ST-8E USB Upgrade Evaluation

May 11, 2002

In mid-April, SBIG offered me the opportunity to beta test an ST-7E USB prototype. As a result of that experience, I changed my earlier position and decided to upgrade my ST-8E camera for USB operation. I am glad I did. The upgraded camera came with a USB cable, CCDOPS software version 5.10 and included updated drivers for USB operation. The following report gives some data and impressions of USB in action. Remembering that the entire camera electronics were replaces as part of the upgrade, this seemed like a good opportunity to evaluate electronic differences with the same imaging chip. Unless otherwise identified, all data is from my ST-8E.

Equipment Used

My observatory is remote from the operating computer so I used three 16-foot USB extenders and located a powered USB hub in the observatory. No other USB devices were attached to the hub. The computer is a 1 GHz Athlon with 512 MB RAM running Windows 2000 SP2. In addition to installing CCDOPS, I also have CCDSoft v5.00.055 beta and Maxim v3.06.

{ST-7E} The camera was attached to my 10 R-C scope with an Astro-Physics 0.75x Focal Reducer, giving an effective image scale of 1.0 arc-sec./pixel. The scope is on an Astro-Physics 900GTO mount on a permanent fixed pier.

Installation

CCDOPS installation was used for initial runs and went smoothly after carefully following the USB installation instructions. After setting it to USB, it recognized the camera and I was able to set the cooling and do some dark field downloads. One of the features of this version of CCDOPS is to set the entire monitor screen to black during image acquisition and download, with only a progress bar showing. I do not consider this desirable. This does not happen during an autograb sequence.

I also successfully got the camera recognized in Maxim v 3.06 by copying the SBIGudrv.dll file from the CCDOPS sub-directory to the Maxim 3 sub-directory. CCDSoft v5.00.055 recognized the camera directly.

Image setup and calibration

{ST-7E} The automatic focus routine, @Focus, has to be seen to be believed when running at these kinds of download speeds! It took around 75 seconds to run through the entire sequence, using the stepper focus motor on the secondary mirror and a RoboFocus controller. It was a delight to operate and seemed very repeatable.

{ST-7E} Taking flats is always tedious, waiting for a long download after a brief exposure. With this camera, I ripped through 10 light box flats, 10 flat-darks, and 10 bias frames in almost no time. I was able to capture sets of sky flats for each color filter with minimal average value change, due to the fast download times. Using CCDSoft, I set took a single image to determine the exposure then quickly took 5 frames with auto dark enabled. I was able to take four filters in less than 10 minutes. With practice, Im sure Id be able to get even more. There is no doubt that high speed download will greatly improve your image setup time.

Flats and Bias frames were taken in order to assess gain, readout noise and background artifacts, since the internal electronics were completely replaced during the upgrade. Bias frames were used for this evaluation because of their very low signal level and with the expectation that any electronic artifacts would be more apparent. The camera was operated at -10C. Gain was measured, using a methodology Stan Moore adapted from "Handbook of CCD Astronomy" by Howell. Since the same CCD chip was used before and after upgrade, I was able to see that the effect of the upgraded electronics would be. Here are the results:

 

ST-7E USB

 

ST-8E Parallel

 

ST-8E USB

Gain (e/ADU)

2.6

3.0

2.6

Readout Noise (e RMS)

15

12

15

Gain is a bit closer to the theoretical 2.3 with the upgrade and was very similar to the ST-7E prototype. There is a slight increase in readout noise but still within spec. However, there is a definite improvement in how the new electronics handle streaking due to hot pixels. 10 bias frames were taken at each binning before and after the upgrade using Maxim. These images are shown below:

1x1 bin, Parallel                                                           1x1 bin, USB

The left image shows low level streaking from hot pixels that is significantly reduced with the USB electronics. Also, the shading, which goes from roughly upper left to lower right is exchanged for a much-reduced left to right shading. These two phenomena are plotted below. The vertical axis scale is 5 ADU per major division.

Average value of each column

Note that the magnitude of streaking and the overall noise is lower with the USB electronics. The vertical value of column 273 was 154 ADU with the parallel electronics and 111 ADU with the USB electronics. This reduction is greater than that due to the gain difference of 3.0 to 2.6, indicating better processing in the USB electronics.

Average value of each row

Note that the vertical shading is substantially reduced with the USB electronics. This is apparent in the images as well.

Here are some comparison frames at 3x3 binning.

3x3 bin, Parallel                                                                              3x3 bin, USB

Again, note the reduction in streaking from the hot pixels with USB.

First Image

{ST-7E} I decided to image NGC4319 to see how well guiding interacts with image acquisition. First I started the autoguider. I was able to get an acceptable star with an exposure of 3 seconds. I set the aggressiveness at 9 with no backlash compensation. I set up a script to take 21 10-minute exposures. Here is a plot of the tracking log during one-600 second interval.

The standard deviation of the tracking error for RA was 0.30 arc-sec. and DEC was 0.18 arc-sec. Seeing averaged 3.6 arc-sec. With the high-speed download, there was none of the guiding transient that one normally sees with a parallel download. The old rule of thumb of leaving a delay of 3 to 4 times the guide exposure can probably be significantly reduced.

