LEGO Test machine #2 is dead after 1 test

Well, that didn't last long.

The machine did 1 full test and then died at about 1/3 of the second test.  The first test result is 32,066 iterations, supporting the result from the Test Machine #1 which ended with 37,112 iterations. So the average is at 34,589 for the moment.

So, why is the machine #2 dead? It didn't died as much as it was slaying DC motors and it all came down to a bad design. One of my goal was to build the machine with only (as much as possible) the Makeblock parts that TLBRC* sent me.  It turns out that I found my nemesis in the motors that where included in the two kits. DC motors.

For something like this build, every molecules in my body was screaming stepper-motors or, at least, servos.  But NoooooOOOoooo, it had to be DC motors, so I looked around for ideas to use them in the same way I would use stepper-motors.   The result was to build physical barriers to stop the motor in known positions.  I knew this would not be good on the motor's gears so I monitored the current going into the DC motors and “tried” to stop them before they were straining. At this point in the story, all the engineers are rolling on the floor laughing. Well, 3 dead motors later, I've learned my lesson: Engineers are cruel and, most importantly, DC motors and made to run freely like wild mustangs in prairies.

The following video was made after the machine killed its first motor.  At that point I thought I could fix it:

Here is my sad video when I decided to stop the machine for good:

What now?  Thanks to Galaxy Quest, my motto is “Never give up, never surrender” and so I have 2 plans:

1. Build a simple machine using stepper-motors.
2. Find a DC motor friendly design and try that too.

But, in order to keep my sanity, I will take a short break from this and work on other projects.


R.I.P. LEGO Test Machine #2.  At least you did 1 full test.




*Again a big thanks to The Little British Robot Company for the 2 Makeblock kits that were used for building this project.  I will reuse those parts on a project pretty soon

Edit:
This story was covered on:

• Ars Technica

Profiling DC motors

For the second design of the LEGO test machine #2, I stopped using rotary encoders an switched to ACS712 lowcurrent sensors.

I initially hooked them to a multimeter and an Arduino to tweak the 2 potentiometers of the sensors. Following some of the steps from this MobileAPES post, I've set the voltage (Vref potentiometer) to 2.5v without any problems but the gain (GAIN potentiometer) was initially tricky. Back when I was using the rotary encoders, I was running the motors as slow as possible (speed of 65-85) in order to not stress the parts. At those speed the current fluctuations are too large and I had to find out what is the optimal low speed of the motors.

I decided to log the data out to better see what was going on. The initial test was to log the current sensor values every 1ms while raising the motor speed from 64 to 210 by increment of 5 every 20ms. 

Yellow=speed / Blue=running freely / Red=stalled (prevented from turning)

Clearly the values I get from the motor is erratic when the speed is below ~110. So I profiled the motor again from 130 to 230.

Yellow=speed / Blue=running freely / Red=stalled (prevented from turning)

Now I know the how slow the motors can go and how to detect a stall using the current sensor. This also told me how long the motor takes to ramp up to speed when initially set at a speed of 130.

Here's the diagram of the machine made with Fritzing:

I don't have the Adafruit motor shield [fritzing] part so I used the Arduino motor shield rev3 instead. The missing details are the following: The servo is connected on SER0 (pin 9), the push-down motor is connected on M2 and the lift-arm motor is connected to M3.

The code and the Fritzing diagram for this new version are available on git.

You can see the new machine here

Once more into the [legotest] breach

Here's a video update about the progress of the LEGOtest machine #2 (#legotest)

Here is some extra info and progress starting with something I forgot to mention in the video: the shorter lever beam.

The lever was too long and, out of sheer luck, one of the shorter beams was the exact length needed. After that, I finally strengthened the base by removing the last LEGO legs and installing the Makeblock leg I showed in the video.

Once that was done it just made sense to attach the Arduino to the front leg as you can see in this picture of the whole machine:

The Makeblock parts are provided by The Little British Robot Company

As for the double action on the new motor; it both assembles the LEGO bricks and helps disassemble them.

