Hacking The Benjamin Titan NP Break Barrel

[George Schmermund]

Here's the first attempt to use an accel to measure displacement. The Ch #2 preamp output is 1v/mm. The Ch #1 preamp is still 10mV/g. The top of the screen is the zoom window and the brackets therein show what will be displayed on the main screen. The timebase changes with the zoom so have a care when counting horizontal divisions. There may be some interesting stuff here to get started with. These are all from the same shot.

You're on your own with these images.

[WhatUPSbox?]

It is hard to see the relationship between those two curves. When you mentioned the preamp capability, I tried to find what starts the integration period for the preamp to get to velocity and distance, but could not find it in documentation. My scope can do at least a single integral. I need to read on how that works, or maybe just export the captured waveform and do it in the computer.

[George Schmermund]

The preamp does all of the heavy lifting. The DSO just records the signal. I was surprised myself when I saw the first measurements. Here's some info about the preamp and how it does the magic: https://www.bksv.com/-/media/literature/Product-Data/bp0099.ashx

[WhatUPSbox?]

Yeah, I read that documentation. I'm still not sure of what those integrators are doing, including the low frequency filters they use. But that may be just me.

Maybe try them with a cleaner waveform like a tap test.

[George Schmermund]

The integrators act like low pass filters, but each one also contains a constant of integration. This constant is mainly drift and can go to infinity over time. Using a low frequency cutoff filter and short measurement times will reduce the double integration constants to negligible levels.

The impulses that we've been looking at are mostly in 10 and 20ms wide windows. There is a much lower frequency component that is suggested in some of the DSO shots, but you can't really see them until the time window is opened to 500ms wide. The gain difference between preamps also goes from 10 mV/unit to 1000mv/unit. The DSO's trigger is still on ch#1 so the initial impulse is captured as acceleration, but the displacement signal on that channel is buried much deeper than an 8 bit DSO can see.

I hope this helps to make sense out of what we're looking at with these measurements. 

[WhatUPSbox?]

Thank you, that helps with understanding the preamp. I was still expecting to see some features on the accel curve that correspond to the change in slope of the displacement curve. Maybe it is low amplitude, long duration.

[WhatUPSbox?]

I didn't have full confidence in the friction coupling between the stock and my roller/encoder. I used some thin rubber sheet but even with that I wasn't confident that there wasn't slip occurring. So I went on Thingiverse and found a rack/gear model and scaled it to fit my roller diameter (image 1). I used some strong double sided tape to attach it to both the stock and the roller. I took a couple of shots and the motion during the cycle is very repeatable. The final rest point is influenced by the cog interval.

Here are some results from the second shot:
Image 2 - overview of the cycle. The trigger (time=0) for the trace is the light trap at muzzle detecting the pellet exit. There is a 0.1 msec delay there because the trap is one inch in front.
Image 3  - shows the timing of the cycle start (-14 msec) and the piston impact (-1.9 msec). these numbers have been very consistent for all of the Titan shots
Image 4 - shows the first encoder cycle (slot to slot motion). I measured all of the increments to generate the recoil motion in image 5. Each encoder increment represents 0.63 mm of rifle motion as mounted. The total motion (8.16 mm) matched the length of the sharpie mark to my ability to measure.
Image 5 - Shows a plot of the recoil motion and the velocity with some of the key cycle events noted. It is just my assumption that the second peak in the accel trace is the piston bouncing.

More data to wrap some theorizing around 

CORRECTION: There was a typo in the excel plotting. I replaced the recoil chart (sorry)

[George Schmermund]

Stan - I was just posting this when I saw you new one. I'll put this one up and then got back and look at yours.

..............................................................................

If you look carefully at the images again you'll see what your looking for. First keep in mind that there is a difference of 100:1 in the signal levels after one of them has been integrated twice. Now look at the slope of the displacement curve at the time when the first (ramped) impulse signal drops through it on the screen. The slope of the curve has momentary changed.

The second interesting event is in the next image when the rifle moved forward during the shot and the front accel just sightly touched the face of the block. This was unintended, but fortuitous in the overall outcome. I put one of the cursors there to mark the event. If you look at the third image you can see a the other cursor at the top of a very smooth curve. Looking back at the first cursor you'll see that the curve is flattened out there due to the block having stopped the accel (and gun) from proceeding further forward, therefore limiting the displacement.

If the DSO had more bits to improve the resolution the images would be much more dramatic in presentation.

[George Schmermund]

Stan- I like your electromechanical approach to getting the recoil data. Printing a gear set is a clever idea.

I've been staring at your plot. I think it might help if I go lay down for a while. 

[WhatUPSbox?]

I hope you are looking at the corrected plot, not the original.

[WhatUPSbox?]

