Subject:
|
Re: Brickfest Pneumatic Master and new Pneumatic Gate Circuits
|
Newsgroups:
|
lugnet.technic
|
Date:
|
Sun, 28 Aug 2005 20:34:36 GMT
|
Viewed:
|
4830 times
|
| |
|
|
In lugnet.technic, Mark Bellis wrote:
> In lugnet.technic, Kevin L. Clague wrote:
> > Mark,
> > I have studied and understand your 3+3 AND gate. A few years ago I made a
> > simpler version of the same thing. My design replaces your six switches with
> > one switch and mechanical linkages. Here is a picture of this AND gate:
> >
> > http://www.brickshelf.com/gallery/kclague/Computing/p5280034.jpg
> >
> > In my 1 AND gate, A and B each control a piston. The two pistons are
> > connected base to base, not unlike the pneumatic boom controls in the 8421
> > technic crane. One piston is afixed to a frame that also contains a single
> > switch. The center of the switch is hooked to the pressure source. Given two
> > pistons hooked together in this fashion the total amount of expansion due to
> > both pistons can be:
> >
> > A B Sum Length
> > C C 0
> > C X X 1
> > X C X 1
> > X X XX 2
> >
> > If you extend the handle of the switch by inserting an axle connector and a #2
> > axle between the switch handle, and the #1 Angle piece you use to connect to the
> > piston, the extension length of the piston is the same length as the throw of
> > the switch. The linkage between the pistons and the switch contains rubber
> > bands. When both pistons are expanded, the switch handle is pressed away from
> > the pistons. When one and only one of the pistons is expanded, the switch is
> > pulled toward the pistons. When both pistons are contracted, the rubber bands
> > stretch in an attempt to further pull the switch handle closer to the pistons.
> >
> > By moving the switch closer to the pistons, you can easily make an OR gate.
> >
> > http://www.brickshelf.com/gallery/kclague/Computing/p5280036.jpg
> >
> > This is the most cost effect AND gate that I know of that meets all your
> > requirements. While too fragile to be put in-situ as a muscle piston, for pure
> > computation it has the smallest pneumatic part count.
> >
> > I don't know that we'll ever see eye to eye on how to count pneumatic parts
> > within our gates, but it does not matter to me. Whatever pneumatic circuit I
> > make, I will tend to use whatever gate technology that is cheapest and meets the
> > need.
[SNIP]
> > Kev
>
> Yes, I can see how the 2-cylinder design is more cost-effective, particularly
> since a Backhoe has 10 cylinders and only 7 switches. That allows more
> cylinders to be used for AND and OR gates, saving the switches for XOR gates.
Yes, it is *very* bulky and fragile, but it is a computing solution applicable
in some situations.
>
> I've done some more study in order to investigate the 3-input gate, using the
> 4-switch module in multiple. This has shown me how the original 4-switch module
> from your sketch is an AND gate in its own right, as long as the differential
> pressure input is one input and the switch position is the other input. A
> reverser between stages can invert the differential pressure input, making
> endless possibilities for the logic function of the system, by allowing the
> output to be high in 1, 3, 5, or 7 of the eight combinations of three inputs.
Very cool. I had not realized this addition to the 1+4 could be that flexible.
Polarity reversors are so cool. When switched manually they are like an XOR
gate with one input controlled by non-pneumatic means. When controlled by a
piston, it functions as a full fledged XOR gate.
I have adopted your design because it is much more compact than mine. Thanks
for sharing.
> Seeing the 4-switch module that way, and using it in multiple, makes for a gate
> without any extra propagation delays due to daisy-chaining. It's also more
> cost-effective, since only four switches are required. For a gate with two
> switch position inputs (rather than one switch position and one differential
> pressure), one of the switches driven by "A" can be trimmed from my diagram,
> making a 1+4++4 gate.
Nice optimization.
One plus for your 3+3 gate is that you only release (leak?) pressure between one
switch and the piston that drives it.
In the 4-switch version, if you have four gates in series, you end up
depresurizing through all four switches (their hoses) and the piston. Life is
full of trade offs.
>
> I still like the elegance of my 3+3 design, especially because it's symmetrical,
> but the most elegant solution is rarely the sleekest! In pneumatics there is
> more to be said for reducing parts count because the parts are so expensive.
> That's where 1+4++4 scores over 3+3++4.
For me, they both are viable solutions, and each has advantages and
disadvantages. We then get to pick and choose the right technique for the
situation at hand. More choices are better!
> Overall this means that a set of combinational logic can be achieved with
> 4-switch AND modules and 4-switch reversers such that there is only a single
> propagation delay between states, which is good for machines whose timing is
> critical, such as your walkers.
LOL! SSC1 is the only walker I've made with these kinds of gates! In SSC1 I
decided to give the old tried and true central timing loop (with slave pistons
as muscles) a try, because I wrote it off so long ago. After finishing the
design, and hacking (I mean modifying) it to make it work, I realized once again
that central timing loops are not the way to go.
Other than SSC1, all my walkers use muscle pistons as timing pistons, and I
exclusively use single switch AND gates to compose the sequencer.
Previous to SSC1, I had to have two non-muscle timing pistons because of
problems of muscle pistons depressurizing during walking. I had to have the the
inputs that were not always differential drive a non-weight bearing piston, that
drove a single switch which made things differential again (I think you called
this buffering).
While brainstorming issues about SSClagorpion, in particular the legs collapsing
when the circuit is depressurized, Steve Hassenplug suggested that I put in a
mechanical leg lock, like those on folding chairs and tables. This is a great
idea!
On the way to BrickFest 2005, I realized that this leg lock "fixed" the problem
with the leg lift muscles not always being differential, so I can throw the
buffer piston+switch away in all leg lock designs. This eliminates four piston
transitions out of a 10 transition sequence. I hope it will speed things up.
>
> Mark
SSC1 is way too huge for my taste. As a counter to it, I spent this weekend
working on small piston leg lift design. Each leg has four small pistons (all
tied to together as one larger piston) and two switches. It has a pretty little
leg lock mechanism. Also two pistons expand, while the others contract, which
should reduce inherint assymetry of expansion vs. contraction rates and
strength.
I look forward to hearing more about your adventures in pneumatics.
Kev
|
|
Message is in Reply To:
32 Messages in This Thread:
  
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
