Hacking the CP-2

[George Schmermund]

I'm not sure about the long term sealing on the calibrator. The best test for long term would be to weigh it. My needs are for intermittent calibration, so small leaks over time are not a consideration. I'm willing to spend 50 cents on a powerlet for it's occasional use. I don't think you'll do better that a 2 or 3 pack of canned duster gas at Fry's or some other outlet.

I don't use my shop compressor for general dusting because it always seems to make a bigger mess than what I started with. Then I have to breath the stuff. I use the can duster exclusively around the machine tools. high pressure compressed air blows swarf into places it should never go. I use chip brushes to clean up the bigger piles.

Considering how seldom I actually cleanup any of my messes I can get a lot of mileage from a can of duster gas. Hope this helps.

[WhatUPSbox?]

you are right, duster gas is the way to go......I'm still trying to find a use for that adapter set

[George Schmermund]

I shall consider this to be my piece-de-resistance in the trigger timing challenge. The only thing cheaper and easier then this would be a thought experiment. The parts were free. The laser was added only to comfort Stan. (It doesn't really work anymore.)

The materials list is:

1 small piece of 5mil brass shim stock (could also be a piece of kitchen foil)

2 Pieces of electrical tape (could be Scotch tape, etc)

1 multimeter (could be free at Harbor Freight)

The photos show how it want together. The voltage from the ohms setting on the multimeter is connected to the gun frame and the brass strip. The scope probe is connected to the the meter probes. When the hammer pinches the brass between itself and the valve body the meter circuit closes and the voltage pulse is generated.

The DSO traces tell the whole story.  The switch closes in 2 µS and any real hammer bounce is recorded as a 2nd (or N) full voltage pulse(s). The valve stays open for ~1.5 mS. This test was done without a powerlet installed.

I'm now getting very close to putting the new DSO out on the playing field.

[WhatUPSbox?]

The only thing cheaper and easier then this would be a thought experiment.

Not true a successful thought experiment usually includes a pint in the consumables list.

Nice result. What keeps the shim off of the valve after it closes? Just the spring of the shim? Do you have the hammer spring backed out so it does not preload the hammer after a shot?

Yes the laser adds tech cred.

[George Schmermund]

Well, I consider the pints to be an overhead expense so they're amortized into the cost of life itself. Sort of like eating and breathing.

The shim is resilient and has a small displacement so it returns to it's rest position just above the face of the valve body. The small signals just after the switch closes are probably the settling time of the squish. The hammer spring is at the factory setting with no preload. I expect more interesting things to happen when the powerplant is pressurized.

[George Schmermund]

One of the problems that occurs when doing the barrel harmonics testing is that there is a high probability of getting double strikes caught up in the measurement as Stan has attested to. I decided to do another cheapskate approach to building a better instrumented hammer for doing the tapping. This was prompted by the arrival of the 20 PAK of 12 mm piezoelectric disc ($1.48, free shipping). 

The decision was made to get the next size up hammer from the Harbor Freight watch & jewelry tool collection ($7). I actually like it better then the smaller metal handled one. The size is perfect for the new 12 mm discs. A little lathe time with the new hammer and a small piece of Delrin and I was ready to do some tap dancing on everything within reach of the cabling.

The scope traces shows the difference between tapping the wood benchtop (stored wave A) and the jaw of the vise (channel 2). The disc was wired in behind the brass hammer tip. This effort was a simple proof of concept project that turned out quite well. There are several refinements that will be added to improve the overall performance, but I'm really happy with this new toy for now.

[WhatUPSbox?]

George, Wow, great job on that hammer! I looked and Piezotronics has one for $760... looks like you have enough left over for a pint or two on overhead.

When I was testing the 2240 barrel I gravitated towards the dome or chisel tips to get a single point contact without rocking.

Again, great job

[George Schmermund]

Stan - you're right about the shape of the tapping end of the hammer. I've got to do some machining on the shape and type of material for the faces. Right now I'm working on characterizing the hammer itself. The wooden handle may end up having some desirable traits compared to metal or other composition materials. This hammer project adds a whole new layer of interesting investigation to the barrel harmonics measurements.

[WhatUPSbox?]

OK, it is only fair to show the dead ends and partial fails.

I added the two axis accel near the muzzle. I used a block (a keen eye will recognize it as the base from a clamp-on desk lamp) because I wanted to try to get the two axes normal to the barrel. The block also had some convenient cross holes for mounting a second laser/receiver set for detecting pellet exit. The first image shows the configuration. I used Al tape as shim to make it a snug fit on the barrel. You can also see the laser receiver mounted but not yet wired up (next step).

