Loading ...
Sorry, an error occurred while loading the content.

Re: [beam] Re: An Op-Amp Neuron Equivalent

Expand Messages
  • Michael Demaddis
    I like that way of thinking!   -Michael ________________________________ From: connor_ramsey@ymail.com To: beam@yahoogroups.com
    Message 1 of 29 , May 24, 2012
    • 0 Attachment
      I like that way of thinking!
       
      -Michael

      From: "connor_ramsey@..." <connor_ramsey@...>
      To: beam@yahoogroups.com
      Sent: Thursday, May 24, 2012 12:25 PM
      Subject: [beam] Re: An Op-Amp Neuron Equivalent

       


      > Let's see - do we happen to have an elitist wandering around here?
      > "BEAM" technology certain is "cutting edge" after having sat around
      > since 1994 (date of Tilden's patent) and all told I haven't seen a
      > whole lot more of _five_ years of actual progress in the last 18
      > years. People keep 'inventing' the same stuff over and over (which is
      > great for newbies because it gets them into the very basic parts of
      > this jazz), but a lot of stuff has merely been trying to eke out yet
      > another .01 % of some circuit... which, of course, doesn't really
      > represent much of anything new.

      Yes, exactly. This is why I'm going for things that HAVEN'T already been invented. Personally, my goal is to finish what Mark Tilden started. I want to move BEAM beyond just nerve nets and and try to couple them with definable brains. Brains that can can learn and reason to a primitive degree at least. And if a brain is not required, then at least specialize the neurons to a higher degree of operation. That's my approach to BEAM.



    • S G
      http://www.youtube.com/watch?v=zbUHcwPZqS0 http://www.societyofrobots.com/member_tutorials/node/292  ________________________________ From: Michael Demaddis
      Message 2 of 29 , May 24, 2012
      • 0 Attachment



        From: Michael Demaddis <bmajjik@...>
        To: "beam@yahoogroups.com" <beam@yahoogroups.com>
        Sent: Thursday, May 24, 2012 12:27 PM
        Subject: Re: [beam] Re: An Op-Amp Neuron Equivalent

         
        I like that way of thinking!
         
        -Michael

        From: "connor_ramsey@..." <connor_ramsey@...>
        To: beam@yahoogroups.com
        Sent: Thursday, May 24, 2012 12:25 PM
        Subject: [beam] Re: An Op-Amp Neuron Equivalent

         


        > Let's see - do we happen to have an elitist wandering around here?
        > "BEAM" technology certain is "cutting edge" after having sat around
        > since 1994 (date of Tilden's patent) and all told I haven't seen a
        > whole lot more of _five_ years of actual progress in the last 18
        > years. People keep 'inventing' the same stuff over and over (which is
        > great for newbies because it gets them into the very basic parts of
        > this jazz), but a lot of stuff has merely been trying to eke out yet
        > another .01 % of some circuit... which, of course, doesn't really
        > represent much of anything new.

        Yes, exactly. This is why I'm going for things that HAVEN'T already been invented. Personally, my goal is to finish what Mark Tilden started. I want to move BEAM beyond just nerve nets and and try to couple them with definable brains. Brains that can can learn and reason to a primitive degree at least. And if a brain is not required, then at least specialize the neurons to a higher degree of operation. That's my approach to BEAM.





      • Sweet Riffs
        Well I am quite confused as to how you are creating anything new. If you want to create a brain it may take hundreds or thousands all the way to billions and
        Message 3 of 29 , May 24, 2012
        • 0 Attachment
          Well I am quite confused as to how you are creating anything new. If you want to create a brain it may take hundreds or thousands all the way to billions and even trillions of connections, so you may as well get down to the IC or processor level and create all those networks on a microscopic level. But I still don't quite understand what you have been discussing and I was hoping you could briefly explain the concepts you have introduced, and how I may be able to apply them. I have yet to build any nervous networks or even a neuron because I don't understand them at all. I wish someone would just break it down for me in simple language, without assuming that I know or don't know anything.
        • connor_ramsey@ymail.com
          But I still don t quite understand what you have been discussing and I was hoping you could briefly explain the concepts you have introduced, and how I may be
          Message 4 of 29 , May 24, 2012
          • 0 Attachment
            "But I still don't quite understand what you have been discussing and I was hoping you could briefly explain the concepts you have introduced, and how I may be able to apply them. I have yet to build any nervous networks or even a neuron because I don't understand them at all. I wish someone would just break it down for me in simple language, without assuming that I know or don't know anything."

            Well, the XOR neuron works just like a regular neuron, except it has an extra input that switches it between inverting and non-inverting. Essentially it can switch between "modes". I designed it so that it you could invert the timing between bicores, and reverse a robot without additional control circuitry. After all, simplicity is key.
          • connor_ramsey@ymail.com
            Well I am quite confused as to how you are creating anything new. If you want to create a brain it may take hundreds or thousands all the way to billions and
            Message 5 of 29 , May 24, 2012
            • 0 Attachment
              "Well I am quite confused as to how you are creating anything new. If you want to create a brain it may take hundreds or thousands all the way to billions and even trillions of connections, so you may as well get down to the IC or processor level and create all those networks on a microscopic level."

