Not an April's Fools joke.
This will be a NINE part series.
We use three of the best recoiling spring-piston rifles ever made.
We will publish one part every two weeks.
Hope you all enjoy!
HM
This page preserves a discussion thread from the original AirgunWarriors.com community forum, reproduced here as originally written. If you're a contributor to this thread and would like something removed or credited differently, contact info@airgunguide.com.
This thread originally included member-uploaded photos. They were hosted externally and are no longer available, so they've been omitted below.
Hector, I look forward to reading about this research in your blog. I wish you could test multiple transfer port sizes in one gun. I think that would be insightful. We need some new research to update what Cardew did years ago.
David Enoch
Thanks, David.
Sadly, that is something that cannot be done easily. AND it has been done. What we will show is that the TP geometry is NOT THAT important if other factors are taken into account.
The Cardews' work was exceptional for its time, but it is, now, dated. AND, more importantly IMHO, is that we will provide anyone that's reasonably handy with the tools and knowledge to create a machine to measure the vibrations in THEIR rifles
This will be an experiment in "Crowd-researching" that HOPEFULLy, will de-bunk some "theories" ("truths"?) that have been accepted and prevalent since the old days and that have hampered the real development of the spring-piston airgun.
So, stay tuned . . . 😉
This is great information! thanks for sharing.
Hector,
I look forward to the research. I know Tom Gaylord played with a set of changeable transfer ports at one time. I don't think he had a mechanism to measure movement though.
One thing that has puzzled me is that occasionally I come across a gun that jumps vertically when fired. I don't think it is a rotational movement but that is the only thing I can think of that would cause a springer to bounce vertically.
David Enoch
David;
Most modern spring-piston airguns have a "jump" or "upwards flip" as the Brits call it.
It obeys the fact that the pistons are moving forwad, the rifle is moving backwards, but it is supported by the shoulder at a point where that reaction is ABOVE and offset vertically, so that the gun seems to "jump".
If you have ever shot a Baikal IZH 60 or the copy, or the SPA SLR900, you will note the real absence of this "flip". That is because the piston runs in the stock, from the buttpad to the pistol grip. SO the support point is ABOVE the reaction of the gun.
To the shooter it is a flip, but in reality is a rotation around a point that is relatively far away from the origin of the reaction.
😉
Keep well and shoot straight!
HM
What we will show is that the TP geometry is NOT THAT important if other factors are taken into account.
The Cardews' work was exceptional for its time, but it is, now, dated
I have read Cardews cover to cover and I believe the only thing that is dated are the tools and technology. Same could be said for Einstein's theory of relativity; Given proper context. But dated? That is an EXTREMELY strong statement -given your own context. Same for TP geometry. Are we to believe that Bernoulli's principle no longer holds water as well?
Hector, you are more knowledgeable than I in airgun related terms so my hat is off to you sir. Your efforts are to BE applauded and I will enjoy following along....seriously! Nothing comes without science and its mathematics and using a "thing" in a controlled test also constitutes experience. I personally would love to see someone reinvent the wheel. That would be a pretty cool thing to see....
Best wishes in your endeavors 😀 😀
Bo
Bo;
It's not a question of who's "right" or "wrong". There can be more or less precise measurements. Sometimes, as in quantum mechanics, the simple FACT of taking a measurement, affects the outcome ("uncertainty principle").
In airguns things happen so fast, that the human senses have NO WAY to really unravel the whole picture, so SOME FORM of EASILY ACHIEVABLE instrumentation is important to the furthering of airguns themselves.
And AAMOF, I am NOT saying that our measurements are absolute and exact. On the contrary (and we can talk about this later, if you want).
What I am ASKING of everyone that is truly interested in airguns is to build the instrument, and test his/her guns. Even if they do not follow the calibration procedure that John came up with, that is amazingly simple to follow, just the fact that ANY "home tinkerer" can now COMPARE in very effective terms two different rifles in a non-subjective manner is a huge step forward
My HOPE is that by "Popular Mechanic-izing" the INSTRUMENT, we can all start talking in concrete terms about things that, so far, have been only subjective concepts, like "smoothness of shot cycle", "efficiency of the compression phase", "efficiency of the barrel", and even "pellet dwell time".
In the same way that the advent of, relatively inexpensive, Chronographs on the late '80's came to dispel a LOT of "MYTHS" surrounding the reloading, archery, and airgun world, I do hope that someday, we can all be talking in concrete terms about the firing characteristics of our guns, instead of the old "it kills deader than dead" of the old timers when they were talking about their favorite deer cartridge, or "it is a tack driver with a flat trajectory" of the target shooters. My other favourite one was the "this one bucks the wind, like it's not there" .
After the chronos became available, a lot of those expressions simply fell into disuse, and more solid approached could be followed.
Thanks for the kinds words, I do realized we ALL learn here, nobody is born with innate knowledge, nor with infallibility.
We, together, will drive the future with concrete evidences of what is really happening inside.
😉
Keep well and shoot straight!
HM
Gotcha. The excitement in your words and contributions to this sport are nothing short of well, exciting!
I agree, since the dawn of the internet ambiguous and arbitrary words/terms are/have been thrown about like snowflakes in the wind...guilty here LOL
I'm glad you did not interpret my response in a totally negative way. Large rocks that have been sitting for years can be hard to move. Who's to say who can move them?
With that, I'll shut up and listen since I have nothing to contribute -except that my R7 is "smooth as butter" 🙂
Wish you well,
Bo
for a springer
Not an April's Fools joke.
This will be a NINE part series.
We use three of the best recoiling spring-piston rifles ever made.
We will publish one part every two weeks.
Hope you all enjoy!
HM
Hector, I'm very much looking forward to your treatment of this complex topic, and in particular to learning some of the details of your "INSTRUMENT!"
Thanks in advance, Steve
First Chapter has been posted, where we look at Instruments we can use and instruments we can make to understand the dynamics of the spring-piston airgun:
Hope you enjoy!
HM
AND, more importantly IMHO, is that we will provide anyone that's reasonably handy with the tools and knowledge to create a machine to measure the vibrations in THEIR rifles
This will be an experiment in "Crowd-researching" that HOPEFULLy, will de-bunk some "theories" ("truths"?) that have been accepted and prevalent since the old days and that have hampered the real development of the spring-piston airgun.
So, stay tuned . . . 😉
I'm looking forward to this portion of the series. I made an instrument with an Arduino and a GY521 accelerometer that can be rubber-banded to a gun to test vibration in 3 axes. I can get 4 millisecond readings in its current configuration.
Below is one shot cycle for a .22 rimfire. The axes that should show the most movement do so it looks like it is performing as it should. I want to get the data point times lower but I haven't fooled with it for quite a while.
Nice packages, but I think you'll find the 16g fullscale limit of that accelerometer isn't adequate (like by a factor of 10x) to cope with legendary (scope-busting) springer recoil.
Steve;
Let me be clear: We are not friends.
I respect everyone in the airgun community. It's civil to do so.
You are free to believe what you want, but you are definitely "NOT GETTING IT".
I will repeat it here: The MAIN DRIVER of these experiences is to allow EVERYONE to build something that will allow EVERY TUNER AND TINKERER to compare a "before and after" rifle of HIS doing.
We are not claiming to produce absolute results, we are not aiming to achieve a different design of internals. We are just making it possible for everyone with the will to have an instrument to record the results of his efforts.
You seem to be attached to the idea that the airguns should be redesigned. That's a possibility, but a really long shot one. NO manufacturer is at present doing anything worthwhile as far as the internals are concerned. Architectures have been frozen in time since the 19th century for the breakbarrel and the 20th century for the sliding compression cylinder guns.
Other architectures have been explored and have ended by the wayside.
The sad story of the Walthers (LGU and LGV) is a sobering experience for the industry. Walther listened to the vociferantes in the community that assured they would pay $800 for an OoB "tuned airgun". Sure! about 300 of them paid the full price, ¿after that? You know the result. Apart from the limited supplies of the LGU Varmint still available here and there, everything else has been, or is in the process of being, discontinued.
So, trying to get a "pure" view of the behaviour of the airgun is rather futile. There is no future in that research.
Where there is future is in creating the ability for all to "see" what truly is happening in a gun set in a condition AS CLOSE AS POSSIBLE to the field usage.
Does the human body allow a completely rigid fixation of the gun?
Do your arms and hands have no weight?
Does your torso's mass play no role?
We have already established that whatever happens after the pellet exit is important only to the shooter's frame of mind, or do you still think that the recoil post-pellet-exit has some effects on the pellet's trajectory?
The consistency of whatever happens between the moment of piston release, and the moment of pellet exit is the all crucial time lapse at which the "tuning" efforts should be focused on.
The much vaunted "Smoothness of shot cycle" has been proven to be irrelevant to the accuracy, the consistency, and the trajectory of the pellet.
Again, taking into account a much bigger frame of reference: The pellet's POI relative to the initial "state of rest" will depend more on the architecture of the rifle, than on the internals.
Slight offsets between piston axis, barrel axis, point of support at the rear, and hand support will create rotational momenta in the X-Z plane that will make the barrel to be "looking" at different spots when the pellet exits.
