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

Op-Amp Bicore [was FORMAL CHALLANGE]

Expand Messages
  • Bruce Robinson
    ... OK, Adam, here you are. At first glance, just another bicore variation, but with a higher component count. It seems to reverse the BEAM drive toward
    Message 1 of 11 , Jul 19, 2003
      Adam wrote:
      >
      > I CHALLANGE YOU (EVERYONE, EVEN NEWBIES) TO COME UP WITH NEW CIRCUITS

      OK, Adam, here you are. At first glance, just another bicore variation,
      but with a higher component count. It seems to reverse the BEAM drive
      toward simplicity (i.e., more parts, no apparent increase in
      functionality). But wait ...

      What I've done is use an LM324 op-amp to build a bicore. (Two bicores
      will fit on one chip.) Why use an op-amp? Because you can vary the
      trigger threshold to change the bicore timing. It turns out that this
      can have some advantages over the usual practice of varying the R-C time
      constant.

      The first attached image shows the pinout diagram for the basic bicore.
      The second image is the schematic for the same circuit, and I'll use
      that for a brief explanation of how the circuit works. You should
      recognize a basic bicore configuration -- two IC devices acting like
      inverters, each one with a resistor and capacitor connected to the input
      (R1 & C1, R2 & C2). What's different is the extra input to each
      "inverter" -- this sets the trigger threshold. In the upper "neuron",
      input VT is a variable threshold that controls the bicore frequency. In
      the lower "neuron", I've set the threshold to a constant -- 1/2 Vcc --
      using resistors R3 & R4. The smaller values for R2 and C2 make a short,
      constant-width pulse. You can replace R3 & R4 with a trimpot (or other
      circuit) to give the lower neuron a variable delay as well.

      The values I chose are just about right for driving an LED so you can
      play with the circuit on a breadboard. Raising the value of VT means it
      will take less time for the upper neuron to reach its threshold, so the
      bicore will speed up. Lowering the value of VT makes the neuron take
      longer to time out, so the bicore slows down. You can test this effect
      by connecting a trimpot between Vcc and GND, with the center lead
      connected to VT.

      If your goal is to set the frequency of a bicore using a trimpot, there
      isn't much point in using this circuit -- just make R1 and/or R2
      variable instead, and use inverters instead of op-amps. But if you want
      to vary the bicore frequency using a circuit, my design has some
      advantages.

      The third attached image shows just one way to use this circuit. Here
      I've taken the basic op-amp bicore and connected an R-C pair (R6, C3) to
      the VT input. Pushing the pushbutton will drive the inverter output low;
      when the pushbutton is released, the inverter output will go high,
      charging capacitor C3. This will cause the bicore to pulse very quickly.
      As the charge on C3 drains off through R6, the bicore will slow down.
      I've shown an inverter connected to the output, driving an LED.
      (Technically you can drive a low-power LED directly from the op-amp
      output, but keep in mind that the maximum output voltage from the LM324
      is 1.5 volts below Vcc. I added the inverter to the output to keep the
      LED nice and bright.) The effect of this circuit is to cause the LED to
      flash very quickly when the pushbutton is pushed and released, and then
      to gradually slow down.

      If you attach the same circuit to a motor driver of some kind, the
      result will be a PWM speed control that causes the motor to spin quickly
      at first, and then slow down gradually to a crawl. Sounds a bit like
      Wilf's recent design, doesn't it?

      Other applications come to mind. Use this circuit as the slave in a
      master-slave bicore. Make both thresholds variable and you can change
      the timing relationship between the master and the slave.

      Or create a chain of regular Nv neurons and connect each output to one
      of these bicores, using C3 and R6. As an impulse travels through the Nv
      chain, each bicore will be triggered in turn, causing some interesting
      visual effects. Try it with a 3-neuron chain, or a tricore, and hook
      each bicore output to a 3-colour LED to get interesting multi-colour
      visual patterns (Jenny? Tom?).

      This is a nice little circuit to play around with on the breadboard to
      see just what you can do with it.

      As the maestro says, enjoy.

