[FMF] Free Model F - Capacitive Flippers

[idollar]

Torayca ASHT-18A with 18 percent carbon content

Great !

Thanks a lot !

[idollar]

Dears,

I wonder if there is any China-based member who could help in interfacing with potential suppliers.
You may publicly post or PM as many other member have already done

Any help is welcome.

Thanks

[idollar]

Dears,

---------------------- Resin printing ------------------

Continuing with options I have seen the following video (resin printing) :

0.5% 1% and 5% graphite concentration are tested in the video.

You may want to jump to the minute 6:40 to see how graphite is mixed with the resine.
Conductive testing can be found in minute 11:10. None of the tests are conductive.

I wonder if the capacitor properties would change with flippers made with this mix.

Torayca ASHT-18A with 18 percent carbon content

kmnov2017 proposed 18% concentration.

---------------------- silicon mold + epopxy with Graphite powder -------------------

Graphite powder is cheap: https://www.ebay.de/itm/174501143324

This video show how to make conductive paint with graphite

I wonder if we could create a silicon mold from the original flippers and cast a mix of regular epoxy and graphite
We do not need conductivity but enough conductive material to change the capacitor built in the PCB

Please comment

[DMA]

Great !

Elipse: (Give and take) why don't you please share the information you hold (CAD, springs specs/source and how to purchase from China) with the community to help in producing the flippers in the same way the community has done with the critical components you have (and still are) using in your products ?

It just occured to me - what if Ellipse isn't unwilling to provide CAD files and spring specs, but actually unable to?
I've seen the insides of one "new model F", and what I saw was consistent with a chinese factory getting an original 4704 and copying it cheaply. Here's several observations consistent with such a scenario:
1) barrels are backwards-incompatible with OG Model F (tested on XT, AT and F122 - locator nub is in a slightly, but noticeable enough angular position)
2) Top plate is fixed to fit the new barrels, so they aren't at weird angle.
3) Controller cable is pushing at the case from inside violating minimum bend radius - hard to not actually notice, but easy enough not to care, because it will work for couple years and warranty is usually 1.
4) Grinder marks on the plates - and overall horrible internally (some will say externally too, but externally there's at least _some_ QA effort is shown)

[idollar]

Hello DMA,

I hadn’t considered the possibility that he might not actually have the design. Instead, he could be pretending to organize the purchasing while actually providing copied samples. If that's the case, he should have shared the source of the parts so that others can access them.

But that’s in the past.

I propose that we focus on the future, which is:

• Designing the parts for a FMF.
• By doing this (and sharing knowledge), we can contribute to the survival of DT.

Cheers

[idollar]

I was not aware of this thread !!!!

viewtopic.php?f=2&t=25847&start=60

There is lot of information to be used. An example: viewtopic.php?f=2&t=25847&p=489498&hili ... 8A#p488482

Hi Dick,

Again big thanks. I have a bunch of questions but I'll space them out.

I have been using a 3d printer with a conductive filament to print beamspring capacitive elements. I have also tried copper tape and a conductive paint for model f capacitive elements that I printed in photo resin. The model f reproduction used plastic Torayca ASHT-18A with 18% carbon.

1.
Do you happen to remember what plastic the beamspring and buckling spring elements were as well as the conductive material?

I would think something like graphite would work but given the capacitive nature I wonder if that would create dust etc that would mess up sensing. As graphite itself is quite lubricating and it could flake off. I would guess that the f flipper was likely a self lubricating type of plastic. When you cut a model f flipper it almost shatters like pencil lead. My 3d printed beamspring elements with the conductive plastic worked perfectly and were sensed even by the original PCB.


2. Idollar flipper shape
We can see in this post from user idollar.
Hi Dick,

Again big thanks. I have a bunch of questions but I'll space them out.

I have been using a 3d printer with a conductive filament to print beamspring capacitive elements. I have also tried copper tape and a conductive paint for model f capacitive elements that I printed in photo resin. The model f reproduction used plastic Torayca ASHT-18A with 18% carbon.

