21 replies to this topic

[tommm]

This is on some updates I made to a 22.4” Dobsonian telescope I originally built in 2002 following the David Kriege and Richard Berry design (K&B) in The Dobsonian Telescope. The updates include a new shorter focal length primary mirror, larger Antares secondary mirror, improvements to the truss, reduced weight, and the addition of tracking and go to. 

 

Some parameters are:
Physical diameter of primary: 22.4", 569mm”, clear aperature: 567mm
Focal length: 80.7", 2050mm (f/3.61)
Secondary mirror minor axis: 127mm, 121mm clear aperture and 23% CO with the Astrosystems holder
Fully illuminated FOV: ~15mm
TFOV with ES 20mm 100 deg ep: 0.97 degree, 0.85 with P2

These were my guidelines for the update:
1) Control cost
2) Use as much of the existing telescope parts as possible.
3) Lower eyepiece height, but primary mirror > f/3.5 to enable use of less expensive eyepieces.
4) Ease of transport and setup.
5) Mirror box to double as storage box.
6) On board drive motors for go to and tracking.
7) No tools required for truss installation.
8) Fans for mirror cooling.
9) For visual use only, no AP.

 

Primary mirror
The 22.4” diameter, 2” thick quartz blank was given to me.  I ground out the 0.388” deep curve by hand, mainly using a 10 lb barbell weight (as sopticals mentioned using).  I started with #25 grit, then #40 grit, then #60…Then switched to a ¾ size tool for fine grinding. Polishing was started with a full size tool, and completed with a 10” tool which was also used for figuring along with some smaller tools. 

I tested the mirror with a Bath interferometer and Dale Eason’s DFTFringe software (https://groups.io/g/Interferometry). The test results were verified by a Foucault double pass null test performed at Ostahowski Optics prior to coating.  Terry reported the surface was very smooth and very well corrected. He also said it has a slightly heavy (low) edge, and that is common for large fast mirrors he has tested.  He said you usually can’t figure that well without a null test during figuring, which I think speaks well for DFTFringe. 

 

Secondary mirror
The 127mm (5") supremax 33 secondary mirror was purchased from Antares. PV error: 0.056 wave, rms error: 0.009 wave.  It, and coating of the primary were my largest expenditures for the scope.

 

Mirror cell
The original mirror was held by a sling – the old seat belt design.  The model at Cruxis (http://www.cruxis.co...ecalculator.htm) indicated that a simple two-point support would work fine for the 2” thick, 22.4” diameter, f/3.6 quartz mirror, but a whiffle tree support was better, so I went with that.  I modified the old K&B tailgate design to add an “A frame” with whiffle tree edge support to the mirror cell to make it similar to a JP Astrosystems cell (http://www.jpastrocraft.com/cells.htm). The three whiffle trees of the original 18 point mirror back support were re-mounted on the A frame with aluminum angle and shoulder bolts, and pins were added for the triangles to prevent rotation, mimicking the JP design.

 

 

The pads were cut from a sheet of Acetal (trade name Delrin). They were bolted to the triangles then each triangle was placed pads down on a sheet of #400 sandpaper on the steel top of my table saw and sanded flat until I could not see any light between them and the table surface when illuminated from behind, ensuring they were coplanar and flat.  I also examined the pad surfaces to ensure they were sanded completely over their surface, with no low spots.

 

Alt and Az drives
I took inspiration from Andrei H…’s design for the drives, described here: LINK He said that the alt drive worked fine, and cautioned me about the difficulty of making the ground board O.D. precisely fit the belt glued onto it.  He recommended I just use the more standard approach of using the ground board as a pulley with a belt around it and a drive pulley on the motor, similar to the photo at the above link under “The azimuth drive”.

