Virk I – Open Hardware Scara robot

This is a Virk I, a project I’ve been working for a long time. It’s A small, slim 3-axis scara wooden desktop robotic arm. It was meant to be a learning tool or hi-tech toy. The design will be open sourced once the first phase be accomplished and all the units on the first lot be sold. The aim is to provide a small, portable, quality and beautiful robotic arm alternative, either for the academic or applied hobbyist.

Current version it’s made of Australian Blackwood (Acacia melanoxylon), acetal, pvc and standard components. Most parts where machined on a cnc mill/router/lathe.

On the shoulder and elbow it has custom design, modular hollow axis rotary joint containing three bearings inside. Two thrust bearings give the joint very low play, allowing soft, low play movements. Being hollow means that you can route the wires inside the robot most of the time, so you don’t end with a mess of wires outside, and get a more cleaner look. On the z axis it has standard 8mm linear bearing rails.

Rotary joints are driven by a nema 17 stepper coupled to custom made 3mm pulleys. Z axis is driven by a nema 11 stepper linked to a 8mm pitch leadscrew. There’s currently no gripper on the end of the arm; gripper design and built will be tackled once the first lot of eight robots be finished.

Currently, it’s driven by RAMPS 1.4 electronics. As far as I know there’s no open source 3D printer firmware able to drive a scara robot, and it’s outside the scope of this project to make such a development, so final owners will have a challenge making this thing work. Of course you can use something like grbl to test the robot movements, but to be useful you need a firmware containing scara kinematic support. Maybe this marlin based parallel scara source could be a starting point.

I’m currently starting to test the first working unit and solving some final design issues, so the very first units to be sold will be available, very likely, at then end of this month (September).

You can find more info here.

Video here.

Update: It has been pointed out here and at hackaday that there are several platforms already supporting scara  kinematics:

  • Smoothieware: Morgan-scara: not single arm scara, but close.
  • Machinekit: According to the documentation, it has scara (single arm scara) support.
  • Reprap-helios:  That seems to be a similar arm, so I guess the same firmware could drive Virk I (I haven’t checked current status of this project).
  • Marlin for single arm scara: This seems to be a suitable option. I will test this at some point.

I would love to hear about someone experiences using those platforms to drive a scara robot.

The Perfect Paper

I’ve been building a leadscrew cover for the Sherline’s I’m going to sell. I think this is a must on any cnc machine.

I like the classical accordion-like cover, originally designed by So far, the main obstacle has been to find an appropriate paper. But I guess here is: Fabriano Tiziano Paper. This is a acid-free, 40% cotton, 160 gsm drawing paper. Very high quality and strong paper.


Folding has a trick; you should use a ballpoint pen to trace each line, pushing hard (kind of emboss). Even if you do this, folding is difficult, though not impossible.

And the best thing… you have a lot of color options!. I like the black-red combination.

An Incredible Project

DARwIn-OP is a humanoid open source robot developed by some prestigious universities. It’s not just a toy, but an advanced research platform; commercial version cost USD $12,000.00.


Some time ago I found a guy in my city that is building this thing on their own and I was amazed; that involves some serious metalworking, electronic and programming skills. He had some trouble because the largest part has a size larger than the cutting area of his machine (a 5400 Sherline CNC mill), so I provide my machine to do the cut; that was a lot of fun. So here is his site:


I will start a small (indeed very small) business soon; I mean, begin to create and sell some things. As I’ve said before, I used to work as software developer, and tough it’s not as bad, machines, tools, electronics and creating things is what I love and from what I wish to live. A lot of people around me are a bit sceptical (and sure they have good reasons), but I really don’t care; failing doesn’t really worry me. What worries me is to see time going without doing all can be done to achieve what I want (I know it sounds a bit cliche).

So, I’ve been doing several things in this days: setting up a FreeBSD server, switching to Ubuntu and open source tools, and starting to prototype my first product.

desktopMy Server

A version control system can be very useful to store and keep the track of changes of your work (even if you work alone). I already had a dusty FreeBSD server (I really love FreeBSD) with subversion and other things, but I wish something fresh so I install the last version (10) and setup subversion, websvn, dokuwiki and some other things. These are really nice tools.

