Michael A. Covington    Michael A. Covington, Ph.D.
Books by Michael Covington
Previous months
About this notebook
Search site or Web

Daily Notebook

Links to selected items on this page:
Audiotex "Super Amplifier," 1965
Flickering lights that are on a dimmer
Master Browser (Browse Master) under Windows 10/11
M40 and galaxies
M100 and other galaxies
M102 and NGC 5906
Sharpless 2-101
Many more...

This web site is protected by copyright law. Reusing pictures or text requires permission from the author.
For more topics, scroll down, press Ctrl-F to search the page, or check previous months.
For the latest edition of this page at any time, create a link to "www.covingtoninnovations.com/michael/blog"


In praise of short exposures


If the star images in this picture of M13 look a bit sharper than the ones in the previous day's picture (below), there's a reason. It's a stack of thirty 30-second exposures instead of fifteen 2-minute exposures. What you are looking at is downsampled 3× from the original image, so you may not see much difference. In the original, it's there.

(Don't compare the overall sensitivity, as this is only half as much total exposure. But I don't see much difference.)

As you know, when I take long exposures with my Celestron 8 EdgeHD, the tracking isn't perfect, and the stars are slightly elongated in an east-west direction. There are several possible reasons:

(1) The image scale of 0.6 arc-second per pixel may set an impossibly high standard. Nobody tracks quite that well. Good tracking is 0.6" RMS, not 0.6" peak-to-peak. That remains the case with more expensive mounts and drives.

(2) Tracking in right ascension (west-east) is inherently rougher than in declination (north-south) because there is always a motor running from west to east, to track the earth's rotation, while the north-south motor runs only rarely, when corrections are needed.

(3) There could be differential flexure between the camera sensor and the guidescope. That could be anywhere, and there's surely some of it in several places, but the main culprit is almost certainly shifting of the C8's mirror. The telescope focuses by moving its main mirror, so even though a lockdown mechanism is provided, the mirror isn't immune to movement.

Differential flexure of all types is worst when the telescope is aimed high in the sky (as it was here) because the direction of gravity's pull is changing relative to the telescope.

Since upgrading Losmandy's firmware to give higher-resolution tracking, (2) is much less the case than before. That leaves (3). In fact, the evidence that really incriminates (3) is that there is a shift from frame to frame, not just within frames, and it's in the same direction as the star elongation.

One cure would be to use an off-axis guider. Those are tricky to set up, and I'll probably get into that later. (Cameras with built-in off-axis guide sensors are starting to appear, but the only available one, the ASI2600 Duo, requires a bigger illuminated field than this telescope gives.)

But in the meantime I did something else.

I changed guidescopes, so that instead of a large guidescope riding on top of the telescope, I'm using an iOptron iGuider, 30 × 120 mm, bolted to the side of the mount saddle, so there's much less to flex. With PHD2 multi-star guiding, it guides just as well.

And I switched to shorter exposures. I figured that whatever flexure might be happening in a 120-second exposure, there'd be only a quarter as much in a 30-second exposure. And I was right.

At ISO 3200, at my site, the Canon 60Da still picks up enough light that the sensor noise is swamped by the sky background (as it should be). That wouldn't work so well at a darker site; here the city lights are working in my favor. (Or at least are prevented from taking their full toll.) This is comparable to "high conversion gain" (HCG) on astrocameras, where we gain sensitivity, and cut the sensor noise measured in electrons, at the expense of dynamic range. It doesn't matter, because stacking a large number of exposures gives us the dynamic range back.

I'm going to experiment further with deep-sky imaging with perhaps 100 or 120 30-second exposures instead of a smaller number of 2-minute exposures, when this telescope is in use. It's one more way to get the most out of it.

[Added:] There are two more things I need to try.

One is adaptive guiding. If the shift shows up from one frame to the next, then after a few exposures have been taken, the shift can be measured and the autoguider can be told to account for the shift during the exposures, gradually shifting the desired position of the guide star. To do this, there is an experimental plug-in for N.I.N.A. 3 by someone I'm corresponding with. If it is announced publicly, I will post a link to it here.

The other is automatic deconvolution. That is, figure out what distortion has happened to the image, and undo it by computation.

This is possible because the stars are excellent test instruments. Stars as seen from earth are essentially perfect point sources (under 0.1 arc-second in apparent diameter). So the image of a star on the sensor is, essentially, a record of the system's optical imperfections, including imperfect guiding.

