[FMF] Free Model F - Capacitive Flippers

[DMA]

PXL_20250131_055140261.jpg (1.73 MiB) Viewed 3741 times

(if you want the full source - google "One can already spread it, but not yet eat it.").

Current issue list:

1. Stem barely fits into the barrel. It does fit into OG barrel pretty OK - not too wobbly, not too tight - but it's too tight with printed barrel.
   This is likely overexposure (I accidentally printed with 2.0s exposure, not 1.9s that I found ideal yesterday).
   Also, due to layers perpendicular to Z-axis, printed-stem-on-printed-barrel produces funny buzzing sound. There _is_ some binding, but surprisingly little.
2. It clicks. Once. Doesn't return to original position.
   It does return to original position with printed stem. Sound and feel are pretty close to original, less pingy.
   It also kind of returns to original position with printed flippers - they do go up, but sound after the first click becomes pretty quiet AND it looks like the spring doesn't "unbuckle" into straight configuration.
   So it looks like it's both spring upper point is little bit too low, and spring lower point is a little bit too high.
3. spring holders are a bit too large, had to file some plastic off. Could also be overexposure.
4. Unlike Cheap Chinese Copies, one doesn't need to pay attention how stem is inserted into the barrel - the spring finds it's way onto the in-barrel catch automatically. I'm quite proud of solving that.

Turns out I don't even need rafts - landing bare supports onto build plate should be enough (they will be held in place by elephant's foot) - but I need a bit more supports to prevent sagging of large overhangs.
Also, I already bought some transclusent ABS-like resin, but I just understood I can simply cut windows into model instead - it's not injection molding, I can cut windows wherever I please _before_ printing, not after. Transclusent stuff should definitely help bringing more light in though.

Not sure if I should preserve those first prototypes for some museum - or even for the future shop LOL. "One in Stock! Brand New in Box FSF Barrel And Flipper Prototypes, part 101 or 102"

[vvp]

It looks to me that you can modify the flipper model so that it is printable in vertical position without support. Align the two side-tabs and the spring holder to one plane which will be touching the build plate when printing. Print vertically. This way you can also print much more tabs at once.

You can cut about 20% (along the cylinder axis) from the cylindrical spring holder so that you can put it on the build plate with this cut. Modify the side tabs so that they end in the same plane as the cylindrical spring holder cut.

You can print at least 30° overhang quite easily. 0° overhang being vertical. Using this overhang should be enough to get the axis of ration at the proper places even when the side tabs and the spring holder pin are modified.

[xxhellfirexx]

Not sure if I should preserve those first prototypes for some museum - or even for the future shop LOL. "One in Stock! Brand New in Box FSF Barrel And Flipper Prototypes, part 101 or 102"

Please don't take that page from Ellipse's playbook.

[DMA]

It looks to me that you can modify the flipper model so that it is printable in vertical position without support.

That's easy - just rotating the model 90 degrees will do. Some supports will still be needed (because 0.35mm will be squashed and elephant-footed, and I want to actually use the final model for the injection molding without gross modifications), but way less plastic will be wasted. However, I'm not at the stage of mass printing yet (and hopefully never be - injection molding is just better), and printing the flipper bottom parallel to XY plane has it's advantages (like, it's actually flat when printed, not micro-grooved. I'll try to print a copy in vertical position as a part of the next batch - not yet gotten there between oncologist visit and other errands).

Please don't take that page from Ellipse's playbook.

Oh, c'mon, it will be fun. It will be way funnier if it's THE ONLY SKU in the shop, but I'm too stingy to pay for domain without expensing it as a business expense.

..you know what will be even more fun? Renaming the project "Open Model F" and name the resulting build "OMF 2097"

[vvp]

I sometimes compensate for elephant foots by chamfering the bottom. Sometimes elephant foots do not matter for functionality. In such a case, I do not care.

And elephant foot is often better than uneven surface because of the surface being too close to horizontal and resting on support.

Tiny steps on almost vertical surfaces can be mitigated by antialiasing in the slicer. Though I never did this. It looks to me like a lame mitigation. My guess is that about 20 µm steps on the surface will not matter for the capacitive action.

One more thing I forgot to mention before. Most cheap calipers have resolution of 10 µm and their relative error is in this range as well. But the absolute error is often around ±30 µm. And absolute error is what matters when measuring parts. If your calipers are not exceptional then just ignore any errors in the rage of 30 µm.