Ubuntu and the open source

I had never been a technology enthusiast. My notebook is old, I use my old phone just for calls and up to some week ago I use windows xp on my old PC. Linux never attract me as desktop platform; I had all I need on windows. But when you wish to do some serious work, linux it’s unbeatable. Forget about viruses. Download all you need, free, from a central repository. I can easily do a new install, with all the tools I need, then checkout all my work files from my svn server and a get a working environment in a breeze.

Just one small complaint: Ubuntu’s default desktop environment it’s not resource friendly. So if you have old equipment, it will not run very smooth. Nevetheless the solution it’s very easy: install lxde, a lightweight desktop manager. I love lxde so much now. Also, maybe I give a try to lubuntu in the future.

From Eagle to KiCad

I’m a Eagle fan. But free version is for non-commercial work and has a limited footprint.  So I had two alternatives:  pay a Eagle licence (about $575) or try an open source alternative. The first time I put my hands on KiCad, some time ago, I find it a little polished and not very easy to use tool. Sure, I was a Eagle user. But when you begin to use KiCad and get used to its own way of doing things, you begin to love it. Sure, it’s not bundled with so many components as eagle, but it’s really easy to create a new device/footprint (also there’s a lot of extra libraries out there). And, although the board doesn’t update automatically when you add a new device, updating associations and netlist it’s just plain simple. My only complain: all footprints are in one big list, there’s no filter. But I can live with this. Good bless KiCad developers.

My first product

My first product will be a simple accessory for the Sherline lathe; some small parts and a bit of electronics, nothing special. My main objective it’s to get selling experience. Once I get confidence I will start sell more elaborated things (like a micro table saw kit or something else).

Name and domain

Finding a good name for your business it’s not easy, and finding an available .com domain for a good name it’s almost impossible. It wasn’t easy, but finally I did make a choice that has an available domain; now I should choose a domain registerer and webhost.  It’s sad buying and reselling domains be a common business.


I’m still not sure if I will sell on eBay or in the .com site. Actually there are a lot of alternatives to eBay, but I think eBay is a good starting point. Also, if you wish to create your own online store, there’s a plenty of free software tools. Some things I hate to deal with are legal and accounting issues, so I will tackle them later.

I don’t like to call myself an “entrepreneur” or something like that; I just want to give the effort some things deserve.



I have a high respect and admiration for woodworking, and tough I have the tools (chisels, saws, planners, etc), rarely find myself working with wood. More commonly I build particle board furniture, and tough it’s not easy, I wouln’t call it woodworking.

This is a toys shelves project for my son. I could made the sides of particle board, but I wanted something special and liked the idea of a painted wood frame.

Tough this is a simple half-lap joint structure, It required a lot of work and tools (as always, much more than I expected… this isn’t a weekend project). One thing that makes it harder it’s the fact wood bars aren’t exactly the same width. Also I almost never can cut perfectly square; always had to use the chisel and/or file to correct the cut.

I did all with manual tools except rounding the edges; for that I did use a router. Also I did the all holes with my small microlux drill press.


Not bad. It’s nice to see how practical considerations and a little of devotion cand lead to a beautiful design, without pretending doing something beautiful in the first place (I sometimes envy designers capabilities to create cool designs). May be it’s has to do with the way nature works.

Something about Delrin

Here is a design of a telescopic cover for the Sherline lathe  (please note this isn’t the full drawing).

Why I need that? 1. Because I hate having to clean and lubricate the ways after every use. 2. Because all these debris can increase wearing on sliding parts. 3. Because this is very desired feature on a CNC machine.

I made some delrin parts but drop the design. it I was more committed in doing something cool than functional. Probably it would work, but a simpler design was possible. The simpler, the better, so all these nice parts will go to the bag of sample parts or be reused later.

