Dual Acting Cylinder With Free Piston

Thank you for a real answer and man hugs.

Man hugs (?) Is that a hint of things to come ?

You never hugged your brother, or your cousin?

<...can a 4-port, 2 way valve do this...> A 5/2 valve certainly can. Air in the centre port, exhaust from each outer one, and the two other ports are connected one each to each end of the cylinder.

<...How many air switches...are needed to reverse travel direction at the end of each stroke?...> One at each end.

As to the switches, that are not really air switches, but more likely position indicator switch, do you suppose these could be simple optical interrupted beam type switch?

Another thing: could the piston be a metal or other type ball that rolls in the cylinder?

I can see the Arduino program for this being pretty simple (until one gets into the data logger part), have a do loop that counts up transits, while millis accumulate (milliseconds).

The logger has to log on variable interval where the events accrued reach the total in X time. X is divided into the number of transits then the cal factor is brought into account.

The switch can be what ever you want it to be, simple lever switches to yes an optical interrupted beam switch. Electronic switches will be a bit more complicated than plain air switches because you need to convert them to air. We use a PLC to put out a signal to a Mac air pilot switches, which fires the valve. I imagine you can do this with the Arduino, but with the Mac pilots we use, you need 24v AC to fire them.

I would look on E-bay for your valve and switches, used stuff can be very inexpensive.

I would research it now for you, but dealing with sick kids, some kind of stomach virus.

If you think you can use my help, P.M. me later. I am not familiar with Arduino, but I think I can help you with everything after that. I do have drawings at work on how our cylinders, valves and switches work.

For your piston to be a ball, the tolerances would have to be tight, I think you would get too much blow by, unless you just want to see it go back and forth. My pistons have cup seals on each end facing the air pressure and some also have O-rings in the middle. Kind of hard to do on a ball.

I agree with all you have stated. Kids are No. 1, everything else is back seat.

I too was thinking cup seals, but here I am very concerned with getting motion with the least amount of back pressure (i.e. the fluid being measured into the flow tube).

The fluid is stoichiometric hydrogen-oxygen, so I do have to be cautious about any electrical contacts. I prefer red light optical path switch. That way, if we screw up and get chloride in the electrolysis, the chlorine produced will not get a chance to go pop with hydrogen.

I got a quote yesterday for a Hyde type wet test meter (total seal volumetric flow meter), for new one decked out with output contact (per revolution) intrinsically safe: north of $5800; reconditioned was in at $2900 (decked out). I fired back that I just want the guts out of one, so I hack it into a digital readout. It seems that designing my own profile for the drums/vanes with orifices (or slots) and having protoype 3D printed will be far cheaper (at least, so I think).

That is why I want to build this simple flow tube meter also. The current flow meter is water displacement with about 1-2 psig back pressure due to water lift, and small tube diameter (poor design of tube straighteners). And the solenoid valve (12V) transient running off relay fired by Arduino is causing the Arduino to lose its mind.

Best way to clamp off coil transients? Zener diode? Capacitor/resistor to change amplitude and time constant?

We've used these RC devices very successfully for both coil and motor suppression:

http://uk.rs-online.com/web/p/rc-network-capacitors/2067831/

We use them both for mains and low voltage DC applications. Fo 12 or 24 Vdc you could just connect a 220nF cap and a 470Ω resistor in series across the load.

Series with the load (solenoid valve, 12V), or parallel across terminals of the load?

C and R in series with each other, the 2 together then in parallel with the load.

That is what my mind was actually telling me (left-overs from excellent 2 quarters of three in graduate level physics electronics course, taught by the guy that built the 10 ton Swiss watch (detector ring) on SLAC.

Thank you also, very much. This is all very pertinent information for me, I assure you.

Anyone know the most economical valves in this category? I don't care if the first one breaks after 10000 cycles. It is a prototype, they are never expected to work the first time anyway.

Use a 5/2 valve. If 4/2 are available, I suspect they're quite hard to come by and costly.

Please 'splain to me your valve diagram, I started on chapter eight yesterday but it was mostly Greek to a chemist. Is that a 5/2 you diagrammed?

