Connect with us

Todays blue sky

Discussion in 'Electronic Design' started by N. Thornton, Jan 21, 2004.

Scroll to continue with content
  1. N. Thornton

    N. Thornton Guest

    Todays blue sky

    Todays blue sky is a machine that will save its owners a fortune.
    Large organisations such as councils, hotel chains, and many others
    buy large quantities of electronic equipment: with a suitable machine
    much of this expenditure could be avoided.

    New goods are so cheap because they are made by machines, used goods
    are so costly to repair because they are repaired by hand, slowly and
    expensively. The central idea here is to change that, and develop a
    repair machine that can churn out repaired goods at speed and in
    quantity, with only one human operator.

    It would take the best of the following faulty goods:

    stereos and radios
    washing machines
    hoovers (vacuum cleaners to the Americans)

    and apply a fast mechanised and automated repair process to them.

    As well as junk and low value items, tips and dumps also have a
    regular supply of modern non-working goods in othewise good condition.
    For example Dyson vacs now turn up regularly at dumps, along with many
    items just out of guarantee. Many retail chains also offer to take
    customers' old goods away. A lot of quality material is buried along
    with the junk simply because repair is too costly.

    This machine gets round the problem of repair cost in 2 ways. Firstly
    it drastically reduces labour costs.

    Secondly it almost completely eliminates parts costs. It logs all
    items it checks, and decides what to do with the items it does not
    repair. Most it dumps in the scrap bin, but it will select some to
    keep in a parts pile, the size of which is determined by the operator.
    After the first week it will thus supply itself with parts for
    repairs. Only low value items that must be new are bought, things like
    hoover bags for example.

    So how would this machine work?

    This is what it would need to do with each item:

    1. pick the item up and position it for testing
    2. identify the item, by make, model no, etc
    3. when it cant do that it would test it to see what it does, using
    the item's external connectors, and narrowing the options down by
    optical word recognition.
    4. test it to see how it works and whats wrong
    5. remove covers when necessary
    6. either recall specific circuit information or use general purpose
    fault analysis approaches.
    7. apply the necessary tests to work out what the problem part is
    8. work out how to fix it
    9. either label the item saying which component to replace, and which
    dead machine to source it from, and place it in the repair pile, cover
    ready removed, with donor machine next to it.
    10. Or decide that its not worth repairing, and drop it in the scrap
    11. Or ditto but put it in the parts pile
    12. pick up and function and safety test the item after repair.
    13. Fully stock control the repaired items.

    It would pick the items up itself, and run its processes on item after
    item, churning out a stream of goods labelled with how to repair them
    in less than say 20 minutes, with only the easy repairs being chosen.

    The following goods would be rejected:
    not of sufficient value
    takes too long to repair
    spare parts needed that arent immediately accessible onsite free
    any goods the company already has too many of

    The machine will maintain a folder of choice related data which helps
    it intelligently assess which goods to repair. This folder will
    contain information for example on which items the company has no
    interest in, any it particularly needs, any specific models with known
    problems, make or model preferences, and any other issues that affect
    which goods it should select for repairs.

    Basically this (imaginary) machine takes a slow manual repair process
    and turns it into a conveyor belt operation with high output per
    person, more like a factory.

    A factory makes one item on a line, and thus uses simple machines
    specific to that task. This repair machine must cover a range of goods
    and a range of models, and test and diagnose faults, making it need to
    be much more complex. On the other hand it is dealing with goods that
    have already been almost completely manufactured: they will need only
    one item replacing. All the rest of the manufacture has been done
    already. Not only that but the materials it works with are much
    cheaper than a factory's, because they are simply scrap rather than a
    selection of new designed and made to order parts. Thus this machine
    has cost advantages over a factory as well as its own
    complexity/expense downside.

    I think this is the way forward. It is probably more a question of
    when these machines will come into use rather than whether.

    Can we do it now?