 

I wanted both the alt and az motors in the rocker box for simplicity so I combined aspects of Howard Banich’s (https://www.cloudyni...-a-28-dob-r1089) telescope with Andre’s, using drive belts glued to the alt bearings (Andre) and keeping the mirror box inboard of the alt bearing radius (Banich design) so it passes above the drive shaft and pulleys in the rocker box.  This, and the use of a “drive disc” fastened on top of the ground board permits mounting of both alt and az drive components inside the rocker box. The disk is 19.5” diameter, 12mm (~1/2”) thick Baltic birch, screwed to the top of the ground board, as shown in Fig.4.  Also shown at center is the ½” bolt screwed into a nut welded to a plate mounted on the back side of the ground board. It inserts through a brass bushing in the rocker box to fasten it to the ground board.

 

 

The drive system components are shown in the first photo below.  A close up of the altitude drive is shown in the second photo, and the azimuth drive pulley/belt is shown between the rocker box bottom and ground board in the third photo. The Az motor is mounted on a “slide” which permits adjusting belt tension.  The fasteners are in slotted holes to enable this movement.  The black box houses the drive PC board. Adjacent to the battery are a DC/DC and fuse supplying 12V for the fans from the 20V battery. Also shown in the 2nd photo is a close up view of one of the bearings the altitude bearings ride on.  The two power cords laying in the box are for mirror fans.

 

 

The blocks holding the Alt drive shaft are Baltic Birch and are adjustable up/down to adjust pulley engagement with the belt on the alt bearings. The shaft runs on ingus bearings seen mounted on the block. (tip of the hat to Andre for those too).  The photo below shows a closeup of an Alt drive pulley with the mirror box in place on the rocker box.

 

 

Truss design
The truss is made of 1 ¼” O.D. aluminum tubing from the original scope.  The original truss was the folding design described by Guy Walton in the February 2004 Sky and Telescope, vol 107, issue 2:  (http://connection.eb...truss-dobsonian)
I was concerned that this was not rigid enough to maintain the alignment necessary for an f/3.6 mirror, so I changed the design. I decided to make the truss brackets to save cost, and to ensure rigid mounting of the truss to help hold optical alignment.

 

The mirror box and UTA truss brackets are made from 3/16” thick, 1 ½” aluminum angle and are bolted to the mirror box top and UTA bottom ring.  The angle for the mirror box brackets was cut with a hack saw at 45 deg, filed, jigged up, and taken to a weld shop to make the corner brackets shown in Fig.9. The truss is clamped with the knobs on the ¼-20 bolts that are threaded into the brackets.

 

 

The top truss end brackets, shown in the photo below with the UTA truss brackets, have a pivot bolt with lock nut in the bracket below the pole ends which permits folding each pair of poles together when removed from the scope.  The knobs are kept on the studs in the UTA brackets for easy assembly. The truss brackets were made from 2” aluminum angle with a hack saw, bench grinder, and file. The holes visible at the “corner” on each side are for the bolts that fasten the brackets to the truss poles.

 

 

The bottom truss brackets were made from 1” aluminum angle as shown in the photo below and fastened to the truss poles the same way as the upper ones. The ¼-20 bolts in the UTA and mirror box brackets fit snuggly into the bottom of the notches in the truss brackets. The intersection of the axes of a pair of truss poles is very close to the knob of the UTA brackets and the lower corner of the mirror box brackets.

 

Edited by GShaffer, 02 January 2019 - 10:47 AM.

[tommm]

Mirror box
The original mirror box was designed per K&B and was about 20” deep.  In keeping with guideline (2) I re-used it, but I cut it down to 7” deep to reduce weight, which placed the truss knobs in an opening in the alt bearings for access.  I also added a cover for the bottom of the mirror box as shown in the first post, so the top and bottom of it can be completely closed for dust-proof storage. I wanted to avoid lifting the 57 lb mirror out to store it in a box to protect it from dust, guideline (5).   The fans then have to do all the mirror cooling.  Will have to see if they do well enough.  The fan shown in the mirror box bottom is a 200mm Noctua with PWM speed control.  There are also small fans offset from the mirror box corners (so flow from the bottom two is not blocked by the edge support wiffle trees).