              Well, not necessarily. Go to Bruce Robinson's biography on the BEAMwiki and click on "learning machines". You'll see a circuit of his that is capable of learning. What I have in mind is basically several of those grouped together, and with improved long-term memory, which is pretty much where all the circuit's intelligence originates. Something like that should perform the FUNDAMENTAL functions of a brain, and at that, would be considered a brain in its own right. However, as far as making a brain more directly comparable to that of an insect, an animal, or even a human being, your point stands valid. Mankind is simply not capable of creating something equal to himself, so I don't see the need to try and go that far anyway.
            • connor_ramsey@ymail.com
              ... That s pretty neat. Sounds like its on way to being the first BEAMbot to reach six feet in length, if that s what you re planning.
              Message 6 of 29 , May 24, 2012
              • 0 Attachment
                --- In beam@yahoogroups.com, S G <garvinguy@...> wrote:
                >
                > http://www.youtube.com/watch?v=zbUHcwPZqS0
                >
                >
                > http://www.societyofrobots.com/member_tutorials/node/292%c3%82%c2%a0

                That's pretty neat. Sounds like its on way to being the first BEAMbot to reach six feet in length, if that's what you're planning.
              • Amit
                I d have to agree with zozzles about the rehashing. I did it myself without realising it on the A and P net, and I was doing it when we were working on the
                Message 7 of 29 , May 25, 2012
                • 0 Attachment
                  I'd have to agree with zozzles about the rehashing. I did it myself without realising it on the A and P net, and I was doing it when we were working on the Spider Replica .
                  The XOR neuron is nice, in that there is reduced parts count, not so much in biological mimicry, but years ago Clifford Boerema (droidmakr) made an XOR circuit using Nvs and diodes which works a treat,
                  and you can can use the same HC14 chip instead of having to buy another kind of IC.

                  The MUX's used in BEAM for reversing aren't terribly complicated either, just two tristate buffers and resistors. I'm not saying any of this to knock your circuit, I'd try it if i didn't already have a bunch of HC14s
                  and 240s to interconnect and get the same function a la droidmakr's PLNs.
                  Concepts are nice, but implementation in a working bot is even better, as it is instructive as to how to use those concepts. There are many splendid concepts contributed by Wilf, JWG, droidmakr, Jenner and Bruce Robinson and many others, which have not been used in very many
                  bots. We could use that groundwork and try to make some nice bots instead of churning out the same things over and over again. I finished my 5M2 so I could test out it's "running" behavior which was an implementation of the P-net (rehash of something already known...)
                  JWG had his FEH for the deVries neural tube if memory serves, I had my salamander for multi-master bicores before it was lost in a move, Bruce has Hider, which is the most advanced beam machine thus far and implemented many things
                  from bicore pwm motor control and beam sonics. Take a look through Wilf's folder and you'll see tonnes of circuits waiting to be used for something cool.

                  With regards to a brain, the learning circuits alone won't be of much use.On this scale the ability to sense and preprocess various inputs is what gives you some nice behaviours. You can make CPGs and sequencers fairly easily to achieve different gaits and behaviours.
                  based on the sensory information.
                  I doubt any of us will be sitting around soldering hundreds of components together and interconnecting all these neurons to try and make a brain. You can implement a network in software and get further. I'm not even sure what the fundamental functions of a brain are for a robot.

                  Which finally brings me to the scavenging for parts thing. With the move to miniaturizaion and smd components, you'll be hard pressed to find useful components more salvageable than LEDs, Electrolytics and switches.
                  I got lucky when the TU was throwing out old tape drives and scored a single Maxxon planetary gear motor. I need at least 7 more for Spyder, over twice as much for my barely working YFH bot.
                  A few years ago I ordered a bunch components from Digikey, because you'll not find many logic chips knocking around in consumer electronics.

                  Anyways, It's good to build these circuits, but it would be imho even better to build nice robots which use them.

                  Best Regards


                  --- In beam@yahoogroups.com, "connor_ramsey@..." <connor_ramsey@...> wrote:
                  >
                  >
                  >
                  >
                  > "Well I am quite confused as to how you are creating anything new. If you want to create a brain it may take hundreds or thousands all the way to billions and even trillions of connections, so you may as well get down to the IC or processor level and create all those networks on a microscopic level."
                  >
                  > Well, not necessarily. Go to Bruce Robinson's biography on the BEAMwiki and click on "learning machines". You'll see a circuit of his that is capable of learning. What I have in mind is basically several of those grouped together, and with improved long-term memory, which is pretty much where all the circuit's intelligence originates. Something like that should perform the FUNDAMENTAL functions of a brain, and at that, would be considered a brain in its own right. However, as far as making a brain more directly comparable to that of an insect, an animal, or even a human being, your point stands valid. Mankind is simply not capable of creating something equal to himself, so I don't see the need to try and go that far anyway.
                  >
                • Michael Demaddis
                  Hi Where can I learn more about the spider? I have been tinkering with Tildens circuits for a while now and have maded a few Bi Core walkers. I was always
                  Message 8 of 29 , May 25, 2012
                  • 0 Attachment
                    Hi Where can I learn more about the spider?
                    I have been tinkering with Tildens circuits for a while now and have maded a few Bi Core walkers.
                    I was always interested in his spider.
                    Any info would be appreciated.
                    Thank you
                     
                    -Michael

                    From: Amit <amitjones101@...>
                    To: beam@yahoogroups.com
                    Sent: Friday, May 25, 2012 11:57 AM
                    Subject: [beam] Re: An Op-Amp Neuron Equivalent

                     


                    I'd have to agree with zozzles about the rehashing. I did it myself without realising it on the A and P net, and I was doing it when we were working on the Spider Replica .
                    The XOR neuron is nice, in that there is reduced parts count, not so much in biological mimicry, but years ago Clifford Boerema (droidmakr) made an XOR circuit using Nvs and diodes which works a treat,
                    and you can can use the same HC14 chip instead of having to buy another kind of IC.