      Bruce
    • Martin Jay McKee
      I ve been working with just that but I ve been tying the two op-amp inverting inputs together, changing the threshold of both at the same time. That way you
      Message 2 of 11 , Jul 19, 2003
        I've been working with just that but I've been tying the two op-amp inverting inputs together, changing the threshold of both at the same time.  That way you keep the same ratio's between the pulse lengths.  It does make PWM more difficult (you have to add a stage that doesn't have a changing threshold) but it should allow for some different behaviors.  I wonder if they were combined...
         
        Martin Jay McKee 
      • Bruce Robinson
        ... Yep, that s one of the many possible variations, Martin. If you mean combining the decreasing frequency front end with your common threshold concept, I
        Message 3 of 11 , Jul 19, 2003
          > Martin Jay McKee wrote:
          >
          > I've been working with just that but I've been
          > tying the two op-amp inverting inputs together,
          > changing the threshold of both at the same time.
          > That way you keep the same ratio's between the
          > pulse lengths. It does make PWM more difficult
          > (you have to add a stage that doesn't have a
          > changing threshold) but it should allow for
          > some different behaviors. I wonder if they
          > were combined...

          Yep, that's one of the many possible variations, Martin. If you mean
          combining the decreasing frequency front end with your common threshold
          concept, I suspect you'd get an LED that would give very rapid, short
          blinks at first; the LED would gradually stay lit longer and blink more
          slowly. Add to THAT one of the 3-colour LED's and things should get even
          more interesting.

          If you really want to try something weird, make a suspended bicore with
          both thresholds independently variable. Then play around with the two
          thresholds. I haven't figured out all the permutations of this one yet.

          Bruce
        • wilf rigter
          Amazing Bruce! The opamp grounded bicore. As a design concept, a simple yet completely overlooked beam circuit. Some years ago I designed the perfect
          Message 4 of 11 , Jul 20, 2003
            Amazing Bruce!
             
            The opamp grounded bicore. As a design concept,
            simple yet completely overlooked beam circuit. 
             
            Some years ago I designed the "perfect" suspended bicore
            using opamps.  As the theory indicated, the more perfect the
            suspended bicore, the less likely it would oscillate and after
            proving, that I ended my investigation (I mean what's the point
            of designing oscillator that doesn't oscillate 8)
             
            I have lately posted some analog comparator based circuits
            including the Rainbow LED, and motor controller circuits both
            of which use PWM.   The LM339 is an attractive alternative to
            the LM324 for these pulse circuits although it requires a pullup
            resistor for each output.  Come to think of it, I did post a rather
            similar grounded bicore circuit for joystick reversing  speed
            control some time ago but it was buried deep inside that circuit
            (as shown in the attached) and not presented as a general
            purpose beam solution.
             
            Grounded bicores don't have such problems with perfection
            and have two separate "Nv neurons" with independent time
            constants that can be electronically controlled.
             
            With the exception perhaps of the suspended bicore, 
            designing opamp versions of other beam circuits that
            previously used 74HC14s or 7HC240 inverters (microcores,
            monocores, heads) opens the door for more predictable and
            repeatable beam circuit operation, finally allowing for time
            constants to be based on calculated component  values.
            In addition, the voltage controlled thresholds add a new
            dimension to sensor inputs.
             
            With many beam solar engines operating at 2.5V, it should be
            noted that the LM324 requires a minimum 3V supply but the
            LM339 can operate down to 2V. Both require about 1mA of
            supply current. Maximum operating voltage for both is >30V
            allowing for automitive BEAM  applications that run directly from
            the car battery. These devices also meet another important
            beam requirement of ready availability and low cost.
             
            For special upower applications, there are plug in CMOS
            versions of these chips that operate below 1V and draw
            just a few uA of supply current.  
             
             
             
            I also have some comments on Bruce's description of operation: 
             
            Bruce said,
             
            "Other applications come to mind. Use this circuit as the slave in a
            master-slave bicore. Make both thresholds variable and you can change
            the timing relationship between the master and the slave".
             
