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

Hexapod

Expand Messages
  • Richard Piotter
    I m moving to my next project, which officialy, will be the first robot I build with reversing and turning capabilities built into the main circuitry during
    Message 1 of 10 , Sep 25, 2000
      I'm moving to my next project, which officialy, will be the first robot
      I build with reversing and turning capabilities built into the main
      circuitry during the initial design.

      I may need to cross a few signal lines or reverse a couple polaritys,
      but I think the circuit is about right. If anyone wants to get a look at
      it, go ahead. It uses the 3 motor hexapod mechanics similar to the
      cricket or Lynxmotion's Hexapod I.
      --


      Richard Piotter The Richfiles Robotics & TI web page
      richfiles@... http://richfiles.calc.org

      -- Make Money by Simply Surfing the Net!!! Click below!!!
      --- http://www.alladvantage.com/go.asp?refid=ATL147 ---
    • Ben Hitchcock
      Very nice Richard! ONe thing however: would it be easier to tune if the bicore controlled the two leg motors directly, and the waist motor had that half a time
      Message 2 of 10 , Sep 25, 2000
        Very nice Richard!

        ONe thing however: would it be easier to tune if the bicore controlled the
        two leg motors directly, and the waist motor had that half a time constant
        delay? (Assuming leg motors are the two right hand motors, and waist is
        the left hand motor)

        That is, get rid of the four gates with the 1/2R and R resistors, and put
        two gates in line with the left hand motor? Hmm you still need one extra
        gate to drive the H-bridge properly...
        Well, at least it saves you one gate, and would make it easier to tune
        (the two 1/2R and R branches wouldn't have to be exactly equal) and would
        allow you to change the timings a lot more easily.

        Just an idea,
        Ben

        Richard Piotter wrote:
        > I'm moving to my next project, which officialy, will be the first robot
        > I build with reversing and turning capabilities built into the main
        > circuitry during the initial design.
        >
        > I may need to cross a few signal lines or reverse a couple polaritys,
        > but I think the circuit is about right. If anyone wants to get a look at
        > it, go ahead. It uses the 3 motor hexapod mechanics similar to the
        > cricket or Lynxmotion's Hexapod I.
        > --
        >
        >
        > Richard Piotter The Richfiles Robotics & TI web page
        > richfiles@... http://richfiles.calc.org
        >
        > -- Make Money by Simply Surfing the Net!!! Click below!!!
        > --- http://www.alladvantage.com/go.asp?refid=ATL147 ---
        >
        > To unsubscribe from this group, send an email to:
        > beam-unsubscribe@egroups.com
        >
        >
        Content-Description: Unknown Document

        [image/gif is not supported, skipping...]


        --
        http://www.wollongong.apana.org.au/~ben
      • Richard Piotter
        the bicore portion is always 180 degrees out of phase, and safe for the H-bridges. the branch portions can exist for short times in the same state depending on
        Message 3 of 10 , Sep 25, 2000
          the bicore portion is always 180 degrees out of phase, and safe for the
          H-bridges. the branch portions can exist for short times in the same
          state depending on the timing. Did it that way just to be on the safe
          side, and to reduce the chip count. I'm not sure if the reverse trigger
          will work though. It needs to switch the enables on both reverse chips.
          I think the diode setup should work, but it's been a while since I've
          done Nv work, so if anyone wants to verify the circuit, it'd be great! I
          designed it in half an hour while sitting in a waiting room.

          later!