                    The MUX's used in BEAM for reversing aren't terribly complicated either, just two tristate buffers and resistors. I'm not saying any of this to knock your circuit, I'd try it if i didn't already have a bunch of HC14s
                    and 240s to interconnect and get the same function a la droidmakr's PLNs.
                    Concepts are nice, but implementation in a working bot is even better, as it is instructive as to how to use those concepts. There are many splendid concepts contributed by Wilf, JWG, droidmakr, Jenner and Bruce Robinson and many others, which have not been used in very many
                    bots. We could use that groundwork and try to make some nice bots instead of churning out the same things over and over again. I finished my 5M2 so I could test out it's "running" behavior which was an implementation of the P-net (rehash of something already known...)
                    JWG had his FEH for the deVries neural tube if memory serves, I had my salamander for multi-master bicores before it was lost in a move, Bruce has Hider, which is the most advanced beam machine thus far and implemented many things
                    from bicore pwm motor control and beam sonics. Take a look through Wilf's folder and you'll see tonnes of circuits waiting to be used for something cool.

                    With regards to a brain, the learning circuits alone won't be of much use.On this scale the ability to sense and preprocess various inputs is what gives you some nice behaviours. You can make CPGs and sequencers fairly easily to achieve different gaits and behaviours.
                    based on the sensory information.
                    I doubt any of us will be sitting around soldering hundreds of components together and interconnecting all these neurons to try and make a brain. You can implement a network in software and get further. I'm not even sure what the fundamental functions of a brain are for a robot.

                    Which finally brings me to the scavenging for parts thing. With the move to miniaturizaion and smd components, you'll be hard pressed to find useful components more salvageable than LEDs, Electrolytics and switches.
                    I got lucky when the TU was throwing out old tape drives and scored a single Maxxon planetary gear motor. I need at least 7 more for Spyder, over twice as much for my barely working YFH bot.
                    A few years ago I ordered a bunch components from Digikey, because you'll not find many logic chips knocking around in consumer electronics.

                    Anyways, It's good to build these circuits, but it would be imho even better to build nice robots which use them.

                    Best Regards

                    --- In beam@yahoogroups.com, "connor_ramsey@..." <connor_ramsey@...> wrote:
                    >
                    >
                    >
                    >
                    > "Well I am quite confused as to how you are creating anything new. If you want to create a brain it may take hundreds or thousands all the way to billions and even trillions of connections, so you may as well get down to the IC or processor level and create all those networks on a microscopic level."
                    >
                    > Well, not necessarily. Go to Bruce Robinson's biography on the BEAMwiki and click on "learning machines". You'll see a circuit of his that is capable of learning. What I have in mind is basically several of those grouped together, and with improved long-term memory, which is pretty much where all the circuit's intelligence originates. Something like that should perform the FUNDAMENTAL functions of a brain, and at that, would be considered a brain in its own right. However, as far as making a brain more directly comparable to that of an insect, an animal, or even a human being, your point stands valid. Mankind is simply not capable of creating something equal to himself, so I don't see the need to try and go that far anyway.
                    >



                  • connor_ramsey@ymail.com
                    The XOR neuron is nice, in that there is reduced parts count, not so much in biological mimicry, but years ago Clifford Boerema (droidmakr) made an XOR
                    Message 9 of 29 , May 25, 2012
                    • 0 Attachment
                      "The XOR neuron is nice, in that there is reduced parts count, not so much in biological mimicry, but years ago Clifford Boerema (droidmakr) made an XOR circuit using Nvs and diodes which works a treat,
                      > and you can can use the same HC14 chip instead of having to buy another kind of IC.
                      ...
                      > Concepts are nice, but implementation in a working bot is even better, as it is instructive as to how to use those concepts. There are many splendid concepts contributed by Wilf, JWG, droidmakr, Jenner and Bruce Robinson and many others, which have not been used in very many bots."

                      Thanks for that. I wasn't entirely convinced I was the first to come up with this, but I couldn't find any info on anything similar, so it was a safe assumption.

                      "With regards to a brain, the learning circuits alone won't be of much use.On this scale the ability to sense and preprocess various inputs is what gives you some nice behaviours. You can make CPGs and sequencers fairly easily to achieve different gaits and behaviours.
                      > based on the sensory information.
                      > I doubt any of us will be sitting around soldering hundreds of components together and interconnecting all these neurons to try and make a brain. You can implement a network in software and get further. I'm not even sure what the fundamental functions of a brain are for a robot."

