Author Topic: Let’s discuss gasket mount and why it may not be what you think.  (Read 34415 times)

0 Members and 1 Guest are viewing this topic.

Offline Leslieann

  • * Elevated Elder
  • Thread Starter
  • Posts: 4518
Let’s discuss gasket mount and why it may not be what you think.
« on: Thu, 17 September 2020, 23:27:01 »
Lately we've seen quite a few people wanting “gasket mount” and I think it's time we discuss it.


First off, what is gasket mount.
Gasket mount is a bit of rubber (or similar) between the plate or pcb and the case. The original purpose was to isolate vibration/noise traveling from the plate and/or pcb to the case.  Does it work? As with most things in keyboards, yes, but as always it's not a fix for everything. I’m not sure where the idea came from but along the way some people got the idea it dampens impacts as well. While some forms of it can do this it's usually only on small keyboards such as 65% or less and then it's done more as a hack, not so much a part of the design and on larger keyboards it's a side effect.

251897-0
On top mount (Plate bolted to the top cover) or bottom mount (plate bolted to the case bottom) both depicted here, the damper can be below, above, around the edge or any combination of the 3.  On any keyboard like this it's really more of a side effect and minimal at best for reasons we'll get into.

251899-1
On plate-less or floating plate designs it's found under the pcb. Most of these are just people adding o-rings or foam under the pcb. It does work as we'll discuss later but it has limited applications and it can create some issues. Like I said, it's a hack.


So what’s the problem
Let's start with the most common thing people have with it, the suspension itself. Imagine 3 springs in a row,  1, 2, 3 with a beam across them, like this.

251901-2

Ok, so this isn't a perfect example of what happens but it should get the point across because it's important and something I've seen a few times now.
If you press above spring 1 almost all your load is on spring 1 (50grams), same for spring 3 (50 grams), if you press on top of it almost all your load is on spring 3, in both instances spring 2 becomes almost inconsequential. Depending on the design it may pivot on it like a see-saw or it may slightly compress (60-70grams), but it will not fully compress as far as spring 1 or spring 3. So what happens if you press above spring 2, it’s load gets spread across spring 1, 2 and 3, instead of 50-70grams your spring pressure can be as high as 150grams. There is no way to make the middle feel as soft as the ends even if you lower the spring pressure both sides they still to the pressure you will see. Again, greatly over-simplified, the point is anything in the middle is effected by everything around it. If you add more springs it doesn't change this, it's going to be that spring/damper plus those around it. What if you eliminate spring 2, you still spread a middle press across spring 1 and 3, besides, you have more than 2 keys. What this means is there’s almost no way to suspend a flat object (I.E. a plate) this way and make it consistent across the keyboard. It’s not impossible, but it is in the confines of a keyboard, and certainly not by just sticking foam, gel or some sort of gasket under the pcb or the plate.

Maybe you don’t really care about it feeling consistent, I can see that, certain parts can be beneficial to be lighter and softer. That’s part of the idea behind Realforce, many people use a stiffer spacebar, even my current board is setup that way to an extent with a few slightly stiffer springs in strategic places. The problem is you aren’t able to control it well enough. You need to have it down to individual keys or at least various zones but while you can, it would take quite a lot of design work and testing.

There is still yet another issue and that is mass and the lack of dampening itself on such a large moving object. The stem in your switch is controlled by the spring and to a lesser degree lube, it may be loosely controlled but it’s very low weight so it doesn’t really matter. The pcb, switches and plate are not low weight, the higher the weight the more damping you need to control it. What happens if you don’t? You slam the W key, the assembly goes down then starts to come back up only to slam right into your finger slamming down on S key making the impact harder than it would be had you not had any movement in the first place. So how do they deal with dampening the plate movement? Easy, they either squish the gasket down tight so it doesn't or they simply don't bother. Remember it's meant to stop vibration and sound not soften the 300 pound gorilla pounding on the keys. I think this may be where the idea gets misunderstood. On Top and bottom mount the the plate isn't just softly floating on a cushion, it's clamped, preferably tight. Yes, you could run it loose, which is what they do with the o-rings on plate-less and floating plate designs but what did you just read about consistency?


So if you can't reasonably suspend the plate, how about allowing for some flex?
The plate was never meant to flex, it’s very job is to keep things from flexing and protect everything under it. While you could make it out of something flexible, cut it up or eliminate it entirely, in all cases you're kind of defeating the whole point of it being there in the first place.


That leads to the second issue and it's actually a bit more grim.
Instead of springs imagine a trapeze artist high above the floor, he slips and down into the net he goes. What happens to the net? It droops down, just as you want but what else happens? It draws in extra material from around it.

