IC: Top Shelf, Made in the USA, Custom MX 62g Springs (High Spec)

[Linkbane]

The only reason that this deserves any attention is if it's cheaper than Korean springs. It seems dubious that you could possibly have a problem with a spring; what in hell is a 'meh' spring? It's 62g, if you don't like it, it's not because it's not from the US of 'Murica, but because you don't like 62g.

The ID of the various springs that I've bought as replacements seem to be all over the place. Some sit apart from the stem tube, some are seated on the stem tube so tightly that they have to be pulled off gently so that I don't stretch them too far when removing. Of course this is less of an issue after lubing.

And I'm pretty sure that this is translating into a "scratchier" feeling spring.

Scratchy. Come on, we still go on about nonsense like that? My god.

[Puddsy]

Welp, I'm interested.

[mkawa]

for material, the best springs are made out of custom spring steel alloys. usually these alloys are high carbon quenched for tensile and fatigue strength. the two top spring-makers in car-sizes in the US are Swift and HyperCo. They both have proprietary steels that they use for their top shelf designs.

that said, all you really need is measurable tolerances and some reasonable amount of fatigue strength. with MX springs, you are also actually locked to a small window of ID and OD, so the first thing you're going to want to do is measure those out with mic. once you have an idea of how thick your wire can be, you want to find a drawn wire maker that will do something like t316l and not charge you a billion dollars. high precision springs made out of good drawn wire with tight tolerances usually go into things like medical devices where if the spring is off by too large a delta, lives are lost or people become very uncomfortable.

other than that, all steels are very close in elastic modulus, which is what you exercise when you compress a steel coil spring. so it's really mostly about finding a good manufacturer that you can work with.

for those guys with fingers like hammers, you might want to look at rubber or polyurethane die springs (i'm looking at you joneslee ).

actually, come to think of it, the samples of sorbothane i have coming from that other thread might end up being this very thing. we'll see.

[SpAmRaY]

The only reason that this deserves any attention is if it's cheaper than Korean springs. It seems dubious that you could possibly have a problem with a spring; what in hell is a 'meh' spring? It's 62g, if you don't like it, it's not because it's not from the US of 'Murica, but because you don't like 62g.

The ID of the various springs that I've bought as replacements seem to be all over the place. Some sit apart from the stem tube, some are seated on the stem tube so tightly that they have to be pulled off gently so that I don't stretch them too far when removing. Of course this is less of an issue after lubing.

And I'm pretty sure that this is translating into a "scratchier" feeling spring.

Scratchy. Come on, we still go on about nonsense like that? My god.

Does it hurt?

[Linkbane]

Shhh. Your jimmies are safe with me.

[SpAmRaY]

Shhh. Your jimmies are safe with me.

My jimmies are fine but I'm worried that corncob up your butt is becoming permanent.

[Linkbane]

Shhh. Your jimmies are safe with me.

My jimmies are fine but I'm worried that corncob up your but is becoming permanent.

[n0rvig]

GLWBG!

[Pacifist]

GLWBG!

Good Luck With Buy Group?

[korrelate]

for material, the best springs are made out of custom spring steel alloys. usually these alloys are high carbon quenched for tensile and fatigue strength. the two top spring-makers in car-sizes in the US are Swift and HyperCo. They both have proprietary steels that they use for their top shelf designs.

that said, all you really need is measurable tolerances and some reasonable amount of fatigue strength. with MX springs, you are also actually locked to a small window of ID and OD, so the first thing you're going to want to do is measure those out with mic. once you have an idea of how thick your wire can be, you want to find a drawn wire maker that will do something like t316l and not charge you a billion dollars. high precision springs made out of good drawn wire with tight tolerances usually go into things like medical devices where if the spring is off by too large a delta, lives are lost or people become very uncomfortable.

other than that, all steels are very close in elastic modulus, which is what you exercise when you compress a steel coil spring. so it's really mostly about finding a good manufacturer that you can work with.

for those guys with fingers like hammers, you might want to look at rubber or polyurethane die springs (i'm looking at you joneslee ).

actually, come to think of it, the samples of sorbothane i have coming from that other thread might end up being this very thing. we'll see.

