Tuesday, July 28, 2015

Apache Cassandra

I'm writing a post about Cassandra for Orbitz's tech blog. I just thought I'd leave this here because it may not be appropriate there...

That is all.

Wednesday, March 18, 2015

Find steel locally

I've built equipment mostly from local steel. Stuff I can't find locally is from onlinemetals.com. They are good at small orders. However, for bigger things that weigh over a hundred pounds, local is better. Also some of these might just have big drops they can't find a use for.
SB Specialty Metals Cincinnati Tool Steel co Key Metals

Update: Metal Supermarkets has distributors all over. There's one 5 miles from my house, and I can order anything and have Zero Shipping Costs, as long as I pick it up myself.

Cutting ISO threads (by trial and error, mostly error)

 I recently got a 8" Bison 6-jaw Set Tru scroll chuck for about $100. These are forged steel, made in Poland, and are generally around $2000 for a new one. I figured this one would be worn out, since there are more moving parts in a scroll chuck than an independent jaw chuck. The internal ring gear is pretty well sealed, so it doesn't usually wear out if kept greased, but the centrifugal forces of the spinning chuck sling the grease out and pack it into places where no mechanical engagement happens, and over time the grease hardens and doesn't do much good. The ring gear can wear if the grease isn't replaced, from contamination or just from age. There's not much wear on this ring gear or the turning gears at all. They still have the original machining marks, just a little polishing on top of that.

The Bison scroll chuck (taken apart for cleaning and lube). I didn't show the Set Tru screws :( or the shoulder where the chuck registers. So this is just the chuck, in pieces.

The 8" backing plate blank. Cut from 1 1/8" 1018 round, ~2" length.

Boring/threading 1 3/4 x 8 TPI thread. Shop made boring bar from a 1" 1018 bar with HSS bit ground to 60°. You can just see a few threads about 1" in from the face. The back plate is drilled from the back and bolted to a slotted face plate that's trued to the spindle. 

I don't have ANY kind of thread gauge, or a snap gauge, or any way to measure bores deeper than 1/2". I'm cutting this to 1.80 on the outer bore. An ISO thread should have all of the dimensions below included:

So one way to cut threads (single pointing) is to grind the tool to ~60°, and set the compound angle 29.5°, which is .5 less than half. A hair less than half is basically the idea here, because when you move the tool into the thread by a few thousandths per cut, only the leading edge of the tool has to cut, and the trailing edge is not 'dragging' because of the added clearance angle.  The dragging side of the tool would create heat and ruin the tool and maybe the part. Plus, the tool will try to get ahead of the lead screw (where there is backlash), and tend to make the pitch longer as the thread advances.  

 This is how threading is taught in schools, where the lathe is an old South Bend 9" with a not very stiff lantern tool post. If you plunge cut both sides of a thread in that setup, straight in with no compound angle, you'll get a lot of howling from the tool, broken tools and not very good threads. Especially if you're threading from the end of a boring bar.

Another way to single point a thread is to use a half tool without the back cutting edge and feeding with the compound at 30ยบ.  Any flex in the tool or cross slide will probably make the left face of the thread rough, but the 1/2 degree is taken out of the equation. This thread would require more cleanup.

Regardless of how the tool is ground, there's the question of how much the total compound feed needs to be to complete the thread. The rule of thumb equation is .708/threads per inch. So in this instance .708/8tpi = 0.0885". So your compound needs to feed forward from zero to 0.0885 in increments of a few thousandths per cut. That's one method.

The no calculation involved method, is called the zero-to-zero method.
Touch the tool to the major diameter, set the compound and cross slide to zero. Back out the cross feed enough to clear the part, move the cross slide out free of the part, reset cross slide to zero, and advance the compound until the desired depth of cut is measured. A dial test indicator on the compound touching the major diameter will measure the desired depth. Reset the compound's dial to zero.
Starting your cut from the major dia., advance until the compound reaches zero.

At the end of either method, a really light spring cut is taken using the cross slide feeding into the cut straight to clean up both sides of any marks from wobbly gibs or tool flex. A file is run over the tops of the threads to remove any burs.

I didn't mention anything about change gears, gear box or threading operations besides the angles involved here. Gearing is specific to the lathe, and threading operation is best shown by demonstration.

Friday, February 14, 2014

Blanking and grinding

A few blades blanked and rough ground. 1095 and O1 steel. 

