February 09, 2013

Now it all makes sense

This appears to be the source of most of the extra traffic I've been getting the last few days. In the video, he mentions my coping calculator, and shows an example of its use. I noticed the the joint looked a little loose. One thing that I probably didn't point out clearly enough is that you can get a tighter fit by decreasing the tube thickness to about 50% of the actual tube you're using. That will require a little more grinding to get it snug, but it might be useful for tight tolerance work like this.

Posted by Hal Eckhart at 03:51 PM | Comments (0)

November 20, 2012

Handy Magazine

Mark Simpson found my blog posts on bending tube by hand, and I ended up in Handy Magazine.

Posted by Hal Eckhart at 03:57 PM | Comments (0)

January 20, 2012

Multiple radius bend on 2" 16 ga. square tube

This project required bending some 2" square steel tube for a couple of free-form trusses that hold up elements of a museum display. I wasn't really sure I could do this, but it worked out pretty well. Hand bending large tube with a very thin wall thickness is hard to do without kinking, but I've had some success with using thin jigs centered on the tube that push the wall of the tube in. It can lead to some extreme distortion, but it allows you to get a much tighter bend without kinking.


The other trick here is that the radii were variable; from 150" to 18". I was only able to get it down to about 34", but I was able to cheat the design a bit to make it look good enough.

I followed the same process I use to bend a large ellipse. This involves breaking the curve down into arcs, and then transferring the drawing to a full scale pattern. I used the old theater trick of a brown paper grid and charcoal. Having a full scale pattern is essential, because this always involves a fair amount of tweaking to get it to look right. Besides getting the overall shape correct, you have to make sure there isn't a sudden shift in the radius that would make it look lumpy. Progressing from the longest radii to the shortest seems to prevent this issue.


It was a lot of work, but they ended up looking pretty darn good. No kinks, no welds and no blood spilled. I keep remembering my physical therapist's wise words "your body is not a machine". Well, sometimes it is. At least I didn't break it this time around.

For a more comprehensive article on hand bending, see this.

the jig setup


2 foot long, 1/4" thick steel jig

inner wall distortion

outer wall distortion

the finished bends and a test piece

Posted by Hal Eckhart at 09:53 AM | Comments (0)

September 03, 2011

Bird Nest Bed

This is a project from a few months back. It's a hand-forged steel bed for our bird-themed guest room. The nest is the one the robin left in the wreath above the garage at our old house.




Posted by Hal Eckhart at 08:31 AM | Comments (0)

April 08, 2011

Hand bending 1" square 18 gauge steel tubing to a 4-1/2" radius

The key to this jig is a piece of 1/8" square wire, which is roughly centered on the outside of the jig. The wire could actually be round and anywhere from 1/8" to 3/16" for this job. The idea is to push the inside wall of the tubing in, and once it's going the right direction, it will take care of itself. Without the wire or something even more elaborate, you won't be able to bend this material tighter than about 15" radius without kinking.

If you look closely, you might notice that the straight part of the tube that I'm bending from ends up slightly curved also. If this is a problem, you can overbend the material slightly, and then carefully unbend the side that you used as a lever. But if it has to be perfect, then go buy a big expensive machine.

Oh, and doing this sort of thing is why I now have arthritis. And on that topic, the suggested youtube keywords to add are "rage against your hands".

Posted by Hal Eckhart at 11:12 PM | Comments (0)

March 03, 2011

Big aluminum walls

Jackson Graham made these walls from 2" square aluminum tube for a theater set using my tube-bending post from a long time back, and a few extra words of wisdom (hah!) from me. Nice job Jack!




Posted by Hal Eckhart at 10:24 PM | Comments (0)

January 24, 2010

Big Aluminum Boxes

I recently had the opportunity to make some large display cases from 1/4" aluminum to house some of the Dead Sea Scrolls for an upcoming exhibit at the Science Museum of Minnesota. Aluminum tends to warp and move around a lot from the heat of welding, and the idea of welding all the corners and having it come out perfectly made me pretty nervous. But it all worked out better than I feared. There was certainly warpage to straighten out, but it wasn't so bad that a few whacks with a dead blow hammer couldn't correct it.


clamping the bottom frame


The underside seam here isn't seen, so it only had to be tacked.

I found that welding into a fairly shallow 45° bevel (at about half the plate thickness) was sufficient to join the corners without causing a lot of warpage. Since this had to be more flat and square than strong, it seemed like a good risk to take. The joint was strong enough that I could hammer a sample to about 30° before it would start to break. Keeping the bevel away from the corner kept the corner from melting back.


Mig welding aluminum isn't always pretty. But with my fancy Lincoln PowerMig and its push-pull gun and "pulse on pulse", it worked great. It's at least 5 times faster than Tig welding and it heats up the aluminum far less. Maybe not quite as pretty, but it didn't matter here, as it was all being ground and routed.


I used a 1/8" roundover (climb cut, for a good finish) and 100 grit orbital to even it out.


Gap for the gasket


2 of the finished cases

Posted by Hal Eckhart at 11:05 PM | Comments (2)

September 15, 2009

Dragon Clock Tower


I blogged about building this corten dragon-head clock tower earlier (another project designed by Seitu), but I never got around to posting this picture of the piece when it was installed and had a nice patina on it.

Posted by Hal Eckhart at 09:18 PM | Comments (0)

September 11, 2009

Remote Control


Cheapo, no configuration solution for remote jogging and touch-off for my EMC-driven CNC plasma cutter. Six bucks at Amazon. I haven't tested the waterproof claim yet, but for the price, I'm not too worried. The USB cable cost more than the keyboard, and it was really cheap. The keyboard is flexible, so I put it in a frame to hang it on the side of the machine.

