Dell Inspiron 13 7347 Review

UPDATE 21 June 2015:

The problem has not been resolved, but I have managed to prevent the ghost touches from flaring up by being very careful with the notebook.  Do not change the screen angle by grabbing the screen with one hand - use two hands, and ensure that the screen doesn't bend. Do not carry the notebook by its screen! (this works fine with Macs and Toshiba Z30's). Do not apply pressure to the back of the screen. In other words, treat the laptop like the fragile flower (of lemony-ness) that it is.

In restrospect, the same-priced entry-level Macbook Pro Retina is a far better choice, even with less RAM and a small SSD. The build quality is really up there; the fact that used Macs hold their value over the years really tells you something about Mac build quality.


UPDATE 21 November:

Sadly, I can no longer recommend this particular model.  It has a severe design flaw in the touch screen that causes "ghost touches" - and based on my research and the reports of others on the Dell support forums, this is not an isolated issue.  Unit replacement most often does not address the issue, hence my unit is not a lemon unit.  Since Dell makes refunds very difficult and my feeling is repair or replacement won't help, I am keeping the notebook (but disable the touch screen when it acts up).



I recently found myself in need of a notebook.  I haven't bought any such thing in over five years (the employer-supplied notebooks - a variety of Dell Latitudes and Toshiba Porteges - have proved more than useful).  However my office notebook isn't due for refresh yet, and I was getting tired of hauling its magnesium-skinned bulk around.

As my wife had recently gotten a Toshiba Portege Z30, which weighs an amazing 2.65 pounds - I decided that weight would be my major selection criterion.  But I also wanted a full HD (or higher) IPS screen, and a backlit keyboard.  And the notebook would also need to be cheap.

The above set of criteria are very, very hard to meet (even the Portege Z30 and ThinkPad X240 do not meet all of them, and these laptops cost a lot more money than I was willing to pay).

In the $1000 range, the notebooks which meet most or all of the above criteria are the Lenovo Flex 2 13 (which weighs about 4 lb), the Asus ZenBook UX303LN and recently-discontinued ZenBook UX32LN (the UX32LA and UX303LA don't have IPS screens), the Dell Inspiron 13 7000, and of course the Macbook Pro Retina 13.

I am no Apple fanboy (I have a lot of Android devices) but let's get this clear - the Macbook Pro Retina 13 is overall the best choice. It has the highest-resolution display (although not the highest-contrast); the fastest SSD (that isn't field-upgradable); the fastest processor (a special model i5 that only Apple gets); the best build quality; and by far the longest battery life. However, the 128GB SSD model (the lowest-end one) was at the high end of what I was willing to pay, and the 128GB SSD is small.

On paper, the ASUS UX303LN comes very, very close to the MacBook Pro Retina 13 in terms of specifications, and the ASUS has an Nvidia discrete graphics solution.  However it costs more than the 128GB SSD MacBook Pro Retina 13 (and a bit less than the 256GB SSD model) and doesn't come with an SSD.

The recently-discontinued UX32LN costs less than the base MacBook Pro Retina 13, also has discrete graphics, but no SSD.  Both ASUS devices weigh about 3.2 lb, slightly less than the Apple, but when you heft them in your hand, the difference in build quality is obvious (even though the ASUS notebooks also have aluminium chassis). The ASUS notebooks have decent battery life as well, but they can't match the Apple.

The Lenovo Flex 2 14 is probably the cheapest option you can get at this time (October - November 2014) with a full HD IPS screen.  It is a pentile screen and not as good as the other FHD screens. The large downside with the Lenovo is its 4 lb weight.  Battery life is acceptable.

My fourth option was the Dell Inspiron 13 7000.  It is available in two models, a 1366x768 IPS, and a full HD IPS. The latter model costs more (about the same as the ASUS UX32LN) and has the downside of not having a discrete graphics solution (it uses the Intel 4400).

I made a handy comparison guide of these notebook candidates:


The last parameter is how long (in hours:minutes) each notebook could loop the "Big Buck Bunny" movie before running out of charge.  It is obvious from this comparison that the MacBook Pro Retina 13 is the best overall choice, but also the most expensive, and this is the base model only with a 128GB SSD. Prices are in SGD, but the other ratings are from notebookcheck.net.