 

Rocker box
The old rocker box was made double thickness (36mm) per K&G.  I made a new one single thickness to reduce weight.  What a difference! This, the much shorter mirror box, and thinner alt bearings made the scope much lighter, though not light.  The mirror box, cell, mirror, and alt bearings together weigh about 120 lb.  The rocker box, ground board, and drive system assembly weighs 42 lb.  The bottom surface of the rocker box is faced with 24 gage ss to ride against bearings on the ground board.

 

Altitude bearings
New larger radius (20 ¼ ”) altitude bearings were made to fit the mirror box inside their radius (requirement for having the alt drive in the rocker box) and to permit enough drive reduction with a small reduction between the motor and drive shaft. These were also made single thickness (18mm) to reduce weight.  They are covered with 24 gauge ss to ride against ball bearings. Arcs, 2” wide, made of 12mm Baltic Birch are screwed to the inner surface of the alt bearings and timing drive belts are glued to these - inspired by Andre’s design - as shown below.  The bearings are fastened to the mirror box with three bolts, one of which goes through the square “spacer” near the top of each bearing.

 

 

UTA
I used the original UTA, but moved the focuser so had to replace the shell.  The shell is made of Formica, and is 24” I.D.  The matte black Formica was purchased at Home Depot and painted flat black on its backside. 

 

Ground board
The ground board is ¾” pine plywood. There are 3 pairs of 0.25” wide, 0.875” O.D. ball bearings mounted in counterbores in the ground board, each centered above a foot of the board as shown in a photo in the first post.    The counterbores give a bearing height such that the rocker box clears the drive disk by 1/8”.  With disk thickness of 12mm (~1/2”), the az drive adds about 5/8” to the eyepiece height compared to having a ground board with no drive.

 

OnStep
OnStep, by Howard Dutton, (https://groups.io/g/onstep) offers a relatively low-cost option for a telescope drive based on stepper motors (just the cost of the PC board and components).  It does not use encoders for position feedback, which is why I chose not to use a friction drive for the altitude axis.  The timing belt reduces the risk of losing position.  The azimuth drive is friction-based, but the belt in tension around the 19.5” disk has a fairly high friction force, so it is difficult to make it slip.
In addition to the Arduino sketch for firmware, Howard has written an Android app so the telescope can be controlled from an Android cell phone with a Bluetooth module connected to the main OnStep PCB.  I use Sky Safari 5 Plus to control the scope after initial setup with Howard’s app.  Controlling with a cell phone minimizes the things I have to lug along on a trip to dark skies, guideline four.  The board also includes stepper drivers for focus adjust and focuser rotation, but I am not presently using those (no need for the latter since I am not using it for AP).  The board also has ST4 input.  Many thanks to Howard!
 

Removable handles
The front fasteners for the scope handles were hard to reach in the scope storage cabinet (see below) to remove the handles and close the door.  I cut the handles about midway on the rocker box, added a second bolt to hold the front section with wheels on the rocker box, and two pieces of 1” angle to slide the rear part of the handle into as shown below.  The photo shows a temporary hex head bolt in the handle. I’ll replace that with a knob like the other one. After loading into the cabinet the bolts holding the rear part to the rocker box are removed, and the handles slid out and stored in the cabinet as shown in the storage box photo below.

 

 

Storage cabinet
I live in high desert and have a gravel driveway so dust seems to be everywhere.  Then there are the spiders building their webs… so I built a storage cabinet, shown in below, to store the scope in the garage.  It is made “spider tight” and has two vents, one on the lower right side and one on the upper left side. I put fine mosquito netting in these to keep bugs out and hopefully cut down on dust a little. The door has a 1” wide felt seal. The cabinet has rollers for easy movement for cleaning the garage, etc. Two removable ramps attach to the front for loading/unloading the scope, and are stored in the cabinet. Eyepieces etc., are stored on the shelf. Fortunately, I completed the cabinet just a day prior to smoke arriving from CA fires this summer.