251903-3

The pcb doesn’t stretch, it's not elastic, it only bends (same for a flexible plate) and all that extra length has to come from everything around it and just like above it will happen at different rates. What ends up happening is you are pulling the switch bodies in half vertically and sideways. The plastic will likely be fine, but what else is it pulling, the solder joints and the switch contacts. You may get false triggers, damage the contacts or break the solder joints. A half plate is even worse, while it will flex more easily all that load is now concentrated on the few perimeter switches. Before multiple switches would help absorb the load but here it's all concentrated on fewer switches and it's more switches doing it to them.

On keyboards with a floating plate like Tofu or GH60 or if it's entirely plate-less, mounting the the pcb on o-rings does work to a degree at avoiding the shearing issues and can give you some damping but you’re back to the issue of springs in a row, rebound and and I haven’t yet seen a pcb that had even remotely equal spacing around a keyboard making consistency even worse. There's also the issue of most keyboards over 60% are using a plate and the better ones are typically top or bottom mount not a floating plate. If you do want a cushion though, this is about as close as you will get. If you do go down this road just know it's a limited group of keyboards than can do it and I'd recommend a floating plate over plateless if you want it to last for the reason we'll touch on in a moment. One last important thing to note, doing this lifts the pcb and therefore the USB port, depending on the placement and design the port may no longer fit through the hole. Make sure you have clearance or expect to cut the hold a bit larger to make it work.


And what of the PCB itself?
While fiberglass can and has been used as a spring they generally don''t have copper traces being twisted, pushed and pulled, eventually they can pop off or split. "But it’s only a little bit”.  You don't flex your pc's motherboard so why would you think flexing this is okay? You also aren't flexing the motherboard as often or doing it with a localized impacts or load. What do you think the plate was there for, looks? It's not there to make it heavy, it's there to stiffen the case and protect the PCB. This is why you don't see plate-less full size and TKL mechanical keyboards. I'm not saying this will kill your PCB right away, the point is it's bad for it, smoking kills but people don't usually drop dead after one cigarette. It may last a year it may last several but don't tell me it's good engineering.

By the way, all of this is massively worse with Kailh hot swap, now you have torque on the socket and it's mounting pads on top of sliding the switch around in the socket itself grinding parts together and losing contact. If you have them and you value your keyboard in the slightest do not try doing any of this to it. If you have hot swap, regardless of type, you need a good solid plate, it can be gasket mounted, but it needs to hold things in alignment.


The ironic part of all of this is people spending all this effort trying to put back some of the very properties they ran away from when using a plastic chassis (flex). You could easily add weight to one and avoid all sorts of pitfalls an aluminum keyboard has.  It’s just silly to spend all this money, time and effort to try and replicate the feel of something you could get for hundreds of dollars less. I've said this before, if you want it to last build your keyboard like a race car, bicycle, motorcycle, etc... The keyboard is the chassis, you (generally) want it as stiff and stable as possible, that IS why you bought a heavy aluminum keyboard, right? Treat the switches like the suspension, the part you tune. It’s the only way to make it consistent across the entire thing or adjust specific properties and get a consistent outcome. If you want to type on a cushion with super soft landing get something with more travel and designed for a softer landing, like a rubber dome or Topre because other than the ones mentioned, MX style keyboards (and specifically the switches) just don't lend themselves to this.
Novelkeys NK65AE w/62g Zilents/39g springs
More
62g Zilents/lubed/Novelkeys 39g springs, HK Gaming Thick PBT caps, Netdot Gen10 magnetic cable, pic
| Filco MJ2 L.E. Vortex Case, Jailhouse Blues, heavily customized
More
Vortex case squared up/blasted finish removed/custom feet/paint/winkey blockoff plate, HID Liberator, stainless steel universal plate, 3d printed adapters, Type C, Netdot Gen10 magnetic cable, foam sound dampened, HK Gaming Thick PBT caps (o-ringed), Cherry Jailhouse Blues w/lubed/clipped Cherry light springs, 40g actuation
| GMMK TKL
More
w/ Kailh Purple Pros/lubed/Novelkeys 39g springs, HK Gaming Thick PBT caps, Netdot Gen10 Magnetic cable
| PF65 3d printed 65% w/LCD and hot swap
More
Box Jades, Interchangeable trim, mini lcd, QMK, underglow, HK Gaming Thick PBT caps, O-rings, Netdot Gen10 magnetic cable, in progress link
| Magicforce 68
More
MF68 pcb, Outemu Blues, in progress
| YMDK75 Jail Housed Gateron Blues
More
J-spacers, YMDK Thick PBT, O-rings, SIP sockets
| KBT Race S L.E.
More
Ergo Clears, custom WASD caps
| Das Pro
More
Costar model with browns
| GH60
More
Cherry Blacks, custom 3d printed case
| Logitech Illumininated | IBM Model M (x2)
Definitive Omron Guide. | 3d printed Keyboard FAQ/Discussion

