Hacking the CP-2

[George Schmermund]

Back in early April (#127) I was testing the possibility of using micro LED pairs to make light gates for detecting a pellet's transit down a barrel. The LEDs worked well and had sufficient sensitivity to see a pellet passing through the gate. The down side was that I'd have to drill holes in the barrel at intervals down it's length. This would also entail deburring the holes so that they didn't interfere with the pellets smooth (?) transition. This method would probably bring more uncertainty baggage to the measurement than would be practical to sort out.

After Stan and I were getting some good numbers when timing the first movement of the pellet it was obvious that we were onto a much better way to get timing information with probes down the muzzle than drilling holes in the barrel. The problem now was that for any detection of pellet movement greater than a few mils, the probe itself would grossly interfere with the measurement.

Now that I've put the barrel harmonics onto the back burner while the XCL measurements are being sorted out, I decided to go back to measuring barrel transit times. A couple of days ago I posted some info on making stiff low mass wire probes that didn't need any extra support. Since the barrel diameter is ostensibly 0.22" ID and the wire is 0.027 OD  the ratio is in favor of much better numbers. After several tests to confirm that this new method was viable I made some probes of assorted lengths for a new round of testing.

The method of making the probes was discussed in the previous post. The inadvertent plus to this method is that the wire stretches and maintains its new length, but the insulation shrinks back to its original length when the wire finally breaks at the pulling end. This leaves a perfectly stripped and exposed end of the wire. If you've ever tried to strip the insulation from very small gage wire without damaging the wire itself you'll know what a time and headache saver this is. Also, these probes only take seconds to make and are as expendable as the pellets that impact them.

The photos show a few of the probes and how they are mounted onto the end of the barrel. The muzzle end of the wire is formed over a simple fixture to keep the length fixed and the o-ring keeps the probe in place until it is impacted. What happens after the impact is inconsequential and no longer part of the measurement.

[WhatUPSbox?]

George,
Those probes look great, I'll have to add them to the arsenal so to speak.

I've started to dip my toes into the DSO's built in filter functions. This is not something I have a lot of background in so my controls manipulation is more Etch-A-Sketch than a technical review.

I dusted off my trusty Signal-O-Matic, checked for the NIST traceable cal sticker...nope still not there...and punched up a 1 Khz triangle wave.

I turned on the DSO's internal math and set it to low pass filter (2.5 Khz). I did not see any lag between the two traces. Tried a few other source and cut-off frequencies with similar results. I need to apply it to some of the transient responses to see how it works....and read up a bit on the buttons that should be pushed

[George Schmermund]

Stan - I had overlooked you're inquiry a couple of days ago in post #298 about whether the .22 barrel had gotten the same breech treatment as the .177. Yes it did. Post #20 shows the "before and after" results of the push tests that were done back in January.

[George Schmermund]

The wire probes are coming along nicely. I did a few tests this morning using CPHP pellets. My preference is to use wadcutters because of the flat face they have, but ran out of them last night.

The images show the results of  4", 3", and a 2" probes. The last image is a look at the rise time of the contact signal on impact. As can be seen using a 50 µs /div window the rise time won't even come close to interfering with the time of impact data.

The pressure for each shot stayed at ~ 500 psi. The vertical scale is 100 psi/div. The pressure rise part of the curve shows some interesting bumps that appear to be very quick and small pressure decreases with quick recoveries. I predict that these may be the actual pellet movements. Image 2 shows no distinct pressure reverse steps in the rise part of the curve and the pressure peak is lower than the 1 and 3 curves. Each shot had plenty of time to recover from the previous shot. The temperature was constant in the testing area.

The lack of steps and lower peak pressure of shot 2 may be related to CO2 blow-by before the pellet is fully seated in the rifling. This is the stuff I'm interested in looking at when the Dataq 1110 is up and running.