Excellent feedback mkawa: thanks for this!

[korrelate]

Why not do something a bit different? I would love a spring that has a force curve like clear, but oh I don't know like 15cN lighter. If going to the trouble to have something made, might as well do something that isn't available anywhere.

Good question. I guess the answer why I'm going this route is:
- Whatever I produce I want that product to be useful and well made (first priority).
- I've never done this before so it's baby-step approach: a simple force curve (like on a linear switch) seems an easier target for a first-timer: particularly because I have very little time to spare.
- If this goes well and I learn anything useful I can turn my attention to more complicated targets.

[taylordcraig]

Still wishing you luck. I need some springs. #dogold

[mooswa]

Clear spring force curve is not more "complex" then a spring from a linear switch, it just have steeper slope.  The attraction of a steeper slope is  comparatively lower force needed to reach an activation point and higher force needed to bottom out.  This should in theory allow to float easier.

[yicaoyimu]

I will take at least 200 springs if this ever goes into GB.

[fart_toast]

Would take about 200 or so too

[mkawa]

Clear spring force curve is not more "complex" then a spring from a linear switch, it just have steeper slope.  The attraction of a steeper slope is  comparatively lower force needed to reach an activation point and higher force needed to bottom out.  This should in theory allow to float easier.

Show Image

russ's measurement jig is not very precise and can only give you a regression of what the spring curves actually look like. haata has a another arbor-press based design that is giving slightly better data (he uses an imada force gauge instead of a 5$ micro-scale :P). if you just look at clear springs, it's pretty obvious that the springs are progressive.

further, i think we do ourselves a disservice if we think that the tolerances on MX springs are so tight that they are _that_ linear. drawn wire that thin with many many different manufacturers over the years are going to provide for much more complex and probabilistic force graphs. we just don't have the equipment to measure these things at the moment in the community.

[mooswa]

...if you just look at clear springs, it's pretty obvious that the springs are progressive.

Are you saying that they have varying pitch? I don't think I see it.  Or is there other way to make a coil spring progressive?   Tolerances aside, it looks like all MX springs have a constant pitch so (in my limited knowledge of these things) were designed to be linear.

[mkawa]

yes, my MX clear springs have varying pitch

and there are multiple ways to make a coil spring progressive. you can vary pitch (ie, coils per unit length) or vary wire diameter, add a helper spring or some other resistance. one other very cool thing you can do is vary the metal alloy along the length of the spring.

anyway, springs are one of the cooler bits of basic physics, and we should not assume so readily that they always look like the first order linear approximation that is hooke's equation.

[jacobolus]

russ's measurement jig is not very precise and can only give you a regression of what the spring curves actually look like. haata has a another arbor-press based design that is giving slightly better data (he uses an imada force gauge instead of a 5$ micro-scale ). if you just look at clear springs, it's pretty obvious that the springs are progressive.

further, i think we do ourselves a disservice if we think that the tolerances on MX springs are so tight that they are _that_ linear. drawn wire that thin with many many different manufacturers over the years are going to provide for much more complex and probabilistic force graphs. we just don't have the equipment to measure these things at the moment in the community.

I’m hoping that I can go help HaaTa measure switches/springs in 2-3 weeks. I’m not sure exactly how precise/accurate his measurement device is going to end up being, but hopefully accurate enough to test basic hypotheses e.g. about the force curves of these springs, the variance from one spring to another, etc.

I’m pretty skeptical that the MX clear springs have a substantially different shape of force curve from other springs. [It is true though that they are stiffer and shorter than red/blue/brown springs.]

It would be neat if there were a cheap-ish way to find out the specific type of metal used in various springs. But even without that, we can hopefully make precise measurements of: wire gauge, spring diameter, number of coils, overall spring length and length of the coiled section, and some kind of reasonable measurement of the force curve.

