Split Compression Piston - Crosman Patent

[subscriber]

I am equally skeptical, and for the same reasons, Marty.

For this idea to work, the piston cup seal would require much more mass directly coupled to it (perhaps half the piston mass).  That would keep the seal cup moving, to quickly and "fully" compress the air ahead of it.  Then the second half of the mass could come in to act like a dead-blow hammer - with the rubber acting both as a cushion and a brake. 

Braking too soon (the way it looks in the video simulation), would seem to rob the seal cup of forward momentum.  It looks like the lack of bounce-back in the configuration depicted would not make up for the loss due to the piston cup slowing down prematurely.

The friction of the rubber would depend on lubricant in the system, and the surface finish of the cylinder and rubber.  It would probably also be ambient and recent use history temperature sensitive -  with the rubber stiffness and braking effect being affected.  The exact part dimensions would also matter, as would wear and fatigue of the rubber over life.

I think the idea has merit, but think it could be improved.

Agree, it’s a very simple yet practical idea. I would add that the plasticity and gripping characteristics of the bushing would be very important. If the bushing starts compressing too easily just from the initial forward surge it would rob the gun of power  before it slams the end of the chamber. On the opposite end the bushing needs to maintain its friction coefficient for the life of the gun, so I’m wondering how long it’s anti-bounce properties last given that it is subject to constant frictions on the recoil stroke. Probably wears out quickly...

-Marty

[MartyMcFly]

How about having an inner cavity in the bushing, much like a donut? This would allow more expansion under high impact while allow the use of a stiffer material that would not compress under little pressure.

-Marty

I am equally skeptical, and for the same reasons, Marty.

For this idea to work, the piston cup seal would require much more mass directly coupled to it (perhaps half the piston mass).  That would keep the seal cup moving, to quickly and "fully" compress the air ahead of it.  Then the second half of the mass could come in to act like a dead-blow hammer - with the rubber acting both as a cushion and a brake. 

Braking too soon (the way it looks in the video simulation), would seem to rob the seal cup of forward momentum.  It looks like the lack of bounce-back in the configuration depicted would not make up for the loss due to the piston cup slowing down prematurely.

The friction of the rubber would depend on lubricant in the system, and the surface finish of the cylinder and rubber.  It would probably also be ambient and recent use history temperature sensitive -  with the rubber stiffness and braking effect being affected.  The exact part dimensions would also matter, as would wear and fatigue of the rubber over life.

I think the idea has merit, but think it could be improved.

Agree, it’s a very simple yet practical idea. I would add that the plasticity and gripping characteristics of the bushing would be very important. If the bushing starts compressing too easily just from the initial forward surge it would rob the gun of power  before it slams the end of the chamber. On the opposite end the bushing needs to maintain its friction coefficient for the life of the gun, so I’m wondering how long it’s anti-bounce properties last given that it is subject to constant frictions on the recoil stroke. Probably wears out quickly...

-Marty

[subscriber]

The rubber friction brake is almost certainly already a donut. Else the the front of the piston would have no connection with the rear.  The connection is via a captured but telescoping rod, passing though the donut.

[MartyMcFly]

The rubber friction brake is almost certainly already a donut. Else the the front of the piston would have no connection with the rear.  The connection is via a captured but telescoping rod, passing though the donut.

Yes, but the cavity of their donut is filled by the telescoping rod, which is reducing the expansion potential (I think). What I had in mind is keeping the cavity open and the connections to the forward and rearward assemblies being a press fit grommet sort of like below but bi-directional:

-Marty

[MartyMcFly]

Basically take the picture from above and glue them end-to-end to create the donut/bushing that goes between the front and back of the assembly or you could just put the cavity behind the piston seal...

-Marty

[subscriber]

The sloping angled portions may have some interesting wedge camming effects.

In principle, rubber acts like an incompressible fluid (unless it is open or closed cell rubber).  This means that to generate pressure against the compression cylinder wall, the rubber donut has to general pressure against the telescoping rod.  Else, the rubber will just flow into whatever gap is available, generating hardly any pressure against the cylinder wall.

It may be more important that the steel shoulders meeting the rubber be slanted to drive a wedging effect than the rubber.  When under pressure, the rubber part will conform to the shape of the cavity the steel parts describe.

If the wedge angles are too pronounced, the rubber will battle to unstick itself.  Cocking the springer for the next shot could prove tricky.   