Offline Sup

  • * Exquisite Elder
  • Posts: 1336
  • Doing university was a mistake
Re: Let’s discuss gasket mount and why it may not be what you think.
« Reply #1 on: Fri, 18 September 2020, 08:12:45 »
Thanks for the informative info about gasket mount  :)
current
Filco Zero -  NOS Yellow Alps | Canoe R1 Gateron Red | AEK II JP Cream dampend |Filco Majestouch 2 Tex case Gateron Yellow | HHKB Pro 1 2003 Rev AO Serial 000171 | HHKB Pro 1 2003 Rev A1s|DZ60 OG Panda's with Fei spring and stem. | Sentraq S65_Plus OG Invyr Panda's | A17 Gateron Black TX 65G 3204 | Lubrigrante Wildcard Cherry MX silent blacks 3204 58.5G Springs | Rukia Everglide Tourmaline Blue 58.5G Springs | MGA Standard Greetech brown |
Coming soon
Rest in peace Billy Herrington(William Glen Harold Herrington) 1969-2018
Rest in peace Byron Daniel 1989-2020

Offline Polymer

  • Posts: 1587
Re: Let’s discuss gasket mount and why it may not be what you think.
« Reply #2 on: Fri, 18 September 2020, 09:29:08 »
So basically it doesn't take away the "issues" with PCB mounted...

It isn't clear if this is any worse though..

Offline treeleaf64

  • Posts: 1836
  • Location: United State
  • Traveler
    • treeleaf64
Re: Let’s discuss gasket mount and why it may not be what you think.
« Reply #3 on: Fri, 18 September 2020, 10:17:39 »
Top/ bottom mount good
treeleaf64: https://discord.gg/rbUjtsRG6P

This is the cat and pat!!!!!!!!

Offline Learis

  • Posts: 102
Re: Let’s discuss gasket mount and why it may not be what you think.
« Reply #4 on: Fri, 18 September 2020, 13:24:17 »
Could someone comment on which mounting style is best for dampening the ping/reverberation? I really dislike that sound. I originally got interested in the gasket mount because of this since the rubber gaskets seems to help... But now it seems like more manufacturers are getting smart and including dampening material like neoprene in between the plate and the pcb and also at the bottom of the case.

Are certain mounting styles notorious for causing more ping? I think I saw in a video that an integrated plate style caused a lot of reverb for one particular keyboard (I think it was the dz60 hhkb model). How is sandwich mount for this?
Mend and Defend

Offline treeleaf64

  • Posts: 1836
  • Location: United State
  • Traveler
    • treeleaf64
Re: Let’s discuss gasket mount and why it may not be what you think.
« Reply #5 on: Fri, 18 September 2020, 15:52:20 »
Could someone comment on which mounting style is best for dampening the ping/reverberation? I really dislike that sound. I originally got interested in the gasket mount because of this since the rubber gaskets seems to help... But now it seems like more manufacturers are getting smart and including dampening material like neoprene in between the plate and the pcb and also at the bottom of the case.

Are certain mounting styles notorious for causing more ping? I think I saw in a video that an integrated plate style caused a lot of reverb for one particular keyboard (I think it was the dz60 hhkb model). How is sandwich mount for this?

Depends on plate material lube etc  , yeah integrated plate gives you hollow sound .  In general I don't hear a lot of ping from sandwich mount. But it really depends on plate material. For instance  Acrylic plate sandwich mount you won't hear any ping or reverb because sandwiched acrylic doesn't really have any space to reverb

So it really depends on design and  material that you use
treeleaf64: https://discord.gg/rbUjtsRG6P

This is the cat and pat!!!!!!!!

Offline Leslieann

  • * Elevated Elder
  • Thread Starter
  • Posts: 4518
Re: Let’s discuss gasket mount and why it may not be what you think.
« Reply #6 on: Fri, 18 September 2020, 22:14:10 »
So basically it doesn't take away the "issues" with PCB mounted...

It isn't clear if this is any worse though..
I always default to, keep the pcb stable.
Floating in air, well fastened, whatever, just limit it's flex to as little as possible.

Could someone comment on which mounting style is best for dampening the ping/reverberation? I really dislike that sound. I originally got interested in the gasket mount because of this since the rubber gaskets seems to help... But now it seems like more manufacturers are getting smart and including dampening material like neoprene in between the plate and the pcb and also at the bottom of the case.