1.
Do you happen to remember what plastic the beamspring and buckling spring elements were as well as the conductive material?

I would think something like graphite would work but given the capacitive nature I wonder if that would create dust etc that would mess up sensing. As graphite itself is quite lubricating and it could flake off. I would guess that the f flipper was likely a self lubricating type of plastic. When you cut a model f flipper it almost shatters like pencil lead. My 3d printed beamspring elements with the conductive plastic worked perfectly and were sensed even by the original PCB.


2. Idollar flipper shape
We can see in this post from user idollar.

viewtopic.php?t=9715

How the model f flipper has several variants. Round, notch and cut corners. They are seemingly interchangable. They have even been found in the same keyboards with no pattern or reason for their use. Do you know why these variants exist? Did different factories make different ones?

Why do the springs have slightly different angles between the variants.

3. Floss mod
We discussed how you have seen the "floss mod" method of model f spring ping dampen.

wiki/Dental_floss_mod

As well as how you had experimented with similar sound dampening. Could you elaborate on whatat methods did you try? Fibrous material like floss, foam plugs, grease etc?

Were there any attempts to silence beamsprings? If not, how would you do it now? My personal thought would be putting thin foam on-top of the capacitive plate so when it goes up it cant smack as much.


4.
We can see in
US Patent 3699296
That you had designed a circuit closing buckling spring. This is very cool. I think you briefly mentioned on the phone about a double flip action or similar prototype.

Can you elaborate on that?

Was there ever at attempt at having a conductive surface inside a model f housing to have a double action switch

5.
This is one part of user HaaTa's excellent keyswitch testing data force data.

https://chart-studio.plotly.com/~haata/68

For the IBM switch section we can see that non-angled stem beamsprings like in my IBM 5110 unit are seemingly different in force from angled stem beamsprings. Such as in the 3278. To me the 5110 switches are lighter.

The switches look basically identical to the eye. Do you know why the more common angled stems have a different force curve? Is the beam itself different?
viewtopic.php?t=9715

How the model f flipper has several variants. Round, notch and cut corners. They are seemingly interchangable. They have even been found in the same keyboards with no pattern or reason for their use. Do you know why these variants exist? Did different factories make different ones?

Why do the springs have slightly different angles between the variants.

3. Floss mod
We discussed how you have seen the "floss mod" method of model f spring ping dampen.

wiki/Dental_floss_mod

As well as how you had experimented with similar sound dampening. Could you elaborate on whatat methods did you try? Fibrous material like floss, foam plugs, grease etc?

Were there any attempts to silence beamsprings? If not, how would you do it now? My personal thought would be putting thin foam on-top of the capacitive plate so when it goes up it cant smack as much.


4.
We can see in
US Patent 3699296
That you had designed a circuit closing buckling spring. This is very cool. I think you briefly mentioned on the phone about a double flip action or similar prototype.

Can you elaborate on that?

Was there ever at attempt at having a conductive surface inside a model f housing to have a double action switch

5.
This is one part of user HaaTa's excellent keyswitch testing data force data.

https://chart-studio.plotly.com/~haata/68

For the IBM switch section we can see that non-angled stem beamsprings like in my IBM 5110 unit are seemingly different in force from angled stem beamsprings. Such as in the 3278. To me the 5110 switches are lighter.

The switches look basically identical to the eye. Do you know why the more common angled stems have a different force curve? Is the beam itself different?

[idollar]

https://en.wikipedia.org/wiki/IBM_Techn ... e_Bulletin

The IBM Technical Disclosure Bulletin was a technical publication produced by IBM between 1958 and 1998. The purpose of the Bulletin was to disclose inventions that IBM did not want their competitors to get patents on. The Bulletin was a form of defensive publication. By publishing the details of how to make and use the invention, patent examiners could have a searchable source of prior art that they could cite against subsequent patent applications filed by others on the same or similar inventions.