Harmonics in the barrel now become more important than ever, that is an x-Y-Z space problem, and they will depend more on the architecture, than on the internals.
Efficiency is also now more evidently important because the less energy is wasted, the less energy is available to move the gun in those critical 8-10 ms between trigger pull and pellet exit, and the less the gun will move to change the POI of the pellet.
To close this: We KNOW this to be a first step. You want to go one better than us? By all means! Go ahead! Build a better sled, get better results, publish them for everyone to see. I am sure you are more than capable.
In the next chapter we will discuss the statistical tools needed to analyze the performance of rifles, and why the isolated 5 or even 10 shot group is quite a poor evidence of good performance.
When we swap the powerplants between the LGU and the LGV it will be even more apparent that the actual accuracy and precision of airguns depend more on things that are NOT the internals.
The road is still long ahead, so you have the time to get your improved device built and start putting in the hundreds (if not thousand +) shots that these experiments require to be statistically significant.
Keep well and shoot straight!
HM
Steve;
Let me be clear: We are not friends.
I respect everyone in the airgun community. It's civil to do so.
You are free to believe what you want, but you are definitely "NOT GETTING IT".
I will repeat it here: The MAIN DRIVER of these experiences is to allow EVERYONE to build something that will allow EVERY TUNER AND TINKERER to compare a "before and after" rifle of HIS doing.
We are not claiming to produce absolute results, we are not aiming to achieve a different design of internals. We are just making it possible for everyone with the will to have an instrument to record the results of his efforts.
You seem to be attached to the idea that the airguns should be redesigned. That's a possibility, but a really long shot one. NO manufacturer is at present doing anything worthwhile as far as the internals are concerned. Architectures have been frozen in time since the 19th century for the breakbarrel and the 20th century for the sliding compression cylinder guns.
Other architectures have been explored and have ended by the wayside.
The sad story of the Walthers (LGU and LGV) is a sobering experience for the industry. Walther listened to the vociferantes in the community that assured they would pay $800 for an OoB "tuned airgun". Sure! about 300 of them paid the full price, ¿after that? You know the result. Apart from the limited supplies of the LGU Varmint still available here and there, everything else has been, or is in the process of being, discontinued.
So, trying to get a "pure" view of the behaviour of the airgun is rather futile. There is no future in that research.
Where there is future is in creating the ability for all to "see" what truly is happening in a gun set in a condition AS CLOSE AS POSSIBLE to the field usage.
Does the human body allow a completely rigid fixation of the gun?
Do your arms and hands have no weight?
Does your torso's mass play no role?
We have already established that whatever happens after the pellet exit is important only to the shooter's frame of mind, or do you still think that the recoil post-pellet-exit has some effects on the pellet's trajectory?
The consistency of whatever happens between the moment of piston release, and the moment of pellet exit is the all crucial time lapse at which the "tuning" efforts should be focused on.
The much vaunted "Smoothness of shot cycle" has been proven to be irrelevant to the accuracy, the consistency, and the trajectory of the pellet.
Again, taking into account a much bigger frame of reference: The pellet's POI relative to the initial "state of rest" will depend more on the architecture of the rifle, than on the internals.
Slight offsets between piston axis, barrel axis, point of support at the rear, and hand support will create rotational momenta in the X-Z plane that will make the barrel to be "looking" at different spots when the pellet exits.
Harmonics in the barrel now become more important than ever, that is an x-Y-Z space problem, and they will depend more on the architecture, than on the internals.
Efficiency is also now more evidently important because the less energy is wasted, the less energy is available to move the gun in those critical 8-10 ms between trigger pull and pellet exit, and the less the gun will move to change the POI of the pellet.
To close this: We KNOW this to be a first step. You want to go one better than us? By all means! Go ahead! Build a better sled, get better results, publish them for everyone to see. I am sure you are more than capable.
In the next chapter we will discuss the statistical tools needed to analyze the performance of rifles, and why the isolated 5 or even 10 shot group is quite a poor evidence of good performance.
When we swap the powerplants between the LGU and the LGV it will be even more apparent that the actual accuracy and precision of airguns depend more on things that are NOT the internals.
The road is still long ahead, so you have the time to get your improved device built and start putting in the hundreds (if not thousand +) shots that these experiments require to be statistically significant.
Keep well and shoot straight!
HM
I'm sorry, Hector, that you have chosen to descend from an objective discussion of technical issues to subjective personal remarks.
But the facts are that you have chosen to publish results that were, from the getgo, in complete and obvious contradiction (e.g., claiming only 10s of g's of piston-bounce driven acceleration instead of 100s of g's) of decades of existing study and analysis of springer ballistics.
So in truth, you have produced "absolute results." Absolutely nonsensical and useless results.
"The much vaunted "Smoothness of shot cycle" has been proven to be irrelevant to the accuracy,"
Agreed!
Years ago I found that the new "outta tha box" .177 Beeman R9s I bought would shoot "snug fitting" 7.9 grain boxed Crosman Premiers into a 1/2" group at 30 yards, even in "factory twang mode" when sitting on a bucket resting the guns on cross sticks.
Anywhoo........if accuracy was the only issue, for me tuning isn't worth the effort with the HW springers!
"The much vaunted "Smoothness of shot cycle" has been proven to be irrelevant to the accuracy,"
Agreed!
Years ago I found that the new "outta tha box" .177 Beeman R9s I bought would shoot "snug fitting" 7.9 grain boxed Crosman Premiers into a 1/2" group at 30 yards, even in "factory twang mode" when sitting on a bucket resting the guns on cross sticks.
Anywhoo........if accuracy was the only issue, for me tuning isn't worth the effort with the HW springers!
Words of wisdom -- at least if group size alone is the sole definition of "accuracy."
But if "accuracy" is also meant to include how closely today's zero matches yesterday's and last week's, then maybe the stability of MV a good tune can provide is worth a significant investment of effort because it pays off in improved likelihood of actually hitting what we aim at, instead of merely missing -- albeit with admirable consistency. 😎
"at least if group size alone is the sole definition of "accuracy."
LOL, guilty as charged! I'm thinking that if a springer can't GROUP WELL then other definitions of accuracy doesn't matter. 🤣 I'm thinking that consistent the "inside 1/2" @ 30 yard accuracy with an untuned springer would probably be adequate for a large percentage of shooters but I do use some "adjustments" to make my springers more consistent. Still.......I've found that this SHOOTER (me) sitting on a bucket resting the gun on cross sticks, atmospheric conditions, etc, on any particular day has as much (or more) to do with accuracy than the hardware tune level!
A couple 25 yard groups using a home tuned Chinese .177 B3 cobbled together from two guns bought from a Cummins Truckload Sale for $19.95 each a couple decades ago using "irons" and a 6x scope..........
Here are a couple .177 HW95 groups shot more recently.........
LOL.....even the pellet brand used affects the accuracy. Notice the CPL group in the upper left corner of this target compared to a couple other brands, all shot at 50 yards...........
What would it take to rigidly attach the rifle to the sled. I was thinking about a trigger guard clamp but you also need access to the trigger.
David Enoch
What would it take to rigidly attach the rifle to the sled. I was thinking about a trigger guard clamp but you also need access to the trigger.
David Enoch
That's a very good question, David, that I was hoping someone would ask. Piston bounce generates tremendous chamber pressure and thereby force, as is estimated in the excellent paper that Hector cited and I lifted some numbers from...
...3303lbs of force against the face of a 1" chamber = 182g of reverse recoil...
That works out to 182g x 2.4lb = 437 pounds (typical weight of a baby grand piano!) of force acting against whatever anchors the gun to the 2.4lb sled, and I doubt a trigger guard could ever be expected to stand up to it!
One possibility I was pondering might be to clamp a 1" (or 30mm, depending on the rings mounted) rod in the scope mount, flip the gun upside down, and have a massive pin that would link the rod to the sled.
I really do like the magnetic inductive piickoff of sled motion that Hector and his friend devised -- very clever and elegant!
Something like this (with apologies to Ed for pinching his B3 photo)...
What would it take to rigidly attach the rifle to the sled. I was thinking about a trigger guard clamp but you also need access to the trigger.
David Enoch
Actually, David, recently I'm being nagged by an alternative goofy idea. Ready? Don't laugh. Here goes. Why not dodge the whole issue in its entirety?
Why not just lay the rifle on its side, entirely unconstrained, on a smooth flat surface so it can recoil freely without marring its finish, and let the rifle be its own sled?
At least you gotta' admit -- you can hardly imagine a more rigid attachment than of the rifle to itself!!
Then the only custom fabrication needed would be a miniature free-standing version of Hector's inductive velocity sensor, and data recording system to sense the rifle's movement.
How about suspended like a pendulum? Trigger actuated by a solenoid (pneumatic/electric).
Grabbity sux.
How about suspended like a pendulum? Trigger actuated by a solenoid (pneumatic/electric).
Grabbity sux.
That would work. One complication is you'd need some kind of rig to independently capture the maximum amplitude of the recoil swing to get the calibration constant used to scale the magnetic sensor's signals.