The two movements are difficult to perform with a single motor since assembling needs a lot of strength while disassembling is all about subtlety. The part was first designed using LEGO Technics and it took 2-3 iterations before I found a way to achieve the two movements. To better understand the motion, check out the video.

Unboxing and Doodle cool

The parts for LEGO Test Machine #2 have made it safely across the Atlantic.

Makeblock parts from The Little British Robot Company

You can see me open the boxes and give some initial thoughts and ideas in this video:

When opening the shipping box, I've discovered a DoodleBot which the guys from TLBRC added as a gift. It's robot that can hold a pen and use it to either draw or write. I couldn’t resist so I built it right away and tested it by trying to make it write “Thanks TLBRC”. It 'kind of' looked good on the first try but clearly there is something wrong with one of the servos or sensor. I'm pretty sure that I didn't mount the rotation sensors properly so I'll have a look at it.

Seriously, look at this little thing. How can I stay made at it?

Still it's a really cool little robot base. The Arduino board that comes with the kit contains a ton of features and most are not used by the Doodle Bot. It even has a header to plug a Bluetooth device. Needless to say that I'll most likely reuse it for a future project.

If you ever get your hand on this kit, here are a couple of tips to make it work with the current Arduino IDE. By following the kit instructions you'll get to the right drivers and select the proper board. But to open and build the code in the IDE, you can do this:

1. Open the Doodle_Bot_Text_Demo.pde with the Arduino IDE.
2. Click OK when prompt about creating a ''Doodle_Bot_Test_Demo” folder.\
3. (In a file explorer) Manually copy all the other demo files into this new folder
4. (In Arduino IDE) Manually 'Add' all those files into the project
5. In the Alphabet and Doodle_Bot_Text_Demo files, rename all the letters functions by adding the prefix 'l'. (e.g. F() becomes lF())

That's it, you should be ready to build and upload the code to the DoodleBot.

If you have any questions let me know.

Inside the box

So close I can almost touch it. Parts for the LEGO Test Machine #2 are coming and I feel like a kid before Christmas. Here's some pictures taken by the guys at The LittleBritish Robot Company while boxing all the goodies.

WhoooOOOoooo motors

Looking at the content I got inspired to create a new design focus around the use of the Makeblock wheels.

The top part (1) would go back and forth to assemble the brick by compressing them together. Disassembling the bricks would be done by a claw system (2) able to twist off the bottom brick. Finally the bottom plateau (3) would go up and down using an off-centred wheel, helping in the process of assembling or releasing the parts. It's just a rough idea but feel free to leave your thoughts and suggestions.

Now, how can I make time go faster? I could go to sleep until the package arrive. Naaaaaah I'll probably wait in front of the door like Scott Pilgrim.

Added:
You can see all the 'boxing' pictures on my G+ album

When in doubt... TLBRC

Oh yeah, this project is rolling again, thanks to The Little British Robot Company (TLBRC).

Like many modern relationships, we met on Twitter. They followed me, I followed them. I complimented their product, they answered (knowledge of social media is a good sign). Finally, I proposed a partnership and they said yes. WooooOOOOooooo!!!!!

[pause to reflect on how cool the Internet era is and how lucky I am to be part of it]

Anyway, I will redesign the second set of LEGO testing machines using almost only Makeblock parts (distributed by TLBRC) and TLBRC parts.

To get better results than the first LEGO Test Machine, certain features and constraints will be added to this new project:

• Gather more than one control point (e.g. adding voltage, pressure, time, ...)
• Use a rotation motion to remove the LEGO bricks.
• Keep the test speed at 10 seconds per iteration to prevent heat from friction.
• Perform many tests. Hopefully, many tests at the same time.
• Test bricks from different eras.
• Once the LEGO bricks have fallen the first time, rotate them 180 degrees to continue testing until they fail again.
• Have a way to query the test status while it's running.

Also, instead of machining everything myself, I will be using the following:

• Makeblock
• TLBRC parts
• LEGO parts
• Arduino and other electronic components
• TLBRC custom milled parts (If needed)
  
  

Let's get this show on the road...