Are certain mounting styles notorious for causing more ping? I think I saw in a video that an integrated plate style caused a lot of reverb for one particular keyboard (I think it was the dz60 hhkb model). How is sandwich mount for this?
Ping is... complicated.
It's not so much the case style as to everything or at least vibration as a whole. Case style matters but it's

However, it's not the plate causing it, the springs do, the plate is merely acting as an amplifier. If you can stop the ping at the source the rest won't mater.
Thinner spring wire can cause ping but note that this doesn't mean thin and light needs to be avoided it's only a facto. How a spring is made, the material, coatings, lube (if any) all matter as well so don't avoid light springs just because it's thin metal.

I've had good luck with Mechanicalkeyboards.com springs as well as Sprit and Zeal. Cherry is known for ping, especially lighter springs. In my experience, Kaihl is the worst I've used. I have issues with certain pitch sounds so I have to really listen to hear ping even from Cherry but the Pro Purples were so audible that they actually caught my attention and forced me to deal with them. INSANELY pingy. A spring swap and they're now dead silent.
Novelkeys NK65AE w/62g Zilents/39g springs
More
62g Zilents/lubed/Novelkeys 39g springs, HK Gaming Thick PBT caps, Netdot Gen10 magnetic cable, pic
| Filco MJ2 L.E. Vortex Case, Jailhouse Blues, heavily customized
More
Vortex case squared up/blasted finish removed/custom feet/paint/winkey blockoff plate, HID Liberator, stainless steel universal plate, 3d printed adapters, Type C, Netdot Gen10 magnetic cable, foam sound dampened, HK Gaming Thick PBT caps (o-ringed), Cherry Jailhouse Blues w/lubed/clipped Cherry light springs, 40g actuation
| GMMK TKL
More
w/ Kailh Purple Pros/lubed/Novelkeys 39g springs, HK Gaming Thick PBT caps, Netdot Gen10 Magnetic cable
| PF65 3d printed 65% w/LCD and hot swap
More
Box Jades, Interchangeable trim, mini lcd, QMK, underglow, HK Gaming Thick PBT caps, O-rings, Netdot Gen10 magnetic cable, in progress link
| Magicforce 68
More
MF68 pcb, Outemu Blues, in progress
| YMDK75 Jail Housed Gateron Blues
More
J-spacers, YMDK Thick PBT, O-rings, SIP sockets
| KBT Race S L.E.
More
Ergo Clears, custom WASD caps
| Das Pro
More
Costar model with browns
| GH60
More
Cherry Blacks, custom 3d printed case
| Logitech Illumininated | IBM Model M (x2)
Definitive Omron Guide. | 3d printed Keyboard FAQ/Discussion

Offline duynguyenle

  • Posts: 1388
  • Location: UK - Midlands
  • Personal text? What personal text???
Re: Let’s discuss gasket mount and why it may not be what you think.
« Reply #7 on: Thu, 05 November 2020, 05:54:07 »
I see some good points raised, but I also disagree with a lot of it from a purely mechanical point of view. From an engineer's perspective, your little sketches are fine, but you are idealising the case down to a simple harmonic oscillator analogy, which really ignore a lot of variables. Namely, it's not a system that can be idealised into a simple harmonic oscillator. You're completely ignoring the mechanical damping term of the equation.

So let's address firstly, your statement: "There is still yet another issue and that is mass and the lack of dampening itself on such a large moving object.", this is not entirely true. If you have used a gasket mounted board, there are plenty of mechanical damping for the plate/PCB sandwich. In this analogy, you're idealising said plate/PCB assembly into a single flat plate in order to illustrate your point about how "slamming your finger on the W key, have the plate deflect, come up and then slamming back into your finger pressing the S key". This doesn't reflect how the behaviour works in reality. Firstly, the system itself (PCB/plate assembly) is mechanically damped. they don't flex like a cantilevered plate. Certainly there will be some vertical deflection, but it doesn't bounce up and down like a spring like you're alluding to, since the assembly itself has its inherent mechanical damping term in the equation that prevents it from "bouncing back" like a perfect spring. The gaskets that is under the plate (assuming the keyboard is mounted with gaskets on the bottom side of the plate, I know some boards have gasjets on both bottom and top) add their own mechanical damping term into the equation.

Secondly, even if the system behave like a perfect spring, as you say, you also need to take into account the magnitude of deflection. Simply put, even if the plate was springing perfectly like you said, you need to take into account the fact that the forces we're dealing with here AREN'T REALLY THAT LARGE, we're talking 0.5-1N of force. For a simple cantilever beam, supported on both ends, the maximum deflection under full load (which happens at the centrepoint of the beam) just ISN'T THAT BIG. Yes you have valid point on consistency, different points on the board will have different maximum deflection based on its relative position to the fixed boundary, but a gasket-mount allows for slightly more uniformity in deflection compared to top-mount anyway (imagine if you were doing an FE analysis on a plate, top mount is akin to setting your boundary conditions to fixed stationary in all 6 degrees of freedom, as opposed to a gasket mount, where you're essentially allowing small amounts of compliance at the mount points).