            The nature of the slave bicore is that, unlike a grounded bicore,
            the two Nv neurons are not independent from each other.  Both
            Nvs always fire at the same time depending on which has the
            shorter time constant.  That was demonstrated by Mark Tilden
            by removing one coupling resistor (ie infinite resistance) to show
            that the circuit continued to operate i.e. was robust. This circuit
            behaviour is even easier to demonstrate with an opamp slave
            bicore as the control voltage adjusted on one Nv, its time constant
            can be made shorter but never longer than the second Nv time
            constant. In fact, if the second Nv voltage control is near Vcc/2,
            then as you attempt to lengthen the first Nv TC beyond 50%, it
            starts to shorten again! Can you see why? This little side trip into
            slave bicore behaviour may be useful because otherwise
            experimenting with this circuit could lead to surprising, if not
            frustrating results.     
             
            Bruce said,
             
            "Or create a chain of regular Nv neurons and connect each output to one
            of these bicores, using C3 and R6. As an impulse travels through the Nv
            chain, each bicore will be triggered in turn, causing some interesting
            visual effects. Try it with a 3-neuron chain, or a tricore, and hook
            each bicore output to a 3-colour LED to get interesting multi-colour
            visual patterns (Jenny? Tom?).
             
            This can be done entirely with Opamp Nv's but a microcore, etc. 
            requires inverters with hysteresis (Schmitt triggers) as shown in
            the attached Opamp microcore circuit. The common control
            voltage from the 10K pot adjusts the duration of all Nv together
            each Nv threshold input could be adjusted with a separate voltage.
             
            And Bruce ended his wonderful OpNv introduction urging:

            "This is a nice little circuit to play around with on the breadboard to
            see just what you can do with it."
            Also check out manufacturer's application notes for opamps,
            comparators and other linear devices.
             
            In the past, when there were fewer special purpose analog
            devices let alone DSPs or microcontrollers, there were thick
            books of application notes for opamps available for free from
            National, Fairchild, Motorola  and others. These were linear
            "cookbooks",  complete with lists of ingredients and equations
            for copying or adapting the application note circuits into
            your own designs. 
             
            These elegant and always simple application circuits also
            inspired generations of analog designers, who adapted,
            crossfertilized and evolved those simple circuit ideas
            into many of today's specialized analog building blocks. 
             
            I have several 20 year old (but timeless) well thumbed copies
            of these linear applications books on my shelf.  Afaik many of
            these are still available on request but now probably only on CD.
             
            Thanks Bruce!
             
            enjoy  
             
            wilf
             
            ----- Original Message -----
            Sent: Saturday, July 19, 2003 7:14 PM
            Subject: [beam] Op-Amp Bicore [was FORMAL CHALLANGE]

            Adam wrote:
            >
            > I CHALLANGE YOU (EVERYONE, EVEN NEWBIES) TO COME UP WITH NEW CIRCUITS

            OK, Adam, here you are. At first glance, just another bicore variation,
            but with a higher component count. It seems to reverse the BEAM drive
            toward simplicity (i.e., more parts, no apparent increase in
            functionality). But wait ...

            What I've done is use an LM324 op-amp to build a bicore. (Two bicores
            will fit on one chip.) Why use an op-amp? Because you can vary the
            trigger threshold to change the bicore timing. It turns out that this
            can have some advantages over the usual practice of varying the R-C time
            constant.
          • Greybeard
            ... wrote: (Jenny? Tom?). Huh, what? Sorry, Bruce, I was going over the schematics for Hider, hoping to implement some of your suggested changes and work in
            Message 5 of 11 , Jul 20, 2003
              --- In beam@yahoogroups.com, Bruce Robinson <Bruce_Robinson@t...>
              wrote:

              (Jenny? Tom?).

              Huh, what? Sorry, Bruce, I was going over the schematics for Hider,
              hoping to implement some of your suggested changes and work in the
              color vision. But, let's see now, where can I put one of these
              interesting Op Amp bicores?

              Colored light displays are Jenny's fave. :)

              Keep BEAMing and dreaming
              Tom

              PS the second version of the color eye is on paper, on hold until I
              turn up an IR blocking filter, because the current version only
              works under fluorescent light, which must be lower in IR than
              incandescent or sunlight. Anybody got some ideas (that cost less
              than $20 each!!) for IR blocking? Thanks in advance.
            • Bruce Robinson
              Hi, Wilf. ... Do you recall listening in on a discussion between M.T., Janet Frigo, and a few others back at WCRG 2000? Janet kept saying that BEAM wasn t
              Message 6 of 11 , Jul 20, 2003
                Hi, Wilf.

