metalworking

Just a minor note today. Brute force is rarely the proper solution to any problem. I've re-learned this simple truth not once, not twice, but three times today!

Ever since putting the trunk on the back of my bike I've been somewhat concerned about the brake light being

somewhat obscured. Not much so, but if I had a semitruck tailgating me, I could see how from his elevated position it would be hard to see the brake light under my trunk. Also, a while back I got a "backoff" brakelight modulator, and haven't had the time to install it.

So, here's the latest addition to my bike - it's a LED bar.

A Givi case is now installed. Finding the case was a bit of a problem -- every dealership in the

vicinity seemed to not have the right model in stock. Not only that, but they all seem to reduce their inventory for the winter, so they would need to special order the case for me. Ordering takes 1-3 weeks. Buying online costs an arm and a leg in shipping expenses. Even though online prices are somewhat lower, and there is no sales tax if I buy out of state, the shipping costs still kill the deal.

Some dealers just told me that they did not have the item in stock and wished me luck finding it this late in the season. A couple of dealers that I have visited in person tried to be creative and offered other, comparable products. One suggested I go for a similar sized monokey case. Uh, thanks... but no, thanks. One of the reasons I wanted a monolock was because it's lighter, considerably cheaper and easier for me to install. Monokey is the last thing I need. Another suggested I buy an E360 model instead. After all, the model differs by one digit only, so they are almost the same and, best of all, they had one in stock! Uh, no, thanks. Please, do your homework. I did mine, and know that E360 is designed to be a side case, not a top case. And even if you could mount it that way, it's still definitely not what I had in mind. After those two in-person visits I decided to use the phone instead.

After calling close to a a dozen of local dealers I finally found the case I wanted at a scooter place. Interbay Scooters. If you ignore the fact that there is precisely zero parking anywhere in a couple blocks radius, this is a very nice place. When I arrived, the case was already sitting on the counter waiting for me. Now it's installed and I have carried about 3 kg of cargo in it to the office and back. No adverse handling effects at low or high speed that I have noticed so far. The only question now is whether to

• leave the rack as is
• anodize it black
• paint it black
• give it a brushed aluminum look
• polish it (not likely to last long).

Yes, I think I am becoming vain.

I've got all the parts manufactured. Assembly is trivial -- just a few 8-32 allen button-head screws. The installation, however, was not quite as trivial as I had hoped. I hit a couple of snags along the way.


First was that the bike is apparently not 100% symmetrical. The mount points for the little hooks under the passenger seat are not at exactly the same location on both sides of the bike. When you machine to tight tolerances this becomes immediately apparent. The difference in height between the front and the rear mount point was greater on the "clutch" side of the bike is greater than on the "brakes" side. Also, the angles of the threaded holes in those points differ somewhat, which becomes quite painfully obvious when using a long 6mm bolt. This was solved by carefully offsetting and enlarging the corresponding holes in the rack until it fit properly.


The second snag was making the little brackets to mount the rack to the grab bar. That turned out to be a royal modeling pain, so in the end they were machined by hand. I just eyeballed the material that needed to be removed, put the piece in a vice, milled that material off, put the bracket back on the bike, eyeballed another section that needed to be removed, put the piece back in the vive... rinse and repeat about 20 times for each side. Here are some pictures of the final result:

After the initial design and measurements comes the fun part -- prototype manufacturing. I decided to go with .25" birch plywood for the material. I had plenty of it left over after a recent kitchen cabinet remodeling project. It's close enough in thickness to the 3/16" aluminum plate that I intend to use for the final product, is quite forgiving and easy to work with. Best of all, I can cut it with a 1/4" router bit to full depth in one pass. This makes prototype building quite rapid. The final product, on the other hand, will be cut by the same size endmill, but in multiple successive 0.05" deep passes with a final cleanup pass, making the machining much slower.

Here you can see a partial prototype made of plywood. The goal was to have the rack as low as possible so I could still use the grab bar for it's original purpose. Furthermore, I wanted to rack to be slanted just a few degrees forward when my full weight was on the seat. The easiest way turned out to take a rough plywood model, mount it to the bike, sit on it with my full weight, take the final measurements using a small level, and adjust the model accordingly. When my wife is on the bike behind me (2-up riding) the rack is perfectly horizontal.