LEGO Wear Test #2 - Design

I have the basic design  for the second LEGO wear test machine.  Here is a prototype made with Castilene plastic:

The result will be a long rig that will hold all 10 pairs of LEGO bricks to be tested simultaneously.  The parts will be made of machined metal fitting the brick shape as precisely as possible.  One thing that I learned from the first test is that it takes a lot of strength to hold those bricks, which is why I'm overkilling this one.

The bottom part is composed of two halves enabling the rig to either hold tight or release the piece completely.

Following comments that many viewers/readers have left, here are the new requirements:

• Use a rotation motion to remove the LEGO bricks.
• Keep the test speed at 10 seconds per iteration to prevent heat from friction.
• Test 10 sets at the same time.
• Test bricks from different eras.
• Once the LEGO bricks have fallen the first time, rotate them 180 degrees to continue testing until it fails again.
• Run the machine in the shed to keep the noise down.
• Have a way to query the test status while it's running.

Not following comments that some viewers/readers have left:

• This will not solve global warming
• I will keep playing with LEGO bricks
• My accent should not improve before the next test

One thing I'm leaving out is to have many control points and this test will [again] only count the iterations.  I initially wanted to monitor forces and current but it makes no sense with this simultaneous test. Next time...

For the test I've done my best to select bricks from 3 eras.  The new ones (3 pairs) were the easiest to find while the 2 pairs from the mid 70s posed a challenge.  It started with one pair I took from a set  bought at the flea market.  It was the set #370 from the mid 70s.  Looking at those two white bricks I noticed that they had a unique casting mark on one side.  After a lot of digging, I finally found two bricks with similar marks.  In a very non-scientific moment I declared that this unique feature probably means that they're from the same era.  Shoot me.

Now that the 10 pairs of bricks have been selected they are set aside until the test and other bricks are used during the rig's development.  Also, in a wave of madness, I might even run this new test twice with the side effect that I'm slowly destroying all the 2x3 bricks in the house.

LEGO all worn out

As suggested by Christopher Gaul, I took some macro(ish) pictures of the wear patterns on both the studs of the bottom LEGO brick and the inside walls and tubes of the top LEGO brick.

Clear marks are left on the sides of the studs (bottom LEGO)

The small ridges on the inside walls (top LEGO) are visibly worn down.  There is also damage to the tubes and even the walls themselves.

Here are the 2 pictures in full resolution (sorry for the low quality) without the notes:

CSI case #66531

LEGO's magic number is 37,112

Have you ever asked yourself this question: "How many times can I assemble LEGO bricks before they wear out?"

Well... probably never but I did (on LEGO SE).  The story would end there if didn't had the knack to build a machine to test it.  Here it is:

The first version was using a metal tube attaching the top servo to the LEGO brick.  It was too flimsy and died during initial testing.  So I took an old CPU heat sink...

And using my trusty angle cutter, I made this new arm:

A successful test proved that the LEGO was no match for this chunk of aluminium.  I then made an infra-red proximity sensor to detect if the presence of the LEGO and rigged the whole thing to an Arduino.  This was shot before the test started:

and this was shot about two thirds into the testing that lasted a whopping 10 days!!!

So here are the 2 dead LEGO bricks after 10 days and 37,112 assembling and disassembling.

Don't try this at home. It's long, noisy and cruel to LEGO bricks.

In the end, both pieces failed at 50%. The studs of the bottom brick and the inside walls of the top brick are visibly worn. Both bricks can still hold on (not strongly) to normal bricks but, when put together, they can't hold. In a way, you could say that they are still in working conditions as long as they don't meet again.

Here are some more answers:

This story was covered in the follow media:

• Arduino Blog
• Ars Technica
• Hack A Day
• Reddit (TIL)
• Discovery Channel – Daily Planet
• Make
• Geek Dad
• Tested
• WIRED (Japan)
• Smithsonian
• Gizmodo
• Kotaku
• NBC News
• Bit Rebels
• And probably others so tell me if you find some new ones