Every ballistic pendulum has one. See the "catcher pawl" in this classic design.
http://hyperphysics.phy-astr.gsu.edu/hbasees/Class/PhSciLab/balpen.html
Dear Steve,
Thanks for your insightful comments. Your questions have made think more deeply about our measurements, and as a result I think that I may understand things a bit better now. Please find below some more thoughts on what may be going on.
1. I agree that the accelerations and forces can be quite large here. As the rifle accelerates it pushes/pulls the sled applying a force that is the acceleration times the MASS of THE SLED (to get the sled moving with the rifle). The sled weight is about 2.4 pounds while the rifles weigh over 10 pounds with the scope. Did you use the sled weight or a typical rifle weight when you estimated the forces?
2. The high accelerations occur for a very short time, otherwise, as you aptly put they would send "the entire apparatus flying away." This makes them hard to observe, even with a perfect apparatus and I'm pretty sure our system is rounding the edges and probably missing the peak accelerations. On the other hand, the small impulses (ma*dt) that these accelerations give to the momentum of the rifle will not strongly affect the velocity or the position of the rifle as functions of time. There will be some sharper edges in the velocity traces that our apparatus has smoothed over.
3. I was curious if I could estimate some of the forces on the sled and I think we can do this for the initial force pretty easily. The spring has a spring constant of around 7 N/mm is compressed about 80 mm, so the initial force of the spring pushing on the piston and the piston pushing back on the rifle is around 560N (125 lbs). This causes the rifle to accelerate backward at 71 m/s^2 (assuming 8 kg rifle plus sled). To get the sled to accelerate with the rifle at this acceleration, the butt of the rifle has to push the 1.1 kg sled with around 78 N (1.1 kg*71 m/s^2) which is around 18 lbs. Here you can see the importance of Point 1! Unfortunately, it's much harder to estimate the accelerations later, but we can compare the measured accelerations. The later acceleration peaks are around a factor of three bigger than the initial acceleration dip, but that still puts the forces on the sled at around 50 lbs, which of course is still significant and makes it clear that the rifle needs to be well-secured to the sled. If I could do these measurements over, I would definitely use a stronger attachment system!
4. I saw very different behavior (no oscillations!) when I forgot to tighten the strap sufficiently, so having the rifle move from the sled can be a real problem. However, when I tightened the strap with all my strength, the results were pretty reproducible. Of course, one could argue that the rifle was reproducibly separating from the sled, but I think the separation, if there was any, was pretty small.
5. The key question is whether the sled (whose motion we're measuring) is moving differently than the rifle. I agree that it would be best to mount the motion sensor rigidly right on the rifle, as Jim Tyler does in his articles in Airgun World (please see first figure). Jim mounts the magnet to the rifle in a scope ring and lets the rifle recoil in a cradle. This is an excellent setup with no added weight from a sled to damp recoil and with a motion sensor that definitely moves with the rifle.
In the next figure I compared rifle position vs time from Jim's setup with ours. The results look pretty similar, both qualitatively and quantitatively.
6. My analysis of the piston motion provides results that are pretty reasonable, so I think that our data are also pretty reasonable. We may have missed some accelerations spikes, but since our goal was to compare three rifles using the same testing system and to look at the overall behavior, I think that we've done a reasonably good job. The goal of this work was not to determine the peak acceleration of airguns during recoil!
7. Mounting a sensor rigidly can also cause problems. If you mount motion sensor rigidly into the airgun, it can vibrate at high frequency (small mass on a very rigid spring) and report accelerations that are much higher than the actual acceleration of the center-of-mass of the rifle. Try mounting an accelerometer on a scope ring on a rifle and then hit the rifle with a wooden mallet. I would bet the accelerometer would go off the charts without the rifle moving much. Hector has a lot more experience with this than I do, so this is more of a question than an answer.
I agree that the backward acceleration dip near 0.02s could be due to the rifle moving forward off the sled's rear bracket and then getting pulled back onto the bracket by the strap, which causes the sled to recoil back from the collision. However, there are some arguments for this being a real effect. First of all, I would expect that the spring constant of the Velcro strap is pretty low and I don't think it could snap the rifle back in a few ms. Also the fact the the acceleration keeps ringing, even when the accelerations (and forces) are really small (and there's no question that the velcro should be holding the rifle firmly against the sled bracket) at later times, suggests that the entire sled is actually moving back and forth.
On the other hand, there looks like there's a bit of a discontinuity in the slope of the acceleration for the FWB 124 and LGV right at the start of this negative acceleration dip in Fig. 2.12, so something weird could be happening? I think the LGU acceleration in that figure looks pretty reasonable. Now that I'm looking more carefully, both the LGU and LGV exhibit a small shoulder (maybe even a peak) as the acceleration heads toward the second dip. Maybe we should ask Jim Tyler to share some of his data at longer times? His published traces usually focus on the first ~20 ms of recoil. In terms of practical use of airguns, we don't really care that much what happens after 10 ms, since the pellet has already left the barrel.
All measurements have limitations, but as long as it is clear how the measurement was done and what the limitations are, useful information can be extracted. If nothing else, we can explore fundamental ideas, gain some insights, and figure out how to do things better next time!
Best wishes,
John
Sorry for the late reply, I was doing some serious and real gunsmithing.
I have not turned this into a personal thing, on the contrary, I have disconnected the personal from the objective.
You call "respecting everyone" and being civil a descent?
Again, Steve you are free to think and believe whatever you want. You dodged the simple questions of HOW A GUN IS USED IN REALITY, and I have not seen even a "hammock" test from you.
Challenge still stands: Build your own device and publish.
Keep well and shoot straight!
HM
Dear Steve,
Thanks for your insightful comments. Your questions have made think more deeply about our measurements, and as a result I think that I may understand things a bit better now. Please find below some more thoughts on what may be going on.
1. I agree that the accelerations and forces can be quite large here. As the rifle accelerates it pushes/pulls the sled applying a force that is the acceleration times the MASS of THE SLED (to get the sled moving with the rifle). The sled weight is about 2.4 pounds while the rifles weigh over 10 pounds with the scope. Did you use the sled weight or a typical rifle weight when you estimated the forces?
2. The high accelerations occur for a very short time, otherwise, as you aptly put they would send "the entire apparatus flying away." This makes them hard to observe, even with a perfect apparatus and I'm pretty sure our system is rounding the edges and probably missing the peak accelerations. On the other hand, the small impulses (ma*dt) that these accelerations give to the momentum of the rifle will not strongly affect the velocity or the position of the rifle as functions of time. There will be some sharper edges in the velocity traces that our apparatus has smoothed over.
3. I was curious if I could estimate some of the forces on the sled and I think we can do this for the initial force pretty easily. The spring has a spring constant of around 7 N/mm is compressed about 80 mm, so the initial force of the spring pushing on the piston and the piston pushing back on the rifle is around 560N (125 lbs). This causes the rifle to accelerate backward at 71 m/s^2 (assuming 8 kg rifle plus sled). To get the sled to accelerate with the rifle at this acceleration, the butt of the rifle has to push the 1.1 kg sled with around 78 N (1.1 kg*71 m/s^2) which is around 18 lbs. Here you can see the importance of Point 1! Unfortunately, it's much harder to estimate the accelerations later, but we can compare the measured accelerations. The later acceleration peaks are around a factor of three bigger than the initial acceleration dip, but that still puts the forces on the sled at around 50 lbs, which of course is still significant and makes it clear that the rifle needs to be well-secured to the sled. If I could do these measurements over, I would definitely use a stronger attachment system!
4. I saw very different behavior (no oscillations!) when I forgot to tighten the strap sufficiently, so having the rifle move from the sled can be a real problem. However, when I tightened the strap with all my strength, the results were pretty reproducible. Of course, one could argue that the rifle was reproducibly separating from the sled, but I think the separation, if there was any, was pretty small.
5. The key question is whether the sled (whose motion we're measuring) is moving differently than the rifle. I agree that it would be best to mount the motion sensor rigidly right on the rifle, as Jim Tyler does in his articles in Airgun World (please see first figure). Jim mounts the magnet to the rifle in a scope ring and lets the rifle recoil in a cradle. This is an excellent setup with no added weight from a sled to damp recoil and with a motion sensor that definitely moves with the rifle.
In the next figure I compared rifle position vs time from Jim's setup with ours. The results look pretty similar, both qualitatively and quantitatively.
6. My analysis of the piston motion provides results that are pretty reasonable, so I think that our data are also pretty reasonable. We may have missed some accelerations spikes, but since our goal was to compare three rifles using the same testing system and to look at the overall behavior, I think that we've done a reasonably good job. The goal of this work was not to determine the peak acceleration of airguns during recoil!
7. Mounting a sensor rigidly can also cause problems. If you mount motion sensor rigidly into the airgun, it can vibrate at high frequency (small mass on a very rigid spring) and report accelerations that are much higher than the actual acceleration of the center-of-mass of the rifle. Try mounting an accelerometer on a scope ring on a rifle and then hit the rifle with a wooden mallet. I would bet the accelerometer would go off the charts without the rifle moving much. Hector has a lot more experience with this than I do, so this is more of a question than an answer.