Note that the usual suspect high-end notebooks (Lenovo ThinkPad X1 Carbon, Lenovo ThinkPad X240, Lenovo Yoga 2, Toshiba Portege Z30) are not on this list, for the simple reason that they cost far more than I am willing to pay.  It was quite a shock to learn that high-end notebooks actually cost more than their Apple competition.

What the above spreadsheet does not show is the level of build quality of these notebooks, something you can't see unless you physically handle them.  As I have already mentioned, the MacBook Pro has the best build quality, the best trackpad, and the best keyboard.  It's the most solidly put-together.  The Lenovo Flex 2 14 is also pretty solid, but its weight goes against it.

From the spreadsheet, it looks like if you can't or don't want to spend money on the Apple, the ASUS is the next best choice, and on paper it is.  However, it doesn't feel as well put-together.  The reason I ended up with the Dell is because the Dell simply feels more solid in the hand, for the same price, in spite of the weak integrated Intel Graphics.

I used to have an ASUS with Radeon graphics, and - it's not all it's cracked up to be.  Certainly you can play most modern games, but the battery life with the discrete graphics enabled is a dismal one hour or so.

Now that I've put a reasonable amount of time into the Dell, these are the conclusions I can make:
  • The screen is very good; certainly the Retina is better, but having used 1366x768 matte TN panels for almost a decade, this screen is an eye-opener.
  • The build quality is above average; as the notebookcheck review points out, the build quality is more Latitude-level than Inspiron-level and is the main reason I chose this model over the ASUS UX32LN which has a better spec and significantly longer battery life.
  • The Yoga-like 360 degree rotation (that turns it into a huge ungainly tablet) is actually a useful gimmick.
  • However, Windows 8.1 on this device doesn't quite cut it as a tablet OS. In spite of the vastly more powerful processor, my Nexus 7 2 works better and is more responsive as a tablet than the Inspiron. However the Nexus 7 2 can't run Microsoft Word, so there is that.
  • The touch screen is less responsive than that on the Nexus 7 2 (which also is a Full HD IPS touch screen). It is sometimes necessary to press harder or multiple times (particularly with the included passive stylus) to get click registration.
  • The track pad on the Inspiron 13 7000 is of mediocre quality, particularly the click-over feel is rough and the actual feel as you move your finger over the trackpad is also rough. The track pad on my employer-issued Dell Latitude E6320 is far, far better, and the glass pad on the MacBook Pro is beyond comparison that it's not even funny.
  • The keyboard is also only acceptable, my employer-issued Latitude E6320 is also better. However the Inspiron keyboard is illuminated (only two illumination levels however). The notebookcheck review noted the excessively-small backspace key.  Another niggle is the absence of Page Up and Page Down keys; you need to press Fn-Up Arrow and Fn-Down Arrow. I can still type at a reasonable speed on this keyboard, so it's acceptable. It is important to point out that the Dell Latitudes are a higher grade of notebook and generally cost much more.
  • Of all my candidate notebooks, the Dell has the worst battery life (four hours on Big Buck Bunny).
  • However the Dell Inspiron 13 7000 is reasonably light, practically the same as the MacBook Pro Retina 13.
In conclusion, the selection of (approximately) 3 lb notebooks is rather thin, particularly if your budget is limited to around $1000.  You can't get perfection in this price class, unless you're willing to pay a bit more and live with a small SSD in the MacBook Pro Retina.

Another conclusion I've come to - if you are spending $1500 and up on a flagship notebook, forget the so-called premium notebooks from Lenovo, ASUS, HP, etc.  Just get an Apple device, they actually provide better bang for the buck and unsurpassed battery life and build quality. That's assuming your applications run on MacOS X.

So the Dell Inspiron 13 7000 is a good compromise, but a ways off from perfect.  It was the least compromised choice for myself, but someone who wanted longer battery life and a discrete graphics solution would probably find the ASUS UX303LN or UX32LN to be a better choice overall.

UltraBookReview has a good review comparing the ASUS UX303LN to the MacBook Pro Retina 13. The mere fact that the ASUS is favorably compared to the Apple device is already quite a big vote in favor of the ASUS. The UX303LN simply was a bit too much money (about $300 more than the Inspiron 13 7000) for me, and the UX303LA while about $300 less than the Dell, didn't come with an FHD IPS screen.

Skywatcher Star Adventurer Tripod

I wanted to find a more suitable tripod than the flimsy Benro A350EX, and less clumsy than the Vixen Polaris tripod.  I initially purchased a Sirui N2204X, carbon fibre, with detachable leg, etc. It was not terribly stable (about 4-5 seconds damping time) in spite of its alleged 33lb rating.