    I think the answer is yes, that we already have all the technology we
    need to develop such a machine. Lets look at each step of the process
    and see what we have.

    1. pick the item up and position it for testing
    - this is straightforward with robot arms and video shape recognition.

    2. identify the item, by make, model no, etc
    - cameras, optical character recognition and a database are all well
    within todays technology

    3. when it cant do that it would test it to see what it does, using
    the item's external connectors, narrowing the options down first by
    optical word recognition.
    - it is not difficult to use meters, sig gen, scope etc, and apply
    power to mains type connectors or leads. Text recognition would
    greatly speed things up by identifying keywords like volume, tuning,
    cassette, spin, bio-profile, 40C, and so on.

    4. test it to see how it works and whats wrong
    - once intended functions have been identified it can run a test
    routine to see what the item does. Robot arms can operate the
    controls. It could even apply a series of thumps to check for any poor

    5. remove covers when necessary
    - in most cases fairly straightforward using robot arms, with a
    machine with all the necessary tools. How to open this case also needs
    to be covered, which I expect could be done, and a general plan of
    attack routine might be worth adding for any items it cant work out.

    6. either recall specific circuit information or use general purpose
    fault analysis approaches.
    - such approaches are well known

    7. apply the necessary tests to work out what the problem part is
    - this often requires probing internal parts. Video would need to be
    able to recognise some of the key parts, the more it could recognise
    the more tests it could run. Object recognition is a known science.
    It wont need to get it right 100% of the time, but the more tests it
    can correctly apply the more items it can repair. Electrical probes
    can be applied using todays robot arm technology, with some
    electronics to detect when it contacts.

    8. work out how to fix it
    - these techniques are well known. If the part rcognition is good it
    should be possible to do this with a good success rate.

    9. either label the item saying which component to replace, and which
    dead machine to source it from, and place it in the repair pile, cover
    ready removed, with donor machine next to it.
    10. Or decide that its not worth repairing, and drop it in the scrap
    11. Or ditto but put it in the parts pile
    - these 3 are simple.

    12. function and safety test the item after repair.
    - all known technology.

    13. Fully stock control the repaired items.
    - EPOS and order systems are well established already

    All these methods can be done today.

    Is it worth it?

    The value of such equipment would be considerable. Rather than use it
    to repair scrapped goods and sell them second hand, it would be of
    greater value if it is used instead to supply a stream of goods to
    organisations that are currently buying new, because the goods
    produced wipe out the new goods purchase costs, rather than only
    achieving used goods prices less all the costs involved in selling.

    Once such kit is developed it could output a stream of items every
    day. The development cost is quite substantial, but its use is also
    very large, not just for the first machine itslf, but the whole genre
    of repair machines that will follow. Is it worth it? You tell me.

    Thats my thought for the morning, I've left plenty of debateables in
    there. What do you think?

    Regards, NT
  2. henryf

    henryf Guest

    Sounds great! Perhaps you could enlarge the concept to repair
    old cars? Let me know as soon as you get the prototype working.
  3. I think you should re-adjust your tinfoil hat, and get back on
    your meds.


    Dr. Anton Squeegee, Director, Dutch Surrealist Plumbing Institute
    (Known to some as Bruce Lane, KC7GR)
    kyrrin a/t bluefeathertech d-o=t c&o&m
    Motorola Radio Programming & Service Available -
    "Quando Omni Flunkus Moritati" (Red Green)
  4. Unfortunately, evolution (which eventually tries just about
    everything) has shown that making new beings is more cost effective
    than repairing old ones, after some point. This is a problem of
    arbitrary complexity, since designs keep changing (both in products
    and living things). I think the best solution is similar to what life
    does. Break old units back down to some common denominator bits
    (large molecules at best) and assemble new units from these bits. We
    don't yet design new products with this process much in mind. But
    eventually, we will, I think.
  5. Ian Stirling

    Ian Stirling Guest

    Not very relevant.
    Evolution does not care about the longevity of the individual, after
  6. John Larkin

    John Larkin Guest

    Of course it does. If we were just to breed and die, who would pay our
    brats' college tuition? Not to mention wedding expenses.