I agree that the backward acceleration dip near 0.02s could be due to the rifle moving forward off the sled's rear bracket and then getting pulled back onto the bracket by the strap, which causes the sled to recoil back from the collision. However, there are some arguments for this being a real effect. First of all, I would expect that the spring constant of the Velcro strap is pretty low and I don't think it could snap the rifle back in a few ms. Also the fact the the acceleration keeps ringing, even when the accelerations (and forces) are really small (and there's no question that the velcro should be holding the rifle firmly against the sled bracket) at later times, suggests that the entire sled is actually moving back and forth.
On the other hand, there looks like there's a bit of a discontinuity in the slope of the acceleration for the FWB 124 and LGV right at the start of this negative acceleration dip in Fig. 2.12, so something weird could be happening? I think the LGU acceleration in that figure looks pretty reasonable. Now that I'm looking more carefully, both the LGU and LGV exhibit a small shoulder (maybe even a peak) as the acceleration heads toward the second dip. Maybe we should ask Jim Tyler to share some of his data at longer times? His published traces usually focus on the first ~20 ms of recoil. In terms of practical use of airguns, we don't really care that much what happens after 10 ms, since the pellet has already left the barrel.
All measurements have limitations, but as long as it is clear how the measurement was done and what the limitations are, useful information can be extracted. If nothing else, we can explore fundamental ideas, gain some insights, and figure out how to do things better next time!
Best wishes,
John
John,
Thanks so much for your generous and thoughtful reply to my comments. First of all, I have been inexcusably remiss in failing to adequately compliment you for the ingenuity and elegance of your design and implementation of your inductive sensor and of the data acquisition and reduction hardware and software that process its signal. Congratulations!
Then let me see if I can respond to some elements of your post.
1. ... Did you use the sled weight or a typical rifle weight when you estimated the forces?
I used a combined mass of 18.1lbs (I hope I copied that correctly from your Chapter 1) for rifle + sled, (15.7lbs for the rifle and 2.4lbs for the sled, so that forces generated by piston acceleration would divide between between rifle and sled in a 15.7:2.4 ratio. Taking a number for peak compression chamber of 29Mpa from the Tavella paper that Hector kindly linked to (which by the way is entirely consistent with other studies of spring piston thermodynamics), produced the estimate of 3303lbs of force acting on the whole apparatus, hence 3303 / 18.1 = 182g x 2.4 = 438lbs coupled to the sled.
2. ... On the other hand, the small impulses (ma*dt) that these accelerations give to the momentum of the rifle will not strongly affect the velocity or the position of the rifle as functions of time.
Sorry, but I don't follow. The impulse delivered by piston bounce to the rifle is indeed brief, but in total magnitude is actually larger than the total impulse delivered by the mainspring, since it not only stops but totally reverses the initial recoil motion of the rifle. Therefore, since dt is small, but ma*dt is large, f = ma is therefore huge. Accurate measurement of these quantities would therefore seem to be vital to accurate understanding of springer dynamics. Right?
3. I was curious if I could estimate some of the forces on the sled and I think we can do this for the initial force pretty easily. The spring has a spring constant of around 7 N/mm is compressed about 80 mm, so the initial force of the spring pushing on the piston and the piston pushing back on the rifle is around 560N (125 lbs). This causes the rifle to accelerate backward at 71 m/s^2 (assuming 8 kg rifle plus sled). To get the sled to accelerate with the rifle at this acceleration, the butt of the rifle has to push the 1.1 kg sled with around 78 N (1.1 kg*71 m/s^2) which is around 18 lbs.
Although it may not take into account the amount the mainspring is compressed during assembly (a.k.a., preload) I agree it's a good working ballpark number.
Here you can see the importance of Point 1! Unfortunately, it's much harder to estimate the accelerations later, but we can compare the measured accelerations. The later acceleration peaks are around a factor of three bigger than the initial acceleration dip, but that still puts the forces on the sled at around 50 lbs, which of course is still significant and makes it clear that the rifle needs to be well-secured to the sled. If I could do these measurements over, I would definitely use a stronger attachment system!
Here we must part company. Both established modelling and measurement of springer ballistics set typical ratios of acceleration due to mainspring vs piston bounce forces at, not ~3:1 but ~30:1. The Tavella paper cited above is a good example.
4. I saw very different behavior (no oscillations!) when I forgot to tighten the strap sufficiently, so having the rifle move from the sled can be a real problem. However, when I tightened the strap with all my strength, the results were pretty reproducible. Of course, one could argue that the rifle was reproducibly separating from the sled, but I think the separation, if there was any, was pretty small.
Well, John, not to minimize the significance of "all your strength," but how closely would you estimate the force so generated equates with the Tavella-derived estimate I make above of 438lbs of tension in the strap at the peak of piston bounce, strap tension being the only force that, at that instant, opposes the departure of butt pad from vertical stop?
Given this point of divergence between our understanding of the issues involved, I'll defer further comment for the moment, and urge instead that you peruse perhaps in somewhat greater detail the analysis in the Tavella paper, and better reconcile the numbers found therein with your measurements.
I look forward to your observations.
Thanks again for a fascinating discussion!
KR,
Steve
Hi All,
I'm the Jim whose recoil measuring rig is in the photograph posted by John. First, thanks for letting me join the forum, and a brief description of my progress in measuring recoil.
I started off in 2013 with the rifle action mounted in a sliding cradle with the magnet attached to the cradle but, try as I might, I could not prevent the action from moving within the cradle, so I devised the rig shown. It comprises a static wooden cradle with an upright at either end, one with a small 'U' for the barrel to slide in, the other a larger U for the cylinder; both are lined with short pile carpet. There is of course some friction between the steel and carpet, but it damps out non-axial vibration which proved a problem with low friction bearings.
I currently have a 400G accelerometer, which I use only for recording accelerations, and much prefer my linear generator to record velocities, which integrate into much more accurate displacement graphs, and which comprises a neodymium rod magnet (set as near the action as possible in a scope mount), coil and oscilloscope.
Here's an example of a UK LGU recoil derived from velocity; it differs hugely from John's largely due to the UK available piston stroke being 88mm, which explains why the 7.87 gn JSB pellet exits (at 818fps) post piston bounce (the red line on the graph). In fact, every pellet I have ever tested in a UK springer exits post piston bounce, and therefore during surge, and I am very grateful to John for opening my eyes to the fact that things are different with much longer piston strokes.
As there is some discussion on accelerations, I'll post a graph of the accelerometer data from a .20" HW95. In this, I drilled a tapped an axial hole in the trigger block of the rifle to accept the accelerometer, and tested the bare action (no stock) in my cradle. With 5 mV representing one G, the maximum G at piston bounce is a little over 240. Because of the dangers of trying to cock a bare HW95 action which you cannot brace against your body due to the accelerometer, and mindful of not trapping your fingers between the cocking lever and cylinder, I do not recommend trying this at home.
Dear Jim,
Thanks for contributing to this thread. This forum is greatly enriched by your participation! I looked more carefully at your LGU plot and have realized that there is another important limitation on the system I built, that is probably more important and fundamental than the strap problem. In order to maximize the sensitivity of the pickup coil, I put a lot of turns in the coil. The induced voltage grows with the number of turns, but so does the response time of the system, which is proportional to the inductance (and therefore the number of turns) of the coil. If you look at the first two dips in Jim's plot on the top of the figure below, you'll see that the dips are much sharper and have a bigger curvature (more tightly curved at the bottom, as shown by red "u" in plots). On the other hand, the curvature in the two dips in my data is much lower. The acceleration is proportional to the curvature of the position vs time trace, so clearly my system is recording much lower accelerations. I think that this is due to the slower response time of my system, which averages the signal out over longer times and misses the sharp features that Jim's system measures. Jim told me that he uses a single layer of wire in his coil, so I'm sure that he has far fewer turns and therefore a much faster response time which can catch the sharper wiggles (to use a technical term!) in the recoil response. I should add that my LGU is pretty heavy, over 18 lbs, which would also smooth out the wiggles. If you look at the FWB 124 and LGV plots in Fig. 2.12, you'll see some sharper features, but they probably also have been smoothed out a bit by my slower system.
My data also shows larger amplitude oscillations which could be due to the lower friction of the ball bearing slides on the sled compared to Jim's setup with the rifle sliding on the carpeted cradle, which damps the motion more. Jim, does this sound reasonable?
I also noticed that the pellet exit time in my data occurred well before the first dip. I think that this is real and I found Jim's explanation very enlightening and interesting. I assumed that the UK LGUs use the same piston stem length as US LGUs and therefore have the same piston travel. Assuming that the piston bounce disturbs the muzzle orientation, this could be a big advantage of the longer piston travel of US LGUs and makes me wonder why people in the UK don't use the longer piston travel with weaker springs, which would still keep them under 12 ft lbs. Thanks for bringing up this point, which shows that our sled data may have some uses after all!
Again, I'd like to emphasize that Hector and I are not claiming definitive measurements of recoil acceleration, but just wanted to compare the recoil behavior of three springers. I still think the results are useful and hopefully interesting. For future recoil measurements, we should try fewer turns in the coil and attach the magnet directly to the rifle.