Access to the IBM TBD archive is here : https://priorart.ip.com/

You need to select the gear at the right of the search button:

IBMTDBarchive.png (62.72 KiB) Viewed 6497 times

I will attempt to find the attach pdf file as a test.

------ also

https://web.archive.org/web/20090326055 ... ures.shtml

The January 1998 issue was the final issue of the IBM Technical Disclosure Bulletin published by the IBM Corporation

IBM technical disclosures (from 1958 through today) are available for a fee through IP.com. Documents can be purchased through the following:
IP.com Prior Art Database
Or contact Angel Bender at 585-427-8180 or service@ip.com.

They are also kept on file with Patent & Trademark Offices and U.S. Government Depository Libraries. Searching/copying services are NOT provided by any PTO or Government office.

Copies of IBM technical disclosures published in the IBM Technical Disclosure Bulletin prior to February 1998 are also available for a fee through the following organizations:

Denver Public Library
Magazine Center
Attention: Mr. Jim Pinamont
Department Manager
10 W. 14th Avenue Parkway
Denver, CO 80204-2731
(303) 640-6308
Fax: (303) 640-6358

New York Public Library
NYPL Express
Document Delivery
188 Madison Avenue
New York, NY 10016
(212) 592-7201
Fax: (212) 592-7215

[DMA]

You may want to jump to the minute 6:40 to see how graphite is mixed with the resine.

One note upfront - graphite is way worse than "conductive carbon black". From datasheets it looks like things become "conductive" at about 15% of the stuff by weight, depending on actual stuff. And even then, "conductive" is a relative term. So no wonder sticking 20MOhm meter into the sample doesn't show anything - contact area is way too small for example.

Torayca ASHT-18A with 18 percent carbon content

Consistent with conductive carbon black datasheet from above.

Graphite powder is cheap: https://www.ebay.de/itm/174501143324

It's also useless, and here's why: in conductive paint video, the only non-volatile ingredient of nail hardener in nitrocellulose (which is better known as "gun cotton" - so it's QUITE flammable.). So essentially, "50% graphite powder" is likely 95% graphite after drying, and "30%" is like 80%. Acrylic varnish has probably 20-30% solids in it - and you can see that resulting resistance is thru the roof (difference between 30% and 50% is 10x because there's way more conductive bridges forming when acrylic doesn't stand in the way. No contact - no conductivity!). "Conductive carbon black" supposed to be better - but then again, this is for injection molding process only. For FDM printers there's conductive filament (which won't get you low resistance but should be conductive _enough_ for flippers. Heck, original flippers aren't conductive if you stick a normal ohmmeter into them either!)

I wonder if we could create a silicon mold from the original flippers and cast a mix of regular epoxy and graphite
We do not need conductivity but enough conductive material to change the capacitor built in the PCB

Casting is slow and needs vacuum. 3D-printing and painting bottom layer with 843AR will not only be faster, but also cheaper. Sure, 843AR is ~$0.03 per flipper, which is expensive, but compare that to casting setup..
So for functional prototyping 3D-printing should work, for production I recommend finding an injection molding shop willing to do the job..
Epoxy.. can't say I'm against, but it's kinda strange thought to build switches from glue To my defence, I own two poxipol-encased flash drives which survived 15 years in different cars pretty well!

[macmakkara]

Hi there. Il also post over here as i did few DMs with i$ i proposed resin casting with graphite powder or carbon (fiber). We could 3d print 2 piece mold for testing and do few cast. I don't own 3dprinter but could try to use community one at library and maybe do few test casts. As i have been laid-off from my work for unknown time frame i might try to do something for this project.

[DMA]

As i have been laid-off from my work for unknown time frame

This sucks, man. Hope you'll find something before money runs out.

[DMA]

OK, my small additive manufacturing facility is ready for the pilot.

PXL_20250108_030248799.jpg (3.15 MiB) Viewed 5350 times

See ya after surgery! (which is tomorrow)

[idollar]

As i have been laid-off from my work for unknown time frame i might try to do something for this project.