The second image shows the overall trace for a dry fire, no CO₂. The blue (ch4) trace is the laser/receiver on the hammer probe it triggers on hammer release. The yellow (ch1) is the mic on the tube. The blue (ch2) and purple (ch3) traces are the accel X (horiz) and Y (vert) axes respectively. The hope was to get some clear event timing marks on the accel channels. The accels detect the start of hammer motion well, the valve impact is detectable but the mic trace is probably better. After that there is too much ringing to discern specific events.

The third image shows details of the initial motion. Looks like the vertical accel axis pick up some of the hammer release noise, then both channels respond to the hammer flight, and then respond to the valve impact.

The hope was to be able to use the accels to generate some event timing points but also see if it can detect recoil when hand held. Combined with the pellet muzzle exit, it would make for an interesting timeline. Looks like right now the mic is a little clearer. I may try the accels on the barrel band to see if that quiets it down. In the worst case I'll have to break out the manual and learn about the built in filters.

[George Schmermund]

I like your instrumented pistol, Stan. It looks like you're getting ready with a top level entry for a steampunk convention! The laser pellet exit detector will probably work well. The hammer start detector is clean, but really needs to be at the point where it strikes the valve stem. That would be an easy machining job, but not an easily reversed commitment.

The rest is to busy for anything to be very well defined. It's all a good start, though.

I've played some with my DSO's digital filters and they might have some interesting capabilities. It's worth the time to figure them out just to know how they work and what they will and won't do. I still prefer the analog world for frequency tuning at this point.

It's nice to know that there is someone else out there trying to make some measurements. So far it's been like being stranded on the moon and then finding out that there's actually another person up there trying to signal back home, too.

[George Schmermund]

It's sure easy to get distracted when you don't really know were you're going. This vibration measurement project has been consumingly interesting for the last couple of weeks. The hammer project is turning into a didactic tsunami of practical experiments using transient signal analysis.

The brass tapping end of the hammer has been reshaped to produce very reliable single strikes and the force transducers are being switched alternately between the PZT discs and quartz crystals. The quartz crystals are considerably more expensive, but are very stable with changes in temperature. The cheapie PZT discs are more temperature sensitive, but they're generally quite useful in these experiments.

By good fortune I had a fixture on the shelf that was designed to test mechanical features of certain silicon wafers, but it's also ideal for characterizing these impulse transducers. Preloading can be added at the top of the fixture which is very handy.

The scope image is of a hammer tap at the top of a 5 lb preload weight sitting atop the fixture. The blue trace is the hammer. Of interest in the hammer trace is the lack of any resonances that might interfere with the measurement. I use a small sheets of different types of rubbery stuff to simulate different tips for the hammer. Different tapes from electrical to packaging types are also very instructive. I'm still trying to figure out how to copy a scope image without the menus in the image. All in good time.

[WhatUPSbox?]

Interesting, is the preload weight loose on the threads or is it bottomed out? For the trace, is that the same size transducer in the hammer as in the test stand?

Great work

[George Schmermund]

At this point the weights are just stacked on top of the flat end of the shaft. The separate weights are 5 lbs and 10 lbs plus the 2.3lb shaft and upper platen. This can give a preload of up to ~86.5 PSI on the discs which are all ~1/2" in diameter.

The disc in the fixture is the same PZT device as the one in the hammer. This is all ground zero for these tests, so I'm only after first approximations to get things going. With increased attention to details I anticipate some interesting and reproducible experiments. One of the first test will be to do a free-free measurement of the shaft and platen's natural frequencies using the soft foam method that Stan has already confirmed as being viable.

[George Schmermund]

I decided to get back on track today do some measurements on the pressure/time curves and the pellet transit times.

The image shows the simultaneous events of the hammer/valve signal, the pressure transducer output, and the force transducer at the muzzle.

The yellow trace (#1) is the hammer/valve contact signal. This signal is used to start the timing sequence.

The aqua trace (#2) is the output of the pressure transducer. This sensor is measuring the pressure just above the transfer port and just behind the pellet loaded into the breech.

The purple trace (#3) is the signal from the force transducer at the muzzle and is recording the moment the pellet leaves the muzzle.

The two purple cursor lines (1&2) mark the transit time of the pellet through the barrel.