Thanks,
John
Dear John,
Many thanks for your kind comments, and I think you’re right about the response time.
Just to give everyone an idea of the problems in gaining precise recoil data, this is my 400G wide band accelerometer mounted low on a cut down scope mount. You might think that this arrangement, with the accelerometer axis some 8.5mm above the cylinder, would be rigid, but you would be wrong when the accelerations of the springer are at work.
This is the same graph of my HW95 as in my post above, but with both the trace from the accelerometer mounted direct on the trigger block (blue) and on the scope mount (red). The flex in the mount loses some vibrations from the initial acceleration (slightly reducing apparent recoil displacement), but just look at the traces post piston bounce.
Integrating the acceleration data once for recoil velocity, and again for recoil displacement shows how misleading data can be if there is any less than absolute rigidity between the rifle and sensor.
With the sledge system I would be concerned with the possibility of the recoil pad and the magnet mount compressing minutely, delaying the onset of voltage generation. The light gate would know nothing of this, and faithfully record the passing of the pellet a fraction earlier relative to the compression stroke. Compression stores potential energy, which is later realised as kinetic energy at the onset of piston bounce.
Why do we in the UK persist with shorter strokes? It’s a long story that goes back to the formative years of field target, and not pertinent to this thread, so maybe some other time.
Best wishes,
Jim
Edit: I forgot to add that methods of recording recoil might fall some way short of absolute precision, but can be very revealing when used for comparative purposes, as you and Hector are doing.
... there is another important limitation on the system I built, that is probably more important and fundamental than the strap problem. In order to maximize the sensitivity of the pickup coil, I put a lot of turns in the coil. The induced voltage grows with the number of turns, but so does the response time of the system, which is proportional to the inductance (and therefore the number of turns) of the coil.... I think that this is due to the slower response time of my system, which averages the signal out over longer times and misses the sharp features that Jim's system measures. Jim told me that he uses a single layer of wire in his coil, so I'm sure that he has far fewer turns and therefore a much faster response time which can catch the sharper wiggles (to use a technical term!) in the recoil response.
John,
I'd like to specifically address your concern about inductance-related response times.
1. The timeconstant t of a inductance L (Henrys) working into a load resistance R (Ohms) is:
t = L / R
https://www.electronics-tutorials.ws/inductor/lr-circuits.html
2. The inductance of an air-core (u = 1) coil is given by:
http://hyperphysics.phy-astr.gsu.edu/hbase/electric/indsol.html
I think if you plug the dimensions and number of turns of your pickup coil (e.g., cm and single-digit 1000s of turns) into this formula, you'll likely get an inductance well below 0.1Hy. Working into the 1M input impedance of your 'scope therefore predicts a timeconstant less than:
0.1H / 1M = 0.1us
Therefore perhaps not such an important limitation, after all?
I realize I'm getting monotonous, but I really think the issues here lie more with the "timeconstants" of buttpads and velcro than electronics.
Thanks for looking,
Steve
Hi Steve,
I agree. I should have calculated the response time before posting any speculations. On the millisecond timescale, the sled-rifle system is pretty wobbly and we're only recording the average position of the rifle. I don't think that this creates any new structure in the traces, but simply smooths out the finer structure that happens at shorter timescales. Of course, this finer structure is needed to get the peak accelerations, so it's unfortunate that we're missing it.
Best wishes,
John
... I don't think that this creates any new structure in the traces, but simply smooths out the finer structure that happens at shorter timescales...
Hi, John
Unfortunately, I doubt even that conjecture is true. As I posted earlier in this thread...
I find it unlikely that these peculiar artifacts are telling us anything about what's going on inside the guns during the firing cycle.
Instead, I suspect what's really happening is, in each case, the buttplate is being pulled away from the vertical stop by piston bounce acceleration of the rifle stretching the velcro strap. Subsequently, the butt slams back into contact with the stop, jolting the sled backward and creating the spikes.
Which therefore have nothing to do with any real events in the actions of the guns per-se, and are in fact (LARGE!) "new structure" created by lack of rigidity in the connection between rifle and sled.
Gosh, it never fails to impress-me to-no-end, just-how complex it is to capture-and-measure the dynamic forces involved-with a "simple machine" at work.
It is pure physics ... plain and simple.
To me, it's a wonderment that an engineer can design a relatively-light spring-and-piston air rifle that a person can expect to shoot accurately with regularity.
Gosh, it never fails to impress-me to-no-end, just-how complex it is to capture-and-measure the dynamic forces involved-with a "simple machine" at work.
It is pure physics ... plain and simple.
And, it's a wonderment that an engineer can design a relatively-light spring-and-piston air rifle that a person can expect to shoot accurately with regularity.
Me too, Alex! The deceptively simple spring-piston action accomplishes an (almost) miraculous trick by (reasonably) efficiently coupling energy released by the slooow expansion of the (relatively) massive mainspring into the (MUCH) faster velocity of the flea-weight pellet. The phenomenon that makes possible bridging between these two events separated as they are by an order of magnitude difference in speed and time scale is, of course, our old friend (and scopes' nemesis) piston bounce.
Well, parts are on order (mainly Amazon), so with luck I hope to have some results from the KISS free-sliding-airgun method sometime in the next few weeks. I have at least 5 springers (2 Chinese, 1 Spanish, and 2 British) available for initial testing
I intend to use "Audacity" s/w for data acquisition, at least initially.
Here's a (rough) sketch of the pickup I'll be using, that being the only hardware required -- besides a laptop computer with the usual built-in audio, a rug for the gun to slide on, and a nice thick catalog for backstop duty.
Hi Jim,
Brilliant stuff! Thanks so much for sharing it.
A question: Would you agree that the second forward recoil spike at ~21ms is created by hard contact of the piston against the end of the compression chamber at the end of its to-fro-to motion cycle, most of the chamber air having been expelled through the transfer port and empty bore (the pellet having exited), leaving too little chamber pressure to stop the piston a second time before impact?
Thanks,
Steve
Hi Steve,
After looking at Jim's data (thanks Jim for posting and sending me the data!), it looks like there should be two large and sharp acceleration spikes due to the piston going backwards (and/or stopping its forward motion), first at the piston bounce and then at the piston landing at the front of the compression tube. Both of these spikes cause the rifle to move forward. On the other hand, the accelerations of the piston moving forward (and the rifle accelerating backward) are much more gentle. It seems to me that this asymmetry is due to the fact that there are much "harder" media in front of the piston (highly compressed gas at the bounce and the steel cylinder front wall at piston landing) compared to the softer spring behind the piston. Jim, please let us know if this makes sense. Please see below some beautiful data that Jim sent me from his LGV recoil measurements:
I agree that our acceleration data misses these features and probably adds artifacts such as the Velcro strap pulling the rifle butt against the sled rear bracket. I think our initial acceleration dip may not be too far off. In the LGV, which is closest to the weight of Jim's test rifles, albeit it has another 2.4 lbs of sled moving with it, the first acceleration dip goes down to around -20g. The rubber buttpad will decrease the measured acceleration, but I've tried measurements with the LGU's steel buttplate hardware pushing directly against the steel rear sled bracket and haven't seen much of a difference compared to the rubber buttpad pushing on the bracket. If I'm reading Jim's acceleration plot (posted on May 9) correctly, he sees about 40g's in the initial acceleration maximum (I think he's using positive acceleration to indicate the rifle is moving backward, please correct me if I'm wrong here) and then -240g's at the piston bounce and similar value at the piston landing.
My earlier point discussed the smoothing out of features in the velocity and position traces. I wasn't claiming that our acceleration plots were simply smoothed out versions of the actual acceleration. The velocity and position traces look qualitatively (and even quantitatively, in terms of magnitudes) similar to Jim's results, but the the fine structure in the velocity trace is really critical in determining the acceleration and our velocity data are clearly missing that structure (and may have some extra kinks due to artifacts such as the rifle bouncing in the sled). I think the first 10 ms of data, when the rifle is pushing back on the sled, are pretty reasonable, but we're really missing the rifle's forward acceleration at the piston bounce and piston landing at later times. I think the smaller acceleration oscillations at later times may be real, since the forces are much smaller and the Velcro strap is probably strong enough to handle those forces. Fortunately, the pellet is out of the bore before the piston bounce, so in terms of accuracy tuning, maybe the first 10 ms are the most critical?
I'm glad to see that you're building your own setup. I'm sure that you've already thought of this, but please be aware that pc sound cards are ac-coupled, so you won't be able to use the calibration technique that I described in Ch. 1.
Best wishes,
John
"Fortunately, the pellet is out of the bore before the piston bounce,"
John, on what do you base this statement? I don't think it's true.
Please note Jim's comment: "In fact, every pellet I have ever tested in a UK springer exits post piston bounce,"
Also please consider the following elementary argument:.
1. The pressure acting on the base of the pellet can never be greater (in fact due to the flow resistance of the transfer port, it must in fact be at least somewhat less) than the pressure acting on the face of the piston.