This will be a bit like car brake shoes used with drum brakes in the 1950s and 60s, having a degree of self-activation by means of a wedged path.  Those were used to reduce the pedal force, when brakes were still cable operated.   The snag was that the brakes tended to lock up in hard use, and stayed locked, despite taking your foot off the pedal...  The use of hydraulic actuated discs brakes fixed that, although recently cost cutters have brought back drum brakes on the rear wheels of base model cars...  At least with hydraulics, the wedge angles used on the shoes can be so mild as to make locking up the wheels much less likely.

[Mole2017]

I had to go back and read the patent to see where the braking issue came up. I see now that they consider that to be a key part of the function: by judicious choice of materials and geometry, the bushing expands, drags on the compression tube, and changes the deceleration characteristics, thereby slightly delaying the bounce and giving the pellet the benefit of a slightly longer impulse event.

I thought it funny how the patent then goes on to cover the bases: may this and may that, 1:20 and 20:1 mass distributions between piston body and piston head, spring or gas ram, etc.

I forget which paper it was that I was reading, but somebody actually had what I will call a pressure dependent seal friction, mimicking the action of the parachute seal during compression of the air. I think they even had numbers for normal seal friction vs friction at peak pressure, e.g. 4 Newtons vs 120 Newtons (metric paper...). So, a clever design would get the bushing to expand--possibly with some mechanical wedging, or careful design of the bushing, or both--to create that delayed bounce dynamic.

And yes, choose the wrong wedge and you might jam things badly. However, I think it shouldn't take much and the elasticity of the bushing will still have the mechanical advantage to properly "unlock" itself as soon as the pressure from the piston head and body drops some. Of course, it might be possible to do the design without any wedging. I too have to wonder about the wear on this bushing...

This also means that if one of us wants to write this into a simulation, we might have to add a pressure dependent friction model to duplicate the effect. That is, it isn't is simple as splitting the mass by some fraction and tuning the stiffness between to get a longer "dwell" by dynamics alone. But it would still be neat to see if the friction is really needed or not...

[subscriber]

I take your friction brake and raise you a transfer port reed valve . 

The latter does not care if the piston bounces because the high pressure air is trapped in the TP.  The valve would need to be on the inside of the compression chamber...

I think that braking the piston while it is moving forwards is a mistake.  If the brake could somehow only activate, just at the instant when the piston kisses the front of the compression chamber wall, I would be impressed.

Perhaps a small movable mass in the rear of the piston could be used to activate the brake, by crashing into it, just as the piston seal has reach full travel. 

Then again, a simple movable mass in the piston, that acts as a dead-blow hammer is probably more effective than the patented brake, and much less material dependent...  Problem is, that has already been patented.  The patent is probably expired, but too often, on-paper prestige sells better than actual performance.  A bit like claiming a springer as a 1400 FPS airgun...

[Mole2017]

A reed valve? Interesting...

For a little while some time ago I pondered how one could come up with some sort of mechanical "jam", like a ratchet, to stop the piston bounce, i.e. hold the piston at the point that had developed maximum pressure, but never got anywhere with it. All I could see what some vicious mechanical wear killing the mechanism...

For what it's worth, I've worked out what to add to my simulation to get some basic pellet response (my current version didn't launch the pellet--it just looked at the bounce dynamics of the piston/action/scope). I've also worked out on paper how I'd add a "pneumatic" bushing, i.e. instead of a polymer bushing, I went with a miniature gas ram piston head. Had to start with something...we'll see what I get.

[subscriber]

David,

If your models predict conventional springers accurately, then adding the improvements would be especially interesting.

A reed valve has very little inertia, and thus is easy to open.  It responds fast, but does not slam shut very hard either.  This is why reed valves are often used at the inlet of 2-stroke motorcycle engines. 

In other words, my suggestion is to "brake" the air, not the piston.  It is the air pressure and flow that are important, so why not try to control that more directly?





https://motocrossactionmag.com/ten-things-you-need-to-know-about-motorcycle-reed-valves/

[MartyMcFly]

Hmmm, combine the reed value TP with an anti-bounce piston (dead-blow hammer design), then add a soft spring with high preload and a soft bushing behind the piston seal!

1. Reed value prevents flow back
2. Anti-bounce piston improves efficiency and reduces bounce
3. High pre-load spring provides an additional force against the backwards surge
4. Soft bushing absorbs some of the harshness on impact generated from the high preload

Over engineered yet?

Wait, there is more! Use a long stroke compression tube and ensure you are using a low start pressure pellet. With a longer stroke and low start pressure pellet you begin accelerating the pellet earlier, avoid skirt flair, and get the pellet to leave the barrel before (hopefully) the piston crashes into the back of the TP.