                > ... The opamp grounded bicore. As a design concept, a
                > simple yet completely overlooked beam circuit.

                Do you recall listening in on a discussion between M.T., Janet Frigo,
                and a few others back at WCRG 2000? Janet kept saying that BEAM wasn't
                paying enough attention to op-amps, and I kinda took that to heart. Been
                using them ever since.

                > With the exception perhaps of the suspended bicore,
                > designing opamp versions of other beam circuits that
                > previously used 74HC14s or 7HC240 inverters (microcores,
                > monocores, heads) opens the door for more predictable and
                > repeatable beam circuit operation, finally allowing for time
                > constants to be based on calculated component values.
                > In addition, the voltage controlled thresholds add a new
                > dimension to sensor inputs.

                In fact, I was trying to solve a specific problem with my declining
                frequency pulsing neuron, which gradually raised the "pulldown" voltage
                in a bicore neuron to slow the pulse rate down. As the pulldown voltage
                approached the inverter threshold, the circuit became so sensitive to
                noise that it would start to pulse in synch with any other bicore that
                was on the same chip -- or even on an adjacent chip.

                The op-amp design allows me to vary the threshold to slow down the
                bicore, so I can stay away from that noise-sensitive area on the
                discharge curve. In theory, anyway. I have yet to try running two
                bicores on one op-amp chip.

                > Also check out manufacturer's application notes for opamps,
                > comparators and other linear devices.
                >
                > In the past, when there were fewer special purpose analog
                > devices let alone DSPs or microcontrollers, there were thick
                > books of application notes for opamps available for free from
                > National, Fairchild, Motorola and others. These were linear
                > "cookbooks", complete with lists of ingredients and equations
                > for copying or adapting the application note circuits into
                > your own designs.

                In fact, most of my op-amp applications are based on circuits suggested
                in the present-day datasheets. These are nowhere near as complete as the
                books you mentioned, but I've noticed that different manufactureres list
                different types of applications in their datasheets. By collecting
                several LM324 datasheets, I've managed to get a pretty good selection of
                useful circuits.

                I suppose I ought to draw up a general purpose op-amp Nv neuron.

                Bruce
              • Wilf Rigter
                Hi Bruce, ... Yes, I recall it well, right after my lecture during the Q&A, when M.T. asked some questions (and then got on his lecture box). Turned into a
                Message 7 of 11 , Jul 20, 2003
                  Hi Bruce,

                  > Do you recall listening in on a discussion between M.T., Janet Frigo,
                  > and a few others back at WCRG 2000? Janet kept saying that BEAM wasn't
                  > paying enough attention to op-amps, and I kinda took that to heart. Been
                  > using them ever since.

                  Yes, I recall it well, right after my lecture during the Q&A, when M.T.
                  asked some questions (and then got on his lecture box). Turned into a great
                  free for all with lot's of ideas tossed in the ring. I had the feeling that
                  some of the answers about Nv circuit analysis really hit home.

                  wilf



                  ---
                  Outgoing mail is certified Virus Free.
                  Checked by AVG anti-virus system (http://www.grisoft.com).
                  Version: 6.0.500 / Virus Database: 298 - Release Date: 7/10/2003
                • Martin Jay McKee
                  ... Already done, I built up a 6 neuron core a while back, and varied the frequency with a LDR (voltage divider arrangement). I have yet to build a bot with
                  Message 8 of 11 , Jul 20, 2003
                    >>Wilf said
                    >>This can be done entirely with Opamp Nv's but a microcore, etc. 
                    >>requires inverters with hysteresis (Schmitt triggers) as shown in
                    >>the attached Opamp microcore circuit.
                     
                    Already done,  I built up a 6 neuron core a while back, and varied the frequency with a LDR (voltage divider arrangement).  I have yet to build a bot with an op-amp based brain but they are so easy to influence that I've yet to decide what all I want avaliable in the way of sensors.  One idea that I've been playing with is a current to voltage converter that will feed the motor current back to the threshold input, extending the pulse width if current rises.  I believe I started with a ten to one ratio (1M to 10M ohm) for the hysteresis divider but I found that it can be almost anything without any noticable change in the behavior.  I replaced  the fixed resistors with a pot and was able to vary it to just barely at the ends, and it all ran smoothly. 
                     