These parts looks simple, but there was a lot of hours involved in design and building. One good thing I learn about working with delrin was that, due to internal stresses of the material, you should do a rough cut of the shape of the part, and later finish all surfaces.

For example, to cut the straight bars in the picture, starting from a plate, a possible set of steps are: finish the edge of the plate, cut the bar and then finish the other three sides. The first bar below,  done this way, has a evident warp.

For the other two, I cut the bars slightly oversize and then did several set of passes over the sides, slowly getting close to the final size. This is a matter with thin parts; for compact and simple parts this is not an issue.

I almost finish the new design for the cover, so hope to begin the building phase soon.

Homemade Anodizing

I ‘m been busy doing some aluminum parts, and after you work hard in a part you want it look beautiful and last long. So here is when anodizing comes. Most of the experience here is based on Ron Newman’s Anodizing Aluminum, the best anodizing guide on the net.

A Brief Overview of Anodizing

I don’t now much about chemical reactions, so this overview will be very basic and not fully precise.

Anodizing is a process to colorize and protect aluminum. Through an electrolytic process a fine coating layer of aluminum oxide is grown. This layer has open pores on it, ones that can be filled with color dye and sealed. Aluminum oxide is a very hard material, so though only a few microns depth, this layer protect the part from small scratches and gives it a beautiful and professional finish.

There are at least two anodizing types, depending on the coating thick: Type II (1.8 μm to 25 μm thick), and Type III (thicker than 25 μm) or “hard anodizing”. Hard anodizing obviously is more durable, but also more difficult to do at home, so the anodizing done here will be Type II.

The main steps involving in anodizing aluminum are:

  • Clean: remove any grease or dust.
  • Desmut: remove smut generated from previous step.
  • Coating: create oxide layer
  • Dye colorize: fill oxide pores with special dye.
  • Sealing: seal pores so the dye stay in the surface.

Every step requires a specific chemical, and time and/or temperature control.

What do you Need

The things do you need to anodize are:

  • Some plastic pots.
  • A metal pot.
  • Distilled water (at least 5 lt).
  • A battery charger or power supply of several amps. Note that some chargers have a “auto-shutdown” and can’t be used.
  • Current meter (10 Amp at least).
  • Caustic soda (lye) solution, the one used to clean pipes.
  • Nitric acid solution (10%). I think that this is not really required for 606X aluminum types.
  • Sulfuric acid solution (15%).
  • Aluminum wire. Mine is 1.5 mm diameter.
  • Graduated glass beaker.
  • Anodizing sealer. I got “ALSE22 ” from Caswell.
  • Anodizing color dyes. I got red, block and brown from Caswell.
  • A small balance or method to weight sealer powder.
  • Anode. Aluminum foil will suffice.
  • Some support to hang the parts over the pot. I build a nice stand for this.
  • Electric cable.
  • Rubber globes, security glasses, old waring, etc.

Below is some of this stuff (the funnel was never used).


Here is my stand and anode setup. The holes and screws allow easy mount of the aluminum wire. As can be noted, this pot fits only small pieces (the only ones I can machine).


Mounting a mini-anodizing line

Please note that the chemicals used here are potentially dangerous and that some nasty gases are produced.

First time anodizing will be hard, but once you have a mounted anodizing line and doing some runs, anodizing will be rutine and take only a couple hours. Here is how I build mine:

  • Water (small pot). This will be used to clean when passing the pieces between pots.
  • Nitric acid solution (small pot): 500 ml watter + 100 ml nitric acid (68%).
  • Sulfuric solution (large pot): 1000 ml water + 150 ml sulfuric acid (98%) (NEVER add water to acid, ALWAYS add acid to water).
  • Dye solution (large pot): 1000 ml water + 15.6 ml dye.
  • Sealer solution (metal pot): 600 ml water + 5 grams sealer.

I store the pots for later reuse, but discard sealer (I’m not sure if can be reused; anyway 1 lb package from Caswell will last).