Posted by Hal Eckhart at 11:19 AM | Comments (2)

September 07, 2009

Coping and Bending

Here's a recent project utilizing bending and coping for The Science Museum of Minnesota. Mostly 1-1/2" schedule 40 aluminum pipe.




I modified the pattern to a 3-way joint.




I really need to get a new hand remote.

Posted by Hal Eckhart at 09:36 AM | Comments (0)

August 16, 2009

Before and After

I'm a fussbudget about my work. It may allow me to justify the prices I charge, but I can't seem to help myself, so I sort of have to go with it. To paraphrase Ben Folds, I don't like wasting time on metalwork that won't make me proud.

Of all the facets of metalwork, grinding is probably the one that I enjoy least (with the possible exception of hand unloading plate). It's tedious and often awkward, and it's hard to keep my gloves on because I have to keep running my hand across hot metal to see if it's smooth enough. Which leads to forgetting to put my gloves back on, which often leads to other unhappiness. Anything that makes this process shorter is a very good thing.

Flap discs are a much more effective way to do finish grinding than the way I learned 28 years ago, which was grinding wheels followed by sanding discs. I go through a lot of flap discs, and Lehigh Valley Abrasives (no, they don't pay me) seems to have the best prices around, so that's where I usually get them. I'm looking at a job in stainless steel that will require a lot of grinding and polishing, and they were kind enough to send some samples of their surface conditioning discs. But they included one type of disc in their sample kit that I didn't ask for, and boy am I glad they did. It's called an interleaf quick combination flap disc which "both grinds and finishes in a single step".

It's not really for any rough grinding, but once you get the surface pretty well cleaned up, it will smooth out minor bumps and also polish at the same time. It's sometimes pretty hard to tell if you've got the work smooth enough for powder coating, even with a regular flap disc. I've even resorted to taking a final pass with a brand new disc. This feathers out the grinding nicely, but it can leave grit marks which grab lint off rags like crazy. And that can sometimes leave visible blemishes in the final product, which I really really hate.

The results are really amazing. See for yourself. This is all 1-1/2" steel tubing handrails, with elbows by R&B Wagner.

img_0025_tn.jpg img_0027_tn.jpg

img_0034_tn.jpg img_0036_tn.jpg

img_1856_tn.jpg img_1864_tn.jpg


Posted by Hal Eckhart at 06:54 PM | Comments (0)

August 15, 2009

Yet another bending video

Here's a grainy video from '05 that I just found. I think this was 2" 14 gauge, which is fairly easy to bend. It is a little slippery, though, as you can see. At 0:24 the shim I hammered into place goes flying. I don't have any good videos of me landing on my butt and slamming my head into a cart while the steel I was bending landed on top of me, but that's certainly happened a few times.

Posted by Hal Eckhart at 12:03 PM | Comments (0)

August 11, 2009

Big gear


This is a big DIY spur gear for a turntable. It's 12 feet of 8 pitch gear rack wrapped around a 1/4" disk. I had to anneal the rack just to get it to bend, and even then it wasn't easy.
Posted by Hal Eckhart at 06:36 PM | Comments (0)

August 10, 2009

Bending again

I just wanted to post some evidence that I can still actually do this.

Posted by Hal Eckhart at 11:31 AM | Comments (0)

August 07, 2009

Large-scale CNC tube cutting

I've toyed with the idea of CNC tube cutting for years, but didn't actually get anything done with it until now. I never would have thought that my first paying gig in this format would be four 120 pound, two foot diameter Corten steel barrels. But the CNC method sure did save a lot of time and pain over doing it by hand. My arms hurt enough already.

The big spur gear was made by simply wrapping a piece of gear rack around a disk. The small amount of irregularity is absorbed by the spring on the motor mount. I just swapped out my Y axis for the rotary one, so the cutting file was a simple 2D pattern.

The lids were made by pulling a flat piece of steel down into a doughnut shaped jig. The amount of force required to do this was amazing, and I had to use a 2 foot cheater bar on the wrench. I ended up using a 3/4" threaded rod, because I was worried the 1/2" rod in this picture might snap off and shoot through the roof.

The goal of the project was to create some artful recycling containers for downtown Saint Paul's tiny but beautiful Mears Park. One problem with recycling in public places is that the cans need to look different enough from trash cans so that people don't throw trash in them. These definitely look different. They were designed by Seitu Jones for Eureka Recycling and Public Arts Saint Paul. The tubes were rolled by Linders Specialty.

img_1358_tn.jpg img_1362_tn.jpg

img_1377_tn.jpg img_1415_tn.jpg

img_1419_tn.jpg img_1421_tn.jpg

img_1424_tn.jpg img_1524_tn.jpg

img_1534_tn.jpg img_1807a_tn.jpg

Posted by Hal Eckhart at 09:49 PM | Comments (1)

February 16, 2009

Introducing the Hal-o-Matic

11 years after I first bought a really crappy CNC kit, I've finally rebuilt my plasma table into a nearly professional machine. With parts from Ebay, including a used Hypertherm 1650, this thing cranks out metal over 4 times faster than my old Esab (and without the nasty RF noise problem that zapped my computer). When I get the time, I will add torch height control and a drill now that the gantry is so butch.