          Ben Hitchcock wrote:
          >
          > Very nice Richard!
          > are the two right hand motors, and waist is
          > the left hand motor)
          >
          > That is, get rid of the four gates with the 1/2R and R resistors, and put
          > two gates in line with the left hand motor? Hmm you still need one extra
          > gate to drive the H-bridge properly...
          > Well, at least it saves you one gate, and would make it easier to tune
          > (the two 1/2R and R branches wouldn't have to be exactly equal) and would
          > allow you to change the timings a lot more easily.
          >
          > Just an idea,
          > Ben
          >
          > Richard Piotter wrote:
          > > I'm moving to my next project, which officialy, will be the first robot
          > > I build with reversing and turning capabilities built into the main
          > > circuitry during the initial design.
          > >
          > > I may need to cross a few signal lines or reverse a couple polaritys,
          > > but I think the circuit is about right. If anyone wants to get a look at
          > > it, go ahead. It uses the 3 motor hexapod mechanics similar to the
          > > cricket or Lynxmotion's Hexapod I.
          > > --
          > >
          > >
          > > Richard Piotter The Richfiles Robotics & TI web page
          > > richfiles@... http://richfiles.calc.org
          > >
          > > -- Make Money by Simply Surfing the Net!!! Click below!!!
          > > --- http://www.alladvantage.com/go.asp?refid=ATL147 ---
          > >
          > > To unsubscribe from this group, send an email to:
          > > beam-unsubscribe@egroups.com
          > >
          > >
          > Content-Description: Unknown Document
          >
          > [image/gif is not supported, skipping...]
          >
          > --
          > http://www.wollongong.apana.org.au/~ben
          >
          >
          > To unsubscribe from this group, send an email to:
          > beam-unsubscribe@egroups.com

          --


          Richard Piotter The Richfiles Robotics & TI web page
          richfiles@... http://richfiles.calc.org

          -- Make Money by Simply Surfing the Net!!! Click below!!!
          --- http://www.alladvantage.com/go.asp?refid=ATL147 ---
        • Richard Piotter
          Oh sorry, missed the second paragraph. I m not so worried about it, being that I ll need to fill the chip anyway, requiring a second, and I already have a
          Message 4 of 10 , Sep 25, 2000
            Oh sorry, missed the second paragraph. I'm not so worried about it,
            being that I'll need to fill the chip anyway, requiring a second, and I
            already have a layout for the Nv net. I did consider that, but it
            doesn't mater to me, since the 74ACT139s will "filter" the output anyway.

            Also, is it OK to tie outputs from A and B on the 139 together if the
            enables are tied by an inverter. I forget if it is tristate of not. I
            thought it was anyway.


            Ben Hitchcock wrote:
            >
            > Very nice Richard!
            >
            > ONe thing however: would it be easier to tune if the bicore controlled the
            > two leg motors directly, and the waist motor had that half a time constant
            > delay? (Assuming leg motors are the two right hand motors, and waist is
            > the left hand motor)
            >
            > That is, get rid of the four gates with the 1/2R and R resistors, and put
            > two gates in line with the left hand motor? Hmm you still need one extra
            > gate to drive the H-bridge properly...
            > Well, at least it saves you one gate, and would make it easier to tune
            > (the two 1/2R and R branches wouldn't have to be exactly equal) and would
            > allow you to change the timings a lot more easily.
            >
            > Just an idea,
            > Ben
            >
            > Richard Piotter wrote:
            > > I'm moving to my next project, which officialy, will be the first robot
            > > I build with reversing and turning capabilities built into the main
            > > circuitry during the initial design.
            > >
            > > I may need to cross a few signal lines or reverse a couple polaritys,
            > > but I think the circuit is about right. If anyone wants to get a look at
            > > it, go ahead. It uses the 3 motor hexapod mechanics similar to the
            > > cricket or Lynxmotion's Hexapod I.
            > > --
            > >
            > >
            > > Richard Piotter The Richfiles Robotics & TI web page
            > > richfiles@... http://richfiles.calc.org
            > >
            > > -- Make Money by Simply Surfing the Net!!! Click below!!!
            > > --- http://www.alladvantage.com/go.asp?refid=ATL147 ---
            > >
            > > To unsubscribe from this group, send an email to:
            > > beam-unsubscribe@egroups.com
            > >
            > >
            > Content-Description: Unknown Document
            >
            > [image/gif is not supported, skipping...]
            >
            > --
            > http://www.wollongong.apana.org.au/~ben
            >
            >
            > To unsubscribe from this group, send an email to:
            > beam-unsubscribe@egroups.com