Can anyone recommend a good mechanical engineering book about compression springs?

[CPTBadAss]

Can anyone recommend a good mechanical engineering book about compression springs?

Look for a machine design book. I forget the name of the one I have.

[Melvang]

So far most of the talk here is about package counts.  That can be sorted out later.  What I want to see is springs for MX that have a exponential force curve.  This can be done via varying the turns per inch throughout the spring.  I want something that actuates about half way between a red and a black but then ramps up to stiffer than a clear at bottom.  I think these would be amazing in most switch types.  Reason I want this.

Nice easy actuation force
Help reinforce to not bottom out.

[mkawa]

russ's measurement jig is not very precise and can only give you a regression of what the spring curves actually look like. haata has a another arbor-press based design that is giving slightly better data (he uses an imada force gauge instead of a 5$ micro-scale ). if you just look at clear springs, it's pretty obvious that the springs are progressive.

further, i think we do ourselves a disservice if we think that the tolerances on MX springs are so tight that they are _that_ linear. drawn wire that thin with many many different manufacturers over the years are going to provide for much more complex and probabilistic force graphs. we just don't have the equipment to measure these things at the moment in the community.

I’m hoping that I can go help HaaTa measure switches/springs in 2-3 weeks. I’m not sure exactly how precise/accurate his measurement device is going to end up being, but hopefully accurate enough to test basic hypotheses e.g. about the force curves of these springs, the variance from one spring to another, etc.

I’m pretty skeptical that the MX clear springs have a substantially different shape of force curve from other springs. [It is true though that they are stiffer and shorter than red/blue/brown springs.]

It would be neat if there were a cheap-ish way to find out the specific type of metal used in various springs. But even without that, we can hopefully make precise measurements of: wire gauge, spring diameter, number of coils, overall spring length and length of the coiled section, and some kind of reasonable measurement of the force curve.

Can anyone recommend a good mechanical engineering book about compression springs?

haata's force measurement device is awesome. he just needs a jig that irons out physical error in placement, give, and increases repeatability. if you look at his graphs, he has a lot of noise.

[mkawa]

the other thing to remember about these things is that no matter how accurate or precise your measurement device is, it's completely a statistical property that needs to be modeled as a distribution over a large number of samples. that, in fact, is the major difference between cheap and expensive springs. it's not the spring force, the wire diameter (ok, progressives are hard to get from cheap manufacturers, so there's that) but tolerances. you pay for tolerances. that is what costs money.

[Melvang]

the other thing to remember about these things is that no matter how accurate or precise your measurement device is, it's completely a statistical property that needs to be modeled as a distribution over a large number of samples. that, in fact, is the major difference between cheap and expensive springs. it's not the spring force, the wire diameter (ok, progressives are hard to get from cheap manufacturers, so there's that) but tolerances. you pay for tolerances. that is what costs money.

I would have to agree 100% here.  At my job I somethings have to do a align a shaft between an electric motor and a pump.  Usually 10 to 50 hp motor.  Nothing huge but I need to know RPM of the motor.  The higher the RPM the tighter the tolerance.  I worked on one very slow motor that had an RPM less than 950.  The tolerances for that was around .015" offset and .030" angular missalignment.  I could have gotten that in tolerance in 20 minutes without laser measuring tools.  But the plant wanted .002" tolerance all the way around.  This took an extra 3 hours because it was only a 1hp motor.  The smaller ones actually take longer.  So tolerances can cost exponentially more money.

[kod]

I realize it's pretty unlikely, but I'd be interested in alps-size springs...

[Linkbane]

the other thing to remember about these things is that no matter how accurate or precise your measurement device is, it's completely a statistical property that needs to be modeled as a distribution over a large number of samples. that, in fact, is the major difference between cheap and expensive springs. it's not the spring force, the wire diameter (ok, progressives are hard to get from cheap manufacturers, so there's that) but tolerances. you pay for tolerances. that is what costs money.