Actually I ordered a couple of 5/2's last night off Amazon. Two different brands (don't ask me why), and each was under $15.00, nearer $10.

Yes, it's a 5/2 (which means 5 ports/2 positions). The diagram shows the de-energized state, with the spring pushing the spool to the left. In this state, the inlet (P, at the bottom) is connected to port B, and port A is connected to the exhaust. When the solenoid is energized, the spool moves to the right; P connects to A and B connects to the exhaust.

Make sure they don't need pilot air, as they won't work correctly with such low pressure.

There is the option of increasing the pressure of the gas flow, that I did not mention. We are definitely good up to 15 psig, but around 28 psig, we blow the lid off the reactor, it all depends on how well we cinch down the gasket ring to the concentric dome lid.

That probably would also mean a lot more gas wastage (and a failed metering attempt) if the pressure is raised much more than a few inches water head. I can see that being the Achilles heel of this type of measurement system. The only way around having wastage error, one would have to take a positional reading along the tube several times per cycle, and allow some reset time at the start of the next cycle, yes?

Not sure I follow you, but sounds about right. By "wastage", do you mean what's lost in the cylinder?

If the valve needs pilot air, it probably won't work below about 3barg.

Just another thought - with the free piston, there'll be a nasty tendency for it to skew over and let gas past (or at worst, jam) unless it's quite thick relative to its diameter.

Agree. It just means it has to be engineered to be thicker, and call for a larger void volume correction. Really, I have to say, as a flow meter this design leaves me to think there will be a lot left to be desired.

Are you making a Stirling Engine?

Really, it is more of a flow meter concept, but I suspect the switching losses will rule it out, but we shall see.

Check out AirPot Corp. They make a line of virtually friction free glass cylinders using a precision carbon piston, that may be just what you are looking for. As far as a control valve, as has been mentioned, any 4 way, two position spool valve will be suitable. Be sure the spool is direct acting and not pilot operated and rated for vacuum service. Such a valve is needed to work with low pressures.

I'm not sure I fully understand what you're trying to do here, so forgive me if I'm off course.

Instead of the cylinder and piston: would a spherical vessel with a hemispherical rubber diaphragm do the same job?

They used to have electric beer pumps in the Nottingham area that measured a half pint in exactly that way. I must have watched them working thousands of times, but, I can't find a picture now.

the idea makes sense to me, in that the piston seal is effectively the expanding diaphragm (balloon), and I suppose the ones clowns use to make balloon wiener dogs would be perfect in a longish tube.

The one I saw at the NIST gas flow meter calibration site had a mercury seal on the piston, and a mercury vapor trap on the vent. I am a bit shy of having that much mercury around.

How about Teflon piston rings? Viton rings with light mineral oil? At least paraffin shows an A1 rating. Glycerin shows also A1 and might offer lower friction.

Then there is the aspect ratio of piston to cylinder diameter, I suspect at minimum a 2:1 aspect ratio is required to keep piston straight. I also have some small diameter glass tubing to try, I think 8 mm I.D., and that material should also work.

To reduce losses during switching, since the length of each tube shall be a minimum of L/2 + %L where the %L is the required amount for bend radii, etc., the volume contained in the small gas tubing can only be minimized by choosing the smallest diameter that the required gas flow can travel through without increasing back-pressure out of bounds (< 2 psig, 13.8 kPa.) This means the pressure correction alone would be ~14%. That is fine as long as pressure is included in recorded observables, along with temperature of the flow.

Heck, for that matter what you guys are saying is I could even use a bellows as long as I know the extended and contracted volume precisely and accurately, then the piston could effectively be spring loaded (very light spring or even gravity return) thus eliminating the 2-step cycle, although my first thought was that a two way (double acting cylinder) that produced a measurement on the extend stroke, and on the return stroke would be an elegant solution.

To get a low friction piston i was thinking about using a bellows or balloon on either side of the piston. That lead me to a ventilator. An old gas powered bellows type. I think you work in a hospital, if you do, see if there is a storage room for old crap. I found this on the web. An even older BIRD or BABY BIRD may do all that you want.

Jim

Nah, I work in a power generation municipal utility. This is for my hobby - trying to get something from nothing (the cosmos). We hope to quantify our "nothingness"! LOL!