Area and Time Calculation

In the electrolysis process the oxide coating layer grows up to a maximum thickness; after that, the coating remains the same thickness but the part begins to shrink. Hence electrolysis time is a important parameter. It depends on current and total cathode area. Less time will result in a thinner oxide layer; too much time will result in a smaller part. I use the “720 amp-min per square foot” rule to deduce this simple formula:

Full Thickness Time = (A / 929) *720/ I

Where A is the full cathode area in cm3 and I is the nominal current and the result units are minutes.

There are some area calculator for simple shapes on the net; they can help you to do a rough estimate.

For my first try, the total area was roughly 100 cm2, and with a nominal current of 2.4 Amp this gives 31 minutes.

First Attempt

Of course I will not try to anodize my beloved parts in this attempt. Instead, I machined some scraps of 6061 and 2024. The last is know to not be easily anodizable.

So the procedure is:

  • Calculate total area for the cathode.
  • Clean the parts to remove dust. Rinse and/or use acetone.
  • Cut aluminum wire and make hangs for every part. A good electrical contact is a must for a good anodizing. Also please note that the contact points will not be anodize, hence these should be in a less visible area.
  • Put 2 min in lye solution. After this step parts should not be touched and should not stay out the water.
  • Put 10 min in nitric acid.
  • Put in the electrolysis bath and measure how much current is being draw. Estimate time based on this current.
  • After half the required time has been elapsed, measure current again and recalculate time with this current. The current will go up in the process, so this will be a rough average or nominal current. Temperature should be in the 20-23 ºC range. The solution will heat up after a while, but I that shouldn’t be a problem when you don’t do continued runs.
  • Heat dye solution to 60º and put parts for 15 to 20 min.
  • Boil the sealer solution and put the pieces for 15 to 20 min.

Prior cleaning

I clean the parts with acetone before the lye solution. This is the last time you can touch the parts.


After lye solution

After the lye 6061 parts looks the same. However, 2024 get covered with a blackish smut. As far as I know this is due to the copper content in this alloy type.


The lye solution pot

This is the pot size I use for lye, nitric acid and water.


After the nitric acid

Again, 6061 parts look the same, but in the 2024 ones the smut has vanished. I’m not sure if this step is required for 6061 parts; anyway nitric acid doesn’t eat aluminum, so this will not hurt.


In the electrolysis bath

At the start, the meter measures 2.2 Amp, increasing up to 2.6 after 30 min. So I use a nominal current of 2.4 Amp for the time calculation.


After the electrolisys process

After this process, 2024 get darked as the parts in Ron Newman’s info; but 6061 remains almost the same. This lead me to make a mistake. I once read that 2024 anodizes faster than 6061, so I thought most action was by 2024 and 6061 parts get almost nothing coating. Also, I measure the diameters of the 6061 round parts an were the same before the process. So I repeat the process for 6061 parts alone (recalculating time of course). After the elapsed time, the parts looks the same, so something was wrong I think… I measure again, and they were 0.05 mm less in diameter! Then I realize that the 6061 were already coated, and the the second run only eats the surface.


In the dye solution

For sure the most easy way to heat the dye solution will be use a cup heater, but I don’t have a small enough one to fit my pot. So put my pot in a large metal pot and surround with boiled water. After a while the dye reach 55º, and I put the parts inside. This is bit less that the specified, but enough I think.


After the dye

When the parts left the dye solution, It was clear that something was wrong with the 2024, and that the 6061 don’t get a uniform color. Anyway I decided continue and finish the process.


In the sealing pot

And the final step: boil the seal solution and put the parts 15 to 20 minutes.


The final result

Well , that’s my first anodizing, far from perfect.


Second Attempt

This is the second try. Here is my improvements and corrections:

  • Surface spots were caused, I think, because of time out of the water (or solution). This time I will take the part straight from a pot to another.
  • The first time I don’t use pure caustic soda, but a special cleaner… bad idea. May be this cause the small pits.
  • I clean the parts in normal water, this time I will always use distilled water.
  • 2024 parts were anodized only at the bottom; a bad contact was probably the fail.
  • This time I will heat the die to the specified temperature: 60º.