After final adjustments are made, the real manufacturing begins. Here you can see one of the side brackets being milled.

I've been wanting to make a rear rack for my motorcycle for quite some time now.

There are no after-market options for mounting a hard-case luggage box on the back of the bike. Well, I guess that's not entirely true. There have been attempts, some more crude than others. I've seen designs made from aluminum angle and steel plate hacksawed and bolted together, welded rebar, rubbermaid containers clamped to the grab bar, and a pretty decent-looking rack made from square steel tubing (aka Jeb's rack).

So, I decided to venture out and make a brand new design. The rack is being made primarily out of scrap and various left-over pieces of 3/16" alumium plate. Boeing surplus seems to have an endless

supply of very nice scrap aluminum. Cheap. The most challenging part here is making a mockup, measuring the angles, distances, etc. The goal is to make a sturdy, yet reasonably light and stylish rack. Once that's done, a prototype or two will be milled from plywood, and eventually from metal.

Working on my '93 Accord is always a bit of a pain. Everything is so tight. Getting to just about any part under the hood (with the possible exception of the battery and the air filter) requireds removing just about everything else first. Getting to the timing belt is, however, is a truly out of this world experience. It is very impressive how this car is engineered, and how everything fits together, but man am I happy that I am not a professional mecahnic! Today's biggest challenge was to remove the crankshaft pulley. I prayed the bolt with some penetrating oil the night before anticipating that it's going to be hard. And it was hard. And no amount of penetrating oil helped. You see, there is not good way to make the thing stay in place while applying a few hundred ft-lbs of force to it! There is a special tool, and apparently there's a store around here (Tool Town) that can get it for me, but it's a 4-6 week wait time, and this single-purpose tool costs $100! It's acually amazing that they carry it, but... thanks... no. I'll pass. Alternatively the manual suggests using an old alternator belt in some creative fashion, but I cannot figure out how they suggest it to should used to hold the pulley in place. There's at least 200 ft-lbs of torque there, likely much more. How is the belt supposed to help? Maybe I am not creative enough. An impact wrench may have helped, but I don't have a compressor and my Christmas toy budget has been exhausted for quite some time now...

A couple of years ago I added a leadscrew to my taig lathe, along with a split nut that somewhat resembled Tony Jeffree's design. I made a number of modufications to the design, namely changed the axis of rotation to horizontal, perpendicular to the lathe bed, made it was smaller, somewhat flimsier, made from brass and used a spring to keep it locked in place. It was by no means perfect, but worked pretty well... until a couple of months ago when the splitnut... well... split. Literally. And beyond repair. So, I decided to try a somewhat different design. After receiving some valuable feedback on the taigtools newsgroup, I came up with a new design. Now it is built, and I have used it for about half a dozen hours of machining. So far I like what I see.

I need some studs and square nuts for my taig mill. I often find myself wanting to clamp the work down rather than put it in a vise, and generally this is a rather challenging task. I used to use 10-32 socket head cap screws with square nuts, but they have to be of the "just right" length - a little too short, and they do not engage in the threads very well. A little too long and you risk damaging the table. You need an assortment of 10-32 screws and lots of washers to have any hope of holding a workpiece like that. 10-32 cap screws are also a bit too small to use with the clamping kit that I picked up a while ago. You need some pretty thick and wide washers that fit 10-32 screws to make that work. In other words, there is technically nothing wrong with using 10-32 screws and square nuts to clamp a workpiece to the table, but it's tedious, takes long to set up and adjust, and is not quite as rigid as it could have been.

Ever since I started my mill for the very first time, I knew the spindle motor was gutless. 1/8 HP is gutless. Just to give you an idea of how mindbogglingly gutless it is, consider that the motor was too weak to spin the spindle with no load at the highest pulley ratio until the bearings warmed up. If I needed 10K RPM (for PCB work, for example), I had to run it at the second highest pulley ratio for 10 minutes before switching to 10K RPM. In fact, taig knows that those motors were waaay too weak, and now they ship their mills with, I believe, 1/4HP motor. This is not super-powerful by any stretch of imagination, but that's twice the power that I had when I bought my mill.