            --


            Richard Piotter The Richfiles Robotics & TI web page
            richfiles@... http://richfiles.calc.org

            -- Make Money by Simply Surfing the Net!!! Click below!!!
            --- http://www.alladvantage.com/go.asp?refid=ATL147 ---
          • Bruce Robinson
            ... Nope, it isn t. Select = H, all outputs forced to H. Select = L, one output is active low, depending on inputs. So, it looks as though you re going to fry
            Message 5 of 10 , Sep 25, 2000
              Richard Piotter wrote:
              >
              > Also, is it OK to tie outputs from A and B on the 139 together if the
              > enables are tied by an inverter. I forget if it is tristate of not. I
              > thought it was anyway.

              Nope, it isn't.

              Select = H, all outputs forced to H.
              Select = L, one output is active low, depending on inputs.

              So, it looks as though you're going to fry those '139s as soon you fire
              up the circuit :(

              Bruce
            • Richard Piotter
              I ve followed the advice below, and I think I have a reasonable circuit to work with. Feel free to look at the schematic to see if it looks right now. I d like
              Message 6 of 10 , Sep 26, 2000
                I've followed the advice below, and I think I have a reasonable circuit
                to work with. Feel free to look at the schematic to see if it looks
                right now. I'd like to do the PCB layout as soon as I can once the
                schematic is verified as "doable".

                Bruce Robinson wrote:
                >
                > Hi, Richard.
                >
                > I saw a couple of flaws in your schematic. The obvious one I already
                > mentioned -- the '139 is not a tristate chip, so your circuit will fry
                > it.
                >
                > Here's an interesting (and unappreciated) way of getting the reversing
                > function without increasing your chip count. It uses a 74HC86 (XOR) chip
                > for the reverse function -- a method that is invariably rejected by
                > "experts" because it has an apparent flaw.
                >
                > ___
                > In 1 -----\\ \
                > || |---- Out 1 (pin 2/14 - 74HC139)
                > +--//__/
                > | ___
                > In 2 -----\\ \
                > | || |---- Out 2 (pin 3/13 - 74HC139)
                > Rev --+--//__/
                >
                > So, the flaw is that not only does the circuit reverse the signals on
                > inputs 1 and 2, it also inverts the "both off" and "both on" conditions,
                > which is highly undesirable.
                >
                > The secret that everyone misses is what happens when you tag half of a
                > 74HC139 onto the outputs from this circuit.
                >
                > __ __
                > GND---| V |---(+)
                > 1 ------| |-
                > 2 ------| |-
                > -| |-
                > Hbrdg --| |-
                > Hbrdg --| |-
                > -| |-
                > GND--|_____|-
                >
                > If you build a truth table, you'll see that the only possible signal
                > pairs to the H-bridge are OFF (1,1); LEFT (1,0); and RIGHT (0,1). The
                > 74HC139 effectively knocks out any condition where the inputs have the
                > same values, whether they be (1,1) or (0,0). This neatly eliminates the
                > objection to using a pair of XORs for a reverser function. Best of all,
                > one motor use only half a 74HC86 and half a 74HC139, so you can drive
                > two motors with a pair of chips.
                >
                > That's my "everyday chip" solution to the reversing problem.
                >
                > The other flaw (I think) in your circuit is with the sensor Nu array. It
                > looks to me as though you've built a couple of OR gates, each with a
                > pair of diodes and a resistor. Easy enough to do, and often simplifies
                > the layout. However, your three Nv outputs are normally High, so the
                > contact array will always issue High outputs unless BOTH the backup and
                > turn contacts are closed. I'm not sure that's what you wanted.
                >
                > Two solutions to this -- either invert all the polarity connections for
                > all three Nu neurons (resistors and caps go to Vcc, switches go to GND)
                > -- or reverse all the diodes in the OR gates and hook the two resistors
                > up to Vcc. This will turn your OR gates into AND gates -- but since your
                > Nv outputs are normally high, the effect is to create an INVERTED OR
                > gate.
                >
                > You may want to double check my logic on this -- I'm pretty sure I got
                > it right.
                >
                > I'm looking forward to seeing Hexapod in action.
                >
                > Bruce