 

 

[TonyStar]

and cautioned me about the difficulty of making the ground board O.D. precisely fit the belt glued onto it

Very nice rebuild  

I've also used Onstep for motorizing my dob. I glued a timing belt to the ground board (cut with a router, see my signature). It works great. 

[TOMDEY]

Nice! Sooo... Have you gotten First Light?!  Tom

[tommm]

Nice scope Tony! Very clean design.  The problem Andre mentioned with fitting the Az belt was in matching the I.D. of the belt to the O.D. of the ground board.  So did you just route the ground board to the nominal diameter of the belt, and it fit snug? Or did you use a strip of belting?

I assume the photos were taken using OnStep for tracking?  How long of exposures did you use? Did you use ST4 to autoguide, or were they unguided?

 

Tom, yes it's seen first light, but not a lot of use yet since much of late July/early August I just left it in its storage cabinet because we had smoke from CA fires.  Just recently been using it more since the smoke cleared. Views are good! Still getting used to using OnStep and Sky Safari. Really like the latter with its "best of tonight" feature, catalogs and Goto.  I find I am much more aware of many different objects to look at and where they are in the sky compared to when I used to use a printed sky map. I'm an out-of-sight, out-of-mind person. Sky Safari puts more things in my sight making me aware of what is there to view.

[TonyStar]

So did you just route the ground board to the nominal diameter of the belt, and it fit snug? Or did you use a strip of belting?

I assume the photos were taken using OnStep for tracking?  How long of exposures did you use? Did you use ST4 to autoguide, or were they unguided?

Yes I just route the board to the nominal diameter and the belt fit snug but not too tight so even with a thin epoxy layer it was easy to glue it. I guess I was lucky... I agree if the diameter is not just right it may be hard to install the belt this way.

 

Anyhow, yes all pictures were taken with Onstep tracking but unguided. I set up Onstep for autoguiding with Firecapture thorugh bluetooth but I have to fine tuning the guide parameters. Usually movies are shot at over 100fps so precise guiding is not required.

Edited by TonyStar, 31 August 2018 - 11:40 AM.

[TayM57]

Good re-build. I like how you incorporated the findings from Lockwood's research into your cell.

 

http://www.loptics.c...rorsupport.html

 

I see that you put wheels on the ends of your whiffletrees. Lockwood found that the wheels at the end of the whiffletree produced the most concentrated stars. #8 finding in the link above.

 

You have given me some great ideas on the cell I'm building. Thanks!

[tommm]

Thanks! Yes, I reviewed Lockwood's article and modeled the cell after the JP Astro design, with some additions and modifications to the original K&B cell.  

 

If you want the "tetter totters" recessed like the JP design and don't have a mill, you could just bolt or TIG weld a piece onto the top center of aluminum bars to get the same form.  Some people have placed the bars on the side of the A frame tubing, with the pivot bolt through the A Frame and the sides of the bar. This gives a still lower profile, but has a bending moment due to the one-sided support of the bars.

[Mason Dixon]

Slick rebuild.

 

So you just glued and screwed some T5-10 timing belt onto the alt bearing with gorilla glue?

 

Do you happen to have part numbers/links for the alt motor assembly and pulleys?

 

Does your Skysafari/Onstep setup do everything you need or did you have to assemble the Onstep Hand Controller as well for alignments, general movements, etc?

 

Does Onstep do an initial calibration to account for all the different bearing radiuses/timing belts/pulleys/etc or is all that carefully planned out?

Edited by Mason Dixon, 22 July 2019 - 10:52 AM.

[tommm]

Slick rebuild.

 

So you just glued and screwed some T5-10 timing belt onto the alt bearing with gorilla glue?

 

Do you happen to have part numbers/links for the alt motor assembly and pulleys?