Having such a weak spindle is painful. I've stalled it more times than I want to admit.I've broken milling bits when the spindle stalled. I've had jaggedly chatterly finish when my spindle slowed down considerably resulting in chip load per tooth being way too high. I've destroyed the parts being milled. I've had to slow the RPM (and the feed speeds accordingly) down to a crawl, just to be on the "safe side". I've had to stand by with a brush and a can of WD-40 to keep the chips away and the bit lubricated to prevent stalling. I've tried to (unsuccessfully) drive my mill from a large spare AC motor using a self-made spline shaft. The list goes on...

No more of that! Ladies, and gentlemen, let me introduce a brand new surplus DC motor! 2.5 HP intermittent duty at 130 volts, 1.5 HP continuous duty at 95 volts. The controller that I bought along with the motor does not quite output as much current as the motor might want, so it is not running at the rated 1.5 HP, but it sure beats 1/8 HP! Also, no more pulley changes - now I have a true variable speed spindle. And I've added a Hall Effect sensor to measure RPM.

The last article covered all the chemical aspects of my PCB manufacturing. This one will briefly cover the computer design and cnc drilling.

First of all, here are a few links:

• board schematic
• board layout
• drill rack data
• excellon drill file (very similar to g-code)
  
• excellon drill tool description
• supplemental library
  

I could not find a suitable object for the large current sensing resistor, so I had to explore how to create one in Eagle. That's the only component in the library file. I think I ended up borrowing some people's ideas, and not doing a very good job overall, but in the end my model does the job reasonably well - it does look like a resistor in the schematic, and it is the correct size component with the right size leads in the board layout. All of these files were generated from eagle. If you want to re-generate the drill files, or generate gerber files, follow this link. I am a novice in the area of PCB design and manuacturing, and that page helped me immensely.

It was not long before I realized that just making a schematic and getting an excellon file is not quite enough. If you have an excellon machine, then this might be all you need, but I did not have an excellon machine, and needed to convert the drill file to g-code. The output from eagle looks a lot like g-code, but differences are significant enough that it is not feasible to adjust the file by hand. One needs a utility to do that. Unfortunately, at that time my internet connection was down for several weeks in a row (significant packet loss + abysmal verizon customer service = no internet), so I could not locate a suitable tool. Lack of internet also meant that I could not even find out any information on how to decipher the format. And to add insult to injury, I am not really all that familiar with g-code programming. So, in essense, I needed to take a file in an unknown format, translate it into another unknown format, and feed that to my mill. And do all of the above without being able to use any sources of information that are normally at my fingertips on the internet! All that was before I knew anything about canned cycles that are great for this kind of task. I remembered only the G00 and G01 commands, and I managed to misinterpret even them! Well, after some trial and error and experimentation I finally came up with this PHP script. You are welcome to look at it, laugh and enjoy total lack of understanding of g-code on my part. But... it works! And that's all that mattered! It'd be trivial to make it accept parameters from a web page instead of the command line, so if you want to give it a shot, please feel free.

In the end, I found that benefits of CNC drilling of any reasonably complex board really outweigh any costs of having to learn how to do it. The time invested into figuring that out is well worth it - the consistency of the resulting board is impressive, repeatability is perfect, there are no missed holes, no errors, no holes drilled to the wrong diameter, no broken drill bits, no sore back, eye strain and inhaling of fiberglass dust (because your nose absolutely has to be within a couple of inches away from the drill bit if you have any hope of seeing where you are drilling the next hole). That, and you can drill in subdued light or totall darkness, if you want to. Why is subdued light so important? Because you can spray boards with photoresist, then drill them, and then use the holes to align the artwork to both sides of the board before exposing to light, like I outlined in one of the earlier articles. That is now my preferred way of making PCB boards. It allows to acheive excellent alignment beteen the traces and the holes.