                --


                Richard Piotter The Richfiles Robotics & TI web page
                richfiles@... http://richfiles.calc.org

                -- Make Money by Simply Surfing the Net!!! Click below!!!
                --- http://www.alladvantage.com/go.asp?refid=ATL147 ---
              • Richard Piotter
                I tend to follow the low high stuff in my head. My mapping skills also seem to work to mentaly image schematic and PCB layouts, and convert between the two.
                Message 7 of 10 , Sep 26, 2000
                  I tend to follow the low high stuff in my head. My mapping skills also
                  seem to work to mentaly image schematic and PCB layouts, and convert
                  between the two. Very great "feature"! :)

                  I was just unsure about the diode OR gate opperation. I guess I forgot
                  it was an OR gate. In my head, that's what I wanted, but it never hit me
                  that the Nus were actively high! Hehe!

                  You're also right about moving the outputs on the 139s up a pin? I have
                  a Z-Bridge schematic on my computer, but it also has an elternate
                  version of the circuit for a different application right next to it. I
                  always mix the two up when I work from memory (I was in a waiting room
                  when I designed this).


                  Also, I have an aditional question. There is a slight, fraction of a
                  second, delay between when an input enters an Nv and the output changes.
                  With the two branches, there can be substancial delay, between loading
                  and variations in component value. If I were to put an additional
                  inverter at the end of one chain and cut the other, will the delay time
                  of the second inverter switching after the first have an adverse effect
                  on power coonsumption while the H-Bridge has the same signal for a split
                  second (with no 139 installed). Is it a low enough effect to not
                  adversely effect power consumption or the safety of the H-Bridges over
                  an extended time?


                  I think I'll do the PCB tommorow or the day after. Depends how bored I
                  get and if I have enough dry transfers left for it.

                  Thanks!


                  Bruce Robinson wrote:
                  >
                  > Richard Piotter wrote:
                  > >
                  > > I also could not recall the polarity/outputs of the Nus off the top of
                  > > my head. All I need is a circuit where hitting the switch for the left
                  > > changes the state of the signal to the left motor, hitting the switch
                  > > for the right changes the state of the signal to the right motor, and
                  > > hitting the switch for reverse changes the state both the motor signals.
                  > > I wanted to do this using diodes to avoid requiring added chips.
                  >
                  > The trick for figuring out Nus is this: Assume it's sitting quietly for
                  > a while -- the cap is going to eventually match whatever the resistor is
                  > connected to. If the resistor is grounded, the cap will eventually have
                  > zero volts on it, and that is what the inverter input "sees". So the
                  > quiescent state of an Nu with a grounded resistor is output = High.
                  >
                  > I can never keep track of these things mentally, so when I sketch up a
                  > schematic, I put a little H or L next to the Nu/Nv outputs to indicate
                  > the quiescent state. Helps me a lot.
                  >
                  > BTW, I think in your revised schematic you need to shift the '139
                  > outputs one pin up the chip.
                  >
                  > Bruce

                  --


                  Richard Piotter The Richfiles Robotics & TI web page
                  richfiles@... http://richfiles.calc.org

                  -- Make Money by Simply Surfing the Net!!! Click below!!!
                  --- http://www.alladvantage.com/go.asp?refid=ATL147 ---
                • Bruce Robinson
                  ... Lessee. Switching time for a 74HC14 powered with 4.5 volts is 15 nanoseconds MAX. I d say you don t need to worry about it, Richard :) Actually, this is
                  Message 8 of 10 , Sep 26, 2000
                    Richard Piotter wrote:
                    >
                    > ... I have an aditional question. There is a slight, fraction of a
                    > second, delay between when an input enters an Nv and the output changes.
                    > With the two branches, there can be substancial delay, between loading
                    > and variations in component value. If I were to put an additional
                    > inverter at the end of one chain and cut the other, will the delay time
                    > of the second inverter switching after the first have an adverse effect
                    > on power coonsumption while the H-Bridge has the same signal for a split
                    > second (with no 139 installed). Is it a low enough effect to not
                    > adversely effect power consumption or the safety of the H-Bridges over
                    > an extended time?