                      You're right. Alone, they're not worth much. The point of it is that the "brain" is used as an add-on to a typical BEAM circuit. And to clarify the "brain's" function, it basically remembers data, such as cause and effect of a chosen action, and produces new data based on the stored data. Its job is to fulfill the operations that its nerve net isn't alone capable of. It compensates for the limitations of the nerve net essentially.

                      For say, if you wanted the robot to react to, oh, a green ball, and ignore all the other balls, do something with or to that ball, and then remember how to do that so it can do so again without further instruction, then that's beyond the capabilities of a simple nerve net. A higher cognitive layer such as the "brain" then takes over control, and when the problem is solved, the nerve net resumes control, reacting to tactile sensor input in a traditional manner.

                      And plus, if you want, you can train the robot to now ignore the green ball and target another object. So the "brain" also introduces flexibility that's not normally possible with lower level circuits, while maintaining the robustness and efficiency of the lower circuitry possessed.

                      Plus the "brain" would also up the survivability factor of the bot considerably. It could allow the robot to take proactive measures against hostile stimuli, and to be able to learn from its mistakes, which is far more useful that completely forgetting everything every few seconds and making the same mistakes over and over. That might get you killed.
                    • Richard Piotter
                      ... I too, lament the difficulty in finding consumer friendly components in modern consumer electronics. Furthering the pain for hobbyists, is the prevalence
                      Message 10 of 29 , May 29, 2012
                      • 0 Attachment
                        Begin forwarded message:
                        > From: "Amit" <amitjones101@...>
                        > Date: May 25, 2012 10:57:44 AM CDT
                        > To: beam@yahoogroups.com
                        > Subject: [beam] Re: An Op-Amp Neuron Equivalent
                        > Reply-To: beam@yahoogroups.com
                        >
                        > Which finally brings me to the scavenging for parts thing. With the
                        > move to miniaturizaion and smd components, you'll be hard pressed to
                        > find useful components more salvageable than LEDs, Electrolytics and
                        > switches.
                        > I got lucky when the TU was throwing out old tape drives and scored
                        > a single Maxxon planetary gear motor. I need at least 7 more for
                        > Spyder, over twice as much for my barely working YFH bot.
                        > A few years ago I ordered a bunch components from Digikey, because
                        > you'll not find many logic chips knocking around in consumer
                        > electronics.
                        >


                        I too, lament the difficulty in finding consumer friendly components
                        in modern consumer electronics. Furthering the pain for hobbyists, is
                        the prevalence of "consumer electronics recycling drops" in the past
                        decade that have taken a terrificly huge bite out of the older
                        materials that once DID serve as a source of parts. I have attempted
                        to seek useful materials from these... Needless to say, they do not
                        take kindly to it. At least the drops I went to clearly stated that
                        once an item is left, it becomes their property, and they provide no
                        recourse for reuse... Just send it to the recycler to be broken down
                        and destroyed, sending the bits for appropriate recycling. The
                        manufacturing industry conveniently likes to forget both "reduce" and
                        "reuse", which if you recall the old recycling awareness slogan,
                        always came first, BEFORE "recycle". They certainly don't want you
                        buying less, and they certainly want to take out any pre-owned
                        competition.

                        I have personally taken to collecting older examples of tech, just as
                        a hobby. My favorites are old calculators, with some of my old 1960's
                        and 70s era models being my favorites. While these do have lovely
                        components inside, I dare not disassemble even one single rarity of
                        these amazing machines. Given the collector market, what isn't
                        destroyed by over zealous recycling, is rapidly becoming highly
                        valued, be it for the reseller looking to turn a profit, or like me,
                        the collector who wants a piece of rare tech. I think I spent nearly
                        $480 on my Smith Corona Marchant Cogito 240SR, over $280 on a Sony
                        SOBAX, $150 on a Friden EC132, and $100 on an HP 9825. I spent WAY too
                        much on a Wang 360SE, that didn't even have the keyboard/display
                        unit... Oddly, I do not regret that. I just regret not wining the
                        other half! Most of my more basic models have been acquired by means
                        of lucky thrift store finds... But most of my oldest and higher end
                        machines are just impossible to find without investing and going
                        online. I truly regret disassembling some of my older possessions. I
                        regret taking apart my Commodore PET, or my Colecovision ADAM, or even
                        my first radio, and old tube based AM radio in a white bakelite
                        housing. I find I am reluctant to take apart much of ANY old
                        electronics anymore, save for older televisions. They are not so bad,
                        since there are just SO MANY, though even the older models are
                        starting to become scarce too. They are one of the few remaining
                        places where one can reliably find germanium diodes, which are very
                        nice for low voltage applications. The fact that older TVs have truly
                        been rendered obsolete by broadcast signal changes and the advent of
                        cheap LCDs makes me less apt to feel sympathy for an old 1970s or
                        newer set, though I have a lovely 1959 set in my collection that I
                        actually want to restore. That one, I won't lay a finger on. Radios
                        are fair game for me too. Again, the occasional germanium diode can be
                        found in them, but as is the case, I only attack late 60s or newer. I
                        simply refuse to harm anything older than at least early 60s. Those
                        things are becoming FAR too rare these days too.