I honestly prefer that you keep your job rather than doing something for the FMF !

Hi there. Il also post over here as i did few DMs with i$ i proposed resin casting with graphite powder or carbon (fiber). We could 3d print 2 piece mold for testing and do few cast. I don't own 3dprinter but could try to use community one at library and maybe do few test casts.

3d printed parts would not work for resin casting. The resin would get glued to the parts.
Instead you have to use silicone ... search in youtube for "2 parts mold resin casting".
I have already purchased the resine, the silicone and the clay. I have some graphite at home.
I am planning to cast something this weekend. I will take some pictures to document my mistakes I have never done it before.

[idollar]

OK, my small additive manufacturing facility is ready for the pilot.
PXL_20250108_030248799.jpg

See ya after surgery! (which is tomorrow)

Very nice ...
Recover soon !

[idollar]

Hello,

A quick update on the flippers test during the weekend:

• I managed to create a mold for the flippers using silicone. This is my first attempt at doing something like this. It's a learning curve.
• I usedthe following mix:
  
  • 3 g graphite
  • 14 ml epoxy
• The resulting flipper was not 100% perfect. A part was missing. I believe this happened because I applied too much pressure on the area holding the two mold parts together, which closed the path for the epoxy.
• The good news is that the flipper copied all the features (see picture). Note that I have not post-processed the flipper to improve the edges.
• The epoxy takes longer to solidify than expected. Longer than 1 day.
• I will try again and may create the mold anyway. kmnov2017 pointed me to a YouTube video with a better technique.
• I installed a Leyden jar on my spare FEXT PCB:
  • My own compiled firmware, which allows me to check (and modify) the code.
  • Metal grounded plate under the PCB as DMA suggested.
  • Confirmed working with the Leyden jar testing tool and metal on top of the PCB.
  • Confirmed working with the Leyden jar testing tool and the original flipper.
  • My flipper does not work! The level change happens, but it is too low.
  
  
  Conclusions:
  
  • Producing flippers by casting with silicone molds is possible. It's just a matter of finding the correct procedure.
  • Graphite will not work. Adding more graphite will be difficult.
  • I am thinking of using copper pulver in my next attempt.
  
  Please comment as needed.
  
  
  

  

  Flipper.jpeg (224.28 KiB) Viewed 2515 times

[DMA]

• 3 g graphite
• 14 ml epoxy

[*]My flipper does not work! The level change happens, but it is too low.

3/17 = 17%. 4g graphite+13ml epoxy will be 23% - which might just be the ticket (could be mechanically unsuitable, in which case better carbon black might work - may be even at original 17%). Copper flakes, depending on particle size, might also work.

[*] The resulting flipper was not 100% perfect. A part was missing. I believe this happened because I applied too much pressure on the area holding the two mold parts together, which closed the path for the epoxy.

Could also be that there's no way for air to escape, and air from the empty space forms the higher-pressure bubble which prevents epoxy from flowing into that area.

[idollar]

Hello DMA,

Good to see that you are back on track !

• 3 g graphite
• 14 ml epoxy

[*]My flipper does not work! The level change happens, but it is too low.

3/17 = 17%. 4g graphite+13ml epoxy will be 23% - which might just be the ticket (could be mechanically unsuitable, in which case better carbon black might work - may be even at original 17%). Copper flakes, depending on particle size, might also work.

Copper was ordered yesterday. It is in its way.

I must admit that the intended proportion was 5+5ml (epoxy) + 5 g graphite.
The procedure which I had in mind was to mix 5g graphite with 5ml of epoxy part-A. Then add the part-B
But when checking the graphite I thought that it would not mix with 5 mm or epoxy.
Also I went to far with the part-A ... so I had to mix 7ml.

For next attempt I will try to mix:

• 5ml part-A with 3g of copper
• 5ml part-B with 3g of copper
• mix the two together.

Lets see if I do not make again mistakes.

