Nu Neurons

From BEAM Robotics Wiki

Jump to: navigation, search

We apologize for the need to display ads on this wiki. But somehow we must pay for hosting.


NOTE: This article originally appeared as part of BEAM From the Ground Up", a part of The BEAM reference Library which is made possible by long time BEAM supporter Solarbotics. Original material by: Bruce Robinson with additional material from Wilf Rigter. Originally compiled and editted by Eric Seale with additional editting by: Clifford Boerema. This work is licensed under a Creative Commons License

Contents


The basic Nv neuron responds to a rising input by switching its normally high output to a low one for a varying amount of time. Given an input that goes from high to low to high again, on the low-to-high transition, the output will go from high to low. After a delay (determined by the resistor and capacitor values), the output will switch back to high. This allows you to chain Nv neuron together and have a signal travel slowly along the chain.

The basic Nu neuron responds to a prolonged change in input. If a high signal is applied to the input, after a delay, the output will switch from high to low and will remain low as long as the input stays high.

One essential difference is that the Nv responds immediately to an input, and sends the output for a time duration -- the delay occurs AFTER the output is sent. The Nu responds to an input after a delay and sends the output continuously -- the delay occurs BEFORE the output is sent.

The basic Nu neuronThe basic Nv neuron
Image:Nu_Schematic.gifImage:Nv_Schematic.gif
* Off condition is a high output. * Off condition is a high output.
* A high input causes the neuron to fire -- after a delay. The neuron remains on as long as the input stays high (delay, then "on", stays "on"). * A rising transition at the input causes the neuron to fire (switch on) instantly. After a delay, the neuron switches off ("on" first, then a delay, then "off").

{As far as schematics are concerned,} the big difference between the two is that the capacitor and resistor are swapped. Each one also has a variation where the resistor (Nv) or capacitor (Nu) is connected to Vcc instead of ground.


Variants

Although the basic diagram shows one input through one resistor (the simplest form), the Nu neuron is potentially a very complex device. You can create an Nu with multiple inputs by giving each input its own resistor, often with different values. You can also put a diode in series with each resistor to determine under what conditions the input will have an effect (to create separate "on" delays and "off" delays, so the cap will charge much more rapidly then it discharges, or vice versa). And you can have inhibitory inputs that tend to cancel the effects of the other inputs. Thus, whether an Nu fires or not depends on the combined effects of its inputs. In fact, you will sometimes have a tough time actually identifying a complex Nu neuron in a circuit.

Another major "flavor" of the Nu neuron involves connecting the capacitor to Vcc instead of ground. In this case, input and output conditions are the reverse of what is described above (i.e., "off" is low, a low input causes the neuron to fire).

Tilden's patent itself identifies three variants of the Nu:

Basic Nu
Basic Nu
Slow to Fire / Fast Reset Nu
Slow to Fire / Fast Reset Nu
Fast to Fire / Slow Reset Nu
Fast to Fire / Slow Reset Nu



Slaving Nv neurons to Nu Neurons

Whenever I've slaved an Nv neuron to an Nu neuron, I've used a resistor between the Nu output and the Nv bias point (where the cap, resistor, and inverter input all join). You can do this several ways ...

  1. The connecting resistor is much larger (at least 4 times the size) than the regular Nv resistor. This will influence the Nv timing a little, increasing or decreasing the delay depending on the Nu output.
  2. The connecting resistor is approximately the same size as the regular Nv resistor. This will make the Nu the predominant factor in the Nv timing.
  3. Use only the connecting resistor and eliminate the regular Nv resistor. This means the Nu will switch the Nv between a grounded neuron and a positively connected neuron.
  4. Use a diode between the connecting resistor and the bias point. Depending on the output level of the Nu and the direction of the diode, this will either speed up or slow down the Nv delay.
  5. Use a diode between the Nu and the Nv bias point. The Nu will then either kill a process on the Nv, or inject a new one. This is the original Tilden startup circuit.
  6. The Nu bias point can be used as a ramping analog source to influence an Nv. A typical example using a diode is shown below. Note that the R of the Nu must be much smaller than the R of the Nv. The same methods of coupling as described in 1. to 5. can be used.
  7. In suspended bicores, the Nvs are not referenced to ground or Vcc however an Nu (including 6.) can be connected to the middle of the suspended resistor or directly influence one or both bias points.

The key to understanding the way the suspended bicore is influenced by an external voltage / current source is to recognize that the null (minimum influence) voltage for a suspended bicore is Vcc / 2. This means if Vcc / 2 is connected to the midpoint of the resistor in a HC suspended bicore with equal capacitors, there is zero effect.


For more information...

Bruce's original postings are still saved in the alt-beam archives > HERE <, and in followups. He also touches on them a bit > HERE <. Meanwhile, the discussion of Nu / Nv interconnections are from one of Bruce's postings > HERE <, and a reply by Wilf > HERE <.


This wiki is sponsored and hosted by Interactive Matter
Personal tools
Ads to finance this wiki