So far, with background radiation counts (small diameter Geiger tube that got 700/m count with Fiesta ware about 1 cm distance) of 10/m +/- 2√10, we observed 19-35/m with reactor up and warm. 35/m +/- 2√35, is I think slightly different from background, so maybe our matrix got stirred up (special anthracite treated by electrolytic AC treatment) and the approximate 1 ppm uranium is showing off, or something else is taking place....hmmm.

The flow measurement of electrolysis gases is obviously a part of overall calorimetry on the experiment. Reactor is in an insulated housing that draws in ambient air, and the ventilation fan is on low flow, but constant, with a calibration load resistor bank on its own duty cycle to calibrate the power in watts as heat dissipated from reactor. Soon we will improve this by installing an automotive heat core on the gas outlet, prior to housing exit, thus cooling the gas and condensing any water vapor for return to reactor.

O.K. The ventilator in the pic is already a bellows with gravity return. The pressure is where you want it to be. The flow is electronically controlled by switching on various combinations of solenoids. !l/min, 2l/min (2 of) 4l/min, 8l/min etc.

Cost may be a factor unless you know a colleague or friend who can rummage through a hospitals junk room for you.

Jim

We are trying to measure the flow of gas from our reactor. We are not gassing a patient.

Yeah that might be fun to try, but we are pretty close with our initial displacement meter, just working on circuitry issue (solenoid valve keeps wigging out the Arduino board with its switching transient even though we were using a high isolation valve relay. Solid state relays work better, or even transistors at times, such have to watch out that the output side is on AC and has zero crossings to shut down the output quickly. DC on the output side of those does not work worth poo.

I suppose I should just call the county hospital district (UMC) and demand they find a surplus ventilator in the basement and hand it over.

You can get DC SSRs these days. We use them (for similar reasons).

Apparently, I need to hone up my research some more.

How about modifying an old air operated double diaphragm pump?

Just take the 2 side plates, sandwich them together with one diaphragm between- you could use the regular backer plates to help provide a more consistent volume per stroke.

That might work, as I had also made a thought experiment with a small inner tube.

Why reinvent the wheel? Find your nearest gas utility meter shop and see if you can scrounge an old residential gas meter; very low to moderate flow, very low restriction/back-pressure, temperature compensated, not much engineering required.

The question is whether your gas volume/pressure is sufficient to make the movement work. If you're counting off-gassing molecules from your cold fusion reactor then this probably won't cut it.

Yeah, I already thought out that option, and I would totally have to hack the meter to get impulses or something.

I found an angle meter chip for measuring the angular position (NVG??), and these are not expensive at all. They utilize the giant magnetoresistive effect using a bridge circuit to create two out of phase (sine and cosine) waves that allow simple computation of the quadrature in Arduino program (see if I can hack that part). The rest is just counting rotations if a rotating design, or finding a linear position indicator that is good for counting piston position, and counting traverses.

I still think my idea is dead before it starts due to getting the air switches to work the way I want them to (probably outside their envelope), not to mention blow-off losses that totally kill any measurement scheme. Oh well, they cannot nail me for trying.

Hacking the meter for pulses is the easy part, a simple photo-interrupter is easy to rig up on the register. The iffy part is whether you can get the quantity per pulse to provide the resolution that you want. It might help us if you told us the resolution of flowrate or volume that you're trying to measure.

I see that in my original query I did, in fact, omit that part.

We are currently only running at about 100-300 mL/min, but the current we are using is turned down quite a bit. I anticipate reaching up into the 1 L/m range at some point, but for now 0.5 L/m seems adequate. Unfortunately, that is a bit low for some flow-meters. Presently, we are using a water displacement with load cell and a timed sampling interval that is repeated until the collection jar gets full, then it is supposed to automatically dump. The problem we are having is that the 12 V solenoid valve we are using seems to produce a nasty voltage spike upon switching (through a relay controlled by Arduino board). We plan on introducing capacitor-series resistor across those terminals. Other option is a Zener diode.

"seems to produce a nasty voltage spike upon switching"

Show us your circuit: have you used a quench diode?