In any case is just a prove of concept. Going in a massive production 1000+ this way may not be suitable.
The other sourcing threads are been tracked in paralell.

[*] The resulting flipper was not 100% perfect. A part was missing. I believe this happened because I applied too much pressure on the area holding the two mold parts together, which closed the path for the epoxy.

Could also be that there's no way for air to escape, and air from the empty space forms the higher-pressure bubble which prevents epoxy from flowing into that area.

No, this was not the case DMA.
I made sure that I injected the epoxy from to top (where the spring fits the flipper) and I made a hole to exit the air on the very bottom of the part.
I am sure that I need to play with the rubber bands to:

• put enough pressure
• distribute it across the full mould.

The walls that I used to hold the silicone were not straight neither. It looks so bad that I did not post the images
In my next attempt, I plan to use lego blocks, as I saw in the video from lispnick here. It is always good to listen and learn from the real masters

1.jpg (207.98 KiB) Viewed 1939 times

pressure.jpg (186.89 KiB) Viewed 1939 times

[DMA]

Hello DMA,

Good to see that you are back on track !

Not yet on track - can post but can't really do anything yet, even getting up is a bit of a trouble. But I'm getting there.

I must admit that the intended proportion was 5+5ml (epoxy) + 5 g graphite.

That would be 33% - probably would get good results with that. But yeah, everything needs to be preplanned, once you start mixing time becomes really limited

For next attempt I will try to mix:

• 5ml part-A with 3g of copper
• 5ml part-B with 3g of copper
• mix the two together.

Lets see if I do not make again mistakes.

I like the idea of mixing filler with both parts first, and only activate the epoxy as the last step.
I would encourage you to try that with graphite - but then again, if somebody will try to reproduce it, copper is a better material: it's heavier, so the dust won't fly as far away if you spill it.

In any case is just a prove of concept. Going in a massive production 1000+ this way may not be suitable.
The other sourcing threads are been tracked in paralell.

No, this was not the case DMA.
I made sure that I injected the epoxy from to top (where the spring fits the flipper) and I made a hole to exit the air on the very bottom of the part.

Not sure I understand. The way I see it, whichever way you point the flipper, you'll have either multiple high points, or multiple low points. If you're injecting from the spring attachment - at least one flipper leg has a place for air bubble to be trapped. Can you share mold photos? It's fine if it's ugly - nobody expects perfection on the first attempt.

[idollar]

No, this was not the case DMA.
I made sure that I injected the epoxy from to top (where the spring fits the flipper) and I made a hole to exit the air on the very bottom of the part.

Not sure I understand. The way I see it, whichever way you point the flipper, you'll have either multiple high points, or multiple low points. If you're injecting from the spring attachment - at least one flipper leg has a place for air bubble to be trapped. Can you share mold photos? It's fine if it's ugly - nobody expects perfection on the first attempt.

Hello DMA,

There is only one low point and a high point.
This is the configuration that I have used in my first attempt:

Flipper 1.jpg (54.1 KiB) Viewed 1414 times

I added some wooden sicks to create a channel on "A" and exit air on "B".

------

My plan for the next attempt is to:

• Use LEGO for the walls.
• Design and 3D print a plate with support bars holding the flippers upside down. I can make many at the same time. The flippers will be placed on top of these bars at the same height. I need to find out if I can add some small glue points to hold them in place.
• The plate will be placed at the bottom of the LEGO walls, with the bars pointing up, creating the "A" channels for the different flippers while also supporting them.
• I plan to use syringe needles attached to the same plate to create the air exits ("B") and support the other side of the flippers. They are thinner and will retain some pressure inside. These can be positioned in the middle of the flipper. Once again, lessons learned from Lispnick here.
• I will fill the mold up to the flat part of the flipper with silicone and let it dry.
• I will then spray some liquid on the top surface to allow for the separation of the two parts.
• I will pour more silicone to create the second part.

second attempt.jpg (80.75 KiB) Viewed 1414 times

[DMA]

There is only one low point and a high point.