                        I find that Digikey is now truly and surely my primary source for
                        electronic components. I rarely scavenge anymore, unless I find myself
                        in an immediate need of a particular value of resistor. If I recall,
                        digikey has a minimum order of $20-25, last I ordered, though that
                        could be changed. Even for younger hobbyists, in this day and age, it
                        should be relatively reasonable to save $20-30, compile a list of
                        parts, and find someone willing to place the order. Me, I order from
                        them whenever I need parts. I do recall the days when I was young, and
                        I wanted to order something. It usually involved taking my money to
                        the bank or post office and getting a money order to send by mail.
                        Thank God for online stores, shopping carts, and easy checkout! I
                        swear, I can have as much fun browsing DigiKey as I can Amazon, ebay,
                        or even that old Mall... Remember those? :)

                        Motors remain the absolute biggest hurdle though, and I fear this will
                        only get worse. New gearmotors of any level of quality tend to be
                        expensive, and most of the time, the surplus market is the only way to
                        find the good ones. As for scavenging... Good luck! EVERYTHING is
                        shifting to PURE solid state these days. Solid State was once a badge
                        of technical superiority, and it was used loosely. TVs were never
                        truly solid state until the LED illuminated LCD came on the market.
                        CRTs and CCFTs are still older tube based technologies. Many TVs had
                        1-3 tubes in the tuner, right up until the early 80s. Even computers
                        still mostly rely on mechanical storage of data, with hard drive
                        mechanisms being most common, and BluRay/DVD drives being the other.
                        Yes, we've ditched the floppy disks and tape drives and created SSDs,
                        but the price and density has limited them to special applications.
                        Even looking at FDDs and HDDs, the motors are actually either
                        steppers, servos, or voice coils... Few of which serve our style of
                        robotics well. Tape drives, if you can even find one, may have a
                        suitable eject motor... Maybe. Tape and even CD players are fading out
                        of use as iPods and smartphones replace them with whole music
                        libraries in your pocket. Those devices already rarely had suitable
                        motors for anything but a basic roller. Modern manufacturing methods
                        have truly and surely integrated the gear trains of most devices into
                        it's intrinsic structure. It's simply not cost effective to buy a
                        premanufactured gearmotor and drop it into the eject mechanism of a
                        bluray drive. It's FAR cheaper to drop a few plastic or metal pieces
                        and gears into holes or slots of the drive's frame. The last major
                        bastion of gearmotors were focus and zoom mechanisms, eject
                        mechanisms, and the occasional toy. Even these are fading away, and in
                        some cases, were the earliest to adopt structural gear trains.

                        Your best bet to find motors by means of salvage, is probably thrift
                        stores or garage sales. Look for the boxiest, oldest looking, most
                        massive camcorders you can find. With cheap flash based recorders,
                        cellphone cameras, and digital cameras that take better video than
                        those old camcorders ever dreamed of, they can be found pretty cheaply
                        these days. There is NO guarantee that a salvageable motor can be
                        found in them. Later models use molded plastic structural components
                        with the necessary places for gears molded in. The motors are nothing
                        but cheap gearless motors that mesh into a series of gears in the
                        inner bowels of the camera. The best bet are the older cameras,
                        because they are more likely to have simplified internal structures,
                        that are driven by actual self contained gearmotors. I'd say, stick to
                        VHS and VHS-C camcorders. Even these can fail to provide motors of
                        use. The older the better. My robot Bronco uses a tape eject motor for
                        it's waist. It was removed from an old Minolta VHS-C camcorder from
                        the late 80s. If you're lucky, you might find in older camcorders, the
                        zoom and focus motors may be gearmotors. The iris RARELY is, if ever.
                        I've yet to find one. They are usually apertures that work more like a
                        panel meter. If you found an EXCEPTIONALLY old camera or a very high
                        end camera that has an axially symmetric iris rather than what looks
                        like intersecting "V" shapes, you might find a gear mechanism driving
                        the iris... Maaaaybe... Very rarely, you might find a film camera that
                        has a rewind motor or a zoom motor that is usable. This is very
                        rare... Most of the cameras with these features very quickly shifted
                        toward integral mechanisms like newer camcorders, to save cost, size,
                        and weight.

                        I find, that when I have a rare excess of money, and I see All
                        Electronics, or Electronics Goldmine, or one of the many electronic
                        surplus suppliers with a good looking gearmotor... I tend to buy a
                        handful. I probably bought over $300 in gearmotors once, just to have
                        them. I certainly could not do that right now... Don't have the cash.
                        But I do have a wonderful stockpile that I can rely on, so if I ever
                        DO get an idea, I can simply act upon it, rather than wish.

                        My recommendation for everyone who wishes to still attempt to
                        scavenge... Focus on gearmotors first and foremost. Follow my
                        recommendation of getting old camcorders from thrift stores or garage
                        sales. Don't bother with anything that looks one handed (well, if it's
                        cheap, you can try). You want the bigger, older ones, that have a back
                        end sticking out far enough to rest on your shoulder. Older stereos
                        and TVs can have a variety of resistors and capacitors and diodes, but
                        to be COMPLETELY honest... Those components are DIRT cheap from
                        Digikey. Get your 74HC14s and 240s, and your resistors and capacitors
                        there. You'll get exact values, and they will have leads that you can
                        be ready to work with. If you use prototyping boards, you have the
                        length to place parts where needed, and if you freeform... Well the
                        advantage is obvious! If you make your own PC boards, more power to ya!