Russell Croman's BlurXTerminator software uses a trained neural network to analyze the distortion of stars in an astronomical photograph. The neural network delivers the deconvolution function itself, not just the parameters for it, and that led to concern that it might simply substitute aspects of a star image from the training set, thus making your image unsuitable for scientific work. I've looked into this and am assured it is not the case. All it's doing is deconvolution. I think I'm going to try that.


M13 in moonlight


The globular star cluster M13 looks magnificent even when photographed under poor conditions. I took this picture in moonlight, testing how well my C8 would guide with an iOptron iGuider (a tiny 30mm guidescope) instead of the usual one twice as big; it guided well. I was also putting N.I.N.A. 3.1 through its paces by doing a sequence, where the system would photograph one object, then move to another, check aim by taking a test exposure, and then photograph that. Again, success. This is a stack of 15 2-minute exposures with a C8 EdgeHD and f/7 compressor and Canon 60Da at ISO 800.

Near the top edge of the picture, toward the left, you can barely see the distant galaxy IC 4617, magnitude 12.6.


The dullest Messier object

The dullest object in Charles Messier's catalogue of star clusters, nebulae, and galaxies is M40, which consists of two stars. Not even a double star — the two stars are not close to each other in space — and there is nothing nebulous (foggy) about their appearance. Messier included M40 because Hevelius had reported a nebula in that area, and, finding none, Messier figured the pair of stars may have looked nebulous in a more primitive telescope.

I took aim at M40 last night and was gratified to capture several distant galaxies along with it (none of which are bright enough to have been seen by Hevelius or Messier). Stack of 19 2-minute exposures, AT65EDQ 6.5-cm f/6.5 refractor, Canon 60Da at ISO 800. This was primarily an equipment- and software-testing session, to become more familiar with telescope control by N.I.N.A. 3.0, and the moon was in the sky.

M40 is to the right of center in this picture:



M100 and other galaxies

To test my ability to remote-control the telescope from inside the house, I then selected a galaxy-rich field that includes M100 (the large spiral at the upper right in this picture) and aimed the telescope at it, confirmed its position, and took 30 2-minute exposures, one of which was marred by a satellite trail. So here's the resulting stack of 29. This is not an impressive picture, and it was taken in moonlight, but count the galaxies and think of how much of the universe you are seeing.



And just when I got this far putting N.I.N.A. 3.0 through its paces, I got the notification that N.I.N.A. 3.1 has come out.


Laptops are slow to recognize Windows workgroup
Master Browser (Browse Master) under Windows 10 and 11

Problem: My laptops were often slow to recognize the Windows workgroup that they belong to when I use them at home. After waking up a laptop, attempting to connect to the server's shared files could take as much as five or ten minutes; the first attempt would often report a timeout (network failure).

Curiously, while the problem was present, ping servername would succeed, but net view servername would fail, but a second attempt at net view would generally succeed and would clear the problem, whatever it was.

Apparently, the problem had to do with finding the Master Browser (the computer in the workgroup that keeps track of who's who; the term has nothing do with web browsing). That still works the way it did 20 years ago (despite scant documentation about it for recent versions of Windows), and, apparently, adding the IsDomainMaster registry setting to the server (a Windows 10 desktop) solved the problem, just the way it would back then.

Not to overlook the obvious, before taking that step, I checked that the server was set never to go to sleep, since the initial symptoms seemed to indicate that the server was in sleep mode.


Late-night astronomy

All through college, I never "pulled an all-nighter" to study. In my whole life the only times I had been up all night were medical emergencies, overnight flights, one nearly-all-night entertainment event at Yale in 1980, and one astronomy session at Yale in March 1982, using an old observatory that was soon to be decommissioned.

But on the evening of June 7, I decided, after consulting Melody, to make the most of unexpectedly clear weather by observing very late, then sleeping the next day (Saturday). This is something I may do again, when the sky is clear and I have the following day free (a very rare situation). Much of the time, I actually rested indoors, watching the progress of my astrophoto session from a laptop computer through a network connection to the other laptop that was controlling the telescope and camera. Below are the results.

What do I mean by "processing"?

My friends on Facebook have been waiting for me to "process" these pictures so they can see them. What is meant by "process"? Why can't I give you the picture straight out of the camera?

Well, first, every picture has to be processed for you to see it. The camera records digital signals, not anything you can look at. (Just as a film camera recorded an image chemically, not in a form you could see.) In daytime photography, our cameras usually do the processing automatically and show us a picture immediately. That doesn't mean the picture is unprocessed. It means that a lot of decisions were made automatically by the software in the camera.

Second, astronomy is different from daytime photography. The way the picture should render faint light is completely different and requires manual intervention. Click here for more about this.