-Marty

[mikeyb]

Speculative design discussion is good. It can lead to practical changes that improve performance. Not trying to be negative here, just asking some questions and adding some comments to find out if this can lead to actual improvements. Sometimes these discussions devolve into an exponential spiral of never ending increasing complexity add-ons to fix an initial problem that never really existed in the first place.

Hmmm, combine the reed value TP with an anti-bounce piston (dead-blow hammer design), then add a soft spring with high preload and a soft bushing behind the piston seal!

1. Reed value prevents flow back Where would you put it? There is very little space in the small volume of the transfer port and any increase in transfer port volume to add a valve drops compression ratio and hurts performance.
2. Anti-bounce piston improves efficiency and reduces bounce NP2 & patent = already in production. Why do some folks think this doesn't work?
3. High pre-load spring provides an additional force against the backwards surge Air/gas (NP2) springs already have a much higher "end" force to resist bounce than coil springs do.
4. Soft bushing absorbs some of the harshness on impact generated from the high preload Again NP2 & patent = already in production. Why do some folks think this doesn't work?

Over engineered yet? 1) In my opinion YES. 2)3)4) seems elegantly simple to me. KISS philosophy in action and already here in the NP2 series. Springers were are a very simple design concept. If you want a more complex system skip springers and move right on to PCPs & their support equipment. NOT bashing PCPs at all. Have some and appreciate their advantages.

Wait, there is more! Use a long stroke compression tube and ensure you are using a low start pressure pellet. With a longer stroke and low start pressure pellet you begin accelerating the pellet earlier, avoid skirt flair, and get the pellet to leave the barrel before (hopefully) the piston crashes into the back of the TP. This may be more difficult than you think. How much LONGER do you make the compression tube for the long stroke and how much longer do you have to make the barrel so the pellet can get up to speed? A 6 foot long springer that shoots like a PCP would not interest me at all.

-Marty

Donut elastomer, ref 110 in the patent drawing linked in the first post.

Sounds like the huge range of properties of Polyurethane (PU), from soft light squishy foam to super abrasion and impact resistant hard rubber, are not as well known as I thought. PU is used for bowling balls, shock-strut-leafspring-controlarm bushings, pallet-jack wheels, rollerblade scooter and skateboard wheels, and probably dozens more applications I haven't listed. There are several aftermarket piston seals made from PU which can be impregnated with molybdenum disulphide so you can add "slippery" to the list of possible properties.

https://en.wikipedia.org/wiki/List_of_polyurethane_applications

My bet is the NP2 piston brake-shock/bushing is a 90A-95A (shore A hardness or durometer) polyurethane. A tough polymer like that will laugh (not change shape) at the mellow 20g forward acceleration of the piston. The ~400g deceleration that occurs in the final 1-2mm of piston travel is in the force range were a 90A-95A polymer can absorb some of the shock and change shape enough to expand and grab (brake) against the compression tube. Wear on this bushing is about the same as the piston seal so they both may need replacement at ~5000+ shots. I don't see this as a significant problem, just normal springer maintenance.

I bet the design engineers at Crosman are a little lot smarter than you guys think.

[MartyMcFly]

Mike, I don't have a response to the above yet. But I don't mind negative comments - I like these type of discussions even if I'm wrong half the time 

-Marty

[subscriber]

All good questions, Mike

The "proof" that the Crosman brake donut "does not work" is in the fact that the market does not seem to be raving about it.

The ideas bandied about here are unashamed brain storming.  If the ideas were fully mature and totally practical they would be worth applying for a provisional patent, rather than sharing on open forum...

Springer piston bounce  has been discussed before - even by Marty - with some good food for thought:  https://www.gatewaytoairguns.org/GTA/index.php?topic=149983.msg1531590#msg1531590 

Bonus thread: https://www.gatewaytoairguns.org/GTA/index.php?topic=161010.0

Marty seems overly concerned with the peak pressure reached in a springer, resulting in pellet skirt ballooning.  Reducing the starting pressure will make any springer less efficient.  So, larger and more clunky at a given power level. 

As long as pellet distortion is consistent from one pellet to the next, it need not be something that has to be avoided.  In fact, there is evidence that springers such as the Diana 350 Magnum can shoot .177 pellets accurately at 1050 FPS precisely because they convert pellets from diabolo to cylindrical shape, due to extreme skirt ballooning.

