Learn to learn.

A couple of generations ago the expectation was that you can get an education, develop a given skill-set and then stay in the same job for many years at a time.  For some jobs that is still true, but for an increasing number of jobs, this has not been true for a very long time.

For instance if you want to become a programmer.

If you want to become a decent programmer you will have to be prepared to learn new things every year for the rest of your career.   The most important skill you will have to develop is the skill of acquiring new knowledge.  Preferably on your own.  You need to learn new programming languages, new methodologies, new tools, new frameworks.

If I look back at the software engineering classes I had at the university 20 years ago, almost all of it is either irrelevant, outdated or seen as incredibly naive today.

If you teach software engineering, you have to find some way to prepare your pupils for a life of learning new things.  And for those who are not able to learn new things easily, you should probably advise them that software engineering is not for them.

If you are a software engineer:  try to anticipate what skills will be needed a couple of years from now and start thinking about how you should acquire these skills.  If you expect to just coast on the skillset you have right now, some kid fresh out of school is going to take your job.


Valve prototype 2.

I've shrunk the design quite a bit and made some changes to the rotor.  Now the first 180 degrees of the rotor provides most of the travel and the last 180 degrees represent about 1mm of travel -- giving us resolution where we need it and speed where no resolution is needed.


Non-invasive valve, generation...something, prototype 1.

I'll admit it:  I haven't been in a hurry to work on this prototype since I have a fairly low-tech solution that works surprisingly well based on an ordinary kitchen scale and a screw-on clamp for fastening things to tables.   But I do enjoy thinking about how one goes about solving this type of problem and there are lots of learning opportunities.

Up until this point we have been experimenting with various forms of valves to pinch off medical tubing and stopping flow without compromising the tubing.  The main challenge has been to get the sought after precision, torque, speed and controllability.   The last effort looked something like this:

It translates rotary action into linear movement by way of a nut and bolt.  It provides the needed precision and torque, but at the expense of the servo.  In order to get the needed number of revolutions, the servo was modified so it is now just an electric motor.  Meaning the thing has no idea as to its position.  Well, sort of.  The white switches would take care of that if you write some software to auto-align the thing.  

The valve is simple and it is so powerful it'll take off your pinkey if you are not careful.  Brilliant prototype, I learned a lot from this.

The snail.

For lack of a better name I called the latest prototype "the snail".  Because of the rotor inside it that looks vaguely like a snail's house.

So for the this prototype I wanted the following.
  • We want to use a servo motor since they are small and easy to control from a microcontroller.
  • Most servo motors have limited range.  From 170 degrees to 360 degrees.  Design for max 360 degree action, but be able to adapt design for 180 degrees.
  • Rotary movement to linear movement.
  • Optimize torque -- ie. make use of the entire range of the motor.
  • Make the form factor smaller and more closed.
  • Make allowances for shapes that are easy to 3D print
  • It is okay if some of the parts will have to be glued.
So here is what I came up with.

And here is what the printed parts look like:

Misprint, but still useful.

As you can see the "base" was a misprint;  the bottom and the thing that the rotor sits on are gone.  This was due to sloppiness on my part when exporting the model to STL and importing it into MakerWare for slicing.  The "base" component was in actuality made up of two bodies and I only exported one of them.

Apart from that the print wasn't a complete waste of time.  I was able to attach it to the tubes and to measure a few things.  For instance I now know that I need a maximum of 4.5 millimeters of travel to pinch off the 7.6mm tubing that Fresenius uses.  I haven't tried the Baxter tubing yet, but I suspect it will pinch off completely at about 5.0 mm travel (they use cheaper, thinner,  less flexible tubing).

The difference between 1ml/sec and no flow is about 0.5mm travel of the peg.
Some of the parts mounted together.  The rotor rotates and pushes on a peg.  The peg will clamp off medical tubing that is pushed into the groove.  One of the legs of the top broke off when I was a bit eager to pull things apart.

Non-linear radius.

I had a hunch while designing the rotor that it shouldn't have a linearly increasing radius.  I was right about that.  The above rotor has a linear radius since this makes for simpler testing, but for the finished product we want a non-linear radius change.  The reason is that once you get to low flow rates you want to have more precision.   If you look at the above image you see the two lines I have drawn on the rotor.  The rightmost line represents full flow cutoff.  The left line represents 1 ml/sec flow.   For one, there is still about 25% rotor left when we reach full cutoff, so that part of the rotor is wasted.  Second, we would want the angle between the two lines to be at least 25-30 degrees, and to reach ~5ml/sec at 120-150 degrees or so.

The two horizontal lines drawn on the base represent full open (the lower line) and full closed.

Things we have learned.

  • The operating principle is sound.
  • Double-check geometry before running it through the slicer.
  • The rotor needs a redesign.  
    • Aim for 4.5-5.0 mm change in radius over 360 degrees, 
    • the last 160-180 degrees should result in 0.5 mm change in radius.
  • The back end of the peg might be more durable if not filleted
  • We can reduce overall size if we increase infill.  Try to increase from 20% to 50% next time.
  • 245 degrees extruder temp and 120 degrees base temp works fine.
  • Need to design some sort of axle for manual manipulation as well as testing with servo.
  • Tube holder was spot on.

People want government to snoop on them.

PC Magazine has an article entitled "Why We Can No Longer Trust Microsoft" written by John C Dvorak.  The article says that this is a financial disaster waiting to happen.  That customers will flee Microsoft once they realize that Microsoft is betraying them to spying agencies.

No they won't.  Because most people do not care about their privacy.  If you expose the average consumer/voter/citizen to someone who has a way with words and is not above embellishing the truth a bit, they will gladly accept the government snooping on them.  Gladly.  Every time.

Most people lack the intellectual fortitude to imagine why having no privacy may be a bad idea.  Most people like it in the herd.  Most people like being told what to do and will complain bitterly if you expect them to think for themselves.  Most people are morons.

I'll give you an example.  In Norway we let parliament introduce data retention laws that will make it possible to map your every move and everyone you communicate with.  All the metadata from your phone records.  If you want to know what that looks like for just one person have a look at this example from germany (click the link to see the demo):

Green party politician Malte Spitz sued to have German telecoms giant Deutsche Telekom hand over six months of his phone data that he then made available to ZEIT ONLINE. We combined this geolocation data with information relating to his life as a politician, such as Twitter feeds, blog entries and websites, all of which is all freely available on the internet.
Save for a few scattered protests, the Norwegian people signed up for this.   We want the government and all its agencies to know where we have been and who we talk to.  We do not want privacy.  We elect people who take away our privacy and we do not think twice about it.  If we really cared about this we wouldn't elect these people -- so obviously we want the government to snoop on us.  It is our own damn fault.


Get used to this.

Get used to this, because you are going to see this happen a lot:
  1. New entrant to market finds way of doing something an order of magnitude more efficiently.
  2. Incumbents laugh at new entrant;  point to their own years presence in field and talk about the immense value of experience.
  3. New entrant starts to eat market shares.
  4. Incumbents look on paralyzed not quite believing what is happening.
  5. New entrant takes #1 spot in market.
  6. Incumbent desperately tries to mimic the superficial aspects of what new entrant does without actually addressing the underlying substance.  
  7. Incumbent fails at mimicking new entrant, tries to sue new entrant.  Fails.
  8. Incumbent is irrelevant.  Shrinks, gets sold or goes bankrupt.