                    Old standby op-amps (i.e. LM324) do use slightly more power but I've been working with the uPower versions, small caps, and large resistor values and I've been able to get a microcore to run stably at around 30 uA.  Since I've come to a bit of a stand still on my lizard robot I'm working on twin three motor walkers to test different brains in.  I'll pass on anything I learn.
                     
                    Martin Jay McKee
                  • Wilf Rigter
                    Good on you Martin! Can you post your opamp hexcore circuit? Just curious about the PNC, etc. Btw, the opamp tricore for the rainbow LED application Bruce
                    Message 9 of 11 , Jul 20, 2003
                      Good on you Martin! Can you post your opamp hexcore circuit? Just curious
                      about the PNC, etc.

                      Btw, the opamp tricore for the rainbow LED application Bruce mentioned does
                      not need a PNC and neither should it suffer from hypersaturation because of
                      the limited bandwidth of these opamps.

                      Exciting times ahead as we port more beam circuits over to opamp technology.
                      With opamps you can do true neural networks, seemlessly glued to sensors and
                      beam circuits.

                      regards

                      wilf

                      ----- Original Message -----
                      From: "Martin Jay McKee" <MartinJayMcKee@...>
                      To: <beam@yahoogroups.com>
                      Sent: Sunday, July 20, 2003 7:59 PM
                      Subject: RE: [beam] Op-Amp Bicore [was FORMAL CHALLANGE]


                      > >>Wilf said
                      > >>This can be done entirely with Opamp Nv's but a microcore, etc.
                      > >>requires inverters with hysteresis (Schmitt triggers) as shown in
                      > >>the attached Opamp microcore circuit.
                      >
                      > Already done, I built up a 6 neuron core a while back, and varied the
                      > frequency with a LDR (voltage divider arrangement).


                      ---
                      Outgoing mail is certified Virus Free.
                      Checked by AVG anti-virus system (http://www.grisoft.com).
                      Version: 6.0.500 / Virus Database: 298 - Release Date: 7/10/2003
                    • Wilf Rigter
                      As Bruce suggested you can download the data sheets such as http://www.national.com/ds.cgi/LM/LM124.pdf and find lots of application circuits. Note how similar
                      Message 10 of 11 , Jul 20, 2003
                        As Bruce suggested you can download the data sheets such as

                        http://www.national.com/ds.cgi/LM/LM124.pdf

                        and find lots of application circuits. Note how similar these are to many
                        beam circuits.

                        wilf

                        ----- Original Message -----
                        From: "Bruce Robinson" <Bruce_Robinson@...>
                        To: <beam@yahoogroups.com>
                        Sent: Sunday, July 20, 2003 7:08 PM
                        Subject: Re: [beam] Op-Amp Bicore [was FORMAL CHALLANGE]


                        > Hi, Wilf.
                        > In fact, most of my op-amp applications are based on circuits suggested
                        > in the present-day datasheets.


                        ---
                        Outgoing mail is certified Virus Free.
                        Checked by AVG anti-virus system (http://www.grisoft.com).
                        Version: 6.0.500 / Virus Database: 298 - Release Date: 7/10/2003
                      • Martin Jay McKee
                        ... Not much to post actually. It s been several weeks and I can t for the life of me find the schematic I used. I do know however that I wasn t using a PNC.
                        Message 11 of 11 , Jul 21, 2003
                          >>Wilf said
                          >> Can you post your opamp hexcore circuit? Just curious
                          >> about the PNC,
                          etc.

                          Not much to post actually.  It's been several weeks and I can't for the life of me find the schematic I used.  I do know however that I wasn't using a PNC.  I was manually injecting patterns and playing around to see how well it would hold it.  All in all it was a stable platform.  Since I'm working on the twin three motor walkers I'm going to be rebuilding with a proper PNC, I'll post that when I get it finished.
                           
                          Martin Jay McKee
                        Your message has been successfully submitted and would be delivered to recipients shortly.