The piston does not "sweep out" the air to follow the pellet down the barrel - the piston is not travelling fast enough to do that (once the pellet has started moving).  A spring with lots of preload is not going to "sweep air" either.  Not unless it has several 1000 lb of preload (to match the force on the piston face, due to the compressed air acting on it). 

More practically speaking; consider that a 200 lb preload spring would would first need to re-accelerate the piston to follow the air travelling behind the already moving pellet.  The initial 3000 PSI acting on the other side of the piston head will win the piston bounce shoving match against any practical spring, every time...

The spring launches the piston and stores energy in it.  The piston then crashes into an air column, compressing it.  It would not matter very much if the piston was uncoupled from the spring in the last inch of piston travel.  The piston is going to bounce anyway.  And it only has to bounce 1/10" to diminish the useful air pressure in front of it.   Reducing piston bounce from 1/4" to 1/10" is much easier to accomplish with a heavier piston, or a split-mass deadblow hammer effect.  If the bounce reduction does not occur right close to the compression cylinder wall, it will not add energy to the pellet.


The monumental dual opposed piston thread that the quote below is from, is well worth reading. 

The massive conclusion is that piston inertia does the work, not spring preload.  If you want to prevent acceleration (or piston deceleration and reversal), add mass.  Air pressure from the spring force at the end of the forward stroke is negligible, compared to air pressure due to the (air cushioned) collision the piston sees against the cylinder wall. 

The second lesson from the thread is that short stroke large bore piston springers don't convert spring force into piston momentum very efficiently.  No matter how how much preload is on the spring.  The ratio of piston mass to piston area needs to be high for a springer to maximize efficiency, an minimize losses due to piston bounce.  A small bore piston needs a longer stroke to regain swept volume...

Bore to stroke ratios for springer are often quoted as needing to be between 1:3 and 1:4 for efficiency.  There is apparent truth to this, based on springer bore diameters around one inch being common.  Else it is a nonsensical statement:  It is bore area to stroke ratio that matters; not diameter.  Of course, the units are different; so bore diameter squared to stroke length ratio, if fuzzy logic is not your forte.

Reed Valve:
The design QVTom used for his opposed piston pistol below used a sliding breech to couple the vertical transfer port to the barrel breech.  A reed valve could be placed in that transfer port fairly easily, compared to a break barrel springer - there it might have to live in the piston nose. Any increase in TP volume to house a reed valve will hurt efficiency more than the valve gains...

Compare a steel "springer" against an equivalent gas spring air rifle.  The gas spring version has a faster shot cycle, but despite its much "greater area under the curve" (from the greater cocking force early in the stroke), the energy transferred to the the pellet is not much larger than with the steel spring.

I haven't posted on my project in a while, mostly due to time constraints not a waning interest.  The following are some pictures from the past few months and some conclusions.
\
I'll try to get some finished photos or video up in the next week or so.


Breech design




Gas spring charging contraption


Piston weight adjustment  (bad idea)
Before lightening

After lightening



The gun is shooting and I have been tuning.  I've learned a lot during the tuning process, mostly about how little I understood spring guns when I started the design.

Piston weight is probably the most important design element.  I started with HS aluminum pistons that were discarded very early on due to the coating that was intended to provide hardness and wear resistance didn't penetrate the 7063 alloy to a sufficient depth.  The next set were made from 4130 and they are the ones pictured above.  Initial performance with the 4130 pistons was terrible, the gun was producing about 10 fpe.  That was about a 1/3 of where I wanted to be so I wrongly decided to lighten the pistons.  The lightening of the pistons reduced the velocity by over 100 fps!  At least I was on to something, so I incrementally added some brass rings in between the gas spring and piston skirt.  Each ring weighed almost an ounce and added about 40 fps.  With the brass weights I was able to bring the piston weight from .4 lbs to almost .95lb each. There was no more room for brass, I had reached the max weight with this method.  Next I removed the brass weights and poured lead into the pistons, with a little fiddling  I got the piston weight matched and was at 1.35 lbs each.  Performance was 23 fpe at this point and still far from the design goal.  Charging this gas springs from 1600 to 2300 psi increased power to 25 fpe.  2300 psi equates ~ 550 lbs. of force which made the cocking force uncomfortable. 

My conclusion is..............  Large piston face/short stroke is unworkable. The large piston face area makes them prone to bounce and it is impossible to make them heavy enough to overcome the bounce.  I believe that if I had more space for heavier pistons I could get closer to the design goal.  I have an uncompleted longer stroke version with the same bore that could fit heaver pistons but to have 3 lbs. of piston weight seems silly and will make the gun too heavy. Even with the longer stroke, I believe I'd still be on the wrong side of the swept area/piston face curve.  The 1.79 dia. piston design ends here.  Lesson learned  


Now for some good news..........  Completely sans-recoil, if it wasn't for the noise you wouldn't know that it fired!  Really, the gun has no motion or vibration period.  It's that smooth!