2. Assuming ~30% springer energy efficiency, while the pressure decelerating the piston toward the instant of bounce is absorbing (at most) ~70% of mainspring energy, similar pressure accelerating the pellet is delivering ~30% of the same mainspring energy to the pellet. 30/70 = 0.43. Therefore, if the pellet is assumed to exit the bore no later than the instant when the piston stops, the average force accelerating the pellet down the bore must at least 43% of the average force decelerating the piston.
3. Force = pressure x area. Therefore the ratio of the area of the base of the pellet to that of the piston face would have to be at least 43%.
4. Typical piston face area is ~0.78in^2, but the base area of a .177 pellet is only 0.025in^2.
5. Calculation of the ratio 0.025:0.78 is left as an exercise to the reader.
6. Do you still think the pellet can exit prior to bounce?
Hi Jim,
Brilliant stuff! Thanks so much for sharing it.
A question: Would you agree that the second forward recoil spike at ~21ms is created by hard contact of the piston against the end of the compression chamber at the end of its to-fro-to motion cycle, most of the chamber air having been expelled through the transfer port and empty bore (the pellet having exited), leaving too little chamber pressure to stop the piston a second time before impact?
Thanks,
Steve
In that instance, Steve - yes.
In other cases, though, there's enough air left to cause a second piston bounce.
The mainspring can also be throwing its mass backwards and forwards enough to possibly emulate a piston bounce.
"Fortunately, the pellet is out of the bore before the piston bounce,"
John, on what do you base this statement? I don't think it's true.
Please note Jim's comment: "In fact, every pellet I have ever tested in a UK springer exits post piston bounce,"
Also please consider the following elementary argument:.
1. The pressure acting on the base of the pellet can never be greater (in fact due to the flow resistance of the transfer port, it must in fact be at least somewhat less) than the pressure acting on the face of the piston.
2. Assuming ~30% springer energy efficiency, while the pressure decelerating the piston toward the instant of bounce is absorbing (at most) ~70% of mainspring energy, similar pressure accelerating the pellet is delivering ~30% of the same mainspring energy to the pellet. 30/70 = 0.43. Therefore, if the pellet is assumed to exit the bore no later than the instant when the piston stops, the average force accelerating the pellet down the bore must at least 43% of the average force decelerating the piston.
3. Force = pressure x area. Therefore the ratio of the area of the base of the pellet to that of the piston face would have to be at least 43%.
4. Typical piston face area is ~0.78in^2, but the base area of a .177 pellet is only 0.025in^2.
5. Calculation of the ratio 0.025:0.78 is left as an exercise to the reader.
6. Do you still think the pellet can exit prior to bounce?
Steve, pellet exit is determined, all other things being equal, by the length of the compression stroke, shorter strokes push pellet exit further into surge (piston bounce), longer strokes advance pellet exit.
All the rifles I test are UK spec. The UK LGU available stroke is 88mm, the LGV 90mm, so pellet exit is post bounce - but not a lot. The longer strokes of US spec rifles can indeed see pellet exit precede bounce.
To add a bit of detail, in my TX200 with 85mm of stroke, the piston has 13mm of available stroke left as the AADF pellet starts to move, but in John's LGU, I believe the stroke is 130mm (please correct me if I'm wrong), so there's a tad under 20mm of stroke.
The LGU piston is decelerating over a greater distance, and the pellet only needs a couple of milliseconds and a bit to get clear of the muzzle.
@johnc "It seems to me that this asymmetry is due to the fact that there are much "harder" media in front of the piston (highly compressed gas at the bounce and the steel cylinder front wall at piston landing) compared to the softer spring behind the piston. Jim, please let us know if this makes sense."
In a word, yes, John.
The highly compressed air in front of the piston at the end of the compression stroke is highly energetic, manifest by its highly elevated temperature (itself a measure of the air's internal kinetic energy). The piston has run out of momentum, and the spring is down to its preload plus a mm or two of potential energy - a fraction of a foot pound.
The air wins.
Thanks as always, Jim, for the expert commentary.
One caveat: Please be careful not to confuse folks by inadvertently conflating "pellet start" (when breech pressure first becomes high enough to create the force required to overcome obturation, static friction, etc.) with "pellet exit" (when breech pressure becomes irrelevant to MV).
I'm totally confident that you're using these (and other) terms absolutely correctly, but when they appear in the same post, as here, some folks might lose sight of the difference. Especially if they want to.
Start-up ritual for every accelerometer session. Place the accelerometer horizontal and check that it reads 0v. Place it vertical and remind yourself (if you need reminding) how many volts equals one G.
It's difficult to get wrong, Steve, even for me.
I'm keeping an open mind and trying to dream up a way of accurately recording peak pressure or temperature, or piston stroke, which may be the easiest.
Well Jim, while I admire the elegant simplicity of the theory of your calibration method, I'm not so sure about the practice side.
A long (and sometimes even allegedly productive) career involving frequent hand-to-hand combat with uncooperative instrumentation taught a number of hard-learned lessons.
One of these was that attempting to calibrate a sensor to a signal that is less than 1% of its full-scale range often runs afoul of any number of unexpected nonlinear complications: e.g., hysteresis. Not saying that happened here, but I'd worry.
Start-up ritual for every accelerometer session. Place the accelerometer horizontal and check that it reads 0v. Place it vertical and remind yourself (if you need reminding) how many volts equals one G.
It's difficult to get wrong, Steve, even for me.
I'm keeping an open mind and trying to dream up a way of accurately recording peak pressure or temperature, or piston stroke, which may be the easiest.
Well Jim, while I admire the elegant simplicity of the theory of your calibration method, I'm not so sure about the practice side.
A long (and sometimes even allegedly productive) career involving frequent hand-to-hand combat with uncooperative instrumentation taught a number of hard-learned lessons.
One of these was that attempting to calibrate a sensor to a signal that is less than 1% of its full-scale range often runs afoul of any number of unexpected nonlinear complications: e.g., hysteresis. Not saying that happened here, but I'd worry.
I didn't know that, Steve. Might comparing the displacement from the accelerometer data with that from the linear generator (scaled to fit) be a reasonable check for nonlinear issues?
Way above my head, but thanks for the explanation.
I'm sorry, Jim. Please excuse my incoherent babbling and let me try again.
The question was whether chamber pressure averaged over an estimated 3ms of pellet bore transit time is at least high enough to equal 11fpe / 335mm / pellet-base-area = 407psi.
To answer it, I averaged the 148 g-force readings (i.e. from 1.5ms prior to the pressure peak to 1.5ms after) and got 31mV. I then divided that result by 72.2mV, the amplitude of the peak pressure g-force, got 43%, and multiplied that by the previously calculated 594psi.
Voila! The maximum possible average pressure over the 3ms and 11 inches of pellet acceleration, assuming it to be centered on the peak for the highest theoretically possible ME: 255psi.
Which is only enough for 7fpe.
Better?
(edited for multiple typos)
Much better, thanks, Steve, though if I had to describe your post, I would never use the term 'incoherent babbling'.
I didn't know that, Steve. Might comparing the displacement from the accelerometer data with that from the linear generator (scaled to fit) be a reasonable check for nonlinear issues?
Any sort of cross-check might be revealing.
I didn't know that, Steve. Might comparing the displacement from the accelerometer data with that from the linear generator (scaled to fit) be a reasonable check for nonlinear issues?
On the subject of accelerometer nonlinearity, awhile back you posted this interesting plot, which seems to show both red and blue post-firing-cycle "final" positions accelerating away at approximately the same (albeit in opposite directions) 2mm/s/10ms = 0.02g. Failing to integrate a closed trajectory back to zero is just the sort of nasty trick nonlinearity can play.
I'm indebted to Jim in UK for this lovely example of what an inductively sensed recoil plot ideally looks like when the sensor is rigidly attached to the rifle. Note in detail the respective shapes of the leading and trailing edges of the initial recoil velocity peak, especially the practically vertical rise of the latter due to violence of piston bounce deceleration.
Thanks, Jim. You set the standard. Of course that's only fair -- since you invented the thing!
John,
In your description of the recoil measurement apparatus, you say...
...for an air rifle one can neglect the momentum of the pellet, which in this case is around 40 times smaller than the momentum of the piston or rifle.
On first reading, this made perfect sense, because in typical shooting, you certainly can.
But just now I got to pondering if, in the context of the method of sensor calibration you're using -- in which the ratio of sensor signal to sled velocity is calibrated using the total displacement of the sled -- you really can safely neglect the pellet's momentum?
Please check my arithmetic.
In the course of firing a round, your 18.1lbs = 126,700 grains of rifle+sled will expel ~8 grains of pellet at ~250m/s + ~1 grain of superheated air muzzle blast at ~500m/s. Net velocity imparted to sled will therefore be...
V = (8 x 250 + 1 x 500) / 126700 = 0.0197m/s = ~20mm/s
Unless I slipped a decimal place somewhere, it would seem that, since the total displacements you measure and use for calibration are typically less than 10mm, the potential for pellet-related recoil to move the sled that far in only half a second might be a bit more than you can afford to "neglect."
In fact, I think this ~2mm/100ms residual velocity can actually be seen in some of your position plots.
If the sled continued to coast for a significant fraction of a second, and therefore potentially several mm after the end of data recording, what effect would this have on your calibration calculation?