  7. You're nuts. But I've got some very nice plutonium nitride Hi-Fi
    speaker cable going cheap if you're interested...
  8. Depends on the species. Human offspring have a uniquely necessarily
    long nurture period compared to all other animals so evolution has to
    "care" about the parents surviving for several years after breeding.
  9. N. Thornton

    N. Thornton Guest

    I get the feeling that you might have missed something, unless I've
    misunderstood your point. Ie that this machine would repair the recent
    goods, items still as good as new ones, and not the obsolescing stuff,
    which would be scrapped. There is no shortage of such recent kit

    Regards, NT
  10. Nor do most manufacturers of commercial products. They are generally
    worried that they will last too long.
  11. I think I understand what you are proposing. I am seeing this as an
    entropy problem. Keeping the robot up to date on the entire array of
    product details and making decisions about what to keep and repair,
    and how that can be done versus what to discard is not a trivial
    task. That decision will end up costing more than the repairs,

    At work, I am regularly asked to fix things that are eminently and
    cheaply fixable if only all the relevant info were available.
    Sometimes a 50 cent part can save 10s of thousands of dollars. But
    lacking that info, my cost of researching or reverse engineering the
    information would often end up costing way more than the device is
    worth. Occasionally I perform real heroics to get some essentially
    unreplacable device back in service, and it has earned me a reputation
    of something of a miracle worker. But I hate to think of the money
    spent on some of these efforts.

    How will you encourage every manufacturer to provide the vast library
    of technical details needed to make reasonable decisions about discard
    or repair, when they have every interest in keeping secrets to
    encourage consumption of new production? Then you have to figure out
    how to educate the robot to digest those details.
  12. I design machines for a living and I see a few flaws in your idea.

    You're forgetting that machines are very expensive. The only way to justify
    the capital expense of even the simplest automatic machine is if it is kept
    busy. This means it must process thousands, or even millions, of parts or
    devices per year to make it worthwhile. Most machines in industry perform
    simple, repetitive operations. They must be situated along a conveyor, with
    the part located exactly, to do their designated task, or must be easily
    loaded and unloaded by a human operator who positions the work precisely in
    a custom-made fixture. Some assembly machines can accept parts that have
    been simply poured into large hoppers, sorting them and properly assembling
    them, but only on a relatively simple level.

    Could you really have a constant enough supply of defective devices to keep
    such a machine busy enough to pay for itself, let alone make a profit? And
    if you had that many defective parts, isn't there something wrong? Would
    each machine be dedicated to only repair one flaw? And who finds the flaw
    and routs the product to the proper machine? Do you realize what it would
    cost to design and build a machine that could actually do all the things you
    mention, if it's even possible at all? And what happens when you update the
    product, or improve it, or obsolete it and replace it with something better?
    Is this product the only one sold by the company? If not, do you need one
    of these "repair machines" for each product? I'd think an ordinary repair
    technician would be much cheaper.

    If you think a machine can actually be built that intelligently picks up a
    product, takes it apart, inspects it, troubleshoots it, discovers the
    problem and determines a solution, fixes it and then puts it all back
    together again, then you've been watching too much science fiction, or
    reading Popular Mechanics. This would be a sophisticated system indeed. It
    would practically be a human being! Even fairly simple machines can cost
    hundreds of thousands of dollars...this one would easily cost millions. I
    desinged a million-dollar machine once and it just makes brake levers for

    Don't get me wrong. If you actually have detailed plans for such a machine
    and can get someone to make it happen, then good luck. But you can't just
    throw out a raw idea and expect someone to design it, finance it and build
    it for you while you sit back and collect fat checks.

