need help for a pneumatic sequencer

[mescalinum]

hi all (long time no speak!)

I'm building a pneumatic walker, and I need now to make it move in some automatic way, but I cannot commute the pneumatic switches by the pneumatic actuator themselves, cause the construction is too crowded.

what could be an alternative to sequence the pneumatic switches states?

I have 3 switches to sequence, A, B, and C, and I figured out I need this sequence of states, for each time step t:

t | A B C
---------
1 | 1 0 1
2 | 1 0 0
3 | 1 1 0
4 | 0 1 0
5 | 0 0 0
6 | 0 0 1

(where I indicated with 0 switch to the left, and with 1 switch to the right; switch is never in middle position)

now the easy thing is that all 3 switches have the same period although with different phase offsets (so they can be actuated by some rotating gear + lever).

the complication is that switches B and C have a duty cycle of 33%, i.e. they stay in one state 1/3 of the cycle, then stay in the other state 2/3 of the cycle.

can you figure out a linkage that would achieve the above sequence?

Actually, I think the sequence can be simplified a bit, by doing 2 state at once, becoming:

t | A B C
---------
1 | 1 0 1
2 | 1 0 0
3 | 0 1 0
4 | 0 0 0

I was thinking to some linkage seen in akiyuki creations (like in

), but apart from taking quite some space, I fear it would not be strong enough to actuate pneumatic switches…

ideas?

[S.I]

What if you add an additional cylinder with some sort of resistence which only moves once the leg cylinders for the current step have moved completely and pressure is still rising?

[kieran]

Is nxt control an option? You can code any sequence you want then

[mescalinum]

What if you add an additional cylinder with some sort of resistence which only moves once the leg cylinders for the current step have moved completely and pressure is still rising?

hmm that's an interesting idea. in that case the extra cylinder would rise at the end of each state, so it should trigger a state transition.

we'll see… :)

[mescalinum]

Is nxt control an option? You can code any sequence you want then

yes, I happen to have one 8547 (and planning to buy another one or two!)

but the point is that this little thing would be much cooler if with the same single PF motor pumps air and does the sequence, don't you think? ;)

[Jetro]

Kevin Clague wrote an excellent tutorial on Pneumatic Sequencing and I'm sure it holds the answer to your question.

Unfortunately it is no longer on-line, but you can read the first part in Hispabrick Magazine 013 and we'll publish the second part in the next issue which will be out in August.

[mescalinum]

I think I have solved it!

(I'm afraid the video is a bit dark)

tomorrow I'll put all the pieces together

Edited July 4, 2012 by mescalinum

[mescalinum]

here it is:

unfortunately, crushed by hypergravity, it can't even stand on its own legs (it was designed on anotherplanet hehe)

seriously, after putting the sequencer, the motor, the battery pack, the structure that holds the joints it's just too thin, and now it doesn't walk properly (needs improvement)

[DLuders]

Cool pneumatic "octoped" walker! Do you have any of the 45590 "Technic, Axle Connector Double Flexible (Rubber)" to put on the ends of each leg? They may help with the "grip" on smooth tile floors....

[Burf2000]

Cool pneumatic "octoped" walker! Do you have any of the 45590 "Technic, Axle Connector Double Flexible (Rubber)" to put on the ends of each leg? They may help with the "grip" on smooth tile floors....

I would love to see how you built the sequencer!

[mescalinum]

I would love to see how you built the sequencer!

it's pretty basic. it's the following mechanism serially repeated (3 times in my case).

synchronizing the opening and closing time of all the valves might require some patience :)

[Jetro]

unfortunately, crushed by hypergravity, it can't even stand on its own legs (it was designed on anotherplanet hehe)

This is one of the reasons many Pneumatic walkers have an external air feed.

Other than your structural issues, does the walker behave as expected/designed?

[mescalinum]

This is one of the reasons many Pneumatic walkers have an external air feed.

Other than your structural issues, does the walker behave as expected/designed?

yes, the sequencing was OK, and when lifted from the ground, it behave fine

the problem was a weakness in the structure, that can be seen in the following photo. each leg is attached to only one liftarm, causing it to bend

btw, I tried another implementation of the walker, this time pneumatic only, and it is again a failure.

once again it works in no-gravity, but in the real usage test, the pneumatic cylinders can't withstand the weight of the structure (and it's even lighter than before!) and the tubings continuously pop. check the video for details:

[kieran]

Which tube are you using? do you have any idea how much PSI you get to before the tube blows off? I know when I was playing with LPE I had this issue. I found that cutting short lengths of PVC tube with an ID just bigger than the pneumatic hose over the nipple would hold it on. the idea is you slide the PVC tube over the pneumatic house once you have pushed it onto the nipple. allowed me to hole more pressure before a blow off occurred.

[Jetro]

Looks like the problem may be the hoses them themselves. if you can measure the pressure then please do, for example using the 64068 Pneumatic Pressure Gauge - Manometer:

Another thing you may try is fastening the tubes to the inlets using some string.

[Brickthus]

hi all (long time no speak!)

I'm building a pneumatic walker, and I need now to make it move in some automatic way, but I cannot commute the pneumatic switches by the pneumatic actuator themselves, cause the construction is too crowded.

what could be an alternative to sequence the pneumatic switches states?

I have 3 switches to sequence, A, B, and C, and I figured out I need this sequence of states, for each time step t:

t | A B C
---------
1 | 1 0 1
2 | 1 0 0
3 | 1 1 0
4 | 0 1 0
5 | 0 0 0
6 | 0 0 1

(where I indicated with 0 switch to the left, and with 1 switch to the right; switch is never in middle position)

now the easy thing is that all 3 switches have the same period although with different phase offsets (so they can be actuated by some rotating gear + lever).

the complication is that switches B and C have a duty cycle of 33%, i.e. they stay in one state 1/3 of the cycle, then stay in the other state 2/3 of the cycle.

can you figure out a linkage that would achieve the above sequence?

Actually, I think the sequence can be simplified a bit, by doing 2 state at once, becoming:

t | A B C
---------
1 | 1 0 1
2 | 1 0 0
3 | 0 1 0
4 | 0 0 0

I was thinking to some linkage seen in akiyuki creations (like in

), but apart from taking quite some space, I fear it would not be strong enough to actuate pneumatic switches…

ideas?

Take a look at the circuit and sequence for my robot. A simplification, by taking out one of the four functions, should get you quite close to what you need.

The two building blocks are a gray-code flip-flop and an exclusive-OR gate. My robot extends the flip-flop loop with a third stage in the sequence. The exclusive-OR output cylinder moves a single switch that dictates which of the two (or two sets of) functions in the flip-flop are allowed to get an air supply to allow them to change state. The two flip-flop functions (or the last stage in each set) each have one half of the exclusive-OR logic, one with 1 switch and the other with 4 switches.

Mark