 

Does your Skysafari/Onstep setup do everything you need or did you have to assemble the Onstep Hand Controller as well for alignments, general movements, etc?

 

Does Onstep do an initial calibration to account for all the different bearing radiuses/timing belts/pulleys/etc or is all that carefully planned out?

1) I used contact cement to adhere the belt to the Alt bearings. Screw at each end of the belt. I later epoxied a belt to the drive disk rather than rely on friction. Drive disk diameter must be accurate to within about 1/64" so the belt JUST fits around it with no play. Took me 3 cuts to creep up on it with a router - after failing on the first attempt.

 

2) Belt/pulleys are XL size, 0.1" grooves. Finer groove belts are available but not in the length I needed for the Az drive. Grainger, McMaster-Car and Stock Drive Products all sell belts and pulleys. SDP has the most for smaller sizes. I purchased two of the "precision" planetary gear reduction motors at the link below, but there are many stepper motor suppliers, such as Oriental Motor, on line thanks to robotics.

https://www.omc-step...anetary-gearbox

 

Datasheet for the 30:1:

 steppersonline, datasheet for 17HS15-1684S-HG30.pdf   135.93KB   67 downloads

 

3) In typical use I set up and level the scope carefully, align the optics, then center Polaris in an illuminated reticle ep on the main scope using the OnStep app on my tablet to drive the scope.  I then do an alignment using the OnStep app using 3 or 4 stars.  I run SkySafari at the same time so I can pick stars that are visible to me (the city is north of me so I can't always see alignment stars in the north due to city lights), then go to the OS app and select the first star from the list of alignment stars there.  The app drives the scope to the first star (or close), I center it in the reticle ep, then select "align", then repeat for the next n stars. 

 

From that point on I just use SkySafari to operate the scope. I have custom observing lists there, and you of course can just tap an object on the sky map and Goto it.

 

Some people like the hand controller but I don't use it. I don't see a big net advantage.  I use OS and SkySafari in night mode so they aren't too hard on my night vision, and I don't like the fact that the hand controller uses an ST4 cable so there is a wire to wrap around things and trip over. I also would still want the tablet to use SkySafari to pick objects to view.

 

4) The latter.  OnStep does no calibration for the drives.  You have to determine the drive ratios yourself and enter those into an MSExcel spreadsheet that Howard (OnStep author) created and you can download at his site.  The spreadsheet calculates a few parameters from that info that you must then enter in the config.h file in the OnStep firmware. I just treated the Alt drive as a huge pulley with the radius of the Alt bearing plus part of the belt thickness and calculated the number of grooves it would have to find the gear ratio for that drive. Drive ratio is of course just the ratio of total number of grooves on the belt to those on the mating pulley.  Overall drive ratio includes the gear boxes on the motors.  Both drives are around 860:1 overall.

 

OnStep has a site on groups.io. The wiki there gives a good overview as well as acting as a "user manual". It has most everything you need to know to set up OnStep and use it.

Edited by tommm, 22 July 2019 - 12:08 PM.

[Mason Dixon]

Great info, thanks. Just wanted to get my feet wet before diving into a bunch of reading. Love the open source aspect of this, doesn't look like GOTO can get any cheaper.

[tommm]

After using this scope for about a year I decided to make some modifications to it. I bought a Chevrolet Bolt at the end of 2018 and wanted to be able to transport the scope in it.  That would require cutting down the UTA height from 18" to about 8.5", and cutting about 4" off the altitude bearings or making new ones.

 

Cutting down the UTA would require a new secondary holder and spider since the Astrosystems holder was about 14" long from mirror bottom edge to top of the 1/2" threaded rod.  The secondary mirror would have to be nested up in the spider. 

 

The photos below show the holder and spider, and the 10" tall UTA (which includes the 1.5" truss brackets on the bottom) sitting on the mirror box. Also visible in the second photo are the cut off altitude bearings.