I've considered de-tuning to 18 fpe and using for FT but rejected this idea for a new line of development.  As I said, the 1.79 bore is dead.

A sneak look at my new version.  A longer stroke and smaller bore.  Cocking, sears and breech design will remain the same.  One advantage of the longer stroke is that it simplifies the cocking linkage somewhat.  Yeah



Tom

Tom's very elaborate and very nicely made dual opposed piston springer breech would easily house a reed valve.  See red "cube" that represents space for valve when breech is closed in attachment below (copied from Tom's image above):

[subscriber]

So, Marty; is this project still running?  https://www.gatewaytoairguns.org/GTA/index.php?topic=149983.msg1541192#msg1541192

[MartyMcFly]

So, Marty; is this project still running?  https://www.gatewaytoairguns.org/GTA/index.php?topic=149983.msg1541192#msg1541192

That project has cost me dearly - I ended up getting a lathe so I can make the parts I need for the modifications.  And now I also need to learn how to actually make the parts. Consider the project to be in limbo for the moment.
-Marty

PS. I got side tracked on another top secret project that has sucked me in. Right now I’m trying to figure out the best way to make that fit into a Diana 34 without butchering it too much.

[MartyMcFly]

Adding a link to an old article I dug up from the UK about piston bounce.

https://www.airgunshooting.co.uk/expert-advice/is-it-possible-to-prevent-or-reduce-piston-bounce-1-4870874

-Marty

[Mole2017]

That transparent compression tube experiment was beautiful! Except that I think it was for pictures only--he's estimating the position from piston velocity, correct? Piston velocity, by the way, can be measured by non contact methods: moving magnet in a fixed coil is my guess in his case. Unless he has some dopler laser unit...

[subscriber]

Reading the article at the link you posted, Marty

I have a problem with the opening statement:

The most serious consequence of piston bounce is that it reverses the rifle’s recoil into forward surge, before the pellet emerges from the muzzle...

The piston, running into and decelerating against the column of high pressure air (it created), against the large area of the compression chamber's front wall, is what causes most of the forward surge.  A non-bouncing piston would not prevent piston deceleration.  It only prevents the deceleration force applied to the rifle, after the piston travel direction would have reversed.   What fraction of the total surge force is this? 


As I see it, the reason for de-bouncing a spring piston is that more energy goes into the pellet.  So, for a given pellet energy, the spring piston can make do with less stored and released energy.  That greater efficiency equates to reduced overall harshness in the shot cycle.


That article provides some useful insight:  Make the seal out of material that has very high static friction and very low sliding friction (if it exists).  That way, the seal swelling out against the cylinder wall while the piston is moving won't slow down forward movement too much.   But, as soon as the piston stops against the compressed air, the high static friction would tend to hold it there, rather than allow it to easily reverse direction.

A cup or lipped piston seal is walking a tightrope with respect to "sealing better" as pressure in front of the piston builds; and increasing the friction to the point of wasting energy on friction that could have been imparted to the pellet.

The article is about a low power HW95.  I think that piston bounce distance with that may be more of a problem than for higher power applications, where the higher speed piston is not going to be reversed by a light, low start pressure pellet - while the latter is in the barrel.

For low power applications, if piston bounce while the pellet in the barrel must be eliminated, cut the barrel shorter.  Cut it so that the pellet leaves the barrel, just ahead of the piston's initial forward motion stopping.  If cocking force is too high from a 10" barrel length, add an extension of large LDC, as used by Si Pittaway:
https://www.youtube.com/user/VerminHuntersTV/search?query=imp
https://www.youtube.com/user/VerminHuntersTV/search?query=.25+snipe

Having said that, the air rifle will still surge forwards because the piston is pushing on the front wall of the compression chamber, via the high pressure air.  Stated another way; the forward surge starts as soon as the piston starts slowing down to compress the air.  The surge does not wait until the piston reverses - it starts when the piston experiences rearward acceleration (not rearward velocity).  If this seems confusing, when you are driving forward and apply the brakes, your vehicle is moving forwards, but accelerating backwards.