Wouldn't it result in a significant over-estimation of the calibration constants for both velocity and acceleration?
92g x 11.375lb = 1067lb 1067lb x (25mm /2/25.4)^2 x Pi = 812psi
Thanks to Jim-in-UK's eagle eye catching it, please note that this calculation is wrong. I multiplied acceleration force by piston area instead of dividing. Duh! It should be:
92g x 11.375lb = 1067lb 1067lb / ((25mm /2/25.4)^2 x Pi) = 1402psi
Thanks, Jim!
Progress! The coil is wound and tested -- worked out to about 200' and about 1500 turns of 40AWG, secured with a bit of Teflon pipe tape.
The DS212 scope is on order.
...I'm glad to see that you're building your own setup. I'm sure that you've already thought of this, but please be aware that pc sound cards are ac-coupled, so you won't be able to use the calibration technique that I described in Ch. 1.
Best wishes,
John
I'm also considering possible ways to correct the AC coupled data, depending on what the low frequency response of my computer's audio h/w actually turns out to be.
The simplest model of a 20Hz -3dB per octave rolloff is a single-pole RC 8ms timeconstant with a step-function response such that...
Y(t) = Y(0)exp(-t/0.008)
That could theoretically be corrected back to DC with an exponential summation.
But as the wise(ass) man said:
In theory there's no difference between theory and practice.
But in practice, there IS!
John,
I think I may have made good on that conjecture, making it possible to acquire DC data with a PC mic input with just a relatively simple post-acquisition spreadsheet mathematical massage.
Please consider...
Ai: (i = 1 to N) Array of N acquired AC-coupled input data.
t = Time between input samples (typically 1 / 44kHz for digital audio)
T = RC timeconstant of the audio input, typically T = 1 / (2PiFo) where Fo = low frequency -3dB cutoff = 8ms for 20Hz
Di: (i = 1 to N) = Array of DC corrected output data.
Then Di = Ai + SUM(A1:Ai) * (e^(t / T) - 1)
Below find an example of the algorithm working on a step function. Blue is the raw Ai data representing what's left of a stepped signal after passing through AC coupling, red is the corrected Di array produced by the algorithm, identical to the original signal.
So -- maybe a scope isn't necessary to acquire DC data after all, which might move participation in this project within the bounds of enthusiasm of even more potential participants.
Hi Steve,
Do you have any way to put a square wave with with ~10ms period into your pc audio input? If you had a function generator or maybe another pc where you use the output of the audio jack, you could characterize the ac coupling of your audio input? If you can fiddle with the parameters got get the square wave back, then the problem is pretty much solved. One challenge with modelling is that you can't be sure exactly how and where your ac coupled pc input starts cutting out lower frequencies, so actually measuring the ac coupling response with a well-know input could help.
The DS 2212 scope will make things a lot more straightforward.
I agree that the sled is still moving after 0.1s and this most likely due to the recoil momentum opposing the pellet momentum, as you suggested. In principle the sled should keep moving back forever, but friction eventually wins. When I did the calibration by moving the sled by hand (fig. 1.7), I actually stopped the sled against a steel bracket, so I think the calibration is still reasonably good. I should have done some longer time scans to see where/when the sled stops moving. I don't think the sled moved much after 0.1s, but it certainly could have moved a few more mm. I'd be happy if the calibration was accurate to ~10%, but the it could have been a little worse, but not much worse. Again, the tests here are focused on comparisons of the behavior of three rifles, so I wasn't too concerned about absolute accuracy.
Best wishes,
John
Hi Steve,
Do you have any way to put a square wave with with ~10ms period into your pc audio input? If you had a function generator or maybe another pc where you use the output of the audio jack, you could characterize the ac coupling of your audio input? If you can fiddle with the parameters got get the square wave back, then the problem is pretty much solved. One challenge with modelling is that you can't be sure exactly how and where your ac coupled pc input starts cutting out lower frequencies, so actually measuring the ac coupling response with a well-know input could help.
The DS 2212 scope will make things a lot more straightforward.
I agree that the sled is still moving after 0.1s and this most likely due to the recoil momentum opposing the pellet momentum, as you suggested. In principle the sled should keep moving back forever, but friction eventually wins. When I did the calibration by moving the sled by hand (fig. 1.7), I actually stopped the sled against a steel bracket, so I think the calibration is still reasonably good. I should have done some longer time scans to see where/when the sled stops moving. I don't think the sled moved much after 0.1s, but it certainly could have moved a few more mm. I'd be happy if the calibration was accurate to ~10%, but the it could have been a little worse, but not much worse. Again, the tests here are focused on comparisons of the behavior of three rifles, so I wasn't too concerned about absolute accuracy.
Best wishes,
John
Hi, John,
I couldn't agree more that investing in a scope would avoid the fiddly bits of making a PC audio input do the job.
However, I love your and Hector's model of crowd-sourced every-man's DIY springer science so much that the goal of paring the required investment down to the absolute bone is maybe worth working for. I suspect there are folks out there who would like to contribute, but find a potential investment in the $100s hard to justify for instrumentation they'll use only once.
Anyway, that's the ideal I'm chasing.
Meanwhile, if you get a chance to rerun some of your calibration scans to times long enough to let the sled coast to a friction-limited stop, that would be interesting.
I agree that using a pc audio input would make this kind of testing a lot more accessible. It's really cool that you can reconstruct the dc coupling signal! I'm looking forward to trying it on my pc using a function generator.
The oscilloscope is nice for looking at more channels. I'm now measuring barrel orientation as a function of time and am using all four channels on a scope.
Best wishes,
John
I agree that using a pc audio input would make this kind of testing a lot more accessible. It's really cool that you can reconstruct the dc coupling signal! I'm looking forward to trying it on my pc using a function generator.
The oscilloscope is nice for looking at more channels. I'm now measuring barrel orientation as a function of time and am using all four channels on a scope.
Best wishes,
John
John,
Thanks for the kind remark! There's supposedly a dual channel function generator built into the sound card 'scope freebie s/w I'm using, and if I can figure out how to get it to output actual signals while the microphone input is also active (no idea why that doesn't seem to want to work!), I certainly plan to do the actual experiment too, instead of only simulating it.
So long as this is a realistic model of the mic input...
...it's prettymuch guaranteed to work given the right T value, but then there's that theory vs practice thing again!
When you give the DC reconstruction expression a whirl, if there's any issue with Excel run times (I assume you're spread-sheeting it) with big arrays, please note that (e^(t / T) - 1) is a constant and so need be calculated only once, and SUM(A1:Ai) would probably run a little quicker if allocated its own column and having the Ai data added one at a time as needed.
Of course, with a fast laptop, it all may be academic.
KR,
Steve
Hi Steve,
...When I did the calibration by moving the sled by hand (fig. 1.7), I actually stopped the sled against a steel bracket, so I think the calibration is still reasonably good...
John
John,
Excellent point! I apologize for missing that element of your description of your method and apparatus, and agree that it renders my comment about pellet momentum's effect on calibration mostly moot.
For some reason I imagined you were re-doing the total displacement calibration method for each gun individually.
Steve
I agree that using a pc audio input would make this kind of testing a lot more accessible. It's really cool that you can reconstruct the dc coupling signal! I'm looking forward to trying it on my pc using a function generator.
Blue: All was left by the ADC from a 100Hz squarewave.
Red: Numeric reconstruction of same.
Definitely not perfect. The ADC low-end cutoff turned out to be a disappointingly high 100Hz = 1.6ms, so the necessary correction is correspondingly extreme, but maybe good enough?
Thought for the day: This digital springer stuff ain't as easy as it looks!
First (failed -- but with an explanation!) attempt to acquire a recoil-O-gram.
Equipment...
1. 11lb 6oz (as shown) 16fpe (955fps with CPLs) TX200
2. Dynamic pickup, elegantly zip-tied to an (NIST - traceable!) 2kg (calibrated!) brick, hooked to my long-suffering Dell laptop. Note the aluminum pushrod reaching under to scope to contact the mount directly.
3. Wildly over-ranged PC ADC! I'm just lucky it didn't fry the thing.
4. Next job: A bit of signal attenuation!
PS: The measured (Lyman digital trigger pull gauge) force required to overcome friction and make the gun slide on the rug is 3lbs 8oz, roughly 1/3 of its 11lbs 6oz weight. Consequently, measured accelerations can be expected to be understated by ~1/3g.
Just in case anyone was interested in this aspect of the "free-sliding" approach.
Raw AC-coupled scope trace with 150k signal attenuation resistor added in series with ADC input for ~30:1 (29.5dB) attenuation.
Raw AC-coupled scope trace with 150k signal attenuation resistor added in series with ADC input for ~30:1 (29.5dB) attenuation.
DC corrected velocity plot in blue, Acceleration plot (g-force) in black.
Expanded time axis in vicinity of piston bounce...
A day late!
Sorry 😏
Chapter 4 is here: Swapping powerplants between an LGU and an LGV:
Hope you enjoy!
HM
Position: Vertical axis is mm of displacement (recoil) toward the butt, horizontal axis seconds.