Here an extra step is required: remove the previous anodizing in a lye solution. Please note that the lye solution will get dirty, so don’t use the same you use to clean parts.

After putting the parts in the electrolytic solution, I notice that the current was a bit less that the first time, so adjust time accordingly.

So here is the result. Much better.


After sealing a slight white smut was covering the parts; I successfully remove it from 6061 parts, but remains in the 2024 ones. I think that this was partly due to the fact that after removing previous anodizing 2024 surfaces got a bit porous.


After some anodizing rounds I’ve found that:

  • Bad contact is the most common cause of failure. To minimize this, contact points must have some spring capacity. Play with some wire patterns to get something that work. A god wire contact must be able to support the part as this moves.
  • A large part loosen at the half of anodizing will be a headache as this will change the required anodizing time.
  • No anodizing parts get a dark smut, so after a while they’re easy to identify.
  • If dimensional fit is critical for a part, be sure to don’t have to repeat your anodizing.
  • Some numbers: 167 cm^2 -> 2.7 Amp; 180 cm^2 -> 3.5 Amp. So I think a 6 Amp supply will suffice for the capacity of the pots I use. These are 15x15x8 cm, and I guess the will support no more than 300 cm^2 of parts.
  • I try to use a standard PC supply, but the one I used was cheap, so 12 volts were 10 at 2 amps or so. I think I will buy a standard 13.8 volt – 6 Amp supply.
  • To much parts can be trouble to handle, so splitting in groups will be a more secure way.
  • The metal pot should have the at least the same deep as the plastic pots; mine is lower an that is a problem. Parts should not touch the bottom of the pot and shouldn’t be near the surface (as the water level lowers).
  • Without care, this anodizing will not stay too long without small pits. Though more hard than aluminum, due to its small width, anodizing layer will peel after some rub or shock.

That’s all. If you want to learn how to anodize I encourage you to visit Ron Newman’s Anodizing Aluminum among other resources.

A Lathe Project

This is a project I start to gain skills on design and metalworking. Usually I only do simple pieces.

1. The idea: A precision mini-drill aimed to drill pcb boards (only the drill, not stand). It will hold typical carbide drills with a mini collet and rotate to about 10.000 rpm. Future work will include doing a stand for manual drilling and, why not, using in an always wished cnc machine.

2. The draft: I start about two weeks ago doing some drawing in acad. Below is the current state.

Some tolerances are fairly tight (in the order of several um); for example, the fittings of the bearings must be done with a tolerance of 10 um or less.

I’m sure this is not the best design, with lot questionable things. But I’m also sure it would be a lot better than the cheap drill I actually use for pcb drilling (and even the dremel).

3. The first piece: And this is the first finished (almost) piece I do some days ago. It’s done on 2024 aluminium and looks pretty cool.

I’m satisfied with quality and precision. For example:

  • Diameter vary from to from 23.990 (front) to 23.983 (back). Maybe there’s a very slight misalignment of the spindle, fixable doing a simple adjust.
  • In the rear I measure 15.995 and the bearing fits without pressure, so interference fit was not achieved, but it’s ok.
  • Int the front, the bearing doesn’t fits easily; the measure I get from the hole is 15.975, so it’s a bit more tight than the target interference but I think that heating the piece will do the job. Thermal Expansion Coefficient Al 2024-T3 = 22.68 um/m C -> the hole will expand 18. 144 um raising temperature by 50 ºC.

I didn’t measure all sizes; only the critics ones.

4. The next: The next piece I want to do is the one in the center; but currently I don’t have the required tools (boring tools for deep hole), and after struggle myself if make or not the tools, I decided go the easy way and do an ebuy. May be meanwhile I try the main axis.


The more I detail the draft, the more find myself saying “how the hell will I do this piece?”. But that’s the idea.
Up to this time I have read several books on metalworking, but now I realize how different is theory and practice.