OK, all those that have had their robots arrive at inspection at 129.75 pounds, raise your hands!!!

An awful lot of teams spend an amazing amount of time carving out every extra ounce from their robots to allow them to just barely fit within the weight limits of the competition. Just take a look at the number of machines peppered with “lightening holes” that were obviously added in during the last few days of the build phase (or even at the competition site). So, if there is a way to avoid using a redundant sensors and saving four ounces on the robot, there are a lot of folks that will be interested in doing it.

Using one pressure sensor to detect when the system pressure drops below 110 psi, and firing the compressor for “X seconds” (where “X” is the time required for the compressor to increase system pressure from 110 psi to 120 psi) when the pressure drops below the set point is an acceptable algorithm.

The key is to have the compressor deliver a bolus of air for a fixed period of time – the shortest time sufficient to recharge the system under ideal conditions (as determined experimentally under optimal conditions with a fresh battery and no stress on the system), and not to try to extend the recharge period based on perceived system efficiency. Instead, just repeat the cycle as frequently as necessary – once each time the system pressure drops below 110 psi.

If, due to decreased battery output or other efficiency factors, the system is only charged to 115 psi instead of 120, we don’t really care – that is close enough, since what we are really concerned with is that the high pressure side of the system has enough pressure to recharge the low pressure (60 psi) side as air is used by the cylinders (note: 110 is not a magic number – this set point can be anywhere above 60 psi; based on how extensively your robot uses the pneumatics, adjust as necessary to ensure that pressure is maintained in the high pressure portion of the system to provide a constant reserve pressure to the low pressure side while not causing the pump to cycle continuously). If control system/battery/compressor efficiency drops, the recharge will just approach, but not meet or exceed the 120 psi limit. There is no concern about overcharging the accumulators.

For this set up to fail and cause any sort of unexpected energy release, there has to be at least a quadruple failure. The compressor needs to fail-on for an extended period of time to start to build up pressure in the system. The relief valve, set to 120 psi, has to fail to release at the design pressure. The compressor bypass seals need to fail to release at the overpressure level (if I remember the spec sheets correctly, around 140 psi). And finally, some element of the physical system (tubing, cylinder, or accumulator) needs to mechanically fail and release the pressure (assuming the compressor, which has a design output of 120 psi, is even capable of generating anything close to the 250 psi failure ratings of any of these components before it has a pressure lock and stalls out).

Although a constant-use system would probably not be designed this way, this approach is suitable for the low duty cycle, short lifetime applications like our competition robots.

-dave


p.s. If the boiler explosion referenced is the one that occurred at the Cuyahoga County Fair in July 2001, it actually was not caused by a safety valve failure. The boiler in question was in very poor condition, and had apparently been improperly maintained for several years. As a result, the crown sheet above the firebox was severely corroded. The corrosion in several places near the crown staybolts had reduced the crown sheet from a design thickness of 0.375” to less than 0.125”, and in some locations to less than 0.087”. Although it was later determined that the safety valve was inoperable (it was rusted shut, along with the inspection port and fusable plug), this explosion occurred because of a mechanical failure at a pressure well below the designed operating pressure of the boiler.

Letter excerpts from John D. Payton (Director, certified Boiler Engineers for the Commonwealth of Pennsylvania), who conducted an inspection of the damaged boiler at the request of County Sheriff Neil Hassinger and Chief Dean Jagger (Chief Boiler Inspector for the State of Ohio):

--------------------------------------------

Y = AX^2 + BX + C

Bolus is a neat word, but this is the first time I've ever heard it applied to air.:D

From Websters -

Bo"lus (?), n.; pl. Boluses (#) A rounded mass of anything, esp. a large pill.

It's my weird background showing through. When I used to work on ambulance crews, we referred to a "bolus" in the context of medications delivered as a supplemental I.V. charge. After a while, we tended to generalize it to mean anything running through a tube (and trust me, you don't want to know how it is appropriate to describe large masses of anything running back wards through tubes, particularly after eating college dining hall food!).