Lots to amuse one's self with here. I didn't use any cursors on any of these shots because that info, if anyone is interested in the numbers, can be gathered from the grids. These are only early tests of this probe method and I didn't want to clutter up the screen and disrupt how clean and unambiguous this method can really be.

[WhatUPSbox?]

George,
Yep, not much ambiguity on the timing marks, those curves look great. Is the reason the probe signals roll off because you are powering it with a meter?

I wonder if the bumps on the pressure rise are some LCO₂ spittle flashing

I am amazed how relatively repeatable the pressure curve is. That leveling off at ~2.3ms is consistent and occurs before the pellet exits

Great stuff

[George Schmermund]

Stan - There's lot's of stuff to speculate about in these tests. The contact signal is now using just a 9V battery as a power source. The decay side is after the fact and only of curiosity value, though it may be interesting. The wire is being pushed down the barrel and may or may not be in intimate electrical contact with the pellet. The only thing I'm interested in at the moment is the timing at the point of contact. As an aside, I've fired the pistol without a pellet loaded and then inspected some wires. They stayed straight and unaffected by the blast. I consider this to be a big step forward in getting good numbers from these probes.

The pressure transducer and associated amplifier are much faster than the powerplant so what you see in the P/T curves is real.

I'll wait until the DI- 1110 system is in harness before commenting on more than my current guess about the upward pressure spikes. In the mean time I'm interested in putting another pressure transducer at the muzzle and measuring the P/T curves before and after the pellet leaves the barrel.

[George Schmermund]

 Stan - I've been thinking about the gyrations that we've been going through to do the timing on the hammer launch, valve stem impact, valve opening time, pellet release, etc. It certainly has been a good circus. Now I'm thinking about what's really important in the overall scheme of things.

The shot cycle could be considered as starting at several point. I'm going for compacting everything down to my own most convenient assessment and say that the real cycle, as far as the pellet and I are concerned, begins when the valve starts to open. From then on the action really gets going! Any part of tuning an airgun that is applied to things like hammer or valve mass, spring coefficients, lubes,or anything else that may effect how and when the valve stem is impacted can be measured. This response to the impact can be determined by looking at the P/T curve. The P/T curve also communicates to the pellet what to do and when to do it. The pellet's response can be followed all the way until it leaves the muzzle. Any barrel treatments should also be evident in the curve.

This notion of all encompassing information in one graph may seem like a stretch, but I think that with sufficient resolution in the curve measurements it can be done. I'm interested in any arguments against this proposal.

As can be seen in the DSO images, even with just 8 bit resolution there is still a lot of information there. The main thing is that the pressure transducer signal is clean and fast compared to the other heroic methods Stan an I have been working on.

   

[WhatUPSbox?]

George,
I agree that any of the hammer information is only really important in how it impacts the time initial valve contact and then the stroke and dwell time of the valve.

The pressure curve you are able to generate is where the action is, and it is in general unique in air gun investigations ( I think I found one paper where they instrumented and airsoft gun in a similar way). The curve is also all encompassing and benefits from additional event markers to help dissect what is going on. The hammer/valve contact and the start of pellet motion are two good examples.

The zoomed in pressure plots you posted have some characteristics that with a sensitivity of 100 psi/v represent about 5 psi deviations from a likely curve fit. With the pellet motion starting around 400 micro-sec, these sections of the plot are well before pellet motion. It is hard to further identify their cause. The flow through the valve and transfer port is anything but smooth. There may also be some liquid CO₂ droplets that change phase. Once the pellet starts moving down the barrel, the pressure fluctuations may have an additional contributor in the varying pellet friction and anything that may happen in the air in front of it. How to dissect these combinations gets to be an intriguing challenge. The multi-sampling you did earlier is one step in the process.

I also think the relationship of the pressure curve to the time of pellet start is an interesting area. There is a large collection of bolt probe designs that try to affect the gas flow and the positioning of the pellet relative to the transfer port, that this data would shed light on. Similarly the affect of the pellet fit in the leade and pellet sizing would greatly benefit from direct pressure data as opposed to the more common chrony data.