The image acquisition went smoothly and the resultant image is shown below.

NGC4319

The processing consisted of dark subtraction and flat field in Mira, 5 iterations in CCDSharp, curves, levels and unsharp mask in Mira. This is a fairly tough object and tracking at near the pole has its own set of challenges but everything seemed to go well.

Download speed

I captured 10 bias frames in each program at binning ratios of 1x1, 2x2 and 3x3. All frames were taken with the CCD temperature at -10C. I measured the overall time to capture and write to disk for 10 frames. Here is a summary of that data:

 

1x1

 

2x2

3x3

CCDOPS v5.1 Autograb 10, per frame

6.7

2.7

1.8

Maxim v3.06, series of 10, per frame

7.5

3.0

2.1

CCDSoft v5.00.055 beta, series of 10, per frame

6.8

2.7

1.8

 

 

 

 

{ST-7E} CCDOPS v5.03r11 Autograb 10, per frame

2.4

1.4

1.2

{ST-7E} Maxim v3.06, series of 10, per frame

2.6

1.7

1.5

{ST-7E} CCDSoft v5.00.048 series of 10, per frame

7.4

4.0

2.9

During the beta test, I had noted the relatively poor download speed of CCDSoft compared to the other applications. Also, there were low-level artifacts introduced into the bias frames by CCDSoft. To a lower level, there were some left-to-right stair-stepping with all applications, again at a relatively low level. The ST-8E USB tests were all done with CCDSoft v5.00.055 that is currently in beta test. Not only is the download speed significantly improved, but also I saw absolutely no level of low level artifacts. Software Bisque folks did a great job of cleaning up the software. SBIG also reduced the stair-stepping, at least based on my sample, to an almost non-existent (less than 1 ADU) level. Here is an example of the improvement made by both companies:

ST-7 USB beta                                                            ST-8E USB upgrade

Both frames represent a mean combine of 10 3x3 binned bias frames at -10C. Each combined frame was then median filtered by a 2x2 median to remove some noise and hot pixels and illustrate the patterns. The image on the right is remarkably pattern free, with only the slight left to right gradient remaining. Again, the magnitude of the ST-8E gradient is very small at 28 ADU.

FocusMax

{ST-7E} FocusMax is a freeware focusing application based on measuring the half-flux diameter of a star and developing a predicted focus slope with which to predict the actual focus point. This curve has to be developed for each optical system but, once developed, is an extremely fast and repeatable application. By taking advantage of the fast download, FocusMax achieved focus in 20 seconds! A similar operation using parallel download takes easily 60 plus seconds. Even with taking 10 near focus exposures instead of the default 5, the focus time is lengthened to only 27 seconds.

Color Imaging

{ST-7E} I attached my CFW-8A to the camera and used it for color imaging of M97. Everything worked fine. The acquisition was via script and consisted of 21 x 5m L, 7 x 5 m RG and 7x8m B. RGB was binned at 2x2. Darks were taken as part of the script. Sky flats were taken at twilight. The resultant image is shown below.

M97

Note the very faint galaxy in the bottom center of the image. A good flat field is required to pull a faint object like this out. The ability to take quick sky flats gives the best accuracy flat field correction. I saw no issues at all from using this camera for imaging.

Cooling

{ST-7E} I was curious about the cooling efficiency of the new design. My ST-8E has the two-stage cooler and the ST-7USB has the improved efficiency single stage cooler. Both coolers were set at -35C and given an hour to reach their best cooling point. Using the CCDOPS reported ambient temperature, the ST7USB achieved a delta of 42C and the two-stage cooler achieved a delta of 41C. The advantage of the new design of course is one less power supply and the attendant cable that has to be routed to the camera.

Summary and Impressions

With USB, I did not notice any of the computer cycle hogging that is apparent with parallel downloads. After a while, I let image capture go on while I did other things on the computer and saw literally no impact from the download activity. I would speculate that there is some memory buffering in the camera to allow the computer to take it when it wants, since USB is a polled protocol. In any case, USB puts an end to the parallel port taking over the machine.

High-speed download has a lot of fringe benefits that are not apparent at first blush but will come in handy mostly in image setup. Getting focused, calibration frames, etc. go much faster. It is easier to get a larger number of sky flats close to the same average level and even auto dark them in the process! Focusing, whether manual or automated, goes much faster and it is not a big penalty to take multiple focus runs or multiple hear focus samples to insure accurate focus. During auto guided operation, a mount will have less time to drift during the download period, resulting in a smaller transient when guiding is restarted. A lot of the tedium goes away with fast downloads and one can concentrate more on optimizing the conditions for a great image. It is good to see the major applications such as CCDSoft, CCDOPS (of course) and Maxim work so well with the USB cameras.

The observed low-level artifacts are well within the ability of image calibration to remove. I was also pleased to see that the readout noise did not increase above 15e RMS, even with the increased speed. I was especially pleased to see how the column streaking that resulted from hot pixels in my CCD chip was greatly reduced by the new electronics. I would guess that SBIG used the opportunity for electronics redesign to clean thing up considerably.