[vvp]

I looked at Elegoo Saturn 4 Ultra. I would not go for it. I have seen some bad review of the auto-leveling system. IIRC by FauxHammer. Yes, here it is: https://www.youtube.com/watch?v=7Y-rDu23t-M&t=300

The claim was along the lines of "it is auto-zeroing (a fake auto-leveling where springs compensate for the imprecision) instead of a real auto-leveling like e.g. modern FFF/FDM printers have". I personally do not like also the two level build plate. Cleaning resin from it must be a pain.

Almost every mSLA printer has some problems. E.g. my Mars 4 DLP has two screw build plate attachment instead of the better (more stable) four screw solution. It also does not have heating but I added a custom one - air heating as well but that is plenty enough. I let it heat up before printing and pour preheated resin and it is all good even though I typically print in a room with temperature around 15 °C.

heater-s.jpg (60.03 KiB) Viewed 3498 times

[DMA]

I sometimes compensate for elephant foots by chamfering the bottom. Sometimes elephant foots do not matter for functionality. In such a case, I do not care.

Currently, there's also semi-random squishing of bottom 0.35-0.45mm because of the damn "automatic build plate leveling". Chinese New Year is about to finish, so I should have an answer for that problem soon though.
Plus, getting the model off the build plate resulted in cracks more than once, so supports are kinda cushioning the model. I mean, I can build supports directly in freecad and save some hassle at printing time, but not sure if this saves effort or increases
Thanks for suggestions - it's actually nice to know there's another way to do things, but at this point bumping my own head against things is more fun.

Tiny steps on almost vertical surfaces can be mitigated by antialiasing in the slicer. Though I never did this. It looks to me like a lame mitigation.

*Extremely* lame and will likely cause semi-cured resin blobs in the vat - so I don't use that.

My guess is that about 20 µm steps on the surface will not matter for the capacitive action.

May actually help by holding a thicker conductive layer - we'll find out once I'm happy with mechanical side of things. One variable at a time, y'know.

One more thing I forgot to mention before. Most cheap calipers have resolution of 10 µm and their relative error is in this range as well. But the absolute error is often around ±30 µm. And absolute error is what matters when measuring parts. If your calipers are not exceptional then just ignore any errors in the rage of 30 µm.

Not sure if $470 can be considered "cheap" - sure, it was bought on ebay (don't even know for how much, it was a gift), but it has "origin" button and it passes simple tests like "measure stack of 50 A4 sheets, then measure random selection of those sheets", so I think I can believe what it tells me

[DMA]

I looked at Elegoo Saturn 4 Ultra. I would not go for it.

Me neither, knowing what I know today! But alas, missed my return window with surgery and all that. I'll make it work, just disappointed.

I personally do not like also the two level build plate. Cleaning resin from it must be a pain.

EXACTLY. It can be [url=https://www.instagram.com/p/DFWeGVZvpJK/]improved[/quote], but those damn spacers and that damn angle that's too small for the resin to migrate off the plate under gravity are pretty damn bad. OTOH, squeeging the resin off the plate with spatula while morally preparing for chiseling the stuff off build plate (recommended bottom layer exposure is 30s, I'm down to 16, and I had things torn off the plate by FEP at 15 - although that was before I manually levelled the plate).
Funniest part (as in !!FUN!!) about the build plate is that plastic thing that I removed is HOLLOW - so when you try to wash the plate, couple liters of IPA gets into it - AND there are O-rings on those spacers, so that IPA stays there until you tilt the plate trying to pull it out of the wash station. Then it's cloudy IPA everywhere, and if you don't have a portable 405nm light, once that IPA dries everything is sticky AF and stinky AF.

Almost every mSLA printer has some problems.

Yeah, the stuff is on the market for 10 years, I didn't expect to have early adopter problems! As a Software Engineer ex-Networking-Guy I'm not a stranger to that shit, but I kinda expected to more or less plug it in and print, like FDM machines, especially since build plate aligmnent is "automatic" (NO IT IS NOT, IT'S NOT EVEN LEVEL FROM THE FACTORY)

Oh, and another thing. You probably think that black paint is IPA-resistant. IT IS NOT, if you put a drop of IPA on that black top surface and touch it with your finger - your finger will be black.