Trailing point double edge
Straight tanto point

Thursday, May 16, 2013

Package structure

Package structure is not something to be considered idly.
The first thing I do when I look at an application is
$ tree -d src/
And I get something like this:

 |-- core
 |   |-- dao
 |   |-- domain
 |   `-- engine
 |-- service
 |   |-- business
 |   |-- adapter
 |   |-- exception

What do I know about this application from its package structure? Not much. It has a core, an engine, that's exciting. But not at all descriptive of what it does. Sounds like a spaceship! or an alien weapon!

However, if I were building a spaceship, I'd do something like this:

|-- lifesupport
|  |-- atmosphere
|  |-- gravity
|  |-- replicator
|-- communications
|-- propulsion
|  |-- thruster
|  |-- impulse
|  |-- warp
|-- defense
|  |-- shield
|-- weapons
|  |-- phaser
|  |-- torpedo

OK, so a space ship is much much easier to design than a piece of software! I have all the major features described at the top level. I can tell what the ship does, what its major components are.
The ship's design is organized by Features of the ship. Not by layer or any other attribute. For example, a builder or a command isn't a feature of a piece of software. Neither is an exception, a dao, a domain, an adapter..
These are all layers or programmatic idioms. Grouping classes by layer or type is a common mistake in designing software. It is always taught that one should design by feature, not by implementation detail.

Monday, November 12, 2012

Custom stamping dies

Henry Aevers Co. "EverStamp" Ebay store "If it's worth making, it's worth marking"

Power Hammer notes

Dupont linkage: The total stroke of a Dupont toggle machine is 2x the crank offset, plus the distance between dies and the upward stroke while compressing the spring.
Dupont patent:

little giant

Little Giant sells springs, toggle links, etc for the 25,50,100# hammers. Instead of trying to calculate which truck spring to buy, just get a complete toggle link from them. Use the same crank offset as the LG. Or use the LG crank. I like the later LG upside-down Pitman arm. Buying the above from LG is probably less expensive than making it all. And it would be correct the first time. No need to re-invent the whole thing. Most of the early tire hammers used big structural tube as the frame, in place of the big sand-cast iron frames of the early 1900s power hammers. Instead of structural tube or I-beam, use an A-frame like the Japanese power hammers. An A-frame hammer with a separate anvil can be deconstructed and moved more easily than a monolithic design. An a-frame can be made stiffer than a structural tube or beam, which will tend to develop oscillations and rocking with a small base and high center of gravity. The recommended anvil weight ratio (from IronKiss) is 16:1. A 6x36" bar of 1018 should weigh around 289#, but I've seen different on the scales! 7x36 would be closer to 400#, the ideal weight. For a 50# hammer, a 11" base.. Metals weight calc for dimensions. http://www.machined-castings.com/calc.html

Truck/Trailer springs:
Meh. Bow spring or coil.

Thursday, November 8, 2012

Little Giant Hammer Co


These are the owners of Little Giant company. They bought the rights and tooling a few years ago to help people restore and maintain their Little Giants. Great place to get parts even if you don't have a little giant. Almost everything but the cast frame is available. Most of the parts which were problematic have improved replacements here too.

Wednesday, October 24, 2012

Ford Bookmobile - Prison van?

Ford Bookmobile

This is an ebay auction for a strange bookmobile. It appears to have been either converted to a riot/prison van or just had some bars put in it to keep people out. Check out those bookshelves! It would make a really cool RV.

Friday, October 5, 2012

Augmented Reality for Precision Welding

I finally got a solar auto-dimming welding helmet last year. I never found that with proper technique these were absolutely necessary for quality welding, but they do make it possible to make all tack welds and finish welds without ever lifting your helmet. I hate welding helmets in general, and it is important to get the best ones and maintain them, because the cheap ones are just shit and make welding a pain in the ass. The automatic helmets have some drawbacks though. You have to manually adjust the light blocking and sensitivity to the environment and type of welding. If the photovoltaic cell is obscured by something, which is often the case when working in tight places like under cars etc, the auto-dimming doesn't work, and you end up effectively looking at an arc through a #3-4 shade, when you should be protected by a #10. More expensive helmets, which I would recommend for anyone who welds daily, have auto-sensing filter controllers for ambient light settings, arc amperage, arc time display, other nice stuff. Most of the welding helmets don't have integrated air filters. Welding fumes are toxic, sometimes deadly (galvanized plating will kill you). The automatic helmets are a pretty recent development. The next generation is coming though, and will be a significant improvement over the previous glass filters. These provide a real time HDR projection directly to the eye using video and real time HDR processing: By processing multiple video streams at different dynamic ranges and combining the optimal visible light into an optimal stream, you get complete detail of the weld area, molten metal puddle and arc without the extreme contrast that a #10 shade provides.