-dave

--------------------------------------------

Y = AX^2 + BX + C

David, True, that our little robot air system is unlikely to cause a catastrophic failure, but the purpose of the competition is to teach students proper building practices. I am a programmer, not a pneumatics expert, but using a relief valve as your high pressure sensor does not seem like proper building practices in the industry. Maybe you have some insight that I lack.

PS: Thanks for the correction. I was going off of what I heard in the news.

Animator,

I'm not a Pnuemattics expert either, but I do spend a lot of time around really big pressure vessels (15'x45' autoclaves), and some time running them.

You are correct in saying that it is not proper design to use a relief valve as a "pressure sensor", or even as a "control".

However, in Dave's system the relief valve should never need to do it's thing. By setting the system timing so that at optimal conditions it just barely reaches 120 psi from the threshold pressure, you pratically ensure that it never quite does so under competition conditions. What really happens is that when the pressure drops, the compressor runs for X seconds and then stops. The final pressure will be close to but always less than 120psi. Remember, we are less than optimal here. If at the end of the cycle, the pressure is less than your threshold then the compressor can restart and try again. If the pressure is greater than the threshold but less than 120, it will just sit there until something changes.

I think that part of Dave's point is that the extra psi or two you can get with two sensors probably won't make a difference. If you're drawing down the air system that far you don't have enough margin to start with.

The post was in response to this posting. You really should not feel free to overfill the accumulators.

1 Sensor Code
 

Here is some code that will allow you to use 1 sensor to control your pump if you need to save on weight.

Keep in mind, if you keep your executable code length short, your serin and serout commands will occur more frequently making your robot respond quicker to inputs and outputs.

Also, debug commands slow down the code execution. Comment them out after you're done with testing. If you want proof, watch the Basic Run light when the debug code is in the program and when it is out. That LED is controlled by:

Toggle 7 'Basic Run LED on the RC is toggled ON/OFF every loop.

in the default program. Everytime the code is executed, the LED is toggled. The faster it blinks. The faster your code is executing.

Sorry to go off the topic. Below is the code:

I tested it in RoboEmu 1.07

------------------------------------------------------------------
' Variables

Counter VAR w10
LowPres VAR rc_swA.bit1
PumpMan VAR oi_swA.bit2 'Aux input,
Pump VAR RelayA.bit0
Pump_rev VAR RelayA.bit1
Counterbit VAR Counter.bit0
CounterMax CON 501 'This number needs to be odd
Counter = CounterMax

MainProgram:

Serin...

'Pump Code for 1 sensor

LowPres = ~LowPres
Counter = ((LowPres) * (Counter + 2)) max CounterMax
LowPres = ~LowPres
Pump = ((LowPres | ~Counterbit) & ~PumpMan)
Pump_rev = 0

Serout...

----------------------------------------------------------------

Thanks for the backup Dave,
The primary reason for the delay isn't really to fill the tanks to 120psi, the reason is to get the pressure high enough so that the switch wont cycle on and off (take out the delay and you better have a huge box of fuses). Also yes this may make the code a fuzz longer, but if the competition ever comes down to where 10 lines of code worth of processor time could change the outcome I don't think I would be able to watch the match anyway
:D

If anybody has looked throught the new Pneumatics manual, you'll see that the new pressure sensor is normally closed, like last year, and opens at 110(?) psi. The catch is, it won't close again until pressure drops below 95. Thus, you can safely run the pump whenever the switch is closed (1 in PBASIC).

:ahh: hehe im still devouring the manual, however this does make this whole thread a moot point.

lol... we talked about this the first year that a hysterisis pressure switch would solve all the problems.
hehe.. tnx for pointing that out tho

I haven't read the pneumatics manual yet, but according to the Kit of Parts (pg 36) the opening pressure is 115 psi, and like you said, the closing pressure is 95 psi