If there was high/ultra-high vacuum inside the mold - that would indeed be the case. But there's 1atm air there, and as epoxy flows into the mold, it will eventually create air bubbles in the flipper feet (pointed by bent arrows on the drawing).
In injection molding air naturally escapes via parting lines and around ejector pins. Still, sometimes dedicated vent channels are needed - even though plastic is forced into the mold at tremendous pressures. Here's some random (but decent) internet site explaining in more detail.

So, in your case, epoxy pressure is way lower, and air bubbles won't even be compressed 2x. Which means, in current configuration, that the moment epoxy reaches point "B", all the air remaining within the mold will be forced into the feet. In your first attempt, you got lucky with one foot, but not so lucky with another. You should puncture a very thin channel into the mold at all local high points. After the epoxy reaches "B", you should close it with something, and try to push more epoxy into "A", to ensure there's no air pockets left.

Alternatively, you can remake the mold so flipper is "feet-down". The idea is to fill the mold from the local bottom points, ensuring there's no places for air bubbles to form. IF your epoxy is *very* low-viscosity, you can try it with one inlet and hope the feet will be filled before epoxy cuts them off the exit gate (because if that happens - air will need to form a bubble, and that's hard).

..anyway. I'm still at the "printing exposure calibration targets" phase - resin printing is surprisingly messy and surprisingly hard at the initial setup phase. Will report once I get to something.

[idollar]

[So, in your case, epoxy pressure is way lower, and air bubbles won't even be compressed 2x. Which means, in current configuration, that the moment epoxy reaches point "B", all the air remaining within the mold will be forced into the feet.

I see what you mean. I did not contemplated this option. You are correct.

In your first attempt, you got lucky with one foot, but not so lucky with another. You should puncture a very thin channel into the mold at all local high points. After the epoxy reaches "B", you should close it with something, and try to push more epoxy into "A", to ensure there's no air pockets left.

Another alternative is to enter in the three points at the same time.
As I am trying to 3D print the support, it should not be difficult to create this triple Y

Flipper 2.jpg (49.81 KiB) Viewed 870 times

Alternatively, you can remake the mold so flipper is "feet-down". The idea is to fill the mold from the local bottom points, ensuring there's no places for air bubbles to form. IF your epoxy is *very* low-viscosity, you can try it with one inlet and hope the feet will be filled before epoxy cuts them off the exit gate (because if that happens - air will need to form a bubble, and that's hard).

My epoxy was quite liquid to my surprise. But I see your point (again).
It will become messy if I do it "feet-down". I believe that if I enter in the three points at the same time it will work.

..anyway. I'm still at the "printing exposure calibration targets" phase - resin printing is surprisingly messy and surprisingly hard at the initial setup phase. Will report once I get to something.

I am also interested in this process for resin printing.
I may ask for advice once you are done

[DMA]

Another alternative is to enter in the three points at the same time.

"at the same time" implies luck. Consider yourself very unlucky when designing such things: multiple exits are OK, multiple entry points - not so much..
So, considering you want to keep flipper position in the mold intact (which totally makes sense - you want parting line to be as close to horizontal as possible), I'd advise to make air exits in the flipper feet and keep everything else as it is. It will be more logical to fill from the "B" point and have both feet and stem as exit points - but "B" a high point itself, currently collecting air from complete flipper foot area. Filling from "B" is likely to introduce air bubbles at the sides of the flipper foot. If I was doing this - I would actually fill from behind in the lowest point, and 4 vents. But I'm prone to overcautious approaches.

I am also interested in this process for resin printing.
I may ask for advice once you are done

Current advice is "As you value your life or your reason keep away from the moor."
Resin is sticky, washing is tricky, exposure is impossible to dial in, fume hood (for the WHOLE THING, including washing and curing stations!) is mandatory, everything is expensive (I'm about $800 down already, and not even a single print yet) and slow (about micron per second). I'll make it work eventually, but for now the only good thing is that bottle of 843AR makes for a damn good snowglobe!