Also the clamp is shit - steel on steel contact, will wear out quite quickly. I'm even considering a flexible build plate because of this - but alas, latest firmware doesn't allow to set Z-axis zero even with gcode commands! (but there's hope they'll tell me the new gcodes for that as a part of this "my bottom layers are squished" nagging)

[vvp]

Plus, getting the model off the build plate resulted in cracks more than once, so supports are kinda cushioning the model. I mean, I can build supports directly in freecad and save some hassle at printing time, but not sure if this saves effort or increases

Yes, it is probable that the part will break during removal if it has a lot of surface touching build plate. There are aftermarket flex plates which should help here a lot but I do not have them yet. I often do my own supports in CAD to minimize the surface which is touching build plate.

Not sure if $470 can be considered "cheap"

Definitely not. ±30 µm absolute precision digital calipers are around $60. The specs. of your calipers says: repeatability 10 µm. That looks more like relative precision. Absolute precision is not in the description. Hard to tell. My guess your calipers are much better than mine

[DMA]

I accidentally silent buckling spring. Sadly, it's LINEAR.
Turns out, tactility turns out to be generated by flipper dropping on PCB and vibrating the hell out of the whole mechanism.

PXL_20250204_013712766.jpg (1.79 MiB) Viewed 3296 times

You may notice there's a window where the spring SUPPOSEDLY hits the barrel. SURPRISE, that doesn't affect the sound that much - OG flipper clicks quite loudly, 7-degree and 5-degree flippers barely produce any sound. So, the "spring hits the barrel makes model F loud" hypothesis can be safely retired.

As you can also see, I was able to chop 3mm off the barrel and stem. This still makes the barrel 15mm high, 3.4mm higher than MX switch - meaning, in turn, that using existing MX keyboard cases will not be possible without shortening the spring.

[DMA]

Batch 3 was ruined by washing in the hot water on internet advice. So here's batch 4:

batch3.png (2.07 MiB) Viewed 2183 times

numbers are elevation at rest, in degrees.
5 is whisper-quiet, but also almost linear feel. The spring is buckling, don't get me wrong, but almost imperceptible, even with original barrel+stem. Rarely: flipper goes down before spring buckles.
7 is the weirdest of them all - flipper frequently goes down before the spring buckles. Also very quiet.
9 tends to stick pressed.
11 - half as loud on downstroke, upstroke is still silent.
13 - probably 75% loud, upstroke probably half as loud (but still quieter than downstroke - originals are louder on upstroke if you're not MASHING it)

Looks like I need to look more closely to the feet part - looks like even 0.1mm difference in spring height matters.
Almost dialed in interchangeability of barrels and stems with IBM barrels/stems - not that they'll be used together, mostly to be able to tell they'll actually fit together.
Also, in light of the barrelgate, looked more closely on model M barrels from Unicomp. They definitely put some effort into not having parting lines where stem can contact them!

[DMA]

..I tried photogrammetry to make a 3d scan of the flipper, but developers nowadays are too lazy to write anything without CUDA, so I failed. If anybody wants to help - I can send you the photos and you can use colmap (I hope) to create the model.
I only got to "point cloud" step, which is basically unworkable.

[idollar]

I must report success !!

I have been silently working on the 3d printed flippers using the conductive filament which I bought in amazon: EONO Electrically Conductive PLA Filament

Flipper 11.jpg (122.26 KiB) Viewed 1498 times

This is literally the first text which I am typing using a regular model m, using the FEXT PCB, a Leiden Jar and my 3d printed flippers using conductive filament.

The keyboard which I have used has a particularly bad barrel plate. This was a very bad decision as it has taken me long to ensure that all the keys register properly. I believe that now that we know that it can flippers can be 3d printed, the next step is to replace the original ones which I am using in my FSSK

These flippers are less conductive than the original ones. I had to reduce the threshold in the Leyden Jar firmware. I am typing with activation offset of 7 (instead of the 20 which I have been using with my "regular" FSSK.

Screenshot at 2025-02-14 01-29-05.png (19.84 KiB) Viewed 1512 times

Printing the complete keyboard was less than 3 hours.
I had print the flippers lying on one side, to ensure that the spring pressure is done along with the layers, so it does not break. In order to print efficiently, I designed my own support in the 3d model.