I'm looking at ways I can contribute data with my testing but you have a truly unique and important test capability.

[WhatUPSbox?]

OK, as a side note, it seems fair to also report the fails and dead ends.

A while back, I thought I found a pressure sensor (image 1) that could be put to use in this testing. It had some positive features:
<$20, 0-1000 psi measurement range, simple 5V power, 250 psi/v out, relatively compact with 1/8 npt threads, and "ceramic pressure chip sensor inside". Before I tried to make some kind of a test breech for the 2240, I tried to see what the response time of the sensor is. I thought if I could create a hydraulic shock, I might be able to see a fast enough response.
Image 2 shows the sensor with a 1/8-1/4 bushing, I made a plug that fit inside the bushing, filled with water and tapped on the plug with the George approved tap hammer.

After the first couple of taps I thought we had a winner but tapping it with the brass hammer, I was just putting some electrical noise into the system (image 3). With some insulation on the hammer, I did get a real response but unfortunately it comes in 2 ms chunks (image 4). Too slow

The sensor is nice and I'm sure I can use it for some other slower measurements but for this testing, it is going back on the shelf.

Unless I find something else, George is our pressure guy.

[George Schmermund]

Stan - Your image 3 with a 0.5 µs/div capture would have gotten me all excited, too! < 20 bucks, wow! Sorry it didn't work out. I'll readily admit that the testing of any new transducer of any type is always worth some level of excitement to me.

I'm glad that you've become another active experimentalist in this airgun ballistics arena. It appears to be uncharted territory for the advanced hobbyist.

[George Schmermund]

This morning a couple of shots were taken without using a pellet. I wanted to see what the P/T curve would look like with just the valve opening and closing. The first cursor is at the start of the opening and the second one is at an inflection point where I'm guessing that the valve has closed. This knee may indicate where the plenum is no longer being feed by the powerplant. the time is 2.4 ms. There is another knee that is more like a step at ~ 1.5 ms. My guess is that the valve just turned around and started to close at this point.

Of course there are other things that can be read into this image, but more resolution will be needed to tease out better guesses. I tried to get the DSO's math function to plot the derivative of the curve to accentuate the points of inflection, but without success. I'll have to spend more time figuring out why it wasn't successful.

In the mean time anyone can read anything they want into these digital tea leaves. Keep in mind that this curve was made without a pellet in the mix and therefor probably altered the timing some, but the numbers are real for the situation. My guesses are likely to change with more detailed information.

[WhatUPSbox?]

George, early this month you were thinking about instrumenting the hammer. We talked about piezo and striker plate instead of the accels, another option might be a springier contact shim that's insulated on the hammer side and releases from the valve stem when the hammer retreats. Either of those may be able to identify the valve close event.

I'm not sure if these DSO's do any noise reduction before they take a derivative. If not, the derivative may be busy.

I really like these pressure curves, there is always new data in them

[George Schmermund]

The shim contactor did work well in the trial runs, though it had some noise issues. I got distracted by the piezo/hammer detector idea and stripped down the powerplant to do some of the hammer and main tube mods that would be needed to implement it. The parts are now in a tray of formaldehyde on the dissecting bench.

Since the hole for the shim has already been machined into the main tube I should go back and spend a little more time in reassembling the shim project more carefully in the configuration that you're describing. My problem with that is how embarrassingly simple that method is. As a final insult it will probably work really well.

I'll try to swallow my pride and put the shim experiment back together this weekend. As a cataplasm I'll now refer to the method as being elegant instead of simple.

[George Schmermund]

I did some tests today using the shim impact method and decided that it doesn't have a timing consistency that can compete with the pressure measurements. I'll try again with the piezo detector/striker plate method to see what it can do. I now think that maybe I'm on a fool's errand with valve timing measurements because all of the information about the valve workings up until the pellet moves is best seen in the P/T curves.