But, due to a 20% discount promotion at KEH, I was able to purchase a beat-up Gitzo G1340 Pro Studex tripod, without centre column.  That last part is important, because the centre column induces instability. Coupled with the rubber bungs on the Gitzo's feet, damping time is now 2-3 seconds. The Gitzo is actually more stable than the larger aluminium tripod from my Vixen Polaris. This is a tripod that's rated 20lb.  The total weight of this setup (Stellarvue SV80ED, diagonal, eyepiece, mount, and tripod) is just a hair under 20lb, and can easily be lifted and carried around.

Of course, Gitzo tripods are eye-wateringly expensive new; a major drawback.  A new 2-series Gitzo carbon tripod and the Star Adventurer would cost very close to an Atlas or EQ6, so definitely not cost-effective anymore.


I guess we all can derate Chinese tripod weight ratings by 50% now..

Sirui R2004 Tripod for Astronomy

I was looking for a lightweight tripod that would fit in carry-on baggage and be useful for my Star Adventurer.  The Sirui R2004 tripod seemed like the perfect fit: it is 520mm collapsed (will fit diagonally into a carry-on suitcase), weighs about 1.8kg, and is rated for 15kg.

It also only costs S$85 ($68 US) which is mind-bogglingly cheap.

All of the Sirui 2-series tripods (R2xxx, N2xxx, T2xxx) have 28mm thick legs, only differing in the number of leg segments, whether one leg can be detached to form a monopod, and the material of the legs - aluminium or carbon fibre.  The R2004 is the most basic model, and hence the cheapest.

As payload I used a Sky Watcher Star Adventurer, APM Lomo 80mm apochromatic triplet in a William Optics tube, 2" Televue diagonal, and a 4mm Vixen LV eyepiece giving 120X magnification.  The total weight of all this is 8 kg, well below the alleged 15 kg capacity of the Sirui tripod.

The long and the short of it: at 120X, it takes a mind-boggling 8 seconds for vibrations to damp out after touching the focuser. Hence this tripod is wholly incapable of carrying the mentioned load.

As a control, I used the short aluminium tripod from a Vixen Polaris; this has the benefit of much larger legs, is much shorter, and has a tray to reinforce the legs.  With this tripod and the same mount and telescope, damping time is reduced to 4 seconds, which in my opinion is still excessive.

In short, the Star Adventurer is overloaded with the Lomo 80mm, and the Sirui tripod is also overloaded with an 8 kg load. Of course 120X is equivalent to 6000mm focal length.  So the Sirui tripod and Star Adventurer may still prove useful for astrophotography.  But as a visual setup for planetary, this combination is unusable.

Sky Watcher Star Adventurer, Part II

I had an old counterweight shaft and counterweight for an EQ1 mount lying around (the EQ1 had already broken some years back and I had thrown it away).  The shaft has a large-pitch thread on one side, and an M6 tapped hole for the toe saver on the other side.

I drilled out the M6 tapped hole with an 8mm bit, then epoxied an M8 bolt into the hole.  I then cut off the head of the bolt leaving the M8 threaded portion exposed.  A more robust solution would have been to drill out the M6 hole with a 6.5mm or 7mm bit, then use an M8 tap to cut threads into the hole.  The M8 bolt would then screw into the threaded hole, a more secure fastening than just epoxy.  But I did not have an M8 tap handy.   The epoxied bolt seems secure enough (I used Araldite brand epoxy) and hasn't wobbled through a 2-hour plus testing session.



With the stock EQ1 counterweight, the Stellarvue SV80ED still would not balance; however the result is much more in balance than having no counterweight at all.




However, the counterweight did not really improve the guided performance, or reduce the periodic error to in any meaningful way (and I did not expect it to).  Guided performance is still around 2" RMS, which I believe is the best this mount is capable of.  Hence it should be used at a pixel scale of 4" to 6" per pixel, or 200mm to 300mm focal length.


A longer sampling period has confirmed that the worm fundamental is 10 minutes, hence the worm wheel has 144 teeth (like the Vixen Polaris, Great Polaris, and CG-5).  This was confirmed by a poster on Stargazers Lounge who tore down the Star Adventurer.  The fundamental is about 30" peak-to-peak.



I got a better polar alignment this time (as shown in the minimal DEC drift in the graph above) and in the process noticed that the polar scope is pretty well-aligned - better than my PASILL3.  It's not perfect, but it's good enough that I'm not going to mess with it.