Flipper 11 -printb.jpg (70.16 KiB) Viewed 1498 times

Flipper 11 -printa.jpg (63.47 KiB) Viewed 1498 times

For the springs, I have simply reused the originals from the model m.

There is lot of things to tune, specially with the keyboard which I am using as Guinea-Pig. But, in my opinion, this is a great success which shows that it is possible to create flippers in a simple manner.
It opens, in my opinion, the path, to new ideas that were simply not possible before as they needed to harvest the flippers from original keyboards or buying them at unreasonable price.

It is late here in Europe. I will post additional pictures and details next week.
If you wonder if I plan to make the source files available, yes, I will.
I just want to work a bit more on them to minimise the time expend in removing the support.
I will post detailed information on how to produce these flippers

Cheers

i$

[idollar]

Batch 3 was ruined by washing in the hot water on internet advice. So here's batch 4:
......

Great work !!! You are getting there DMA !!!
We will have several ways to produce flippers.

[DMA]

These flippers are less conductive than the original ones. I had to reduce the threshold in the Leyden Jar firmware. I am typing with activation offset of 7 (instead of the 20 which I have been using with my "regular" FSSK.

Congratulations! I specially approve of the extra step you took to print the spring support across the layers, not in parallel.

The "less conductive" part might be because OG flippers underside is actually not flat - it's a cylinder with some huge radius (I tried to figure out that radius from photogrammetry, but the photogrammetry effort failed miserably, so I'll try to deduce that radius from plain old photographs and counting pixels.), and your model appears to have a flat bottom.
Flat bottom over curved plate will result in a PCB contact patch being one or two narrow bands instead of a theoretically large rectangle with curved flipper.
There's no good way to compare - if you compare on a PCB laying flat on the table, printed flipper should get ahead, if inside actual keyboard - OG flippers should get ahead, and there's really no way to put them into "apples to apples" situation.

[idollar]

Congratulations! I specially approve of the extra step you took to print the spring support across the layers, not in parallel.

Thank you

There is still work to do. I am sure that my design can be improved. Both in improving the shape of the flipper and the support. Now that we know that it is possible to do it, we can reshape it to better match the need.

During all the test which I did, I have also discovered new techniques which helped me to deal with the problems and solutions:

• I have been using a piece of transparent methacrylate which I placed instead of the PCB to see how the flipper moved. I specially used this technique with the keys that did not want to flip.
• The plastic barrel plate had big holes where the screws had to come in. They would not create any pressure, leaving the plastic barrel plate lose at this very point. I used some filament and heat, applied with a cheap solder, to first fill the hole to then drill (with a dremel) a new hole.

The "less conductive" part might be because OG flippers underside is actually not flat - it's a cylinder with some huge radius (I tried to figure out that radius from photogrammetry, but the photogrammetry effort failed miserably, so I'll try to deduce that radius from plain old photographs and counting pixels.), and your model appears to have a flat bottom.
Flat bottom over curved plate will result in a PCB contact patch being one or two narrow bands instead of a theoretically large rectangle with curved flipper.
There's no good way to compare - if you compare on a PCB laying flat on the table, printed flipper should get ahead, if inside actual keyboard - OG flippers should get ahead, and there's really no way to put them into "apples to apples" situation.

The resulting part is less conductive. It is a fact.
It is also factual that with the right adjustment in the DAC threshold it works. I keep typing this text with my test keyboard

As I already said before, there are some adjustments that can be investigated:

• Increase the thickness of the "flat" part of the flipper
• As you said, shape it as a section of a cylinder
• But we can go further, what about designing different shapes for different sections of the keyboard ? I have noticed that the flippers in the middle row register lower values (to be confirmed with the oscilloscope). It should be possible to design different flippers which adapt better to this "difficult" part of the keyboard.

Finally, let me say that there is some manual work needed to clean each and every flipper.

• Removing the quasi-square support it is simple with small pliers.
• Getting rid of the lower long support is more difficult. I need to improve the interface between the support and the flipper.
• The pin holding the spring needs manual cleaning. This is particularly bad; if one "cleans" (cuts) too much, the spring is lose with the associated impact. I am thinking on replacing the self computed support with my in-part design (red parts). This will allow to reduce the amount of filament used, reduce the manual processing and improve the interface with the spring.

[idollar]

Congratulations! I specially approve of the extra step you took to print the spring support across the layers, not in parallel.