The control software is EMC2 (the Enhanced Machine Controller), running on the Ubuntu flavor of Linux; both open-source and free. It's a treat to use, especially after enduring DOS and floppy disks for the last 11 years. The EMC project distributes a CD disc image, "EMC Live", for both the EMC software and Ubuntu. You can install from the CD, or just run it without installing for testing. The most recent version (as of today) is here, if you want to try it.

EMC with the Axis inferface:

Posted by Hal Eckhart at 10:38 PM | Comments (3)

March 22, 2008

Coping Calculator Upgrade

I've added a couple of tweaks to my world famous coping calculator which helps you cut a round tube to fit another round tube. It now uses vectors in the pdf version of the pattern, which should print and scale a little better than before.

And there's also another version that attempts to make the page big enough to put a large pattern on it. It's here:

The operation is a little flaky, but it may help somebody.

Posted by Hal Eckhart at 10:15 PM | Comments (0)

June 13, 2006

How to make steel rust quickly

As usual, a few words of caution. This is merely my attempt to impart some information that may or may not be useful. The potential dangers of following these instructions are your responsibility. Acid will blind you and scar your lungs and copper is essentially a poison. So don't get it on you or in you, or on or in anyone you don't want dead. Do not attempt this if there is any possibility that children will have access to the chemicals. Children are sneaky little monsters that love to drink poisons just to make you feel guilty when you're sitting in jail because you were a bad person who didn't think that they could pick that lock on the shed. And don't come crying to me when the EPA sizes all your assets and you end up on the front page of your local paper and all your former friends shake their heads in disgust at how you made all those spotted owls turn sickly green.

Anyway, be careful, and you might just come out of this alive and sane. Or not.


Rust can be a nice finish for garden art, sculpture, and even furniture if you don't mind the stains on your carpet. For some reason, I've been asked to do it several dozen times over the past 13 years in my business and I've worked it into a few personal projects as well.

Steel will rust all by itself, given a little moisture and time. CorTen steel will surface rust much faster, but it still takes at least a couple of months outdoors before it looks right. CorTen is alloyed with a few other things, but I think the primary element that helps the rusting along (and ironically provides future corrosion protection) is a small bit of copper.

There are commercial patina solutions for rusty steel, but they are usually pretty expensive for tiny little bottles. It's probably because of all the lawyers they have to hire to protect them from all the stupid things that people might do with their products. But by making the patina juice yourself, you might save enough money to hire your own lawyer. But probably not.

My first forays into making steel rust involved muriatic acid, which is dirt-cheap and available at most hardware stores. Used full strength, muriatic acid will pickle steel, removing rust and scale. However, if you treat the steel with a very diluted solution of muriatic acid, it will rust fairly rapidly. This usually takes a few treatments and a couple of days with the weather cooperating (humid and warm is best). It's not a perfect method, and it never really works the same way twice. I don't know if it's the variations in the steel, the weather, or just getting the acidity exactly right, but it always seems to take a lot of attention and messing about to get the finish right. And I always ended up with dead grass followed by mossy patches in the yard.

After several years of pursuing this method, I've found that the big trick to getting steel to rust quickly and well is to use a mildly acidic solution that has a little copper dissolved in it. I doubt that the small amount of copper used would provide any future protection against corrosion. It might even be that the steel will corrode quicker than if it wasn't treated, but I have no idea if it would make for much of a difference.

Getting some copper into the solution is easier than you might think. Muriatic acid dissolves copper, although it's pretty slow. What I did was to take about 2 feet of 14 gauge copper ground wire stripped out of a piece of Romex, wrapped it into a coil, and soaked it in about 2 ounces of Muriatic acid for a week. This will make a concentrate which should be enough for at least half a gallon of patina solution. A sturdy, unbreakable plastic bottle would be a good thing to prepare the concentrate in. Don't screw down the lid tight, or it might explode from the gasses that are created by the acid working on the copper. The fumes coming from the acid will cause rapid rusting on any bare steel nearby, so be careful where you put it. And be sure to label it appropriately. If you're as forgetful as I am, you won't even remember where you put it after a week. And it's not the kind of thing you want to sniff to find out what it is. You'll be sorry if you do. I promise.

After a few days, the solution should be ready to use. The copper wire will still be there, but you should see that the acid has chewed into it a bit. Once the acid has got some copper in it, the greenish acid/copper mixture will deposit a bright layer of copper on a piece of clean steel that you dip into it. This is a pretty cool effect, but it won't last. The coating is extremely unstable and will soon be replaced by rust. You don't really want to use it like that anyway, as it will go a lot farther and be a lot less dangerous if you dilute it with water. You may have to play with the ratio, but somewhere between 20/1 to 100/1 of water to acid/copper solution should work well. If you apply it with a spray bottle, just realize that it won't be long for the world. The acid will eat the spring in the bottle, so it will quit working after a few days. The only thing I've found that really works longer is a little plastic hand-pump Hudson sprayer that has no metal in it.

The rust will appear more evenly if your steel is very clean and free of the scale that you find on hot-rolled products (but I don't really care about that - it can look more interesting if it's not all that consistent). If you can't clean the steel properly, a more concentrated solution may work better, but be very careful. You definitely want to wear rubber gloves and eye protection, and do it outdoors standing upwind of your work. You probably want to do this even if you're using the diluted solution, at least if you care about your life.

Hot rolled tubing and hand forged sheet and round, partially ground, prior to spraying with acid/copper solution.