So now, let's address the second point you raised (about the "trampoline" or the "gymnast and the net" analogy). I see your point about this as fundamentally wrong due to incorrect assumptions. Let's start with the first sentence "The pcb doesn’t stretch, it's not elastic, it only bends". This is incorrect. Every material behaves elastically up to a point. All materials respond to deflection via elastic deflection, and only permanently deforms (plastic deflection) once its elastic limit has been exceeded. Coming back to my point above about the MAGNITUDE of the forces involved here, within the confines of a keyboard application, the forces involved (1-2N at most, for normal typing), the PCB and plate essentially behaves elastically. You're simply are not going to exceed the elastic deflection region of a FR4 PCB substrate in normal typing.  That's not even counting the fact that switches and solder join do also have their own elastic deflection range, and againt, I cannot stress this hard enough, you're not going to exceed this in normal keyboard use. Solder is a relatively soft material (usually tin/lead or tin/silver mixture), and metallic compounds are pretty elastic. If a solder joint in a hammer drill can withstand the harsh vibration and impact load in its normal operating range, the solder joints holding your switch to the PCB pads aren't gonna fail because you're typing on it.

Your assertion that there are no full-size or TKL plate-less mechanical keyboards is also wrong. Cherry G80-3000 is full-size, with no plate, and they have been making them since the 1980s, Cherry G80-1800 is TKL-sized (-ish) and also have no plate. I have several samples of each of those boards (one the early 1990s, the rest are from late 1990s), all heavily used, and the solder joints on them don't really look like they might be giving up any time soon.

All in all, while I understand your arguments, and some of them are coming from seemingly valid first-principles, as an engineer, I can tell you that you have neglected to actually consider the implications on first-principles reasoning in a real-world scenario. All of the points you have raised are more or less a non-issue. Cherry designed the switches and PCB specs to work either with or without a plate, and in fact, they still to this day manufacture their G80 keyboard lines without plates.

Stop worrying about it, gasket mount is the way the market is moving. I will leave it to the users whether they like it or not compared to top-mount, because it really is personal preference, but from a practical product engineering point of view, there's nothing wrong with gasket mount in its many implementations on the market.
| QFR            | Leeku 1800    | Raptor K1      | Dolch Pac

Offline Leslieann

  • * Elevated Elder
  • Thread Starter
  • Posts: 4518
Re: Let’s discuss gasket mount and why it may not be what you think.
« Reply #8 on: Thu, 05 November 2020, 09:23:41 »
I see some good points raised, but I also disagree with a lot of it from a purely mechanical point of view. From an engineer's perspective, your little sketches are fine, but you are idealising the case down to a simple harmonic oscillator analogy, which really ignore a lot of variables. Namely, it's not a system that can be idealised into a simple harmonic oscillator. You're completely ignoring the mechanical damping term of the equation.

Stop worrying about it, gasket mount is the way the market is moving.
I think you ignored some of the parameters regarding the plate, you're argument works if it's being dampened by a couple loosely clamped o-rings but that's not usually the case on a top or bottom mounted plate and while that may be the case in some instance, it's far from the rule.

Yes I oversimply the heck out of it so even a child could understand it, that was the point. I'm using MS Paint and your trying to explain orbital mechanics, yes, everything is elastic and all that but it lends pretty much zero to the conversation so there was no point getting into it.

And I never said it was a bad design, it's a great idea, it's just misunderstanding by a lot of people (I want to type on a cloud!).
Novelkeys NK65AE w/62g Zilents/39g springs
More
62g Zilents/lubed/Novelkeys 39g springs, HK Gaming Thick PBT caps, Netdot Gen10 magnetic cable, pic
| Filco MJ2 L.E. Vortex Case, Jailhouse Blues, heavily customized
More
Vortex case squared up/blasted finish removed/custom feet/paint/winkey blockoff plate, HID Liberator, stainless steel universal plate, 3d printed adapters, Type C, Netdot Gen10 magnetic cable, foam sound dampened, HK Gaming Thick PBT caps (o-ringed), Cherry Jailhouse Blues w/lubed/clipped Cherry light springs, 40g actuation
| GMMK TKL
More
w/ Kailh Purple Pros/lubed/Novelkeys 39g springs, HK Gaming Thick PBT caps, Netdot Gen10 Magnetic cable
| PF65 3d printed 65% w/LCD and hot swap
More
Box Jades, Interchangeable trim, mini lcd, QMK, underglow, HK Gaming Thick PBT caps, O-rings, Netdot Gen10 magnetic cable, in progress link
| Magicforce 68
More
MF68 pcb, Outemu Blues, in progress
| YMDK75 Jail Housed Gateron Blues
More
J-spacers, YMDK Thick PBT, O-rings, SIP sockets
| KBT Race S L.E.
More
Ergo Clears, custom WASD caps
| Das Pro
More
Costar model with browns
| GH60
More
Cherry Blacks, custom 3d printed case
| Logitech Illumininated | IBM Model M (x2)
Definitive Omron Guide. | 3d printed Keyboard FAQ/Discussion