                        Finally, there is circuit boards. Recently, I actually bumped all my
                        supplies to include SMT parts. I standardized on the 1206 size SMT
                        components, which while still kinda small, is more than large enough
                        to easily hand assemble with tweezers, and pass a trace or two between
                        pads. The nice thing with SMT, is when you make your boards, you get
                        to skip drilling out all those holes! Not too big a deal on small
                        boards, But My boards tend to be bigger. My Spyder robot, for example,
                        has 4 PC boards, with 7 chips and 48 drive transistors. Add in the
                        board interconnections, all 16 pairs of resistors and capacitors for
                        the Nv net... and you see that there were a LOT of holes to drill,
                        over 400 holes just for the basic components, wire interconnects, and
                        support components alone. I don't mind working with SMT parts. If I
                        redid that robot in SMD, I would have likely, only needed around 68
                        holes! Just order chips with the larger pitch leads. I can make my own
                        boards, but as soon as I have a PC up and running, I will likely just
                        order boards from a board house from now on, as it is simply the most
                        ideal use of my time. There are some places that have good prices on
                        small quantities. As for those out there who simply either don't have
                        the software, money, or experience to order custom boards, making
                        their own PC boards is easy enough to do. If even that is too much,
                        most BEAM circuits can be constructed on a small solderable breadboard
                        or perfboard. Of course, there's also free-form circuits. "Dead-bug"
                        style prototyping is always very popular with BEAM. One can also still
                        salvage headers from computer equipment still. A nice use for these
                        strips is to act as "ladder" scaffolding for resistors and capacitors.
                        Most BEAM circuits can actually get away without this though. Using
                        certain types of female connectors CAN make it possible to swap out
                        resistors though. Salvaging these can make for nice additions to an Nv
                        net, bicore, or other BEAM circuit.

                        Richard Piotter
                      • connor_ramsey@ymail.com
                        I ve not had to buy anything so far. In my attic, I ve everything from 20-30 yr old computer remains, VHS rewinders, maybe camcorders, there may or may not
                        Message 11 of 29 , May 29, 2012
                        • 0 Attachment
                          I've not had to buy anything so far. In my attic, I've everything from 20-30 yr old computer remains, VHS rewinders, maybe camcorders, there may or may not still be a broken TV, there's gold up in there. And the computers are a great source of capacitors, resistors, hex inverters, octal inverters, octal buffers, multiplexers, demultiplexers, line drivers, line receivers, you name it. Even a few transistors, though not nearly as many as I'd like. There are also a few voltage triggers. I have plenty of dead solar patio lights, perhaps dozens of them. Can anyone tell me what those weird 4-terminal transistors are?
                        • Richard Piotter
                          Not sure. Could they possibly be bridge rectifiers, and not transistors? Those have 4 terminals. Google search the markings on them. Also... Be aware that some
                          Message 12 of 29 , May 30, 2012
                          • 0 Attachment
                            Not sure. Could they possibly be bridge rectifiers, and not
                            transistors? Those have 4 terminals. Google search the markings on them.

                            Also... Be aware that some older computers are becoming rare enough
                            that you are better off not destroying them. You can buy all the
                            components you like for what a few rare models will sell for. Just be
                            aware of what you are dismantling. I myself am quite interested in old
                            calculators from 1974 and earlier in particular, and older computers
                            that predate PCs and Apple IIs.

                            As for the 240, that one's popular with BEAM as well. It's commonly
                            used for Bicores and motor drivers for small motors. Very popular.
                            Ideally, people use HC, AC, HCT, or ACT versions. The LS version is, I
                            believe uses bipolar TTL logic instead of the CMOS logic of the ones
                            mentioned above. You may still find uses for it. LS chips require more
                            current, and typically 5 volt supplies.

                            Richard Piotter


                            Begin forwarded message:
                            > From: "connor_ramsey@..." <connor_ramsey@...>
                            > Date: May 29, 2012 11:07:26 PM CDT
                            > To: beam@yahoogroups.com
                            > Subject: [beam] Re: An Op-Amp Neuron Equivalent
                            > Reply-To: beam@yahoogroups.com
                            >
                            >
                            > I've not had to buy anything so far. In my attic, I've everything
                            > from 20-30 yr old computer remains, VHS rewinders, maybe camcorders,
                            > there may or may not still be a broken TV, there's gold up in there.
                            > And the computers are a great source of capacitors, resistors, hex
                            > inverters, octal inverters, octal buffers, multiplexers,
                            > demultiplexers, line drivers, line receivers, you name it. Even a
                            > few transistors, though not nearly as many as I'd like. There are
                            > also a few voltage triggers. I have plenty of dead solar patio
                            > lights, perhaps dozens of them. Can anyone tell me what those weird
                            > 4-terminal transistors are?
                            >


                            Begin forwarded message:

                            > From: "connor_ramsey@..." <connor_ramsey@...>
                            > Date: May 30, 2012 12:54:48 AM CDT
                            > To: beam@yahoogroups.com
                            > Subject: [beam] I Can't Believe No One's Stumbled Upon This!
                            > Reply-To: beam@yahoogroups.com
                            >
                            > I was looking for more parts on a port driver card from that old
                            > computer again, and I found maybe the coolest IC ever! It's called
                            > the 74LS240N, and it's basically a tristate octal inverter, but with
                            > SCHMIDT INVERTERS. You heard me! A tristate octal Schmidt inverter!
                            > How cool is that!? They're apparently still in production, so you
                            > guys should get yourselves some if you can. I'd be contemplating
                            > what to do with it right now if I weren't busy working on Junkman 1,
                            > a one motor-two legged walker with an N/2 neuron bicore, I'll give
                            > more on him when he's finished. When he is, he'll be my first
                            > functional walker.
                            >
                            > Take care, Connor.
                          • Amit Tiwari-Jones
                            The AC240 is non schmitt so good for bicores. The one that he found would be nice for making Nvs in a large microcore I suppose.I ve been meaning to order the
                            Message 13 of 29 , May 30, 2012
                            • 0 Attachment
                              The AC240 is non schmitt so good for bicores. The one that he found would be nice for making Nvs in a large microcore I suppose.
                              I've been meaning to order the 74HC7540 or HC540, it's an inverting octal schmitt trigger, 3 state CMOS. Should better to use than
                              the LS chip.

                              I agree with saving and selling the antiques. There are some nice things out there which will continue to appreciate in value. I was browsing
                              ebay and magnetic core memory sells for a lot more than a pretty penny.You can buy parts from digikey quite cheaply, in great quantity
                              and you won't have to labour to salvage them. I've damaged so many nice parts trying to salvage them it's not funny.
                              Speaking of old tape recorders, I have an Sony WM-D6C and these don't have the escap mu915L motors . Leads me to wonder where exactly
                              Tilden got his from.

                              Gear motors are quite expensive, I've not continued work on spyder or the other bots because I can't afford them at the moment. Ordering from SB and
                              having them shipped here is bad, because the taxes amount to another motor (GM13as). The GM23 and 24 are attractive though at only  $5, but I've not
                              found any European distributors.Besides that, Food, Text books, bike maintenance and my broken RC heli are sucking funds away.
                               I'm holding out hope that the TU throws out some more tape drives. They have one nice maxxon motor for the ejects.

                              I have some SMD components now. For my prototypes I prefer DIP and veroboard with point-to point wiring. There's something about it I just like. The
                              final versions I design smd pcbs for. I've just never made the boards. I made a clone of the BC4 hextiles, maybe someone wants to look over them and
                              make a bunch?

                              B.R.
                              Amit


                              To: beam@yahoogroups.com
                              From: richfiles1@...
                              Date: Wed, 30 May 2012 06:43:56 -0500
                              Subject: Re: [beam] Re: An Op-Amp Neuron Equivalent

                               
                              Not sure. Could they possibly be bridge rectifiers, and not
                              transistors? Those have 4 terminals. Google search the markings on them.

                              Also... Be aware that some older computers are becoming rare enough
                              that you are better off not destroying them. You can buy all the
                              components you like for what a few rare models will sell for. Just be
                              aware of what you are dismantling. I myself am quite interested in old
                              calculators from 1974 and earlier in particular, and older computers
                              that predate PCs and Apple IIs.

                              As for the 240, that one's popular with BEAM as well. It's commonly
                              used for Bicores and motor drivers for small motors. Very popular.
                              Ideally, people use HC, AC, HCT, or ACT versions. The LS version is, I
                              believe uses bipolar TTL logic instead of the CMOS logic of the ones
                              mentioned above. You may still find uses for it. LS chips require more
                              current, and typically 5 volt supplies.

                              Richard Piotter

                              Begin forwarded message:
                              > From: "connor_ramsey@..." <connor_ramsey@...>
                              > Date: May 29, 2012 11:07:26 PM CDT
                              > To: beam@yahoogroups.com
                              > Subject: [beam] Re: An Op-Amp Neuron Equivalent
                              > Reply-To: beam@yahoogroups.com
                              >
                              >
                              > I've not had to buy anything so far. In my attic, I've everything
                              > from 20-30 yr old computer remains, VHS rewinders, maybe camcorders,
                              > there may or may not still be a broken TV, there's gold up in there.
                              > And the computers are a great source of capacitors, resistors, hex
                              > inverters, octal inverters, octal buffers, multiplexers,
                              > demultiplexers, line drivers, line receivers, you name it. Even a
                              > few transistors, though not nearly as many as I'd like. There are
                              > also a few voltage triggers. I have plenty of dead solar patio
                              > lights, perhaps dozens of them. Can anyone tell me what those weird
                              > 4-terminal transistors are?
                              >

                              Begin forwarded message:

                              > From: "connor_ramsey@..." <connor_ramsey@...>
                              > Date: May 30, 2012 12:54:48 AM CDT
                              > To: beam@yahoogroups.com
                              > Subject: [beam] I Can't Believe No One's Stumbled Upon This!
                              > Reply-To: beam@yahoogroups.com
                              >
                              > I was looking for more parts on a port driver card from that old
                              > computer again, and I found maybe the coolest IC ever! It's called
                              > the 74LS240N, and it's basically a tristate octal inverter, but with
                              > SCHMIDT INVERTERS. You heard me! A tristate octal Schmidt inverter!
                              > How cool is that!? They're apparently still in production, so you
                              > guys should get yourselves some if you can. I'd be contemplating
                              > what to do with it right now if I weren't busy working on Junkman 1,
                              > a one motor-two legged walker with an N/2 neuron bicore, I'll give
                              > more on him when he's finished. When he is, he'll be my first
                              > functional walker.
                              >
                              > Take care, Connor.