I well understand the siren call of cheapskate measurements because I try my hand at it often. At some point, though, it has to come down to what are we really trying to do? Stan and I have offered up several ways to make affordable airgun ballistics measurements, but it hasn't drawn anyone into the hobby on that level yet. What's the incentive? At some point I'll just return to doing the experiments with the best instrumentation I have available and continue to report my findings.

 

[WhatUPSbox?]

George, Sorry to hear the shim-flexure didn't give clean data. I had hoped that with the shim-flrxure, heavily biased towards the hammer, would mark the valve closed event. I don't think of it as competing with the pressure curve, the valve-closed point would be a great addition to the Pt curve.

As far as drawing anybody into testing, maybe not, but Lloyd is doing some interesting testing in the Bob & Lloyd gate. At some point the data streams will cross.

I'll dabble in the low cost end, If you got better instrumentation for some of this, apply it, getting the data is the key

[George Schmermund]

Since the previous post #202 shows that the valve timing can be done with a shim switch it will be worth returning to that testing at some point. What the switch shows is a clean signal of when the valve face is contacted. This is the time that the valve is fully open. The total time for the valve to both open and close is yet to be determined. The method is easy to implement and will work if it is broken into 2 separate shim switches. The problems that I ran into yesterday had to do with using too thin of a shim stock.

I've had to remind myself that the valve timing is of little interest to me at this point except maybe the interest in making the measurement. I'm still an adherent to the idea that all of the important timing information is contained in the P/T curve including valve timing.

This all needs to be sorted out and the ballistics and DSO are now sending me back to the calculus books to get rebaptized. This is a good thing. 

[WhatUPSbox?]

OK George, Since I only had a weak lager with dinner instead of the memory enhancer, how is the cosine function related to the first derivative of the P-T curve?

[George Schmermund]

Well, it looks as though the strong ale memory enhancement only got me back as far as the barrel harmonics testing. Since the Diff function on the DSO didn't seem to be doing it's Job I fell back on making the sin(x)'s derivative, cos(x), to be the stand in. If we were doing accelerometer graphs it would have looked pretty good.

This morning I came partially to my senses and realized that I wasn't getting the slope information I wanted. I went back to the DSO Diff function and tried again. The problem I was having, aside from ignorance, was that the DSO was working overtime on the limited  low resolution 8 bit data. I lowered the number of data points for it to collect and it really works well now. There's hope yet. I still think that the real shot cycle truth is buried in the derivative curve.

 The image is of the DSO's calibration square wave and the slope information the diff function is capable of retrieving. It appears that I'm the weak link for the time being.

Thanks for continuing to keep me awake along the path, Stan!

[WhatUPSbox?]

To be honest, I had a quick try at the DSO's calculus functions last week when I was trying the filter and was not successful. I think I need to follow your lead on how much I ask the DSO to process. The derivative will be interesting in trying to manage the noise without smoothing out the features of interest.

[lloyd-ss]

Hello George,
About a week or two ago, Stan said your thread was something I needed to look at.... and he was right!   I had totally missed it as I've been off line for a while.  Great work George, and a wealth of information, and of course, a wealth of new questions, too. I have only gotten through about half of the thread... it does take some time to digest, and honestly, some of it is out of my knowledge comfort zone, so I have to read up on what the heck you are saying, LOL. But, a day without learning something new is a day wasted.
Your explanations are great, and I can certainly appreciate your stream of consciousness side trips where you get off into the weeds, but eventually make it back to the task at hand, whatever that really is.
Your arsenal of electronic toys is impressive, but what is more impressive, is that you actually KNOW what to do with them, or dream up new ways to put them to practical use on this project.
I'll be catching up and following along to see what yummy tid bits you are throwing out to us. Kudos on your passion for the effort, and your willingness to share.
Lloyd-ss