Sky Watcher Star Adventurer Portable EQ Mount

I purchased this portable mount in London from The Widescreen Centre in London (very close to Baker Street!) instead of lugging my old Vixen Polaris to the UK.  I got the "Astro-Photo Bundle" for £299.00 (about US$ 500) and am still waiting for the £40-odd VAT rebate, which would bring the price down to about US$ 400.  This mount is also available from Perseid in Malaysia for 1700 MYR, or about US$ 540.

The build quality is better than expected, with some nice touches such as the worm-and-sector drive for the altitude axis of the equatorial wedge, a design very reminiscent of the Astro-Physics Mach1 GTO, albeit with much slacker tolerances. There even is a ratchet on the altitude locking bolt (again, just like the AP). Here I have secured it via the 3/8" bolt to the aluminum tripod of the Vixen Polaris.  The mode dial and power switch is also visible here:



Right Ascension and Declination clutches are the large knurled plastic wheels.  On a less sturdy tripod, it is quite easy to knock off the polar alignment when tightening or loosening these clutches.


The declination slow-motion knob is visible in the photo below.  Note that the equatorial wedge, declination slow-motion, counterweight shaft, and counterweight, are all part of the "Astro-Photo bundle" and are separately priced if one purchases the base package.  On the other side of the Star Adventurer body are two electric RA slow motion switches (12X sidereal) to assist in centering objects, since tightening the RA clutch is a fiddly affair that tends to throw objects out of the field.


There is a fairly nice polar scope, although the reticle illuminator is attached to the far end of the polar scope bore, which is obstructed by the declination assembly.  Hence use of the illuminator requires removing the entire declination assembly (and any payload on top), which throws off the polar alignment.  The 4x AA batteries are under the top cover, and are supposed to last for up to 72 hours of tracking. Maybe that's with lithium batteries.


Below is a 100% crop of the area around Deneb, 2-minute exposures with a Canon 70-200mm f/2.8 lens at 200mm.  There is a bit of RA drift, and a much smaller DEC drift.  The RA drift is a combination of periodic error  and polar misalignment in altitude (as I only used the polar scope for alignment, this was in Lancashire in the UK).

Notice the nasty diffraction spikes on the bright star. A nifty trick I learned is to use a lens filter step-down ring to stop down the aperture (for tighter stars) while retaining nice round stars. If you use Canon L lenses, most of them have 77mm filter rings, so a 77mm to 67mm (or similar) step-down ring would be appropriate.


On returning to Singapore, I decided to load up the Stellarvue SV80ED, and a Meade DSI.  Widescreen hasn't received their shipment of counterweight shafts and counterweights yet, so the mount was severely unbalanced (no counterweights).


Using PHD2 I was able to measure a bit less than 2" RMS error in RA, after guiding.  The Star Adventurer has a standard ST4 guide port, but only can guide in RA, since the DEC is not motorized.



Unguided performance shows about 23" peak-to-peak periodic error, and there is no PEC.  The worm period is 10 minutes (144 teeth).  This implies that peak periodic error over a 2-minute period would be approximately 9" - so unguided exposures at 200mm and of 90-second duration should be possible at zero declination (the duration would be longer at higher declinations).

Guided performance should be acceptable at around 2" to 3" pixel scale, limited only by declination drift.  So a scope like the SV80ED would be a good choice if guided.  However, the mount is close to its limits with such a payload. A 200mm range camera lens or small refractor like a Takahashi FS60 is probably a better choice.



In summary, this mount is a far cry from my Mach1.  Although it looks better-built than other Chinese mounts, and some of the parts are CNC machined and not cast, the illusion of quality falls apart under close scrutiny.  That said, if you want something supremely portable for up to perhaps 300mm focal length, this mount will do very well.  It says something that the entire mount weighs not much more than a Mach1 Eagle half-pier (and also doesn't cost a lot more than said half-pier).

If I ever travel to Gran Canaria,  I will certainly not be able to bring the Mach1 along, but this mount will fit perfectly in check-in luggage.

I did manage to get about 30 minutes of the area around Deneb with the Star Adventurer from Lancashire, but I missed the Pelican Nebula by a small amount.  The wispy nebulosity next to Deneb that I thought was a DSO, turned out to be dirt on the DSLR sensor. D'oh!  image scale with the 200mm lens was about 6" per pixel.