                    Lessee. Switching time for a 74HC14 powered with 4.5 volts is 15
                    nanoseconds MAX. I'd say you don't need to worry about it, Richard :)
                    Actually, this is less than the combined delay times on the transistors
                    in the H-bridge.

                    Getting away from the datasheets and into the real world, if you think
                    about it for a moment, lots of people drive H-bridges directly from
                    bicores with no ill effect.

                    Bruce
                  • Wilf Rigter
                    The gate delays are not a problem since short h-bridge glitches have low energy. Anyway the turnon / turnoff times of the h-bridge are also different so only
                    Message 9 of 10 , Sep 27, 2000
                      The "gate delays" are not a problem since short h-bridge glitches have low
                      energy. Anyway the turnon / turnoff times of the h-bridge are also different
                      so only a deliberate deadtime can cure all glitches.

                      By the way, your original post with the 139 "reversing" Z-bridge idea was
                      quite brilliant. You almost got it right but because it does not tristate,
                      you have to use a resistor to isolate the HC139 output to the h-bridge input
                      . In other words, use 2 input resistors for each h-bridge input as shown in
                      the PNCfree microcore with "reversing" 74HC139 h-bridge. Note the nifty
                      control of the enable line.
                      <<u139CORE.gif>>
                      There are several other variations on your hexapod circuit worth pursuing.
                      For example, your design uses a grounded master bicore and two slave
                      branches to generate a pair of waveforms each delayed from the master bicore
                      output waveforms by 90 degrees. These branches act as a slave bicore but
                      have independent delays and independent pulsewidths. However there is no
                      particular advantage to using branches compared to a slave bicore. In fact,
                      as you noted with branches you can get overlapping waveforms so then you
                      need a HC139 to avoid frying the h-bridge. Still independent control of the
                      slave wave could be useful.

                      So how about a bi-phase slave circuit similar to that Nv/Nu circuit used in
                      the powersmart head. It can be made phototropic with the addition of some
                      photodiodes just like a suspended master bicore as shown.

                      For a complete Hexapod circuit combine all these elements: the single ended
                      bi-phase slave output can be connected to the two halves of a HC139 and two
                      h-bridges each with independent reverse. This design gives you a complete
                      hexapod control circuit in TWO chips!

                      enjoy

                      wilf

                      <<hexapod1.gif>>


                      > -----Original Message-----
                      > From: Richard Piotter [SMTP:richfiles@...]
                      > Sent: Tuesday, September 26, 2000 7:50 PM
                      > To: BEAM
                      > Subject: Re: [beam] Hexapod
                      >
                      >
                      >
                      > Also, I have an aditional question. There is a slight, fraction of a
                      > second, delay between when an input enters an Nv and the output changes.
                      > With the two branches, there can be substancial delay, between loading
                      > and variations in component value. If I were to put an additional
                      > inverter at the end of one chain and cut the other, will the delay time
                      > of the second inverter switching after the first have an adverse effect
                      > on power coonsumption while the H-Bridge has the same signal for a split
                      > second (with no 139 installed). Is it a low enough effect to not
                      > adversely effect power consumption or the safety of the H-Bridges over
                      > an extended time?
                      >
                      >
                      > I think I'll do the PCB tommorow or the day after. Depends how bored I
                      > get and if I have enough dry transfers left for it.
                      >
                      >
                      >
                    • Wilf Rigter
                      Here is the detail on the biphase slave.
                      Message 10 of 10 , Sep 28, 2000
                        Here is the detail on the biphase slave.

                        <<biphase.gif>>
                      Your message has been successfully submitted and would be delivered to recipients shortly.