Thursday, August 2, 2012

Monday, July 30, 2012

Mini elm tree

I pulled this little elm out of the ground a few days ago. I chopped it short and cut some of the roots. Then I put it in some "recycled organic material" plus a little nitro.. I didn't figure it'd survive, but nitro makes every plant happy. This is 1 week after planting the stump.

Wednesday, July 25, 2012


Writing words is in the blood. I'm the son of a journalist and accomplished author, and the brother of an accomplished author. I've been told I'm a good writer, then a bad writer. For example, the entrance exam at a community college scored me very low on the essay portion, due to my not following the format. In fact, they scored me so low, I didn't even get into a required 101 level writing course. Bad writer. Academically speaking, I'm too informal. Maybe that doesn't count.

I found this simple script that removes wordiness from text by removing or simplifying words and phrases.
Zinsser transform:

And the nice, complete with library of transforms you can add to, Python script:

Monday, July 23, 2012

Flashback F100s

I love this site: 

All USA-made restoration, reproduction and original parts for old Ford F100s.

Tuesday, July 17, 2012

Timing chain guide rails

The V8 in the 500SEC (Mercedes M117 engine, 5.0 liter) is interesting. It has a forged crank, pistons, rods, light but stiff aluminum block and heads, double row timing chain, some other interesting bits for that time. OHC, 2 valve per cylinder, and like most benz engines have stainless valves with the exhaust valves sodium filled, hardened seats, polished ports, internally balanced and factory cc'd.
They do have a serious flaw though. The SOHC heads require a very long, very heavy timing chain, which is common on German cars of the 'investment' grade, which drives the valve and ignition timing.    The chain is guided and tensioned by a number of plastic guide rails. These are consumable items, since they are friction surfaces, subjected to wear by having a heavy steel chain under tension dragged over them at 1/2 the engine's RPM. I've seen some worn through to the aluminum backing underneath, which means bits of aluminum and plastic have ground off into the sump, and that the timing is waay off, chain probably wearing badly, etc.
This would be a problem on engines with bad cam oilers, bad chain tensioners, people not getting oil changes, bad/wrong oil, other simple things like that which shouldn't happen to a well maintained car. However, these were built more than 25 years ago, and nobody knew then how long the engines would be used or what time and chemical breakdown would do to the plastic guide rails.
BTW I don't think anyone had trouble with the rails in the 80s, not at least like they did with the weak single row timing chain on the 380 engines, which was later replaced with the double row chain.
Long story short, the guide rails will get brittle and break off, fall down into the timing cover and get stuck between the chain and teeth of a cog. This will destroy the engine beyond any reasonable repair, as the timing chain can jump a few teeth, break, or otherwise cause valve/piston interference, after which you will have nothing left but a block that may be salvageable.

In the above picture, the front timing cover is off, exposing the crank timing sprocket, idler, tensioner (left side) upper guides (white plastic sticking out of the heads). This engine has been conveniently pulled out of the car..
I think the labor on changing out all of the guide rails (at a good mechanic) will come to around $7k, plus all of the "While you're in there" shit that you'll want done, somewhere around $9k.

I didn't have any power problems or slapping noise at startup, or any indication that there was anything wrong with the engine except that I hadn't actually opened it up to take a look. That could have been a bad engine blowing mistake.
This is where it pays to know what you're getting into when you buy a used car. Know EVERYTHING about it. Become an expert in that model before you buy it so that you know what to ask, what to look for, how much it will cost to fix (time+money), whether or not you can drive it or have to trailer it home.
The above sequence shows what the guide rails look like when they've failed. It just broke in half, one half of which is bumping around down there on the ignition sprocket, just waiting to get chewed under..
A new set of guide rails (upper only) costs about $18, and they're white as the driven snow, unlike this one, which is brownish-oil colored, indicating it's chemically transformed over the years into a fragile, brittle time bomb.
To swap out the upper rails took about 4 hours, including the 'while your in there' shit, but without removing the front cover and getting at all the lower rails. The timing chain is showing no stretch, which is nice and unexpected, the timing gears aren't worn any more than they should be at this point.
One thing that's a concern with certain engines is one wearing part being of a dissimilar material or less wear resistant than the part wearing on it. For example the cams and sprockets are cast iron, but the chain is an alloy suitable for chains, the lifters of a hardened steel like 4150 or something even harder.
Of course all these parts are protected by a sheet of oil, but it's not a 100% friction barrier in any case.