Thank you

There is still work to do. I am sure that my design can be improved. Both in improving the shape of the flipper and the support. Now that we know that it is possible to do it, we can reshape it to better match the need.

During all the test which I did, I have also discovered new techniques which helped me to deal with the problems and solutions:

• I have been using a piece of transparent methacrylate which I placed instead of the PCB to see how the flipper moved. I specially used this technique with the keys that did not want to flip.
• The plastic barrel plate had big holes where the screws had to come in. They would not create any pressure, leaving the plastic barrel plate lose at this very point. I used some filament and heat, applied with a cheap solder, to first fill the hole to then drill (with a dremel) a new hole.

The "less conductive" part might be because OG flippers underside is actually not flat - it's a cylinder with some huge radius (I tried to figure out that radius from photogrammetry, but the photogrammetry effort failed miserably, so I'll try to deduce that radius from plain old photographs and counting pixels.), and your model appears to have a flat bottom.
Flat bottom over curved plate will result in a PCB contact patch being one or two narrow bands instead of a theoretically large rectangle with curved flipper.
There's no good way to compare - if you compare on a PCB laying flat on the table, printed flipper should get ahead, if inside actual keyboard - OG flippers should get ahead, and there's really no way to put them into "apples to apples" situation.

The resulting part is less conductive. It is a fact.
It is also factual that with the right adjustment in the DAC threshold it works. I keep typing this text with my test keyboard

As I already said before, there are some adjustments that can be investigated:

• Increase the thickness of the "flat" part of the flipper
• As you said, shape it as a section of a cylinder
• But we can go further, what about designing different shapes for different sections of the keyboard ? I have noticed that the flippers in the middle row register lower values (to be confirmed with the oscilloscope). It should be possible to design different flippers which adapt better to this "difficult" part of the keyboard.

Finally, let me say that there is some manual work needed to clean each and every flipper.

• Removing the quasi-square support it is simple with small pliers.
• Getting rid of the lower long support is more difficult. I need to improve the interface between the support and the flipper.
• The pin holding the spring needs manual cleaning. This is particularly bad; if one "cleans" (cuts) too much, the spring is lose with the associated impact. I am thinking on replacing the self computed support with my in-part design (red parts). This will allow to reduce the amount of filament used, reduce the manual processing and improve the interface with the spring.

I will post pictures of the full process, in the form of a manual/tutorial, to allow everyone to 3d print, process and install the flippers in their model m
I just need to find the time. This weekend is already fully booked with activities. It will need to be next week.

Note addressing those who just take ideas without offering any idea nor solution (no need to address the individual as he knows his name):

thanks a lot for all your missing support in this journey. You can now continue misusing DT for your sole benefit. Do not worry, you will be able to steal the ideas and share no benefit with the author. I just hope that the new border tariffs battle help in your business

After reading the above, you may continue with your great contribution to destroy this place, where some others share information in an open manner, for the benefit of the community.

[DMA]

[*] The plastic barrel plate had big holes where the screws had to come in. They would not create any pressure, leaving the plastic barrel plate lose at this very point. I used some filament and heat, applied with a cheap solder, to first fill the hole to then drill (with a dremel) a new hole.

There's a thing called "3D pen" - which is basically handheld FDM printer extruder with a manually operated filament feeder. Should be ideal for this - with less burning plastic smell at least

The "less conductive" part might be because OG flippers underside is actually not flat

I wish to retract this. Spent couple hours today trying to make a high-res photo - which I actually made, see below - only to find out that that cylinder radius is HUGE and only noticeable in how light reflects from the bottom of the flipper, actual non-flatness is like 0.01mm over the flipper area

DSC_0140.JPG (2.18 MiB) Viewed 1110 times

While I was writing this down, it occured to me that I actually have a resin 3d-printer with supposedly 16x24um horizontal resolution and so I can just print curvature gauges (because I have some idea about said curvature radius being around 300mm) and measure by simply putting two parts together and looking at it at magnification (being old sucks, damn presbiopia). Sure, not NIST-traceable, but should be good enough.

The resulting part is less conductive. It is a fact.

Original flippers are not conductive at all. Their surface resistance is >20MOhm/mm, that's well within insulator territory (I have a megohmmeter, but it's currently broken, so I can't currently measure more than 20MOhm.). Your filament is advertised at 7 ohms/cm - you can check actual resistance by sticking multimeter probes into it 1cm apart (and if it displays "O.L" - 1mm apart).