The hammer signal is bouncing, but that seems to be the shim and not the actual valve. The pressure peaks out at ~500 psi in ~600 µS. The pellet exits after 2.66 mS. The temperature was 74º F.

Some adjustment on the hammer's signal shim can probably clean up the bouncing. I'd still like to put another pressure transducer at the muzzle to get the full pre-exit pressure profile since I've now got a 4th channel to collect more data on.

It may also be possible to put a micro hot wire anemometer into the transfer port and get the actual flow rate simultaneously with the pressure curve. Then, of course, a high speed micro thermistor could be added on each side of the transfer port and measure the CO2's cooling rate as it expands behind the pellet.

These extended measurements would require setting up the HP data acquisition and control system. Unfortunately the system runs with LabVIEW and I'd rather have a root canal then get involved with that system again. 

[WhatUPSbox?]

George, that looks great

[George Schmermund]

I got to thinking about the valve open and close time when the hammer's trigger signal is being used. Maybe the timing is different when viewed from the pressure curve numbers.  Since the pressure sensor is located just above the transfer port it seemed to be a good place to do another test.

The barrel and 2 of the clamping screws were removed to reduce the flow impedance as much as possible. The bolt and probe were still in the breech. The test was done with a new powerlet installed.

The photo and scope image show the setup and response of the pressure sensor. I'll assume that the noise on top of the signal is turbulence from the probe and bolt being between the transfer port and the sensor. The time interval between the cursors is 2.40ms. The end of the turbulent portion may be when the valve is closing. The rest of the curve may be bleed down. The negative dip just past the 2nd cursor might be a pressure drop due to the measuring geometry and the refrigerant property of the CO2.

Actually, this test result is anybody's guess. What I'm interested in finding out is the release time when the pellet starts moving. This would give the real transit time for the pellet to get to the muzzle. I think that if the whole pressure/time curve's derivative is taken a much better picture will be seen.

[George Schmermund]

Here are a couple of more images of the valve pressure/time measurements. This is the same setup as the last post, but with the transfer port removed. The pressure has dropped as can be seen on the vertical scale. The time between the cursors is 3.16 ms, but they are set wider than the last ones.

The 2nd image shows what might be some valve or hammer bounce. Both images now show a pressure drop on each side of the curves. These results are all quite sketchy and will have to be reassembled when all of the results are in. You're all welcome to guess as to what any of these tests might mean.

The next test will be with the bolt and probe removed. I'll try that tonight if I get time.

[George Schmermund]

The bolt w/probe was removed from the breech and several shots were fired, again without the barrel and it's clamping set screws. It seems now that the valve dwell time is in the vicinity of about 2 ms. It may really be back down to ~ 1.5ms that was indicated by the original hammer/valve stem contact measurements. I'm not sure what the actual decay time from the peak pressure value is considering how open the test volume is now. Keep in mind that we're measuring the timing of a valve and not a switch.

The noise on top of the main pressure curve can't be from the bolt/probe turbulence. My current guess is that a breech resonance might be the cause. The next test will include an accelerometer on the breech and we'll see if the frequency matches the pressure curve noise.

Anything that can be gleaned from these images is up for grabs. I'll just keep on testing and we can all have a go at what might be happening.

[WhatUPSbox?]

George, do you still have have the hammer/valve signal hooked up? Maybe if the shim material is tweaked to more explicitly show hammer/valve contact, it may help in understanding the pressure curve. As you have it set up does the hammer spring preload the hammer against the valve at rest?

[George Schmermund]

The shim is sitting just above the valve's face. The hammer is not touching the valve stem when the spring is not cocked so there is no preload. I'll be able to measure the hammer/stem offset later today. I'll also measure the actual valve displacement when it's fully open.

Looking back at the DSO image in post #213 there is some interesting info at the foot of the left hand side of the curves. Using the increased acuity of beer vision I've started to see an overlooked part of the record. It shows that the hammer/stem strike lags the pressure curve by about 100 µs. This would mean that the valve takes that long to fully open.

A new set of experiments can now be done to expand that small time window and rearrange the shim to be struck between the hammer and stem. Then I'll redo the shim position at the valve face and pay more attention to detail. Fortunately, the pressure transducer's rise time and sensitivity are up to the task.

I'm most interested at the moment in finding out the time of release of the pellet at the breech and what the pressure is at that moment. I thought it would be easy to see in the P/T curve, but if it's there it's not distinct. It should be quite simple to make this measurement in a much more direct way. I do enjoy having a menu of experiments to choose from.