Offline treeleaf64

  • Posts: 1836
  • Location: United State
  • Traveler
    • treeleaf64
Re: Let’s discuss gasket mount and why it may not be what you think.
« Reply #9 on: Thu, 05 November 2020, 09:31:13 »
I think it is fine if the whole assembly  goes down and up with no flex. It is called bounce and this means that no parts bend, only move up and down by compressing a gasket. I think it's done well on Geonworks F1 8x.

This is not a good system on something like Cherry g80, where the PCB flexes and bends when you press down on one part. It's damaging and putting strain on the soldering joints.
treeleaf64: https://discord.gg/rbUjtsRG6P

This is the cat and pat!!!!!!!!

Offline duynguyenle

  • Posts: 1388
  • Location: UK - Midlands
  • Personal text? What personal text???
Re: Let’s discuss gasket mount and why it may not be what you think.
« Reply #10 on: Thu, 05 November 2020, 09:59:01 »

I think you ignored some of the parameters regarding the plate, you're argument works if it's being dampened by a couple loosely clamped o-rings but that's not usually the case on a top or bottom mounted plate and while that may be the case in some instance, it's far from the rule.

Yes I oversimply the heck out of it so even a child could understand it, that was the point. I'm using MS Paint and your trying to explain orbital mechanics, yes, everything is elastic and all that but it lends pretty much zero to the conversation so there was no point getting into it.

And I never said it was a bad design, it's a great idea, it's just misunderstanding by a lot of people (I want to type on a cloud!).
[/quote]

You fundamentally misunderstood my point. I was TRYING to include parameters that you left out in your argument. When I say 'damping' I don't just mean whatever damping provided by the foam/silicone/poron pads/gaskets itself, I meant the INHERENT mechanical damping of the plate/switch/PCB assembly, think of it like a classical honeycomb-cored composite laminate. The whole point of top mount and its advantage compared to tray mount is to introduce small amounts of flex/compliance to the assembly and avoid a complete dead stop at bottom out. In this sense, a top-mount board ALREADY encourage some amount of flex/movement when typing (again, the max deflection is still very small comparatively, negating one of your arguments based on strain-induced failure of solder joints, PCBs etc... but I digress, let's stick with the purely mechanical analogy for the moment).

Comparatively gasket mount is not all that different from top mount, except it allows for slightly more flex by allowing some additional compliance at the mounting points itself. Mechanically speaking, this is more akin to moving your fixed boundary conditions from the mount points to the edges of the plate...

As for your second statement "And I never said it was a bad design, it's a great idea, it's just misunderstanding by a lot of people" I think now you're just moving the goalpost. Here is the actual quote from your original post: "It may last a year it may last several but don't tell me it's good engineering."

As I have already addressed in my previous post, this is false, PCB flex is not an issue and will not lead to failure under normal typing loads (maybe people like punching their keyboard, I dunno, I'm talking about typical typing force here, around 0.5-2N based on the usual spring weights we typically see in switches). As I said, I own several G80 boards that are 20 years old, no plate, have seen heavy use, and none of the solder points have failed, and I have not had to reflow any of them. PCBs and solder points are a lot stronger than you're making them out to be. Cherry switches have always been designed to be used with or without a plate, it's not needed or necessary for a keyboard to work. You're approaching gasket-mount as a concept from the wrong direction, and with incorrect assumptions, as I have pointed out above (and also in my last post)
| QFR            | Leeku 1800    | Raptor K1      | Dolch Pac

Offline Leslieann

  • * Elevated Elder
  • Thread Starter
  • Posts: 4518
Re: Let’s discuss gasket mount and why it may not be what you think.
« Reply #11 on: Thu, 05 November 2020, 10:15:04 »
"It may last a year it may last several but don't tell me it's good engineering."