And third, there's a lot more going on. Each of my astrophotos is a stack of dozens of individual pictures, taken one after another and then combined by computer. Along the way I reject the ones that are marred by trails of airplanes or satellites. Also, I apply corrections ("darks and flats") to counteract measured imperfections of the telescope and camera. Even a speck of dust on the sensor will be removed this way if it stays in the same place throughout the session, as specks of dust normally do.

M102 and NGC 5906, Spindle and Splinter

Here you see, on the right, the galaxy M102, and on the left, NGC 5906. They are known, respectively, as the Spindle Galaxy and the Splinter Galaxy.


You can also see a few other galaxies. Here's the same picture, annotated by PixInsight:


Stack of 30 2-minute exposures, Canon 60Da at ISO 800, AT65EDQ 6.5-cm f/6.5 refractor, Losmandy GM811G mount, iOptron iGuider (30×120mm) guidescope, controlled by N.I.N.A. 3.0 and PHD2 under Windows 11.

Recall that M102 was once thought to have been a duplicate observation of M101 (and I did some of the historical research to finish establishing that it wasn't). Comparing this picture to M101 below, you may wondering how in the world anyone could mix them up. Well, the answer is that in a small telescope, you see only the bright central region, which looks very much the same in both galaxies; in fact, M102 is easier to see.


The magnificent spiral galaxy M101 was my first, and main, target for the night; I used the same equipment as above but combined 49 exposures. Here you see first the central region of the picture enlarged, then most of the field, then an annotated version of the image.




In the annotated image, you can see that several bright spots in M101 got NGC numbers of their own; nobody at the time knew they weren't separate objects.

Star clouds and Sharpless 2-101

By 3 a.m., Cygnus had risen high enough that I was able to photograph objects in the plane of our own galaxy. Stellarium (my computerized star atlas) showed an interesting-looking nebula but did not name it. I found out from other atlases that it is Sharpless 2-101 (it doesn't have an NGC number). It's the red object near the center of the picture below. Elsewhere you see clouds of stars partly obscured by dark clouds of interstellar dust. The bright star is Eta Cygni.


Dad's Electrical Hints, #1:
How to fix flickering lights that are on a dimmer


If some of the lights in your house tend to flicker, and they are on a light dimmer (whether or not the dimmer is turned all the way up), here is something that may fix the problem.

Simply run the dimmer quickly back and forth from low to high, through its whole range, about 20 times. You can do this with the lights turned on or off; it won't harm them.

This worked for me in no less than 3 places tonight.

The cause of the flicker is something I had long been familiar with in other electronic equipment: dirty contacts in the potentiometer (variable resistor) that controls the dimming. Running it rapidly back and forth through its whole range cleans the contacts to some extent. If the cure doesn't last, you probably need a new dimmer.

Same goes for volume controls in your built-in speaker wiring, if any. Turn them back and forth all the way, twenty or thirty times. They are likely to come out sounding a lot better!

There are other reasons lights can flicker. LED bulbs often flicker because of poor electrolytic capacitors in their current regulators, and, paradoxically, improve with use — an electrolytic capacitor renews itself electrochemically when a voltage is on it, but not when it's idle.

Or it may be a wiring problem. If lights in different rooms flicker in unison, that's definitely what it is, and you should get it fixed immediately, since it's likely to be a fire hazard.

Dad's Electrical Hints, #2:
Getting the words right

This is a plug:


This is NOT a plug:


It is called an outlet, receptacle, or socket. Definitely not a plug.

Yesterday I watched a very confusing video by someone who called both of them by the same name and was replacing the "plug" (the outlet) because the "plug" wouldn't make a good connection in it.

Please, people. A one-word vocabulary is not enough!


More about the Super Amplifier

After more thinking and more circuit simulation, and some conversations by e-mail with Gene Greneker, K4MOG, I've come up with another candidate circuit for the Super Amplifier. This one has fewer parts, is more a 1960s-style design, and, according to the simulation, should work with either silicon or germanium transistors.


This isn't the only way they could have done it, but it's probably close, or at least closer than yesterday's proposal. It's a high-gain direct-coupled amplifier, as advertised. It expects a low-impedance microphone (using a speaker was recommended). And it would be damaged if an 8- or 4-ohm speaker were connected to the output — but would drive a 45-ohm speaker with reasonable volume.

Q3 was included to account for those last two observations. If the output were always going to a 2000-ohm earphone, the earphone could go in place of R5, and Q3 would not be needed. But the manufacturers probably didn't want to make the gain of the amplifier depend on the resistance of the earphone.

(Embarrassed note: How did I get the designation R5 in a circuit with only four resistors? By simplifying the earlier circuit and not renumbering.)