Time elapsed: 20 minutes

Time elapsed: 1 hour (with a second spray)

Posted by Hal Eckhart at 08:51 AM | Comments (73) | TrackBack

May 07, 2006

Dan Hopper's TUBEFIT Program

About once a month someone asks me for a standalone version of my Coping Calculator, so they can run it on their computer out in their garage where there are no internets. Since the script relies heavily on the power of PHP, ImageMagick, and GD, it's not really something that can be easily translated into a separate program. At least not one that I could create.

However, Dan Hopper has written this little standalone program that's vaguely similar to the Coping Calculator, but without all the bells and whistles. It runs in a terminal window and sends a postscript file directly to the printer via DOS. I really don't know how to set up a printer in DOS, so I had to find the spooled file (which went to LPT1, but that might just be my problem) and then drag it over to my USB printer. If anybody knows a better method, just let me know and I'll post your tip.

Dan's coping method differs from mine in that he's designed it to cut the tube with two passes of a power hacksaw, and then grind off the excess. However, the pattern looks very close to mine. For all I know, his math is better. He certainly explains it better anyway.

Unzip this file and put the contents wherever you want. The OVL file needs to be in the same folder as the EXE for it to work. It seemed to work for me, but there are no guarantees or promises from either Dan or me that it won't blow something up. Obviously, that's not the intent, but consider yourself warned.

TUBEFIT.zip (updated 5/19/06)
Derivation of Dan Hopper's TUBEFIT Algorithm
A brief bio of Dan
Posted by Hal Eckhart at 11:05 AM | Comments (0) | TrackBack

March 16, 2006

Another snow day

Here's some arty metal I made covered with snow. It looks pretty and all, but I'm getting a little sick of shoveling.


Posted by Hal Eckhart at 08:25 AM | Comments (0) | TrackBack

March 11, 2006

Parabolic Microphone

This is a user-focused parabolic microphone prototype for a museum project.

The interesting part of doing projects like this is trying to figure out how people are going to break it.

img_1422_tn.jpg img_1423_tn.jpg
Posted by Hal Eckhart at 09:06 AM | Comments (3) | TrackBack

February 26, 2006

How to make the continents on a big metal globe in 10 easy steps

Or maybe not so easy. I had such a hard time figuring this stuff out that I thought somebody else might benefit from my experience. I don't want to drive down business, but a few more big funky metal globes in the world wouldn't be a bad thing. And if you have the wherewithal to do it yourself you probably wouldn't be hiring me anyway.

The following is a rough and incomplete outline of one method of doing this. There are other methods. Easier methods. There are also probably open-source (and free) replacements for Photoshop. I just haven't found one that actually works for this. Gimp can output the PPMs and convert back to PNG (probably quicker than ImageMagick) but it doesn't create paths and it doesn't open RAW files without a plugin. I couldn't figure out how to make the plugin work, so I gave up. If anybody can really do this (as opposed to saying they know that it should work), just let me know. I'll try to keep this page updated with anything useful, and you can always add your own notes to the comments.

This "tutorial" (if you can call it that) assumes that you know your way around the block. I'm too lazy to make this into "The Idiot's Guide for Globe Construction". However, it should help point you in the right direction or at least give you a few ideas.

Minimum Tool Requirements


  1. Go to the Globe project at NOAA, and get the tiles.
    Yes, they are big, but they are free (unless you buy the CDs for $260). One of them is nearly 60 MB, but most are 20 to 30. Better make sure you have either broadband or a whole lot of coffee.

  2. Import to PhotoShop and combine the whole thing into one big image.

    Lose the irrelevant channels and save the image as a two-color PNG.
    Your file will now look sorta like this:


    PPM is a very inefficient format, because the file is written as a rectangular grid. A 1.1 MB PNG can become a 667 MB PPM, so make the PNG as few pixels as you can get away with, or the next steps will be really slow.

  3. Convert to PPM with ImageMagick like this:
    convert in.png out.ppm

    And then go have some coffee. Across town. Walk there.

  4. Split the PPM into gores with Perl and make_gores.pl
    Usage: $0 in.ppm out.ppm number-of-gores
    Like this:
    perl make_gores.pl in.ppm out.ppm 12

  5. Convert back to PNG like this:
    convert in.ppm out.png

    Your file will now look sorta like this:


  6. In Photoshop, create a path by selecting some white, select similar, new path from selection, save path, export path to illustrator, yadda yadda. You may have to screw around with various settings to make it work.

  7. Open the path in your CAD program and clean it up. I managed to get the path open with Rhino , but only after saving it with DOS line endings in BBEdit (it doesn't suck™). Otherwise, Rhino wouldn't open it and just crashed.

    You will want to be careful to get rid of the tiny islands, or you will have a lot of little dots that you couldn't use even if you could figure out where the heck they go.

    You can reduce the detail, but don't get too carried away. Look at Japan and England. They should be more or less recognizable.

    Some things may want a little redo. Greenland can generally be one piece, but it gets cut up because it's so close to the North Pole. Either patch it back together in CAD, or go find a map online. It's not hard. And Antarctica. You really don't want 12 or 24 pieces. Find a map and scale it. And then cut maybe 4 darts in it so you can make it spherical.

  8. Export to your favorite CAM format and cut it out. I saved the file as a DXF and converted it for my cutter with SheetCam, which is a fantastic cheap 2-1/2D CAM application with G-Code and HPGL output.

