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

Re: Variable Voltage Self-adjusting Neuron

Expand Messages
  • Amit
    Will you use this on one of your bots?
    Message 1 of 7 , Jul 1, 2012
      Will you use this on one of your bots?

      --- In beam@yahoogroups.com, "connor_ramsey@..." <connor_ramsey@...> wrote:
      >
      > It's come to my attention that there is a particular type of diode
      > called a varicap diode, which acts as a capacitor in reverse bias, and
      > the capacitance is controlled by the voltage at the anode, hence the
      > name. It's also come to my attention that these are considerably easy to
      > attain. I recently pulled three out of a DVD player and didn't even know
      > it. What I had in mind was to construct a neuron consisting of an
      > inverter(duh), a resistor(again, duh), a varicap(d... wait, wuh?), and
      > three more resistors. While the resistor is tied to ground, it's also
      > tied between the two other resistors in a voltage divider setup, as well
      > as an extra, higher value resistor between the inverter input and the Nv
      > input.
      >
      > Whereas a regular neuron with fixed resistor/capacitor values will
      > change behavior when supply voltage varies, namely, a solar powered
      > walker will move slower as the storage cap empties, only in part because
      > voltage goes down, but more importantly because the ratio between
      > component values to Vcc value increases, causing each Nv to have a
      > longer delay. Now this isn't necessarily the case because any type 1 or
      > 2 solar engine cuts power before it gets that low, but wouldn't you love
      > for the robot to use all of the stored charge in the cap, rather than a
      > fraction of it, not only making it more efficient, but also longer lived
      > per charge? A type 3 solar engine accomplishes this, but the voltage
      > dissipates as the cap empties. So to make a control circuit that
      > maintains the same time constants even near "death", or during a power
      > spike, I want to implement the aforementioned circuit.
      >
      > At normal Vcc, the resistor and varicap assume normal values for proper
      > timing. The resistor actually assumes half of it's marked resistance
      > during an active process, because the voltage divider decreases the
      > charge difference across the resistor's terminals. However, as the
      > supply voltage drops, the charge across the resistor's terminals during
      > an active process increases, so that it takes longer for the charge to
      > drain. The fourth internal resistor should keep a fairly constant output
      > voltage. Also, the internal charge of the neuron determines the
      > capacitance of the varicap. As Vcc drops, the R1's resistance increases,
      > causing a higher voltage to build up at R2, causing the varicap's
      > capacitance to decrease during an active process. Or the opposite could
      > happen if Vcc increases beyond normal parameters, but that's not an
      > important characteristic for a solar application. Here's a diagram of
      > this circuit:
      >
      >
      >
      >
      > >-----||<|----+--R2--|>o----->
      > |
      > R1
      > |
      > +-R3---+---R4->
      >
      >
      > If anyone likes this idea, then go ahead and do something with it. It's
      > not necessarily designed for extra functionality, just to achieve higher
      > efficiency and to make a more voltage flexible design.
      > Hope you enjoy,
      >
      > Connor
      >
    • connor_ramsey@ymail.com
      ... First solar walker I build will have these.
      Message 2 of 7 , Jul 1, 2012
        --- In beam@yahoogroups.com, "Amit" <amitjones101@...> wrote:
        >
        > Will you use this on one of your bots?


        First solar walker I build will have these.
      Your message has been successfully submitted and would be delivered to recipients shortly.