When high-gain direct-coupled amplifiers are mentioned, one thinks of Darlington pairs. But although they make great emitter followers, a Darlington pair does not actually make a good high-gain common-emitter amplifier. The reason is that the gain is limited by the emitter resistance re, which is no lower than with a single transistor. Also, my understanding is that for leakage reasons, they work much better with silicon transistors (although Mr. Darlington invented them during the germanium era).

Also, with its 1.4-volt base-emitter drop, a silicon Darlington pair is hard to use with a 3-volt supply. If, as the brochure seems to indicate, the Super Amplifier also operates on 1.5 V, then a silicon Darlington pair is completely ruled out.

This circuit still hasn't been built and tested. It is not even the top project on my workbench at the moment. But stay tuned...

Additional note: The best free circuit simulation software that I can find comprises two packages given away by Texas Instruments, Tina-TI (not revised since 2018) and Cadence PSPICE-for-TI (current). (These are for serious analysis, not beginning hobbyists.) Both support a wide variety of devices, including the TLC555 timer, and both can import SPICE models from other sources (as I did to get germanium transistors). I'll probably post some notes about these in the near future.


Remembering the Super Amplifier

My father and I did three memorable electronics projects together in 1965 and 1966, before his untimely death. The last and most elaborate was an analog computer; click here to see the details. Before that was a crystal radio that I plan to write about (I lost the original but built a replica in the 1990s). But the first of them was the Super Amplifier.

Click to see the whole page

This happened in the summer or early fall of 1965, when we were living in a rented house on Ridgewood Drive in Valdosta. At Specialty Distributing Company, my father bought for us a module called a Super Amplifier described as follows:


I've learned that the list price was $7.95, which is something like $90 in today's money, so it was not a trivial project.

We mounted it on a scrap of 2x6 lumber from our new house under construction on Lake Drive. It was powered by two C cells in a Keystone metal battery holder of a type that I think is actually still made. I don't recall an on-off switch; to turn it off, we just took out the batteries.

We used a speaker for input and a dynamic headphone (from my Knight 100-in-1 kit) for output. Sure enough, we could hear at great distances — the sound of my dog (Blondie) sitting and panting, six feet away, almost overwhelmed it. We also heard voices that we thought were a radio or TV in some house nearby; I now think it was probably stray pickup of WGAF, whose signal in northwest Valdosta was very powerful.

Later, in the new house, we built a crystal radio and used the Super Amplifier to amplify the output. WGAF came in loud and clear. We found that the Super Amplifier would drive a speaker (which was a 45-ohm intercom speaker, though I don't think either of us realized at the time that it was different from common 4- and 8-ohm speakers). After several experiments with this, the Super Amplifier stopped working, and that was the end of it. My dim recollection is that it failed when a regular 4- or 8-ohm speaker was connected to it.

In later years I've found that this Super Amplifier module was surprisingly little known. The manufacturer's brochure has somehow made it onto Archive.org (click here to see it). This line of modules was advertised very briefly in Radio-Electronics in 1965 (September, page 80) and later popped up as a Radio Shack product (Radio-TV Experimenter, October-November 1967, page 13). Our module was gray, but apparently they were also made in red, as shown on the cover of Popular Electronics, March, 1966. And that's all the knowledge the world has preserved about them. Maybe the total number made was unexpectedly small, or the prices were too high.

So... How did it work? The brochure describes it as "direct-coupled," which is an important clue about the circuit. We also know that it failed when driving a low-resistance load, but first, it successfully drove a 45-ohm load for a while with good volume; that suggests that the output was an emitter follower. And we know the overall gain. And it ran on 3 volts (the diagram in the brochure, intriguingly, says 1.5).

With all that in mind, I used Tina-TI to simulate some circuits, and here's a plausible candidate. This is a bit anachronistic because it uses silicon transistors; they were on the market then but cost over 50 cents each, so unless Audiotex got a quantity or surplus discount, they more likely used germanium. Germanium transistors are now rare and hard to get; electric-guitar enthusiasts have slurped up the world's remaining supply to make sound-effects devices. So here's a circuit using silicon 2N3906's. It should work with germanium transistors if two resistors are changed as marked. The other slightly anachronistic thing about this circuit is the large number of resistors for stabilization; that is late-1970s practice, probably considered a bit pedantic in 1965, so they may have cut some corners that I didn't.


This has not yet been built and tested; I hope to get to that soon. (The last time I hoped to get to it soon was 5 years ago, so don't expect too much.) At least I wanted to record the information shown here, in case this is as far as this project goes.

See also the next entry, above.


If what you are looking for is not here, please look at index of all months.