Motorizing a Vixen Polaris Mount, Part II

I finally got around to re-making the motor bracket, and made a tripod attachment to the Vixen ring mounting points (there is no saddle).  The Polaris carries the Stellarvue 80ED and DSLR quite well.  I also used an intervalometer from e-bay to allow repeated bulb exposures.  Power is (literally) provided through the USB port on the Arduino; because the stepper motor is only 4V, providing 12V through the barrel jack on the Arduino causes a lot of noise and motor heating.


I took some 10-minute long exposures with the DSLR (an exposure this long is normally not practical, as thermal noise in the DSLR becomes unmanageable).  The reason for these long exposures was to measure the periodic error.


Unfortunately, the total PE is a massive 52.5" peak-to-peak over the 10-minute worm period.  This means that the maximum error is 0.175" / second or about 0.012X the sidereal rate.  With my (old) Canon 40D and the SV80ED, the plate scale is 2.1" / pixel. Assuming star FWHM of 4" and a maximum eccentricity of 0.60, this means stars would be 4" x 7" at worst - a drift of 3".  This corresponds to a maximum exposure time of only 17 seconds, which is pretty bad.

This means that PEC is not optional, but mandatory.

Astro-Tech 8" Imaging Newtonian, Part VI

The never-ending saga of the Astro-Tech imaging newtonian continues.  After my last update, I

  • purchased a Feathertouch focuser and motor, and made an Arduino focuser controller
  • used a much larger, 2mm thick aluminum sheet to reinforce the tube, and epoxied the aluminum sheet to the inside of the AT8IN steel tube
  • purchased an AstroSystems 2" laser collimator, because the $30 Seben was hopelessly miscollimated
  • purchased a CatsEye XLKP autocollimator and TeleCat cheshire, and learned how to critically collimate using the autocollimator, because the AstroSystems laser wasn't accurate enough

In spite of these efforts, and in spite of collimating with the OTA in the final position on the mount, I could still not get round stars across the CCD. So after eight months of fighting the ASA Keller reducer, I simply gave up.  I could not control the flex in the tube, and the thought of spending another 500 Euro to purchase a carbon fiber tube - with no guarantee that things would work - simply did not appeal.  So I sold the Keller reducer and played with the bare newtonian for a while.

Thanks to my well-honed collimation skills, the newtonian produced perfect, round, small stars - in the center of the CCD sensor.  Just 10% off-axis, massive amounts of coma reared its ugly head.  Now I know what coma looks like! and it's not pretty.

I found a Type 1 photographic Paracorr for a ridiculously low price on astromart, so I purchased it. Unfortunately.. it required a tremendous amount of extension to reach focus. That's a 2" extension tube there.  The Keller did not require much extension (i did have to add a 35mm focuser base extension to the Feathertouch, which is visible in this photo below).

I had known that the AT8IN had this issue; in fact when using it visually, the extension tube must also be in place and the focuser racked out quite a bit.  But that long imaging train bothered me, and it can't help the stability of collimation (incidentally this ridiculous setup produced nice round stars almost to the far corners of the ST8300M).



Based on some suggestions from the Cloudy Nights forum, I did some additional experiments (not at infinity, but pretty close - two kilometers or so away).  With no Paracorr in place, the DSLR does reach focus
(below) with the focuser racked out about 70% or so.  The 35mm focuser base extension is visible here as well.



The DSLR also reaches focus without an extension tube (unlike the CCD) but the focuser is racked out almost completely, and I had to raise the Paracorr as well (it isn't sitting flush on the shoulder of the focuser).  This obviously was a non-ideal situation and resulted in the extra-long imaging train in my first photo.



I figured that the focuser base extension was 35mm, and the extension tube 50mm.  So if I lengthened the tube by 85mm, I could get rid of both the base extension and extension tube, and significantly shorten the imaging train.  However I decided to be a bit more ambitious, and lengthened the tube by 100mm.

To do this, I hand-rolled some 2mm aluminum sheet into an extension.  Since the aluminum I had on hand wasn't long enough to form a full circle, I had to do some ugly joining to form the complete 360 degrees.  I also epoxied the aluminum at strategic points, and riveted everything to the existing tube (the aluminum extension is actually two coaxial cylinders).  I also learned how to use a pop rivet tool.