  9. Hammer your pieces into shape and weld to the globe's frame.

    Which might look sorta like this:

    This 8 foot CorTen steel globe (to eventually be part of a sculpture in front of my building when I have time) was made with 24 gores. It should have really been more, or perhaps segmented somehow, so that there were more segments near the equator. Oh well. Next time maybe. And maybe thicker material (this was 11 ga.) would have helped. The North America on the wall above the globe was 1/4" stainless (same scale) that I did as a test. It came out remarkably smooth, although it's higher up from the equator, so the width of the pieces was less.

  10. There is no step 10. Unless you want to go back to the coffee shop. You don't have to walk this time. You can ride your bike.

Ancient History

The first globe with continents that I was hired to make was a challenge and a bit of a nightmare. It was a large globe (16 feet across and 10 feet tall, to make a little room in a theme restaurant bar), so I thought I could just beat some aluminum into shape and piece it together. After an hour of hammering, I was so frustrated that I was ready to cry. I'd made a huge amount of noise and an unsightly lump in a 4' x 10' sheet, but there was no way it was going to be usable. I didn't have a clue what to try next.

I ending up making 1/12th vertical gores about 4 feet wide with horizontal darts in the edges about every 16 inches. Then, after hammering and welding them into shape, I eyeballed the continents from a 12" globe and cut everything out. It was a ludicrous amount of work, and the result wasn't as smooth as it should have been. Here it is.

A couple of years later, I got asked to do the same exact thing again. In the interim, I'd built a CNC plasma cutter, so I imagined I could cut it out easier. Since I couldn't figure out where to find good geographic information, I used the same 12" globe, traced the continents on vellum, and scanned it into the computer. I then had to manually draw all the shorelines, which was another fantastically labor-intensive task. And thus, my lifelong hatred of CorelDraw.

Finding a way to import the basic geographic data without a lot of hand work has been the global holy grail for me. The process isn't perfect yet, but it's a whole hell of a lot better than my first pass.

* Can't I just open the path in Illustrator and save it as a DXF?

Don't bother.

For whatever idiotic reason, Adobe Illustrator has seen fit to export its DXF files with only Macintosh style (\r) line-endings. This is completely insane, because 99% of CAD programs that people actually use run on Windoze, and most of them will only read Windoze (\r\n) or Unix (\n) style line endings. So if you want to open the friggin file, you have to run it through a line-ending converter. Yes, they're easy and available everywhere, but it's not something that the ordinary guy is going to know about. And it's completely avoidable. For Chrissakes, a DXF file is supposed to be "a CAD data file format, developed by Autodesk as their solution for enabling data interoperability between AutoCAD and other programs." Since AutoCAD won't even run on a Mac, what the hell is the point of a DXF with Mac line endings? I don't know what they're smoking, but I want some of it.

To make matters worse, every curve is written as a spline, which often causes problems with CAD/CAM systems. And the only way to fix this in Illustrator involves turning the whole thing into about a zillion tiny little straight lines. Which will frequently cause other CAD/CAM issues. Sheesh.

However, Illustrator does have some nice features. It can simplify curves and reduce points, which is helpful. If you have it and can deal with the problems, it might be helpful after all.

(This applies to the Mac version of Illustrator CS. I have no idea what the latest version does.)

And finally, if there was a decent CAD program that would just import SVG files, all of this bitching would likely be irrelevant.

Posted by Hal Eckhart at 10:00 PM | Comments (6) | TrackBack

February 22, 2006

The Universal Tube Coping Calculator

There's now only one version of my Tube Coping Calculator. Trying to maintain various versions scattered all over the internets was too confusing, and it seems that Google doesn't like content that appears to be nearly indentical. It will be interesting to see if it breaks.

I've hacked it to save your last entered info in a cookie. Cookie/form/PHP tutorial to follow.

Posted by Hal Eckhart at 10:36 PM | Comments (0) | TrackBack

February 21, 2006

Steel Blocks

More cute metal. 20" and 16" steel blocks for a children's hospital. 11 gauge steel sheet, tig welded and heavily ground. The corners are really smoother than they look.

The clever part was the little triangular patches in the corner, hammered and ground to fit.

The patches looked like this: corners.jpg

And the blocks came out like this:


Posted by Hal Eckhart at 12:57 PM | Comments (0) | TrackBack

February 20, 2006

Careful with that axe, Eugene

Here's a fun little chunk of metal. Half inch water jet cut aluminum, chamfered with a laminate trimmer, ground with a customized belt sander and a flap disc on an angle grinder, and then dressed up with an orbital sander.

Handle carefully.


Posted by Hal Eckhart at 03:53 PM | Comments (0) | TrackBack

February 05, 2006

Metal bending video

Here's a little video of me bending a chunk of 2" - 12 gauge steel tube by hand. This is one instance where having a little extra body mass is useful. 8^)

Someday I'll have to get a video of me falling on my ass when the jig breaks. 8-(

Despite my joking about getting hurt, it's a serious danger. I've spent the past week recovering from a wrenched knee caused by a clamp that slipped when I didn't expect it. And even though I'd prepared a clear path for falling, I managed to stumble backwards before I fell, so smacked my head on a heavy steel cart. Luckily, the cart was not damaged.

Hal bending steel

Posted by Hal Eckhart at 10:13 AM | Comments (0) | TrackBack

May 01, 2005

How to Bend Tubing and Pipe by Hand

Most people seem to think that you need expensive machinery or heat to bend metal. It isn't generally true. Basic bending can be done with nothing more than ordinary shop tools, a bit of elbow grease, and some ingenuity.

