Thursday, September 2, 2010

Testing Invention Prototypes on a Small Budget

In the development of my asparagus harvester invention I’ve had nothing but problems with the air cylinders for years. Ever so slowly the cylinder evolved though, and we’ve finally reached the point were I am happy with the air cylinders.

Problems with the air cylinders included the blade mounts falling off of the piston rods. We tried every kind of locking nut, locking washer, and chemical thread locking adhesives without success. When we field tested the machine we left a breadcrumb trail of parts behind us. We spent more time repairing cylinders than we spent harvesting asparagus.

We solved the problem of the blade mounts falling off by using a ¼” NPT tapered pipe thread on the piston rod and blade mount. We had a similar problem with the piston rod and piston. Again, the tapered pipe thread did the trick. In fact, it worked so well that I decided to try and obtain a patent on it.

I was going to try to get a patent on the method of locking threaded parts together using parts with mating tapered threads. I was all set to file a provisional patent on the method when I discovered a patent issued in 1956 describing exactly what I was doing with an air cylinder and piston rod. Provisional Patent Application
When we used tie rod cylinders the tie rods would stretch and the tie rod nuts would come loose. Even when we Locktite the tie rod nuts would come loose. The same thing would occurred with all of the threaded parts. We did not have a problem with the parts that had pipe threads.

Life Expectancy

A heavy yielding asparagus field might produce around 75,000 spears per acre per season, with the center 6 cylinders doing most of the cutting. The machine should be able to do about 50 acres per day. 50 acres times 75,000 spears per acre per season = 3,750,000 spears per season. Divide that by 6 cylinders and you get about 625,000 spears per cylinder per season.

The time it takes for the cylinder to reach it’s fully down position is fixed. Therefore accurate cut timing depends on the forward speed of the machine. To program the microcontrollers I need to know the down stroke time as accurately as possible.

Other useful information would be what the terminal velocity is and how long the blade stays in the ground while the machine moves forward.

Seal and bushing wear

Obviously I need to find out if the cylinders will survive 650,000 strokes without suffering from metal fatigue somewhere or some similar structurally related problem.

Bimba air cylinder company has on their website information about cylinder life. The Bimba cylinders should wear the same as mine as far as the seals and rod bushings go.

“Bimba cylinders have been designed and tested for a rated life of 1,400 miles of travel when properly applied and lubricated per recommendations. Bemba’s option E has been designed and tested for a rated life of 2,800 miles of travel when properly applied in an un-lubricated environment.”

My cylinders travel 20 inches down and 20 inches up 650,000 times per season on a heavy yielding field. At that point they will have traveled 410 miles. Doing a little more math, if the cylinders will make it to the 1,400 mile mark, they will have survived over 2 million strokes.

Summing it all up, I need to obtain the following information about the pneumatic cutting cylinders:

• Stroke extend time

• Stroke retract time

• Time blade spends in soil

• Number of strokes until failure

• Effect of Lubrication

• Effect by blade mount mass

Testing Constraints

Unfortunately I do not have a well equipped lab or shop…only my garage. I will need to build the testing apparatus and measuring equipment myself from my junk box. I think all inventors should have a well stocked junk box for prototyping their invention ideas.

Designing the electronic controller

If I am gong to do a life test and measure the speed of the blade, then I will need a controller that will periodically fire the air cylinder for a long period of time. I will also need to be able to vary the length of the pulse that causes the solenoid on the air valve to stroke the cylinder and the period of time that will elapse between firings.

My air compressor is one that I purchased from Home Depot or Lowe’s. It puts out something like 6 CFM at 145 psi. I have to limit the firing to about 1 cycle every 15 or 20 seconds so the air compressor can keep up.

I built a controller by using a microcontroller chip, a computer on a chip. I used a Microchip 12F675 controller, two potentiometers, a couple of resistors and a Darlington power transistor.

The two pots control the on and off times allowing me to produce a pulse that depending on the position of a pot is from 1 millisecond to 100 milliseconds long in 1 millisecond increments. The second pot controls the time between pulses from 1 second to 30 seconds.

I used one pin as an output to drive a Darlington power transistor which in turn supplies 12 volts to the air valve solenoid. Designing the electronic controller

Measuring the stroke time and cylinder speed

Now I need a counter that I can control with the optical switches… back to the inventers junk box.

I again used a 12F675 chip. I programmed the chip to count pulses after one input pin gets a pulse, and to stop counting after a pulse from a second input pin. The chip counts at a rate of 1 count per millisecond. The output is sent to a LCD display which shows the count in milliseconds. When the chip gets the second pulse and stops counting, it then waits for a reset pulse on a third pin supplied by a push button. Until it is reset the chip won’t do anything when additional pulses come in from the optical switches.

Doing the air cylinder testing

Now I have the air supply ready, the optical switches, the counter, the cylinder controller, and the air cylinder ready to go. Since I now have pretty good instrumentation and I have the time to do it, I’m going to learn all the things I’ve wondered about but didn’t have the time or resources to actually find out. Like what affect does the weight of the blade and blade mount have on the speed of the cylinder, and what difference would there be in the speed of the cylinder with different types of seals.

I set up the test by putting the laser beam from the first optical switch slightly in front of the end of the piston rod. With the first 1/16 of an inch of extension the rod will cause the switch to send out a pulse. I placed the second switch 20 inches further down the path of the piston rod.

The results… what I learned from the initial testing

Well, the results of my initial testing were surprising. As is often the case my hypotheses turn out to be wrong on several counts.

I found that the mass of the blade mount makes less difference in the cycle time than I expected. I tried several mounts, the lightest being 3.4 ounces and the heaviest being 8.2 ounces. Even the heaviest mount increased the down stroke time from about 80 milliseconds to about 85 milliseconds.

So far I have only measured the down stroke, but I did take some movies of the cylinder in operation and by reviewing the movies in windows movie maker I can view them one frame at a time. From the movies it appears the upstroke has about the same speed as the down stroke.

So far I’ve only got data on the time it takes the cylinder to extend to 20 inches of stroke at various air pressures and with various amounts of mass attached to the piston rod. Here are those results:

At 80 psi supply pressure and with a 56 millisecond output pulse to the valve the cylinder travels the 20 inches in about .081 seconds (81 milliseconds).

Adding a 3.4 oz mass blade mount increased the extend time to about 94 milliseconds.

Switching to a 4.2 oz mass did not measurably affect the extension time.

Switching to a 5.8 oz mass resulted also did not increase the extend time.

Increasing the air pressure to 100 psi with no blade mount mass resulted in an extend time of about 84 milliseconds.

At 100 psi and an 8.2 oz blade mount mass the cycle time was about 95 milliseconds.

Raising the air pressure to 120 psi did not seem to have an affect on the cycle time or rod speed.

The average rod speed for an 81 millisecond extend time would be about 246 inches per second, and for a 95 millisecond period the average speed would be about 231 inches per second.

At one point my extend time slowed to 100 milliseconds. At that point I tried lubricating the piston rod which had no affect. I then squirted some oil into the cylinder to lubricate the piston seals. The extend time immediately returned to 84 milliseconds.

This makes sense because the rod seal is a single seal with a circumference of about 1.9 inches making contact with the moving rod. The piston however has two seals each with about 3.14 inches each in contact with a moving surface. The lubrication to the piston seals is much more critical than the rod seal. Good to know this stuff.

Soon I can begin letting it cycle all day every day for a few months and see what breaks first.

Testing your invention prototypes sometimes requires a bit of creativity it you don’t have money.

A more detailed version of this can be found here:  Air Cylinder Testing - New Invention