I would have to agree 100% here.  At my job I somethings have to do a align a shaft between an electric motor and a pump.  Usually 10 to 50 hp motor.  Nothing huge but I need to know RPM of the motor.  The higher the RPM the tighter the tolerance.  I worked on one very slow motor that had an RPM less than 950.  The tolerances for that was around .015" offset and .030" angular missalignment.  I could have gotten that in tolerance in 20 minutes without laser measuring tools.  But the plant wanted .002" tolerance all the way around.  This took an extra 3 hours because it was only a 1hp motor.  The smaller ones actually take longer.  So tolerances can cost exponentially more money.

So speaking of tolerances, I know that there was a thread a long time ago about the very large range of tolerances in Topre switches. I'm thinking that's not because of low-quality springs, but does having a conic spring influence force required to depress it?

[kenmai9]

Not Korean? No interest




Kidding. Would be down for a set of 60

[jacobolus]

the other thing to remember about these things is that no matter how accurate or precise your measurement device is, it's completely a statistical property that needs to be modeled as a distribution over a large number of samples.

Absolutely. My plan, if HaaTa can get me set up measuring, is to try to measure at least 10 of each type of switch, and ideally more from several different keyboards/batches.

For better comparisons among Cherry MX and among Alps-mount switches, I’d like to pull switches out and put them just in a bare plate clamped down to something, and make sure to use the same type of keycap to test different switches. Then test each switch at least 10-20 times, pressing it at different places on the key top, and different speeds, if possible. I dunno.. we’ll see what’s possible in a few weeks.

After a large amount of data has been collected, we can try to figure out what the best way is to graph it, to avoid measurement noise and effectively demonstrate the properties of the switch, but also show variance from switch to switch and press to press. Maybe I can make some kind of javascript driven interactive graphics on a webpage somewhere.

Even more fun would be to capture a video and audio of the key being pressed, and try to sync those to an animated bar moving across the force graph.

If that works out well, would be nice to add various switch modifications like orings, landing pads, lubricant, replacement springs, etc., and try to show what effect those have.

[ideus]

Not Korean? No interest

Kidding. Would be down for a set of 60

When I read your first sentence I wondered why the hate against US made products. I smiled after the second. But a bit of the bitterness remained. 

[Linkbane]

Not Korean? No interest

Kidding. Would be down for a set of 60

When I read your first sentence I wondered why the hate against US made products. I smiled after the second. But a bit of the bitterness remained. 

The US doesn't really strike me as making goods of particularly high quality. Compared to places like Japan with Nikon and Canon, Apple (not really a fan, but still) having its products made in Korea, Nintendo making games in Japan, I'd honestly trust their quality more than the United States. I'm unpatriotic in the sense that I'd buy something if it came from out of country if it was cheaper or if I knew it was better, even slightly.

[Melvang]

And some people are in a different boat.  I try and purchase as much stuff that is US made as I can feasibly get away with.  I do realize that is almost impossible with most electronics though.  I guess I was kind of raised on the premise that if we keep buying foreign products, where will our kids work?

So my vote would be for US made springs if we can manage.  I can't see the price being that much different.  Plus in the grand scheme of things shipping would probably be a touch cheaper as I would be willing to bet most customers of a possible GB would be of US origin.

[Linkbane]

And some people are in a different boat.  I try and purchase as much stuff that is US made as I can feasibly get away with.  I do realize that is almost impossible with most electronics though.  I guess I was kind of raised on the premise that if we keep buying foreign products, where will our kids work?

So my vote would be for US made springs if we can manage.  I can't see the price being that much different.  Plus in the grand scheme of things shipping would probably be a touch cheaper as I would be willing to bet most customers of a possible GB would be of US origin.

True, many certainly are. But as far as this topic, it'd be very interesting to know what sort of springs we're looking at in terms of specs. I, too, assume that it will be made in the US for a variety of factors.