The end result is that eyepieces now can reach focus nicely without an extension tube; the CatsEye XLKP autocollimator is much easier to use because when it is resting on the shoulder of the focuser, it is nearly at the focal plane; and, the DSLR reaches focus with very minimal focuser out-travel with the Paracorr in place (below).  I expect the ST8300M will require more out-travel but well within the range of the focuser.  I also removed the 35mm focuser base extension, which makes the setup very low-slung indeed.


The only downside is, the DSLR won't reach focus anymore without the Paracorr in place.  But I question who would use a DSLR for imaging on a newtonian without a coma corrector anyway.

Motorizing a Vixen Polaris Mount

I purchased another Vixen Polaris mount.. for $80 (plus another $100-odd for shipping from the US). Should never have sold the first one.. (this one, with a William-Optics Zenithstar 70ED that I also sold)


The second one arrived in sorry state; rust everywhere, no counterweight (and counterweights for the threaded shaft are impossible to find); and the HA/RA setting circle was busted.  I knew all this, of course - the mount was $80! Vixen Polaris mounts in good condition usually go for $150 to $200.

I ordered a Vixen GP2 counterweight shaft (with white locking nut) which thankfully fits perfectly.  Now I can use standard Vixen (or Celestron) counterweights.


Because I intend to use this Polaris as a travel mount or at least back in the Philippines where Polaris (the star) is actually visible, I purchased a CG-4 polar scope for it.  The CG-4 polar scope fits in the bore, but does not thread in. And it isn't tight, so I had to wrap teflon tape around the polar scope tube for a snug fit.  It works now, but is obviously non-stock.


I had a bunch of Vexta PX243 gearhead steppers (expensive, high-end steppers actually) that I had lying around from my AP600 GoTo conversion (which I have since rescinded).  I hammered together a mounting plate out of 2mm thick aluminum in order to attach the motor to the Polaris.  The motor has a 1:18 gearhead and 200 ppr, giving 3600 steps per revolution of the worm, and 2.5" per step. I don't think I'll bother to half- or microstep the motor.  It is a hybrid stepper though, I hope my Arduino L298 motor shield can drive it without undue heating.


I used the existing bolt hole in the mount for attaching the aluminum bracket. I believe it is an M5 SHCS.


Unfortunately the motor interferes with the mount at low latitudes, which means that there's a large portion of the sky just to the west of the zenith, which is unavailable (kind of like Dobson's hole).  On the other side of the pier things are better, but I intend to mount a Bourns ACE 128 encoder on that side (so I have absolute indexing of the worm position) which will also limit the mount's travel.

The motor is coupled to the worm shaft with a 5mm to 6mm shaft coupler from ebay.


My eventual goal (once the basic motor drive function is working) is to implement some form of absolute PEC (using the Bourns encoder) and then further add atmospheric refraction correction.  This will eliminate the need for RA guiding altogether (and there's no DEC guiding anyway because there's no motor drive in DEC).

Hoping for first light this Saturday.

Dual-Axis Custom Tracking Rates for Long-Duration Unguided Exposures


I have refined my plate-solving and path-modeling software for my Astro-Physics Mach1.  The principle remains the same: the mount and SBIG camera are under the control of Linux software that I wrote (in Perl).

I documented a previous effort here. However, that version used a polynomial fit (that proved inaccurate) and actively sent guiding commands. This version simply uses the variable-rate tracking feature of later AP GTO controllers to cancel the computed RA and DEC drift.

Performing long-duration unguided exposures requires the following:
  • getModelData2.pl which captures ten (10) images each of 30 seconds, one after the other, at the RA/DEC of the target object.  No slewing is done in RA, hence the successive images show the drift in RA and DEC due to polar misalignment.  The assumption is that periodic error is minimal (which it is for my mount). This script only works for SBIG cameras and Astro-Physics GTO mounts, and relies on the Astrometry.net blind solver (local installation) for the plate-solves.
  • calcCustomRates2.pl takes the solved RA/DEC from the previous script as dependent variables, with the HA (hour angle) as the independent variable. It then does linear fitting of RA versus HA, and DEC versus HA, yielding the slope in RA and DEC which are then converted into custom RA and DEC tracking rates for the AP mount.  The script then programs the custom rates into the mount similar to Ray Gralak's PulseGuide.
  • the curve-fitting algorithm only applies the custom rate if the R-squared (correlation) parameter in the line-fit is sufficiently high.
Here is a local model captured of M13 with the telescope on the west side of the pier. The first seven columns are all captured by getModelData2.pl. The RA and DEC rates at the top-right are calculated by calcCustomRates2.pl (the R-squared value is also displayed).  I also calculated the RA and DEC slopes with Excel (in the box).  The DEC rate calculated by the script is of opposite sign to that of Excel because the DEC rate needs to be inverted when the telescope is on the west side of the pier, in the Northern hemisphere. Note that the confidence level for RA is not that high.