Somewhere around 1978, I took a theater materials class and learned how to weld. As part of a sample prop I'd decided to build, I needed to make some 3" rings out of 3/8" steel rod. The only way I could figure out how to bend them was to weld the rod onto a piece of pipe, clamp it in a vise, and twist it around the pipe. Before I did this, my teacher walked by and told me I'd need to heat up the rod with a torch in order to bend it. "Oh yeah?" I replied, as I wrapped the rod around the pipe three times. I suppose I've felt a certain smugness about my metal mangling abilities ever since.

When I first started working in professional theater shops, the only way I saw metal tubing bent into shapes was by cutting partway through the tubing at intervals, bending it, and then welding the kerfs closed. It's a simple method that's easy for a hack like me to understand, but it's tediously slow and usually produces a weak and ugly product. And welding the kerfs closed causes a lot of warpage, so you never know exactly what you'll end up with.

Paying a vendor to do roll bending is one alternative, but it slows down the creative process, and it's not useful for artsy shapes or ellipses. Machines (unless they are very expensive) usually only bend one radius at a time. You end up having to weld a bunch of pieces together, so it's not much better that the kerf and weld method.

After years of frustration with this, I finally saw someone bend tubing with a homemade jig. It seemed magical at the time, but it took me a few more years of intermittent fiddling to understand the process well enough to be able to produce consistent results. In the years since then, I've learned a lot more through trial and error. And spilling some blood. I hope that the following ramble will help you understand the process and its pitfalls a little quicker than I did.

A Word of Warning

I mentioned blood. Metalwork is dangerous. You will hurt yourself. How bad is up to you.

Bending metal requires some strength, but mostly it requires that you figure out a way use the metal you want bend as a lever to bend itself against a form that may be destroyed in the process. When that happens, you need to be prepared for the outcome. Flesh is weaker than metal, and concrete is harder than your ass or your elbow. If you do much of this, all these things will become acquainted with each other eventually.

Consider yourself warned.

Tools and Jigs

What sort of equipment you'll need for bending depends on the size of the tubing you want to bend. For anything 1" and up, you'll definitely want a sturdy, well braced table bolted to the floor. You can get by for a while with a plywood top, but it will eventually get destroyed. My table has a 1/4" steel top, which allows for permanent holes for jigs and stops that won't tear out without serious abuse. This will work well for any bending up to about 2". For anything heavier than that you'd probably better call a professional anyway.

I've mostly always used 1/2" steel pins and bolts to locate the jigs. 5/8" would be better for heavy bending, but it isn't usually necessary. Bent pins are easy to fix or replace. A few large C-clamps will help keep the jig from slipping and tearing out the holes.

Spend some time thinking about where you're going to bolt down the table. Ideally, you'll want at least a 20' radius (from the jig) of clear space on two sides of the table, and a good 10' on the outfeed side. The best way I've found to anchor the table is 1/2" threaded rod and anchoring cement. Normal anchors, even big ones, always seem to fail after a while.

You'll also need a heavy duty stop to hold the tubing against the jig and a bunch of holes in your table for bolting it down. The stop and the jig must both be extremely square to the table or the tubing will twist. My favorite stop is made from a very heavy piece of 1-1/2" tubing with a bit of pipe welded on the end. The holes are offset so that I can swivel it to get a tighter fit against the tube.

jig_stop_tn.jpg trammel_tn.jpg

One other tool that's invaluable is a sturdy router with a large trammel. Routed jigs will bend smoother and with less kinking or twisting than a jig cut with a jigsaw. If you can't get your hands on a router, just make sure that the cut is as smooth and square as you can make it.

Your trammel can be anything you want, but basically what you need is to attach a stiff bar with holes in it to your router. It can even be as simple as a strip of plywood. Most routers have plastic plates on the bottom that can be removed. Just use those screw holes to mount the trammel. I've made trammels up to 35 feet long, but I have to admit that was pushing the envelope.

The jig material that I use the most is 3/4" medium density fiberboard, or MDF. 1" or 1-1/4" is even better, but I don't have a convenient source for it. 3/4" plywood will suffice, but you might have some trouble with kinking, as it's not quite as consistent as MDF.

3/4" MDF will make a perfect jig for 1" square tubing. For 1-1/4" or 1-1/2" square, you'll want to prop up the jig with some shims so that the jig is more or less centered on the tubing. With round tubing, centering the jig is even more important. If the tube is too heavy, or the jig too soft, round tube may destroy your jig. One way around this is to bend a thin strip of steel to protect the edge of the jig.

unbent_tn.jpg bent_tn.jpg
This is 1-1/4" MDF and 1" tube

For bending big tubing or pipe (like 2" tube or 1-1/2" pipe), you'll want to use a steel jig. The easiest way to do this is to bend two pieces of 1" tubing with a wooden jig and then weld and brace them together. This is actually better than a bent 1 x 2 because it's more consistent and the seam between the two pieces of 1 x 1 is the contact point where the 2" round hits the jig. And since the jig is 2" thick, it's the perfect size for a different sort of end stop. Just take some 1/4" x 1" strap, make a U around a piece of 2" round, and weld it to one end of the jig.

heavy_jig_tn.jpg wall_jigs_tn.jpg floor_jigs_tn.jpg
Some jigs

Small Bends

Homemade hand-bending may not be the best technique for really short radii. For example, 1" x 1" - 18 gauge steel tubing can be bent to a minimum diameter of about 30" before it starts to kink. This isn't an exact number because of the differences in temper and metal quality. And there are a couple of ways to cheat that number even lower. The easiest is to bend several times using successively smaller jigs. How far you can take this is something you'll have to find out for yourself. The other way, which only works for square tubing, involves pushing in the inside wall of the tubing. Here's a 5" radius jig I made for 1" tube that works like a conduit bender.