After staring at this data for a while, some observations (speculations) come to mind.

1) Fairly confident in the major events (start, piston impact, pellet exit) in the timeline. These are distinct and don't depend on absolute amplitude calibration. The timing is in general similar to George's data.  In my trace, my calling the second peak piston bounce is speculation. The timing of the accel zero crossing prior to piston impact is similar to George's trace and the overall shape of the pre-impact trace is similar as well.
2) The displacement data from the encoder is interesting (i.e. I don't have it figured out). The encoder mask was made with a CNC encoder (though as a trial it was made from a piece of CD-ROM disk). I calibrated the spacing to rifle motion with an indicator at 5 cycle intervals and it was uniform, though cycle to cycle variations could look like ripples on the velocity plot. The encoder can't tell direction reversals, but since the overall span matched the length of my sharpie trace and there was no indication of reversal I'm going with single direction motion.
3) Given all that, the velocities derived from the encoder motion are lower than you would expect from rigid body acceleration either as measured in the front part of the cycle or from the spreadsheet (F=ma) estimate that Michael provided for a 180lb force gas ram.

All good head scratching stuff.

By the way if someone has a kid/grandkid that needs a politically incorrect STEM project for school, well, step right up, this will keep them busy

[George Schmermund]

 A general idea about what might be going on with this recoil testing is starting to materialize. The highest forces that could damage a rifle's scope are most likely in the first impulse set. I'm going to focus(?) on these for the first attempt to rein them in. This choice is driven mostly by the fact that the power plant may be readily accessible by removing it's rear end cap. I think. There might be some butchering needed to the stock in that area, but if easy and repeated access can be gained by just removing the end cap and not having to dismantle the whole gun every time, I'm willing to make some sacrifices. That's why these are called "Hacking" threads.

I'm going back to putting the piece of foam behind the butt plate to act as my proxy shoulder. I want to record what forces the scope is really exposed to during the shot cycle. The pivot yoke now has a piece of sheet Teflon between it and the stock. This should reduce any of the drag.

The photos show the reconstructed testing arrangement. 

[WhatUPSbox?]

I have not taken it apart but I thought there is some significant preload on the gas ram. Looking at the parts drawing, it may be tricky to get access. https://support.crosman.com/hc/en-us/article_attachments/201592640/BW8M22NP_EVP___PL.pdf

[WhatUPSbox?]

That will be an interesting test. I think the acceleration spikes we've been measuring are due to the internal pressure spike decelerating the piston and are transmitted to the scope through the ring mounts. It will be interesting what an external force measurement on the rifle through the scope looks like.

[George Schmermund]

I was looking at that parts list a couple of weeks ago and thought about just buying some parts to build most of the rifle sans stock. Then I decided to just do exploratory surgery on what I've got since I'm not really much interested in using it for anything but bench experiments. The thing is like handling some sort of elephant gun compared to the CO2 carbines I've gotten used to.

Here's a link to a YouTube teardown of basically the same powerplant. It looks pretty straight forward which is what got me to thinking about expedients. https://www.bing.com/videos/search?q=rebuilding+the+benjamin+air+rifle&docid=608033129985672106&mid=8E672C9716D77B0925DF8E672C9716D77B0925DF&view=detail&FORM=VIRE

[George Schmermund]

Reassured that the 2 accels and preamps have been previously confirmed to produce ostensibly identical signals, we can trust that the signals can be massaged without altering the original information. This will allow a look at the bigger picture which would include the shot follow-through.

Here are 2 images with longer time windows. Ch#2 is still 1mm/V. The differences between the peak distances I'll attribute to the foam at this point.

[WhatUPSbox?]

I thought you were going to try to make the piston/ram removable in the stock. That looked hard. Yes bench top it is straightforward.

[George Schmermund]

At some point I'll have to do a total disassembly to be able to machine the transfer port for a pressure transducer. In the end (pun?) I just want to be able to remove and replace the things behind the rear of the gas spring while the rest of the gun is fully assembled. I'll probably have to do some machining and woodwork at that point. 

[George Schmermund]

Just to keep things moving along here I decided to machine a mount for the force transducer to fit the butt plate. This will allow it to measure the recoil force at the shoulder and see how well it tracks the displacement signal. I would expect some phase shifting during the shot cycle.

[George Schmermund]

Here's A pair of images for Stan.

I tapped the front of the muzzle with a hammer. There was a thin piece of wood shim between the hammer and muzzle to control the impulse spectrum. Ch #2 is still using the displacement setting on the 2635 and Ch #1 is using the acceleration setting.

In one image the digital filter on Ch# 1 is set to be low pass and the cutoff is at 50 HZ. As can be seen the displacement information is there using just the filtered acceleration signal and it's 180º out of phase with the displacement.