Final prototype

17 views
Skip to first unread message

Brain Higgins

unread,
Aug 31, 2021, 12:40:16 AM8/31/21
to hbrob...@googlegroups.com, Brian Higgins
How do you design a final prototype? Maybe you don’t! I asked Cal-Poly for a sensor for my bike, doesn’t work.! The housing holding the rasp pi & battery (made w/ PLA filament ) is warping and falling apart in the sun.
The VA hired a global company to design the final prototype of the smart white cane for the blind. (allegedly spent a lot of money) It works but not as good as what I have already designed. They went astray, created their own design and didn’t improve on my design at all.
They all got the cheapest part they could. Their battery doesn’t last a week, mine I haven’t changed it in three years.
Do I need to be a dictator on these projects. Was Steve Jobs difficult to work for? Is there a class on designing a final prototype. I’m going back to Cal-Poly again and another university this class year for two separate but, similar sensors for the low vision person projects. Disabled Veterans can have engineering graduate students design Items that the veteran needs.


Richard Dyson (Dyson Vacuum) says it took fifteen years and 5,127 iterations to make his vacuum cleaner.

Brian


Sent from my iPhone 12 Pro Max

Mark Johnston

unread,
Aug 31, 2021, 9:37:04 PM8/31/21
to HomeBrew Robotics Club
Yet another 'classic'  oxymoron.  

But seriously, I think your question is 'When do you decide you are ready for a product release'.

This is a judgement call so in general accept that you will always decide at some later point that something should have been done different.  That is for revisions.

Decide when to ship when you know the advertised features work well enough to match the pricepoint and that being honest with yourself you feel it is 'ready'   
It must be capabile of delivering the MVP or minimum viable product features that you advertise and it must not be known to break in some timeframe that you yourself would expect it to last if you were buying it yourself.

Things like optimization of cost and product enhancements and solving issues that you did not know were issues at shipment time are for later revisions.

There is no clear answer to your question, it is one you must decide upon.  Use the rule of 'If this were for sale for this price would I buy one and what would I expect it to do and how long should it last'.  If you are honestly happy with saying yes it is ready then yes it is ready.   

Something funny you have said is  'Was Steve Jobs difficult to work for'   This has HIGHLY documented stories and sometimes perhaps urban myths that states he was not 'Mr Easy Going, Everyone makes mistakes, go ahead and try it again'  sort of manager.   (Im being as kind as possible here).  There have been numerous well known high tech leaders that were lets just say 'quite a bit easier to work for'.  Nuff said.

You mention Richard Dyson.   The bit about tons of iterations is a well known path to improving things.   Many innovators well back to Edison and more say this same thing.

Mark

David Hough

unread,
Aug 31, 2021, 10:53:51 PM8/31/21
to hbrob...@googlegroups.com
If your goal is mass production then there's a path from initial
prototype to production, and there isn't really a concept of a "final
prototype". At some point one you think you've got it close enough, you
ideally start talking to a manufacturer (if you're not already doing
that) and you do a small build with them for engineering validation
testing (EVT), which makes sure that all the functionality is there. If
you're happy with that you progress to Design Verification Testing
(DVT), which is a larger build where you're correcting errors but not
making any changes to the feature set. Production Verification Testing
is a large-enough build done on the real production line to check that
all your documentation is correct, that the thing can really be built
and tested and there aren't any bottlenecks or yield issues. Ideally you
sail through this seamlessly because at the end of DVT you should have
it all figured out and any required changes have been made. If you're
satisfied with PVT then you're into production. All three of these
stages repeat until you're satisfied, so you can have EVT1, EVT2, etc,
followed by DVT1, DVT2, and so on.