I don't recall anyone ever commenting on how remarkable it is that the instantaneous maximum rearward recoil position of the gun is farther back (here ~1.5mm farther) than the rest position, even though it's the latter that represents the ultimate rest of the piston at its farthest forward position in the tube. If we focus solely on the piston's mass as the source of momentum pushing the gun back, this shouldn't be possible.
But of course, the piston isn't the only internal mass that moves forward as the gun moves backward.
In fact this proves (if proof were necessary) that at the instant of bounce, the spring comes totally free of the rear guide and stacks up behind the piston, adding its mass and momentum to the piston before later relaxing and allowing the gun to partially re-rebound forward.
Have we discussed this peculiar sequence before and I missed it?
I've mentioned before Jim-in-UK's unparalleled expertise in interpreting plots of dynamically acquired spring recoil data. Today, after briefly inspecting an example of my recent efforts to acquire them, he said: "I'm wondering what the notch approaching peak velocity is?"
Good question! After pondering it for awhile, I have a theory.
All my spring guns are break barrels -- except the TX. And in the TX, unlike the BBs, besides piston and spring, there's an extra major internal component with significant mass capable of pressure-related movement. Wait for it...
The compression tube.
I'm theorizing that "the notch" is caused by movement of the tube as rising chamber pressure forces it forward to impact with the breech.
Plausible, Jim?
If so, this is perhaps a good example of the kind of detailed insights into springer physics that are possible with a dynamic sensor properly coupled to a spring-piston action.
You could test one of your break barrels and see if the notch disappears, Steve.
You could test one of your break barrels and see if the notch disappears, Steve.
What?! And put my beautiful theory at risk of contradiction?
But all seriousness aside, of course I do plan to run a few BBs thru, but first I'll be rigging a haywire (literally) arrangement where one of the sensor-signal carrying conductors will pass in front of the TX muzzle so that it will be cut by the pellet within microseconds of the latter's exit.
Therefore, if exit really does occur before bounce, the bounce signal won't be seen at all.
And vice-versa.
16fpe TX200 calibrated G-force acceleration plot.
Goal for today: Pellet exit time measurement.
Pellet Exit Time Experiment
The rig: Aluminum foil tape carrying Linear DynamicSensor signal to laptop audio ADC at muzzle of 16fpe (953fps with CPLs) long-stroke TX200, where pellet must cut it at the moment of exit.
...Before...
...After...
...The result: Loss of signal (hence pellet exit) at ~1ms AFTER piston bounce.
Afterthought: You might want to subtract 300us of so to allow for the fact that the length of the TX baffled shroud separates the end of the actual bore from where I could place the foil.
First break barrel: 14fpe RWS94 (Cometa 400)
8lbs 4oz, 9mm recoil distance, very twangy
Another Breakbarrel: 13fpe Gamo Shadowmatic
843fps = 12.9 fpe with 8.2gr Meisterkugelns, 9lbs 2oz as tested, 7mm final displacement, VERY twangy!!
In case anyone's wondering about the photo above in which I've connected the recoil sensor to a plastic trigger guard, which would seem a terrible idea from earlier discussions about the necessity for rigid connections between gun and sensor, let me explain why it's okay.
The total moving mass of the sensor (aluminum pushrod, magnet, and spring) only adds up to 7 grams. So even hundreds of g's of acceleration require only a few pounds of force to make the sensor accurately track the gun's motion.
Note how this differs from John and Hector's instrument where their kilogram sled would need 100s of pounds of force to accomplish the same task.
In the free-sliding case, just a plastic trigger guard is rigid enough.
Third Breakbarrel: 15fpe Webley Tomahawk
Thought for the day: Breakbarrels are peculiar critters -- especially when you measure the motion of the breech block!
Comparing SuperLube to Krytox.
Chapter 5 is here:
Hope you enjoy!
HM
Thank you Hector for your hard work and for presenting this !
A true learning experience.
Thank YOU for reading!
Thanks should really go to John Cerne and Yogi, and also to Steve Herr.
I am merely the "publisher"
😉
Keep well and shoot straight!
HM
Chapter 6 is here, analyzing the effect of rifle weight on precision and accuracy:
Hope you enjoy!
HM
Chap 7 is here, and we look at how power level affects different guns.
Hope you enjoy!
HM
Chap 8 is here:
John Cerne shows us now how the distribution of weights along the barrel affect precision and accuracy, and concludes that manufacturers have "broader" priorities than shooters.
Enjoy!
HM
Thank you, Hector, for another interesting article. I have a question.
In the article, John makes this intriguing observation about the effects of removing the endcap from his LGU's moderator, "This greatly increased the report of the rifle..."
I wonder what this change in the sonic environment of the rifle might might imply about changes in near-muzzle airflow as it moves along with the pellet? In particular I wonder about changes in airflow with and without the restriction of the endcap, how said changes might have contributed to the changes in POI that John observes, and why barrel vibrations were the only possible causation considered?
Now I hesitate to mention the research in this area that the Cardews included in Trigger to Target, because you've repeatedly said how "dated" you regard their work to be. But the effects of near-muzzle airflow on the flight of the pellet is a topic they treat in some detail. Recommended reading.
Please note, for example, this sequence from page 197 in which near-muzzle airflow is shown actually upsetting the pellet. The likely effect on accuracy of changing anything that might affect said airflow (e.g., removing a moderator baffle) is, I think, pretty obvious.
The air flows at JUST the moment of pellet release are a whole different can of worms from the barrel vibrations, yes you are right. During the preparation stage of the article I offered two different "devices" to Jhon: a baffle series to be housed in the cavity, and a "floating weight" that was designed to reduce the barrel vibrations. He opted for none, but still I sent him a "floating weight" because I had had good results in at least 4 LGU's previously tuned.
John DID test the device, but he still opted to keep his "no cap" arrangement as the "best possible results at the target" option. So, what he presented is what he believes to be the MAIN effect.
Now, to explain why the process of pellet release to vibrations of the barrel timing needs to be INDIVIDUAL to the gun lie VERY deep in the making of a barrel. But to explain it in just a few words: The process of rifling a barrel introduces radial stresses in the barrel being made that are NOT uniform, nor equal to all barrels. So, EACH barrel will vibrate differently from all other barrels produced. In SOME PCP's you have the option of "indexing" a barrel, not in spring-piston airguns.
In the specific case of the LGU, the way it was explained to me by the designer, was that the cavity had been devised (he didn't say "designed", so for a German engineer that means that it was a trial and error procedure) as a "Resonant Cavity". A resonant cavity is a cavity that uses the length of the longitudinal air wave to either amplify (constructive interference), or reduce (destructive interference) the sound signature.
Why use that in the LGU? Two reasons:
Commercial.-MANY countries forbid the use of sound moderators/baffled devices in ALL guns (airguns included) and, so, the use of ANY baffled device would have reduced the overall possible market for the rifle.
It was also a push along what SHOULD have been the "Leit Motif" of the gun series: "The sound of silence".
Technical.- In a CONSTRICTED environment, the air flows are not exactly as the Cardews published them, there is a speed of flow where a cylindrical column of air can gather enough difference in speeds (as the surrounding air is being held by the turbulent flow and the cap, and the caliber sized column is not) that it "punches" a core through. So the pellet is not really travelling through the turbulent flow of the "mushroom" generated by the conditions developed under "open air" situation.
Now, it is true that by REMOVING the front cap John gave back to the air flows one degree of freedom they had lost under the original design intent. But evidence suggests that while that definitely had some effect, the overall effect of removing the weights at the end of the "barrel/cantilever beam" reduced the excursions produced by vibrations even more.
Now, if we were to talk about the LGV, then the conditions are completely different and you (and the Cardews) are absolutely right. It doesn't help the gun's accuracy and precision that, even with the 1" bore (approx.), the LGV still has a LOT of swept volume for the 12 ft-lbs of the major market objective and, so, the LGV is one of those guns that responds FANTASTICALLY well to the addition of an air stripper at the muzzle. For most of the UK shooters, the moderators they use perform the same basic work as a stripper, plus the added work of containing the high pressure air release to reduce the sound signature.
Yes, the EXACT moment when the pellet is released from the rifling is a crucial moment to precision and accuracy, but that is true for ALL gas/pressure driven projectiles and was established by Greener MANY years before the Cardews.
It is also the DRIVING reason behind getting the BEST possible "crown" you can in a barrel, as well as getting the most uniform/flat ring of release projectiles you can.
In the Cardews' picture you illustrate, it is clear that while the pellet is being upset from its perfectly linear travel, the air flow (revealed by the "Schlieren-type" lines of shadow) also demonstrate that the air flow does not overtake the pellet. That WAS true in their days at their power levels. At current power levels and "pop/diesel gun" regimes. the air flows DO overtake the pellet and create a "mushroom" through which the pellet must travel. Again, this is why air strippers work, specially in PCP's that use a TON of air for each shot, but also true for any airgun with a compression chamber bore larger than 20-23 mm's ID.
It is also the reason why the 34 EMS has threaded barrels 😉
Thanks for your comment, thanks for reading critically and in detail. Hope I have answered the question.
Keep well and shoot straight!
HM
Tags
Ballistics · Shot Cycle · Springer · CTC Custom