This does cause some distortion, but the tube retains most of its strength and it's a pretty darn fast technique. The ridge inside the jig is only 1/8" square wire, but the tube wall collapses about 3/8".

Manual Power

If you're capable of lifting 100 pounds, you shouldn't have much trouble bending tubing up to about 1-1/2" by yourself. For 2", a friend is usually necessary. Spend some time thinking about how you're going to land when something breaks. Something will break. You will end up on your ass.

The forces involved in this may not be entirely obvious at first. In addtion to the weight of the metal and the force that you're exerting, don't forget about potential energy. The metal being bent is a potentially dangerous spring that really wants to smack you in the head. If your jig gives way, metal and jig parts can come flying off the table with surprising speed.

Cheater bars can help you make difficult bends. But they also increase the danger. Landing on your ass with both a 50 pound piece of 2" pipe and a 2-1/2" cheater bar heading for your face is not where you want to be.

Some Tips

Bend with the seam toward the jig. The weld seam is a little harder than the rest of the tube, so there will be less chance of twisting or kinking if it's on the inside of the bend. And it's usually less visible there.

If your floor is too slick, you may not be able to push hard enough. A very light misting with some nasty spray adhesive will help.

Use the tools that are available. Difficult bending can aided by the proper application of a forklift. Jigs can be mounted vertically on shop pillars so you can use your body weight. It looks silly bouncing up and down, but it's effective.

It's always better to slightly overbend than underbend. It's easier to pop it back out a little than to bend it just a little farther. A stub of 2" tube welded to your table 2" off the floor gives you a convenient notch to unbend your mistakes.

Bending Data

Metal always springs back from the jig. Here are a few charts that show results I've gotten. Some of the data may not make sense, and some of it may be wrong. Metal hardness can vary, even from the same mill. Aluminum is more unpredictable than steel.

Hopefully, these numbers will give you an idea of where to start. If you have a better idea, just let me know.

Happy bending!

* All material is steel unless noted otherwise.
* Frequently, bending using progressively smaller jigs will produce slightly larger radii than if you just use the smallest jig first. It's just one of the weird things about how metal reacts to force.

1/2" - 20 ga. round tube
desired outside radius jig radius
144" 54"
120" 50"
32.5" 25.5"
30" 24"
27" 22"
21.75" 18"
21.5" 17"
19.5" 16"
12" 10"
1" - 18 ga. square tube (Central Steel)
desired outside radius jig radius
192" 108"
144" 85"
114" 75"
96" 65"
80" 60"
72" 55"
65" 50"
58" 45"
53" 41"
49" 38.5"
45" 37"
42" 35"
39" 33"
37" 31"
35" 29"
33" 27.5"
31.5" 26"
30" 25"
29" 23.5"
28" 22.5"
26.5" 21.75"
25.5" 21"
1" - 18 ga. square tube (MetalMatic)
desired outside radius jig radius
67" 50"
49" 38.5"
34" 28"
29" 25"
24.5" 21"
1" - 14 ga. square tube (Ryerson)
desired outside radius jig radius
456" 144"
300" 120"
258" 114"
240" 108"
228" 105"
216" 102"
126" 72"
108" 66"
90" 60"
72" 50"
42" 34"
39" 29"
31" 25"
17.5" 15"
11" 9.75"
1" - 14 ga. square tube (Discount)
desired outside radius jig radius
168" 96"
158" 90"
104" 66"
63" 47"
60" 45"
57" 42"
34" 26.5"
22.5" 19"
18.75" 16"
1-1/4" - 14 ga. round tube
desired outside radius jig radius
33' 12'-6"
20' 10'
16'-6" 9'
15' 8'-6"
14' 8'
13'-6" 8'
12'-6" 7'-6"
12' 7'-6"
11' 7'
10' 6'-6"
9' 6'
8' 5'-6"
7' 5'
6' 54"
5' 52"
63" 50"
54" 44"
47" 38-1/2"
44" 36"
42" 35"
37-1/2" 32"
33-1/2" 28"
30" 26"
25" 21"
1-1/2 - 18 ga. square tube
desired outside radius jig radius
180" 120"
168" 114"
141" 102"
108" 84"
84" 72" first, then 66"
1-1/2 - schedule 40 alum. pipe
desired outside radius jig radius
204" 96"
108" 68"
132" 75"
67" 48"
48" 36"
2" - 12 ga. round tube (Ryerson)
desired outside radius jig radius
228" 144"
168" 120"
156" 108"
144" 96"
84" 75"
44" 36"
28" 25"
2" - 12 ga. round tube (Discount)
desired outside radius jig radius
173" 108"
115" 96"
96" 75"
84" 68"
75" 50"
70" 50"
54" 46"
36.5" 29"
31.5" 25"

These tables are admittedly incomplete. I will update them when and if I've got more data and more time. If you come up with a reliable list of your own, feel free to send it to me.
Posted by Hal Eckhart at 02:34 PM | Comments (87) | TrackBack

October 03, 2004

Deflection Calculator for Square Tubing

Important Caveats:

This calculator may give you a rough idea of the deflection of metal tubing, but it's not accurate enough to do any more than estimate it.

The steel weights are tweaked for mechanical tubing, and will be less accurate for structural tubing.