As I have already addressed in my previous post, this is false, PCB flex is not an issue and will not lead to failure under normal typing loads (maybe people like punching their keyboard, I dunno, I'm talking about typical typing force here, around 0.5-2N based on the usual spring weights we typically see in switches). As I said, I own several G80 boards that are 20 years old, no plate, have seen heavy use, and none of the solder points have failed, and I have not had to reflow any of them. PCBs and solder points are a lot stronger than you're making them out to be. Cherry switches have always been designed to be used with or without a plate, it's not needed or necessary for a keyboard to work. You're approaching gasket-mount as a concept from the wrong direction, and with incorrect assumptions, as I have pointed out above (and also in my last post)
I sincerely question that spring pressure has a massive impact on how hard people type or bottom out when they start just hammering keys. They get angry and hit keys harder and some people are just heavy handed. Springs may help, but on a hard impact that spring isn't going to change the impact much, particularly in switches where the spring pressure falls off after activation rather than going up.

As for your Cherry board, it's false equivalency.
I never said you needed a plate for the keyboard to survive. My comment about it not lasting was in regards to people designing the housing specifically so the pcb flexed and offered dampenning, that's not how your Cherry board was designed, plate or no plate the housing still keeps the pcb relatively stable. So unless you stuck o-rings under that pcb to suspend it years ago it doesn't change anything.
Novelkeys NK65AE w/62g Zilents/39g springs
More
62g Zilents/lubed/Novelkeys 39g springs, HK Gaming Thick PBT caps, Netdot Gen10 magnetic cable, pic
| Filco MJ2 L.E. Vortex Case, Jailhouse Blues, heavily customized
More
Vortex case squared up/blasted finish removed/custom feet/paint/winkey blockoff plate, HID Liberator, stainless steel universal plate, 3d printed adapters, Type C, Netdot Gen10 magnetic cable, foam sound dampened, HK Gaming Thick PBT caps (o-ringed), Cherry Jailhouse Blues w/lubed/clipped Cherry light springs, 40g actuation
| GMMK TKL
More
w/ Kailh Purple Pros/lubed/Novelkeys 39g springs, HK Gaming Thick PBT caps, Netdot Gen10 Magnetic cable
| PF65 3d printed 65% w/LCD and hot swap
More
Box Jades, Interchangeable trim, mini lcd, QMK, underglow, HK Gaming Thick PBT caps, O-rings, Netdot Gen10 magnetic cable, in progress link
| Magicforce 68
More
MF68 pcb, Outemu Blues, in progress
| YMDK75 Jail Housed Gateron Blues
More
J-spacers, YMDK Thick PBT, O-rings, SIP sockets
| KBT Race S L.E.
More
Ergo Clears, custom WASD caps
| Das Pro
More
Costar model with browns
| GH60
More
Cherry Blacks, custom 3d printed case
| Logitech Illumininated | IBM Model M (x2)
Definitive Omron Guide. | 3d printed Keyboard FAQ/Discussion

Offline Domantas

  • Posts: 200
  • Location: Lithuania
Re: Let’s discuss gasket mount and why it may not be what you think.
« Reply #12 on: Wed, 19 May 2021, 07:58:18 »
I see some good points raised, but I also disagree with a lot of it from a purely mechanical point of view. From an engineer's perspective, your little sketches are fine, but you are idealising the case down to a simple harmonic oscillator analogy, which really ignore a lot of variables. Namely, it's not a system that can be idealised into a simple harmonic oscillator. You're completely ignoring the mechanical damping term of the equation.

So let's address firstly, your statement: "There is still yet another issue and that is mass and the lack of dampening itself on such a large moving object.", this is not entirely true. If you have used a gasket mounted board, there are plenty of mechanical damping for the plate/PCB sandwich. In this analogy, you're idealising said plate/PCB assembly into a single flat plate in order to illustrate your point about how "slamming your finger on the W key, have the plate deflect, come up and then slamming back into your finger pressing the S key". This doesn't reflect how the behaviour works in reality. Firstly, the system itself (PCB/plate assembly) is mechanically damped. they don't flex like a cantilevered plate. Certainly there will be some vertical deflection, but it doesn't bounce up and down like a spring like you're alluding to, since the assembly itself has its inherent mechanical damping term in the equation that prevents it from "bouncing back" like a perfect spring. The gaskets that is under the plate (assuming the keyboard is mounted with gaskets on the bottom side of the plate, I know some boards have gasjets on both bottom and top) add their own mechanical damping term into the equation.