 

 

 

 

The scope is now limited to 18 deg or above altitude. I never viewed lower than that anyway. The photo below shows the scope loaded into the back of the Bolt with ramp boards in place. 

 

 

The ramp boards are hinged. The 2"x3" braces on the backside slide out of the brackets permitting folding of the ramps in half. The photos below show the braces on the backside of the ramp boards and the folded ramp boards loaded in alongside the scope.  The truss poles and observing chair fit in along the side also, and there is plenty room for the eyepiece box and other stuff behind the front seats on the floor.

 

 

 

I also added draw latches to hold the scope in altitude at the optimum angle for loading, and a bolt latch to prevent rotation of the ground board when moving the scope.

 

 

 

I run a strap through the altitude bearings and over the UTA to hold it in place, and also one through the altitude bearings and through the seat back metal retainer brackets on the Bolt to prevent the scope from sliding forward during a sudden stop.

 

A video of the scope executing a Goto is shown here. Please be patient, it takes several seconds for it to start - had to juggle my tablet and my cell phone. The drive system is described earlier in this thread.

 

 

[tommm]

I also checked orthogonality of the scope axes. I set the scope about 17 ft out from the side of my house, on a line orthogonal to the wall, and carefully leveled it.  I then fixed a V block to the mirror box cover and set my aligning laser in it with its front end at the azimuth axis.  The back of the mirror box was toward the wall so when it was rotated in altitude away from the stop (south) the laser scanned vertically up the wall.

 

I marked the point near the bottom of the wall where the laser pointed with the mirror box up against its stop, and drove in a 1" long wire nail there.

 

I rotated the scope south to scan the laser up vertically and marked a point 93.0" above the bottom point, and drove a nail there.

 

I rotated the scope back down to the mirror box stop, then scanned in azimuth +/- 11 ft from the bottom point, as measured with a tape measure hooked to the bottom point and pulled taut, then marked and drove nails in at those points.

 

If the axes are square I now had two right angle triangles on the wall defined by the top, bottom, and left/right points, and the two hypothenus should be sqrt(132^2 + 93^2) = 161.47 inch. 

 

I measured the hypothenus of both triangles with a tape measure hooked to the nail at the top point and pulled taut to the right/left points. Result: left side triangle hyp: 161 15/32 inch, right side: 161 1/2 inch, very close to 161.47.  I figure measurement error is around 1/32" so the difference from ideal seems to be around 1/32". 1/32" divided by 93.0" is 3.4 x 10^-4 radian, so about 1.2 arcmin error from orthogonality.

[tommm]

Just an fyi in case anyone is interested in required motor currents for a larger Dob. This is using the OnStep system with two NEMA17 motors, overall reduction about 919:1, 1/16 usteps, 1/8 usteps for GOTO.  I played around reducing currents starting at IHOLD = 400 mA, IRUN = 500, and IGOTO = 600 mA, then decreased each by 100 mA steps testing GOTO  at 3.0 deg/sec and slewing at 1/2Max speed under Guide/Focus, 2.0 deg/sec each time.

I ended up at IHOLD = 100 mA, IRUN = 200, IGOTO = 300.  Nothing indicated I couldn't go lower. I just stopped there because the 5Ah DeWalt battery was easily lasting me for 2 observing sessions at around 4 hours each at the old settings of IHOLD = 500, IRUN = IGOTO = 800, so I didn't see a need to go lower.  I did the testing in my garage and couldn't hear the Az motor as it slewed at 3 deg/sec standing right next to it. Alt had a low humm at all the current settings. 

 

The UTA/mirror box/rocker box assembly weighs about 170 lb, with the 22.4" primary mirror in place.   The ss underside of the rocker box rides on 3 pairs of 7/8" diameter roller bearings. I am using 5160 motor drivers, MaxPCB.  It's really cool to see that size scope just slewing smoothly and silently around at 3 deg/sec on 300 mA.  Once again I have to say Howard sure has created a nice system!