PCB making is not hard. But it is a tedious multi-step process, where each step can adversely affect the final result. It can be as easy or as or as complicated as you want. You can take the easy route, submit your layout to a PCB prototyping shop, like pcb123.com, and receive a great-looking functional board in the mail a few days later. Or you can take the hard route, and start with a blank piece of copper clad. Or you can pick something in-between. I think that the amount of work that you choose to do by yourself is roughly proportional to the amount of money you save, inversely proportional to the square of frustration you will experience, inversely proportional to the quality of the final product, and may or may not take less time, depending on your skill and which steps you decide to do yourself. I think that I'll use the services of a prototyping shop if I need several (10+) boards, but will do everything in my garage if I just need one board, and need it fast (and cheap). I will show how to make a two-layer board with 0.010" traces and isolation between traces from blank 1 oz copper clad.

I've wanted a metal-cutting bandsaw for quite some time now, but have not been able to afford one. Too many hobbies... too many toys on the Santa list... the usual story. For the past couple of years I have been stuck with a hacksaw and a reciprocating saw (aka sawzall). Hacksaw has been giving me such major workouts that I sometimes think twice before embarking on a new project. I've cut doezens of square inches of aluminum, and square inches of steel. Enough said! Reciprocating saw is hardly any better - it does not cut relatively thick material (thicker than one full stroke) very well because there is nowhere for the chips to go. When I use it my whole workbench vibrates like crazy, and so do my arms giving me numb and hurting forearms and hands after just a few minutes of sawing. Not to mention that making accurate cuts with that beast is a formidable challenge.

After adjusting the backlash, cleaning, oiling, greasing and and putting the mill back together, I decided to install my new latest-and-greatest revision of the stepper motor driver and give it a try. Now my backlash was reduced to 0.001-0.002 " in Y direction, and 0.003-0.004" in X. I am not sure why I failed to reduce the X backlash further. Maybe I can play with it a little more some other time, but for now it is "good enough". I did not play with Z due to lack of time. Besides, there's a pretty heavy motor with headstock to minimize any adverse effect of backlash in Z.

I was experimenting with X axis, running a 2-stack 208 oz*in pacific scientific motor on it. The controller was set to provide 4.60A @ 48 volts, which is the maximum recommended current, but only 2/3 of nominal voltage for the motor. So, I expect to get good torque at low RPM, but performance to be sub-par at higher RPM. I used mach2 software for this experiment.

If I set acceleration to the maximum 31.25(?) in/sec^2, I can jog at up to 110 IPM without loosing steps. At that point I start noticing lost steps at the right end of the X axis. Reducing acceleration to 5 in/sec^2 allows me to overcome that problem and reliably operate at 120 IPM. I cannot stall the X axis by hand no matter how hard I try at this point. Binding problem reappears in the same region at 135 IPM with the same acceleration. If I stay away from the problematic region, I can increase the speed a little further, just short of 150 IPM, at which point I can stall the table by hand.

Based on the results of that experiment, I set the table to move at 90 IPM in X and Y and 30 IPM in Z (it is too scary to jog Z faster -- it is too close to the table for comfort). It looks like that should be a safe setting, and should work well. Time will tell if I am right...

First of all, if you want an expert guide to backlash adjustment on a taig mill, go here. Nick knows what he is talking about. I am just a novice, so I roughly followed his steps, deviating from his recommendations here and there, trying things out, and finally arrived at the conclusion that his guide is excellent with only one fairly significant, in my opinion, omission (please, read on if you are interested). I think every taig owner should tear down their mill and put it back together at least once. Not only will it run better after you do so, but you will get a much better understanding of how it works, what it can do, and why things work they way they do. I think taig should sell disassembled mills, just like they sell their lathes, so users can put them together, if they want to. Assembling your own machine is an invaluable experience. After painstakingly adjusting my mill I managed to get backlash down to .003-.004 along X axis and .001-.002 along Y, while still being able to jog at 100 IPM! Maybe not as impressive as some other people have it, but that's quite an achievement for me.