The article is interesting.  It seems a mixture of physics and speculation.  It even predicts Crosman's patent won't work:

Many people have thought about trying to harness inertia (resistance to start to move) weights to crush the seal, or some kind of frictional band, and increase its friction during the bounce. Sadly, it seems improbable that they will work as intended, and to understand why, we need to look at piston velocity.

This is actually my gripe with the Crosman patent. too:

The inertia weight would probably stand a better chance of raising piston seal friction during the final few millimetres of piston travel at the end of the compression stroke, when the deceleration is ferocious (in excess of 1,000g), and that would almost certainly reduce the length of the compression stroke, causing significant loss of muzzle energy. Having said which, the only way to tell for certain whether any modification will work is to build it and test.

I would claim that shifting the resonant frequency of the system is the mechanism by which "surge reduction" improves shoot-ability.  This includes the person shooting the springer and their hold - hinted at, here:

Although surge is primarily caused by piston bounce, it can be exacerbated by the rifle bouncing off your shoulder

My Summary:
* Forward surge is cause by the rearward piston acceleration (slowing down to compress the air).  Eliminating piston bounce (travel reversal) will not eliminate forward rifle surge - but it should reduce it.  (If a vehicle runs into the rear of a stationary vehicle, such that they become attached to each other, both vehicles move forwards.  The fact that the rear vehicle did not rebound does not prevent the front vehicle from surging forwards.)
*Unless you have dual opposing pistons, forward surge cannot be eliminated from a "springer".
*If you are using an air rifle designed for 15 FPE, detuned to 11 FPE, then gaining some or all of that back is easy - piston bounce or not. 
*A detuned spring air rifle probably has so much less vibration and harshness already that if what remains is bothersome, get a PCP.
* If the component of forward surge generated by the piston travelling back hurts your ability to shoot accurately, cut the barrel shorter, and add a good air stripper to deal with the higher muzzle blast that will result, and could cause more initial pellet wobble.
* Piston bounce distance is measurable, but may be irrelevant if it is repeatable.  What matters is how easy the springer is to shoot well. 
* Don't assume that the one thing you changed last is responsible for the improvement in shoot-ability that you measured.  It is often the interaction with what you deem unimportant aspects about the rest of the system that affected the system performance.


Here is a patentable idea that I gift to the airgun community:
The crude sketch below is for a quarter section of an elastomer piston seal that incorporates a reverse braking system. 

It works in the same way that a bronze bristle brush resists reversal when travelling down the bore of a gun barrel:  The bristles are too long stiff to travel perpendicular to the bore, so they bend over such that their contact with the bore trails their attachment to the stem of the brush.  The way the bristles lean over prevents them from applying too much drag when you sweep the brush through the bore form the entrance to the exit (breech to muzzle - or other way around).

If you stop midway down the bore and reverse the brush travel, the bristles will have to lean over the other way first.  The process of standing up the bristles and jamming them into the bore to flip them over increase their friction coupling with the bore dramatically.

Now, my seal idea has rubber rings that are molded to face backwards.  They look similar to the forward facing sealing ring (parachute seal), but are optimized for friction when the piston is forced to reverse direction.  This means the rings may need to be shorter and stiffer than shown in my hastily drawn image.  The molded in trail angle for the piston travelling forwards may need to be a bit closer to vertical.  There may even be a range of rings that are molded at different angles to provide more constant braking that is less abrupt.

The manner in which the seal is attached to the piston may have to be improved to prevent the seal from being pulled off the piston.  Yes, the seal is driving the piston back, but any rearward travel imparted to the mass of the piston as the "bristles" flip up, may be enough to rip off the seal from the piston face, when the seal "brakes" come fully into effect.  So, attaching the seal by means of a single machine screw, clamped via a large washer to most of the seal face may be required.  That washer can be covered in rubber, if the piston's front face needs a cushion for eventually running into the comp-cylinder end wall.

The potential downside of the rearward facing rings is that they will tend to sweep lube backwards, away from the front of the compression cylinder.  Some astute observers might call this an anti-diesel feature; rather than a problem.

How long might such a braking seal last?  Long enough to make it through the warranty period, and serve as a replacement part money spinner...


Before solving a problem, the most important thing is confirming that you have correctly identified the actual problem...

 
EDIT:  Added mirrored half section image to make the shape more obvious:

[subscriber]

This rubber piston seal may be a bit longer than required, but its "bristles" are sure to drive a lot of friction as its motion reverses due to piston bounce.

It is designed for insert molding over a steel part, so it can handle forces in both directions, without stripping off the seal.  Again, perhaps overkill with the total internal shoulder area, but that can be refined.