Use at your own risk.     Or you'll be sorry.

square tube in inches      
decimal gauge                  
length of beam in inches  
center load in pounds      

type = steel   ends = free
est. weight = 1.7105# per foot
OD = 2"  ID = 1.87"
Moment of Inertia = 0.3143

Deflection = 1.9745"

Here's another place to find this calculator.

Posted by Hal Eckhart at 09:08 PM | Comments (0) | TrackBack

August 26, 2004

The Dragon

An early sketch by Seitu Jones

It's only a model

Rhino 3D wireframe

The flattened facets of the trunk

One rendering

Another rendering

The trunk armature

2 powerhammer dies for curving the facets

The facets after hammering

The armature with half of the lower skin on

Tabs helped to align the halves

75% wrapped

This strip was difficult to force into place

One way to cheat



Lying down

Standing up

The head

Lying down

The head again

Lying down

Me and my dragon

This is a project that really gets me going. It's a sixteen foot tall dragon head clock tower designed by Seitu Jones for the FAIR School in Crystal. It's being produced by Art-Tech Productions in Minneapolis, who hired me to do the metal work.

The tower is being constructed primarily from CorTen Steel, the rusty-looking stuff you see nowadays on big power poles, bridges, and sculpture. The small amount of copper in the steel makes it the surface oxidize very quickly, and then protects it against further corrosion.

The only thing really difficult about this project is the tapered, curving trunk of the tower. (Perhaps it's actually a neck more than a trunk, but nobody's corrected me yet). Bending a tube that size (even if you could get it in CorTen) would be enough of a challenge, but it tapers from 9 inches down to 5, which would make bending it almost impossible without equipment I can't even dream of.

So I decided to construct it from two sets of 6 facets, calculating and exporting the shapes with Rhinoceros, which has a couple of clever features that help in this situation.

The first is the ability to extrude a profile (called a curve in the program) along two "rails" (two curving lines, in this case) to create a flowing 3-dimensional shape. I used the option "do not simplify", which helps the next step work correctly.

The second clever feature that Rhino has to help me out here is "unroll developable surface". To use it. it's first necessary to "extract" the surfaces of the 3D object, which separates it into faces. Then, the unroll command flattens the face, and tells you how much distortion it caused. The distortion hasn't ever been a huge deal for me (if the command succeeds), but it can be difficult to get the faces pulled together.

One way that I've found to draw sheets together is this situation is to weld a pair of wide jaw vice-grip pliers to the steel, and then use another pair of the same sort of vice-grips to squeeze the first. Then I just tighten one after the other until the gap is close enough or the weld breaks. It's really amazing how well this works.

- - - - - - - - - - - - - - - - - - - - - - - -

more later

Posted by Hal Eckhart at 06:35 PM | Comments (4) | TrackBack

June 26, 2004

New toy, old toy

I'm in love with my new toy, a digital camera that even shoots video. Here are a few low res screen captures and a small movie of tapering a 1" diameter steel bar in one heat in one minute. It wasn't really white, although it was a skosh too yellow.

The old toy is the one making all the noise, a 50 pound Little Giant power hammer. It sure saves a lot of wear and tear on my arms, as well as speeding up production. Which is a good thing, since I keep finding things to take pictures of.

Jeff with the Little Giant

taper_1.jpg taper_1.jpg taper_1.jpg
taper_1.jpg taper_1.jpg taper_1.jpg

taper video
837K Quicktime video

Posted by Hal Eckhart at 08:58 PM | Comments (0) | TrackBack

April 06, 2004

Coping notes

This is just a validation of the latest rendition of my Tube Coping Calcutor thingie. Using a PDF turned out to be a pretty good idea after all. It scales perfectly. Here's a 20 degree joint on 1-1/2" schedule 40 aluminum pipe. Plenty close enough for rock + roll.

If anybody uses this to make something, send me a picture, and I'll post it. I seem to have a small following among off-road vehicle customizers, but I haven't seen what they're making yet.

And here's something I forgot earlier. It just shows the difference that various wall thicknesses make in the pattern. The calculator thingie plots the pattern by figuring the intersection of the inside and the outside of the cut tube, and then taking the smaller number. It's true! This basically squares off the sharp edge of the joint, which makes it quicker to cut and better for welding anyway. Unless you're brazing, which takes too much patience for me anyway.

Posted by Hal Eckhart at 06:11 PM | Comments (3) | TrackBack

March 26, 2004

Acorn Die

acorn_die.jpg This was so quick, dirty and effective that I just had to write it up 1. Roughly form acorn from 3/4" stock using a belt sander. 2. Mash it between 2 pieces of hot 3/4" x 1" welded at the other end. 3. Quench. 4. Heat some more 3/4" and bash it good in the die, rotating until it's mostly round. acorn_railing.jpg
Posted by Hal Eckhart at 08:06 AM | Comments (0) | TrackBack

March 25, 2004

Potato Chip Truss

chip.jpg Here's another big goofy thing I made a little while ago. It's not quite as complicated as it looks; basically just 4 ellipse segments at tangents to each other. It's all 1 - 1/2" and 3/4" schedule 40 aluminum pipe, hand bent cold using jigs bolted to a table. One of the many things that I someday hope to do with this blog is post some tips and tricks about tube bending. Someday.
Posted by Hal Eckhart at 09:16 PM | Comments (0) | TrackBack

January 01, 1956

12' aluminum pineapple

pinapple_s.jpg **Here's a big thing I made a few years ago.** [John Gruber](http://daringfireball.net/) is a really smart guy. *Isn't it cool that I posted this before there even was an internet?*
Posted by Hal Eckhart at 01:01 AM | Comments (0)