[Spartinix]

Impressive! I'm hoping to learn more as I go, especially concerning installing motors.

[dave brock]

The ramp boards are hinged. The 2"x3" braces on the backside slide out of the brackets permitting folding of the ramps in half.

If you had the hinges underneath, you wouldn't need the braces.

[tommm]

If you had the hinges underneath, you wouldn't need the braces.

I don't trust the 3/4" thick pine boards to hold the 170 lb scope. Soft pine is not very strong.

[gatorengineer]

Very interesting secondary concept. Couple of questions, how is the mirror attached? I don't see any up down adjustment is that designed in? Is it offset?

[tommm]

Yes it's offset, and if you look closely at the first two photos is post #12 you can see a knurled knob at top center on the back of the plate behind the right angle part holding the mirror, and you can see more of them in the view looking down the top of the UTA.  There are 4 to adjust tilt. They have to be short to be in the shadow of the secondary.

 

There is no up/down or rotation adjust. I mounted the secondary in the UTA and attached a "temporary" focuser board - basically a couple of rails I could slide the focuser around on. I inserted a Catseye sight tube in the focuser and moved the focuser until the secondary was concentric in the sight tube, marked the position, then drilled the focuser mounting holes in the real focuser board at the recorded position. When I did the optics alignment it just needed a little tilt tweak and was ready to go.

 

The holder is nothing new. It is a kind of hybrid of an Arjan design (post #72 here) and bratislav design (posts #44, 49, 53 here).

 

The mirror is adhered to the triangular backing plate with silicone. Description of MoR measurement and stress is in post #39 here. I estimated about 4.3 nm movement of the mirror backside for my set up. So far it is working fine.

Edited by tommm, 15 May 2020 - 11:01 PM.

[gatorengineer]

Thanks for your thorough and thoughtful response.

[Augustus]

How'd you make that spider? Looks like the Highe design. I'm puzzled by how to cut and attach the steel vanes mostly.

[tommm]

Yes, most of the offset spider/holders are variations on that theme. The changes I made were to make it lighter and simpler to make using hand tools.

 

My metal working tools are hacksaw, bench grinder, hand drill, bench top drill press, and file, so nothing special required. The vanes are made from 22 gauge (0.029" thick) steel from Home Depot. I chose that thickness because it fits snugly into the kerf of a hacksaw, which is what I used to cut slots in the coupling nuts for the vane ends (I made a little jig out of wood to cut them centered and straight. Basically a wood block with a hole drilled in it which the nuts fit snugly in, and a kerf cut across the block through the center of the hole with a backsaw - a mini miter, which I held in the vise). The coupling nuts are attached with one 1/8" diameter spring pin. The hole for the pins was drilled prior to cutting the kerf for the vanes.

 

I bent the vanes in a vise, using a square to ensure they were square to the vise prior to bending, and taking care to ensure they were all bent the same distance from an end. Each vane is clamped between the 1/4" thick aluminum plate and a 1/8" thick aluminum strip with 4 bolts as shown in the second photo of the secondary/spider assembly. 

 

The difficult part of course is getting them all matched.  The steel is fairly cheap though, so you can afford several attempts if necessary.  The two 1/8" strips were drilled first, clamping them together in the vise after carefully aligning them, and drilling the two outermost holes  with a hand drill. Then they were bolted together using those holes and the other two holes were drilled in a bench top drill press.  Each 1/8" strip was then clamped to the 1/4" thick plate to act as a guide to drill the holes in the plate with the drill press. The 1/8" plates were also clamped to the vanes to act as a template for drilling those. 

 

Oh, I also use my table saw to cut aluminum using a blade for non-ferrous metals. It gives me the willies each time I do so. I cut very slowly and carefully, wear safety glasses plus a face shield and stand off to one side in case the piece catches and gets thrown out the back. But the 10" saw cuts through it like butter.

Edited by tommm, 24 May 2020 - 12:27 PM.