                            • connor_ramsey@ymail.com
                              ... Thanks. If they re not transistors, then oh well, they basically come in prebuilt solar engines. And as far as preserving old technology, it s really too
                              Message 14 of 29 , May 30, 2012
                              • 0 Attachment
                                --- In beam@yahoogroups.com, Richard Piotter <richfiles1@...> wrote:
                                >
                                >" Not sure. Could they possibly be bridge rectifiers, and not
                                > transistors? Those have 4 terminals. Google search the markings on them.
                                >
                                > Also... Be aware that some older computers are becoming rare enough
                                > that you are better off not destroying them. You can buy all the
                                > components you like for what a few rare models will sell for. Just be
                                > aware of what you are dismantling."

                                Thanks. If they're not transistors, then oh well, they basically come in prebuilt solar engines.

                                And as far as preserving old technology, it's really too late for that. All the computers I have available to me were already torn apart. Therefore anything I find in them is free game. And I'm pretty sure that video rewinders were rendered obselete when they were invented.
                              • connor_ramsey@ymail.com
                                ... I ll see if I can find some of those. For now, I m perfectly content with what I have. Oh, and I ve been meaning to ask this, does anyone think I could use
                                Message 15 of 29 , May 30, 2012
                                • 0 Attachment
                                  --- In beam@yahoogroups.com, Amit Tiwari-Jones <amitjones101@...> wrote:
                                  >
                                  >
                                  >" The AC240 is non schmitt so good for bicores. The one that he found would be nice for making Nvs in a large microcore I suppose.I've been meaning to order the 74HC7540 or HC540, it's an inverting octal schmitt trigger, 3 state CMOS. Should better to use than the LS chip."
                                  >

                                  I'll see if I can find some of those. For now, I'm perfectly content with what I have.

                                  Oh, and I've been meaning to ask this, does anyone think I could use a resistor-diode AND gate as an analog switch, if I use the diode as the enable input? Any way I figure it, that should work fine. Because I don't have a 4066 or anything similar and it's going to be a while before I can just go and buy these things.
                                • yahmez_alpha
                                  Conner, The 4 legged transistor is an IC that I could not find any info about online, but I have used them for some projects. If you wire them like the solar
                                  Message 16 of 29 , May 31, 2012
                                  • 0 Attachment
                                    Conner,

                                    The 4 legged 'transistor' is an IC that I could not find any info about online, but I have used them for some projects. If you wire them like the solar light, they take the solar panel current to charge the battery (1.5v). When the solar panel current drops off enough because of lack of light it then provides the battery's current to the LED. There seems to be a built in joule thief inside too, outputting enough voltage to drive a white LED from a single 1.5v cell. It's a pretty cool little part. I have one running a bi-core blinking eye bot that I made for my 3 year old son... he calls it his night night bot. As I recall, there is an inductor you have to wire with it as well. I drew up a wiring schematic of it some time ago, I'll see if I can dig it up, but otherwise just wire it like it is set up in the solar light and replace the white LED with your circuit.

                                    -James

                                    --- In beam@yahoogroups.com, "connor_ramsey@..." <connor_ramsey@...> wrote:
                                    >
                                    > ..... Can anyone tell me what those weird 4-terminal transistors are?
                                    >
                                  • connor_ramsey@ymail.com
                                    ... I ve already looked them up, and Piotter was right,they are bridge rectifiers. The solar lights I have actually use an LDR to measure light levels, but I
                                    Message 17 of 29 , May 31, 2012
                                    • 0 Attachment
                                      > Conner,
                                      >
                                      > The 4 legged 'transistor' is an IC that I could not find any info about online, but I have used them for some projects. If you wire them like the solar light, they take the solar panel current to charge the battery (1.5v). When the solar panel current drops off enough because of lack of light it then provides the battery's current to the LED. There seems to be a built in joule thief inside too, outputting enough voltage to drive a white LED from a single 1.5v cell. It's a pretty cool little part. I have one running a bi-core blinking eye bot that I made for my 3 year old son... he calls it his night night bot. As I recall, there is an inductor you have to wire with it as well. I drew up a wiring schematic of it some time ago, I'll see if I can dig it up, but otherwise just wire it like it is set up in the solar light and replace the white LED with your circuit.
                                      >
                                      > -James

                                      I've already looked them up, and Piotter was right,they are bridge rectifiers. The solar lights I have actually use an LDR to measure light levels, but I plan on replacing these with NPNs, with the base wired to an Nv input, through the rectifier to the solar cell. Now that I think about it, that would actually qualify as a type 3 SE, which are my favorite of coarse.
                                    Your message has been successfully submitted and would be delivered to recipients shortly.