CCD clocks, run start, focus changes, monitor of supply voltages and device temperatures, scan total angle commands, number of frames to be taken, etc..
A second processor will mind the camera sequence. It will watch a clock that determines the stepping rate. It will close a shutter at the end of a frame and initiate a data scan push. It will generate stepping motor sequences to return the camera to it's starting position. It will open the shutter and start the next frame.
Again, the TEC cooling will be controlled by analog circuitry. Somewhat larger TEC modules will be used. With more experience I hope for a little more cooling that possible with the Mark III.
The simplest possible memory module is being designed. It will reside in a PC and will use only the ISA bus. It will be designed around 72 pin sims. The first units will hold at least 8 Meg of memory, but we will probably design the card with much more capacity. Byte wide data will be pushed from the camera to the PC memory card. We will handle at the minimum a 20 MHz byte rate from the ADC. There will be minimal controls. Probably just a line from the memory module that tells the camera that it is full. Probably a line to the I/O bus that says the same thing. We plan on a single byte wide unload over the I/O bus, but might implement a two byte wide path if it is not too much work.
This design should solve the problem for those of you that want to run remote over a telephone line. The primary control will come over the RS-232 cable. No reason to go to a more modern version of this. The slowest possible link should be fast enough. But there is no way to send images over this path. The control could come from the data munching computer or from somewhere else.
The data munching computer has a different but equally simple job. It just eats data. It must unload the 8 MB from a frame and store it away before the next frame arrives. I don't expect to take frames any faster than every 100 seconds, so this should not be so hard. Possibly this computer will also put up a display. One could possibly put this computer on the internet and put up an occasional display. It must dispose of an 8Mb buffer every 100 seconds. Other that that it is free. Even I can do that with QBasic.
A typical evening's run might generate 2.3 Gb of data.
A single frame from the Mark IV should be mag 2.5 less sensitive than the Mark III. We lose a factor of 2 from the smaller lens and a factor of 5 from a shorter exposure. The lens size is just what is possible for the short focal length, and the shorter exposure is required to prevent pixel saturation from the sky background. A star will appear in about 80 frames. By stacking these we should get close to the Mark III sensitivity.
A single Mark IV will cover 35 by 35 degrees. This is over 200 times the area covered by a Mark III camera. This should be a good tool to search for nova, near earth asteroids, gamma ray bursts, and of course, variable stars.
I presently have 11 2k x 2k Ford chips that I hope will be usable for the Mark IV. If they work, then I will be able to build 10 or so Mark IVs at a similar cost to that of the Mark III. If the Ford chips don't work, then I will have to buy Loral chips at $2000.00 and will only be able to build a few until I get help.