Lower levels might be because flipper's bottom surface isn't smooth enough - which can be rectified by sanding OR just sliding the flipper over the flat part of the iron at "wool" setting (careful about melting flipper legs and burning fingers!) - probably wrapping the flipper in tracing paper to prevent it sticking to the iron and melting completely.

[*] Increase the thickness of the "flat" part of the flipper

supposed to be 0.8mm. But may result in LOUDER CLICKS, so why not!

[*] But we can go further, what about designing different shapes for different sections of the keyboard ? I have noticed that the flippers in the middle row register lower values (to be confirmed with the oscilloscope). It should be possible to design different flippers which adapt better to this "difficult" part of the keyboard.

Backplate curvature not supposed to change.. I have one laying around (in fact, typing on it now) - so I'll be back after I print those curvature gauges.

Finally, let me say that there is some manual work needed to clean each and every flipper.

This isn't really much of a problem with 3d-printing flipers for oneself. May be even considered a benefit, when stuff is too easy people don't value it that much

Seriously though - I would consider reducing cross-section of the points where support touches the part itself, and reducing overhangs - say, by making the part "v"-shaped where it contacts the support. Supports don't have to be thick, especially in FDM world.

PS: photogrammetry attempts made me semi-ruin one flipper (I think I'll be able to clean it, but..)

DSC_0099.JPG (1.44 MiB) Viewed 1106 times

DSC_0116.JPG (1.69 MiB) Viewed 1106 times

[idollar]

Hello,

Just a quick post to state that I keep typing with my 3d printed flippers.
The firmware of my keyboard needs some adaptation (the layout of the keyboard that I used is ANSI). There are two keys which need some adjustment, probably because the back screws are too tight. Otherwise, it works as my other FSSK/FEXT keyboards.

I believe that after some testing, we will have a solution to transform our model M keyboards into model Fs.

In the meanwhile, I have received some new FEXT (x10) and FSSK (x100 PCBs. I updated the FEXT to avoid the issue with the hole.
I will post about this, adding some pictures, as soon as I find the time.

Cheers

i$

[DMA]

This is a set of convex and concave curvature gauges, R290 to R330. R315 is there because value calculated from the whole backplate gave me R315.

PXL_20250219_040552976.jpg (1.66 MiB) Viewed 344 times

Turns out, the keyboard part (R2 to R6) is R310, and the upper part is >R330, may be >R400 - at least on my model M backplate. I have XT, AT and F122 backplates, but my lifting restrictions currently prevent me from accessing those. Expect more results in about two weeks.
Flipper also seems to be most closely matching R310 - so R310 is the golden value I guess (NOTE: that R310 arc must include the pivot point too!)

[idollar]

Turns out, the keyboard part (R2 to R6) is R310, and the upper part is >R330, may be >R400 - at least on my model M backplate. I have XT, AT and F122 backplates, but my lifting restrictions currently prevent me from accessing those. Expect more results in about two weeks.
Flipper also seems to be most closely matching R310 - so R310 is the golden value I guess (NOTE: that R310 arc must include the pivot point too!)

Very clever movement DMA! Thanks for the information.

I would like to report my experience with the 3D-printed flippers:

• Positive: I had no problems with under-sensing keys. All keystrokes are well registered.
• Positive: I had no problems with over-sensing keys. There is no cross-pressing contamination either.
• Positive: No flipper has broken. This does not mean that it will not happen. But after heavy typing during a week, it has not happened.
• Negative: The feeling of the keys is heavier. This is probably due to the reuse of the springs, which means that we need to test different springs.
• Pending: I would like to take this keyboard apart to check the status after this week. It will help to adapt the 3D design. My intention is to update it to reduce the post-processing of the flippers coming out from the printer. DMA, I can use your measurements above in this process.
• Pending: I am 3D-printing a "Rock / Print Tumbler": I plan to smooth the flippers in this device to check whether it helps with the key feeling (which is already good).
• Comment: I can confirm this is not a quiet keyboard. The flippers hit the PCB directly.

This idea of the "Print Tumbler" has to do with the fact that I feel that every day the keyboard gets better. We may be able to accelerate the wearing of the flippers to ensure that their shape is soften from day one.