    And I mean actual, buildable plans, too, not just ideas. I can get out a
    piece of paper and draw up a sketch of my "time machine" that shows a chair
    and a big lever with "PAST" and "FUTURE" written above it, but I have to
    come up with exactly *how* it works before it's a true invention.
  13. N. Thornton

    N. Thornton Guest

    Updating on product details is not needed. The robot will diagnose
    goods with no circut diagram available by following as many as poss of
    the same kind of procdures that any repair tech would follow when they
    dont have the circuit diagram. Product details are just an extra, they
    will result in a higher repair rate for that product, but they are not

    Only the very occasional product update would be wanted, for when a
    product comes out that works by a differing principle to what the
    machine knows. An example might be when the VHF radio band came out:
    you cant repair that using the same procedures you would for am.

    Deciding what to repair, I'm unclear how thats such a tough task. Its
    cal;culated from simple things like:
    how long are we allowed to spend per washing machine, per tv, per...
    do we have too many or too few of these
    is this model average inferior or superior quality
    how much inventory storage space do we have
    do we have the parts in the scrap pile (machine decided)
    and occasionally: is this item a special job, and if so how long can
    we allocate to fix it?
    Could you explain?

    This isnt needed. I thought this was explained.
    Not a problem since it isnt needed. But if you do want a team to
    continue working on updating product databases (at considerable cost),
    the team would design new action sequences to follow through and
    probably deliver them to the machine over the net. Probably this would
    only happen once per several years to keep the repair rate up as tech

    Regards, NT
  14. Ken Taylor

    Ken Taylor Guest

    Take the example of a modern DVD or CD player - what can you buy one for,
    twenty bucks? Is the cost of even sending it to the repairer going to exceed
    repair costs? And that's without the repairer being able to get enough money
    back on the job to pay for the machine, whether you believe it's low cost or

  15. N. Thornton

    N. Thornton Guest

    Perhaps thats why such items arent included on the list for this
    hypothetical machine.

    Regards, NT
  16. N. Thornton

    N. Thornton Guest

    Good, as there are substantial issues with it. But you dont seem to
    have understood what the thing is yet. The outline is that it is one
    machine that can diagnose faults in a wide array of items, not a suite
    of product specific machines.

    If I get more time I'll address more points. Please also see the post
    title, I think you missed that.

    Thanks, NT
  17. Ken Taylor

    Ken Taylor Guest

    Now I think you are arbitrarily creating a list to suit your desired result.
    If you only had devices with no moving parts I'd be less skeptical, but a
    fairly random list of cheap items which are going to use spare parts which
    will be too expensive in smaller quantities to make repair economic for a
    'machine' like this makes me a tad skeptical of it's promise, to say the
    least. I'll wait till I see you using EER techniques, then I'll give the
    idea some credibility.


  18. N. Thornton

    N. Thornton Guest

    Ken, did you read the original article? In which I explained why there
    would be no payment for spare parts used, in which I explained that
    the more profitable goods would be the ones selected? DVD players were
    not on that list for a reason.

    Am I crap at explaining things?

    Regards, NT
  19. N. Thornton

    N. Thornton Guest

    Hi Ken

    I'd be interested to hear what you mean by EER techniques. I'm
    familiar with EER amplifiers and equal error rate, but presumably you
    mean something else?

    Regards, NT
  20. Ken Taylor

    Ken Taylor Guest

    I suspect you realise I was being unnecessarily obnoxious with the oblique
    reference to Feerguy - sorry.

    My main beef(s) are the glossing over of the not-so-straightforward task of
    getting a robot to take apart a case, then also the inherent contradiction
    in repairing an item which is useful enough to warrant repair, but not
    modern enough to be cheaply produced (which is a summarization of modern
    electronics and mass production, I'm in a hurry to go to bed! :).

    I think 'Blue Sky' is the kindest description until you more than flesh out
    some of the nitty-gritty.


Ask a Question
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Electronics Point Logo
Continue to site
Quote of the day