Secondly, even if the system behave like a perfect spring, as you say, you also need to take into account the magnitude of deflection. Simply put, even if the plate was springing perfectly like you said, you need to take into account the fact that the forces we're dealing with here AREN'T REALLY THAT LARGE, we're talking 0.5-1N of force. For a simple cantilever beam, supported on both ends, the maximum deflection under full load (which happens at the centrepoint of the beam) just ISN'T THAT BIG. Yes you have valid point on consistency, different points on the board will have different maximum deflection based on its relative position to the fixed boundary, but a gasket-mount allows for slightly more uniformity in deflection compared to top-mount anyway (imagine if you were doing an FE analysis on a plate, top mount is akin to setting your boundary conditions to fixed stationary in all 6 degrees of freedom, as opposed to a gasket mount, where you're essentially allowing small amounts of compliance at the mount points).

So now, let's address the second point you raised (about the "trampoline" or the "gymnast and the net" analogy). I see your point about this as fundamentally wrong due to incorrect assumptions. Let's start with the first sentence "The pcb doesn’t stretch, it's not elastic, it only bends". This is incorrect. Every material behaves elastically up to a point. All materials respond to deflection via elastic deflection, and only permanently deforms (plastic deflection) once its elastic limit has been exceeded. Coming back to my point above about the MAGNITUDE of the forces involved here, within the confines of a keyboard application, the forces involved (1-2N at most, for normal typing), the PCB and plate essentially behaves elastically. You're simply are not going to exceed the elastic deflection region of a FR4 PCB substrate in normal typing.  That's not even counting the fact that switches and solder join do also have their own elastic deflection range, and againt, I cannot stress this hard enough, you're not going to exceed this in normal keyboard use. Solder is a relatively soft material (usually tin/lead or tin/silver mixture), and metallic compounds are pretty elastic. If a solder joint in a hammer drill can withstand the harsh vibration and impact load in its normal operating range, the solder joints holding your switch to the PCB pads aren't gonna fail because you're typing on it.

Your assertion that there are no full-size or TKL plate-less mechanical keyboards is also wrong. Cherry G80-3000 is full-size, with no plate, and they have been making them since the 1980s, Cherry G80-1800 is TKL-sized (-ish) and also have no plate. I have several samples of each of those boards (one the early 1990s, the rest are from late 1990s), all heavily used, and the solder joints on them don't really look like they might be giving up any time soon.

All in all, while I understand your arguments, and some of them are coming from seemingly valid first-principles, as an engineer, I can tell you that you have neglected to actually consider the implications on first-principles reasoning in a real-world scenario. All of the points you have raised are more or less a non-issue. Cherry designed the switches and PCB specs to work either with or without a plate, and in fact, they still to this day manufacture their G80 keyboard lines without plates.

Stop worrying about it, gasket mount is the way the market is moving. I will leave it to the users whether they like it or not compared to top-mount, because it really is personal preference, but from a practical product engineering point of view, there's nothing wrong with gasket mount in its many implementations on the market.

+1
Agree. Who are mashing their boards? If you type normally nothing will happen just don't use hammer. Also pcbs are much more stronger than one might think. Especialy if you use pc plate and pcb, plate takes most of the flex.


Coming a bit away from the main topic: Right now I am testing stacked mount with hotswap pcb and it works perfectly three months already
« Last Edit: Wed, 19 May 2021, 08:08:11 by Domantas »

Offline Volny

  • Posts: 235
Re: Let’s discuss gasket mount and why it may not be what you think.
« Reply #13 on: Wed, 19 May 2021, 09:35:42 »
I'm surprised no one's mentioned Weirdflex PCBs yet. I know next to nothing about them myself, but they do seem like something that belongs in this conversation.

Anyway, despite the debate circling around various specific tangents, I gather that most people here would actually agree with one of the foundational points of the OP, which is that typical gasket mounting achieves more in the realm of vibration and sound dampening than it does to create a lusciously cushiony soft typing feel.

So what avenues *are* available for those who do want to feel like they're typing on a cloud? (let's assume that they've addressed the switches side of the equation, but are still unsatisfied). It seems to me that an obvious place to experiment would be on the underside of the keyboard, by placing the keyboard on a layer of foam, rubber, or perhaps even some spring-mounted gizmo (or just go for gold and place the keyboard upon a giant heap of marshmallows). It's certainly an easy, cheap and safe avenue to explore.

Needless to say, many people already do this to some degree of course, with deskmats, rubber feet and/or foam underlays. But it's not something I've seen discussed in much detail. What would be the best practices for this kind of thing? What material and/or setup could give you the most effective cushion while remaining sensibly thin and minimising see-saw tilting? What about a keyboard-shaped recess in the desk that can allow the keyboard to sit on a generous layer of padding without gaining extra height? Could something akin to a stabiliser be used to allow the full keyboard to bounce whether pressed on the side or the middle?
« Last Edit: Wed, 19 May 2021, 09:38:10 by Volny »