This is a ten-minute (600-second) unguided exposure of M13 at approximately 400mm focal length. The trailing in both RA and DEC are apparent.

The same object M13, but with DEC custom rate enabled (calculated from the model data), which should eliminate declination drift:

And with both RA and DEC custom tracking rates enabled (calculated from the same model). RA correction is still non-ideal; probably the calculated RA rate is not correct due to the 0.75 confidence level in the fit.

A meridian flip was then performed, and another 10-minute exposure done, without changing the tracking rates.  There is some trailing present.

Another set of model data was captured with the telescope on the east side of the pier, data is below. Note the better confidence level in the RA data.

Another 10-minute exposure, with RA and DEC custom rates enabled and with the new model on the east side of the pier. Better results than using the west-side model and not recalibrating.

And for validation purposes, a 10-minute unguided shot with the telescope on the east side of the pier, and custom rates disabled. Trailing is very obvious compared to the above exposure.

Source code is available here.  You will need the latest Linux developer kit from SBIG for the "testapp" program that captures images from the SBIG camera.

ED Doublet Versus FPL-53 Triplet

I just received my (used, $350) Stellarvue SV80ED doublet refractor.  I'd sold off a bunch of off-axis guider stuff and my ST80 to buy it.  My main goal was to have a decent guide scope, but also a portable imaging telescope because the Astro-Tech AT90EDT is too heavy and bulky for air trips.  The hope is that the SV80ED plus the Vixen Polaris that I am refurbishing would make a decent portable astrophotography rig.

I had thought of buying a Takahashi FS-60C to serve as a posh guide scope and portable imaging rig, but the FS-60C is quite expensive and the accessories and adapters to actually make it usable add even more to the cost.  So I went for the used SV80ED even though I knew it would not have perfect color correction.

So how bad, exactly, is an f/7 ED (probably FPL-51) doublet?  to answer this question, I piggybacked the SV80ED on my AT90EDT, and transferred my DSLR (a Canon EOS 40D) from one tube to the other.  In both cases I used the Altair Astro Lightwave 0.6X reducer/flattener.

First off: Sirius.  Admittedly a tough test for an ED doublet.  SV80ED:



and here is the AT90EDT. 10-second exposure.


Second: M44, the Beehive Cluster.  This image is from the SV80ED:


and this image from the AT90EDT. Also a 10-second exposure. Because the AT90EDT is only f/6.7 the focal length of the two scopes are very similar (540mm and 600mm).


I was unable to see any violet fringing on the Moon at prime focus on the SV80ED, however. It was very instructive to do fine-focusing with the DSLR Live View at maximum magnification.  On the FPL-53 triplet the star remained pure white with beautiful diffraction rings inside or outside focus; while with the ED doublet I immediately saw color aberrations once out of focus.

So the quick answer: even a non-critical imager will quickly find fault in the SV80ED, namely the blue halo around bright stars.  For visual, or for deep space objects, I expect the ED doublet would perform perfectly adequately.

Arduino-based Motor Focuser Controller

Everybody it seems has written their own Arduino-based motor focuser controller, of which the most popular is probably the SGL Observatory Automation version. There also is ArduinoFocus, which ambitiously reverse-engineered the Robofocus protocol; the sirJolo ASCOM focuser; and the E.J. Holmes Arduino focuser.  A common characteristic of all these projects is that they provide their own ASCOM driver.  I only tried the SGL version; sadly the ASCOM driver is in an almost-unusable state for me (basically it doesn't work - Sequence Generator Pro can't drive the focuser in an inward direction, which is pretty bad). So I decided to implement my own Arduino focuser, but make it compatible with an existing protocol.  This would allow me to use one of the commercial focuser drivers and avoid the problem of building my own ASCOM driver.

After some thought I decided to implement the Moonlite (otherwise known as the Lacey or EasyFocus) protocol. Since both a stand-alone and ASCOM driver are supported - and the stand-alone program echoes all the serial commands - it would be easier to debug the protocol.