As for things like the batteries, you need to have a clear, written
specification that states that the operational life on a set of
batteries (or on a full charge if rechargeable) shall be greater than X.
that way they have to pick a battery that meets your spec. If they chose
a cheap one that met the spec but is not as good as the one you chose
then either you agree that they've cost-reduced it so it meets the spec
or decide that the spec was wrong and you want it to last for three
years, not a week. You need a durability and environmental spec too, the
thing will be dropped, suffer impacts, get wet, get left in a hot
vehicle, someone's going to try to bend one, etc. Engage the idiot part
of your brain too, switch off common sense and think of stupid things
that people might do to or with your product. Then shut up the idiot and
turn the common sense back on and marvel at what your product is going
to have to survive before figuring out which idiocies it is
cost-effective to prevent by design and which ones you explicitly
exclude in the warranty.

You do need someone who is on top of things. Students are good at many
things but what they usually don't have is the hard practical
experience. Without relevant guidance and instruction, they are going to
make the same mistakes that those of us who've been around for longer
have made and learned from. The real world is a lot different from
academia. You do need someone with experience of taking a design from
concept to production who can look at things and point out the
weaknesses and suggest improvements and who has the authority to insist
that something is corrected before progressing to the next stage. This
is either something you pay for, or given the end users in your case,
you may find someone who will volunteer some time. Eventually you learn
enough from them that you can be that person.

Oh, and weren't you talking about James Dyson, he's the vacuum guy.

Dave

Chris Albertson

unread,
Sep 1, 2021, 12:34:02 PM9/1/21
to hbrob...@googlegroups.com
You ask how to develope a prototype.  I think in this case, I'd recommend a "spiral" development approach.   THis is where you go in circles and each time around you build it a little more complex and better.  Each cycle is informed by the successes and failures of the previous cycles.

So you start with simple requirements and then build the cheapest things that works and then go and use it.  Only by using it do you know if you idea was good or not.  Also, by using it daily you will be able to think of ways to make it better.   Expect this to oake a few cycles.  The first two or three will not be what you end up with.  You knw you are done with the changes made with each cycle are small and nearly trivial because you start running out of ideas for improvement.

The key point here is that each prototype does not start from scratch.  You vstart prototype #3 with the design files of prototype #2 in hand.  You edit, you don't start from the beginning and make a few changes.      Your product is not really the physical device but the CAD files and source code.   These are kept in some kind of version control system that allows for versioning and problem reporting and tracking and gives you a way to back out changes or have experimental forks that can die or be folded back in.

You really have to track the files if the project is big enough that there are multiple developers

The hardest job is YOURS.  Project management is not easy and is a skill set in itself.

Each sysle has about four stages and you can't skip ahead.   A spyril plan repeats these steps until you have what you want.
  1. Modify your requirements based on what you learned using the last version.
  2. write up a test plan you will use after the next prototype is finished to verify that (a) it still works at least as well as before and (b) the new requirements are addressed
  3. build the thing, hardware and software.
  4. execute the test plan

About using student interns.    I've worked with some.  Sometimes I can see these people are smarter than I ever was or will be but at the same time they have no experience and require constant hands-on supervision.   They are very good at solving very well defined problems but fall apart on larger ill defined tasks and REALLY fall off the cliff when using a formalized design process that involves some of the steps outlined above.  Students generally are not as good at requirements analysis, formal testing  or conceptual design.

The spiral system really needs discipline.  You have to capture the design and log specific problem reports on that design, one issue per report and then base the next prototype off the existing design files with specific problem reports either addressed or deferred to the next cycle.  Students are NOT used to such a slow and disciplined approach. The tend to ALWAYS make the mistake of jumping to a final design and then "hacking" it until it works.

If you build each prototype from scratch then success depends on luck.  But the above system forces forward progress and will eventually get you closer and closer.  It is the managers job to explain and enforce the development process and find the people and resources needed for each of the steps.

--
You received this message because you are subscribed to the Google Groups "HomeBrew Robotics Club" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hbrobotics+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/hbrobotics/602b91d1-e7d9-4302-b57a-77da7919ee73n%40googlegroups.com.


--

Chris Albertson
Redondo Beach, California
Reply all
Reply to author
Forward
0 new messages