On the hardware side, I wanted a design that can be readily-assembled, i.e. no soldering required. To achieve this, I used the Adafruit Motor Shield (AFMotor), version 1.2 - this version is now discontinued by Lady Ada, but China-made clones can be purchased for $5.00 shipped on ebay. This shield uses the L293D H-bridge controller, which is only capable of 600mA. This means that driven steppers must necessarily be small, and with fairly high resistance (in other words hybrid steppers cannot be driven).

Because my focuser is a Feathertouch, it uses a Starizona Microtouch MSM20 motor, which is a 300-step per revolution motor (with 48ppr and an internal 6.25:1 gearbox) giving 6.5 microns per step.  One benefit of the Microtouch motor is that unlike the Robofocus (which uses a DB-9 connector), the Microtouch uses a 6P6C (RJ-12) connector. It is very easy to purchase an RJ12 patch cable from any number of sources.  Then cut the RJ12 in the middle and bring out the correct wires.



During prolonged movements (i.e. a homing command), the L293D chip on the motor shield gets fairly hot; more so because I implemented full stepping with double-coil excitation to increase the torque (note therefore that my focuser does not implement half-stepping).  To alleviate overheating, it is a good idea to attach a heat sink to the L293D, I used an Aavid Thermalloy part number 580200B00000G which clips on top and under the L293D. You must make sure to buy an Adafruit Motor Shield (or clone) where the L293D is socketed, otherwise it would be impossible to attach this particular heat sink. Although the motor shield has two L293D's only one is used and therefore only one heat sink is necessary.



A final modification is required: most ASCOM drivers, when connecting the the focuser, immediately send a query or interrogate string. The problem with the Arduino is that when software connects to its serial port, the DTR line is pulled low, which resets the Arduino (this is a feature, necessary to implement programming the Atmega chip). It takes about 2 seconds for the Arduino to reset.  This becomes an issue because the ASCOM driver times out when it doesn't receive a response quickly enough.

The work-around is to prevent the Arduino from resetting when DTR is pulled low.  There are several ways of doing this, but on the Arduino Uno the easiest way is to connect a capacitor between RESET and GND. When the Arduino powers on, the capacitor is a short circuit, and the Arduino is stuck in a reset state. However once the capacitor charges up, RESET is now high, and if the capacitor is large enough, DTR pulling low (and pulling RESET low) will no longer work, because the capacitor provides enough voltage that DTR cannot pull RESET low anymore.  Alternatively you can omit the capacitor; the focuser can still be driven by the Moonlite stand-alone program, but not by the ASCOM focuser driver.


The side effect of the capacitor is that you cannot program the Arduino anymore! (more information in the last link above). In my case, I soldered the capacitor to the RESET and GND lines on the motor shield. Hence, you must detach the shield from the Arduino in order to upload the Moonlite focuser program; after the sketch has been uploaded into the Arduino, power everything down and re-attach the motor shield (with its capacitor).

So the total Bill of Materials consists of the following (aside from the Microtouch motor):
  • Arduino Uno or similar
  • RJ12 6P6C cable
  • Adafruit Motor Shield version 1.2 or clone
  • Aavid Thermalloy heat sink
  • capacitor, any value from 10uF to 100uF will work
Assembly is extremely simple: solder the capacitor to the motor shield; cut the RJ-12 cable, strip the four colored wires that are not black or white; tin these wires, and screw them into the left (M1/M2) motor connection block on the Adafruit motor shield.  Then upload the Moonlite sketch into the Arduino, unplug the USB, attach the motor shield to the Arduino, and you're all set.



In my tests, the Arduino plus shield can rack the Feathertouch focuser in and out under USB power alone (no need for a 12V power source on the barrel plug or on the auxiliary power connector on the motor shield). Results might vary depending on the torque required from the focuser.



Source code with some fixes for :GI# and compatibility with INDI; tested and working with both the stand-alone Moonlite focuser program, and the ASCOM 64-bit Windows driver loaded from inside Sequence Generator Pro. Note that this program does not implement speed changes (the AccelStepper library implements ramping, which makes speed changes un-necessary).

Selecting the Moonlite DRO Focuser Driver, and selecting the Arduino focuser. This program also does not implement temperature compensation, it just returns a hard-wired temperature. Also note that the speed setting, full step / half-step settings and temperature compensation commands are also ignored. I have not tested what happens if half-stepping is set or if temperature compensation is enabled.


Exercising the focuser through Sequence Generator Pro's Focus Control window.