[korrelate]

And some people are in a different boat.  I try and purchase as much stuff that is US made as I can feasibly get away with.  I do realize that is almost impossible with most electronics though.  I guess I was kind of raised on the premise that if we keep buying foreign products, where will our kids work?

So my vote would be for US made springs if we can manage.  I can't see the price being that much different.  Plus in the grand scheme of things shipping would probably be a touch cheaper as I would be willing to bet most customers of a possible GB would be of US origin.

True, many certainly are. But as far as this topic, it'd be very interesting to know what sort of springs we're looking at in terms of specs. I, too, assume that it will be made in the US for a variety of factors.

I'm a math guy so I pretty much believe in the normal distribution and that talent in just about any field of endeavor is likewise distributed. I'm not on a hunt to prove anyone is better than anybody else I just think about it like this: korean springs are already on the market. I don't speak Japanese, Chinese, Taiwanese (you see where this is going) nor German, French, etc. and I don't want to piss around with customs, so why not try to source US springs and put those on the market too.

I know I said that the springs that I had tried were "were kind of meh" but that's just those springs: has nothing to do with country of origin.. it's just those springs.

[mkawa]

the other thing to remember about these things is that no matter how accurate or precise your measurement device is, it's completely a statistical property that needs to be modeled as a distribution over a large number of samples.

Absolutely. My plan, if HaaTa can get me set up measuring, is to try to measure at least 10 of each type of switch, and ideally more from several different keyboards/batches.

For better comparisons among Cherry MX and among Alps-mount switches, I’d like to pull switches out and put them just in a bare plate clamped down to something, and make sure to use the same type of keycap to test different switches. Then test each switch at least 10-20 times, pressing it at different places on the key top, and different speeds, if possible. I dunno.. we’ll see what’s possible in a few weeks.

After a large amount of data has been collected, we can try to figure out what the best way is to graph it, to avoid measurement noise and effectively demonstrate the properties of the switch, but also show variance from switch to switch and press to press. Maybe I can make some kind of javascript driven interactive graphics on a webpage somewhere.

Even more fun would be to capture a video and audio of the key being pressed, and try to sync those to an animated bar moving across the force graph.

If that works out well, would be nice to add various switch modifications like orings, landing pads, lubricant, replacement springs, etc., and try to show what effect those have.

this is starting to get a bit off-topic, but i'm a bit fan of measuring things in-situ. things like case and pcb and plate rigidity all make a difference w/rt how the switch feels in practice. further, as you know, a statistically significant sample would have to be uniformly picked from a large number of manufacturing batches across all the production factories etc. etc. etc.

i'd rather see us design a device that is precise but not accurate but that can be assembled at low cost by anyone who can build a phantom/gh60/ergodox. i think this is particularly important because it allows the user to see, along a given metric, how their modifications to their boards affect that metric.

anyway, back on topic. two ways to get custom die springs to play around with:

1) order a small batch from mcmaster

2) determine your wire diameter and coil diameter. buy wire of desired diameter in a couple of different alloys. buy small rod that matches your coil diameter. wrap wire around rod. cut wire. ta-da! spring!

[TheSoulhunter]

I’m hoping that I can go help HaaTa measure switches/springs in 2-3 weeks. I’m not sure exactly how precise/accurate his measurement device is going to end up being, but hopefully accurate enough to test basic hypotheses e.g. about the force curves of these springs, the variance from one spring to another, etc.

Looking forward to see the results...
Are you gonna measure the springs standalone AND in the switch?
This way we could do a simple subtraction and get a idea of the force added by the slider etc.

[jacobolus]

2) determine your wire diameter and coil diameter. buy wire of desired diameter in a couple of different alloys. buy small rod that matches your coil diameter. wrap wire around rod. cut wire. ta-da! spring!

Have you tried this? From this http://web.archive.org/web/20130116000212/http://home.earthlink.net/~bazillion/compression.html it looks a bit more difficult than you’re making it out to be, but it’s possible the difficulties come from trying to make springs out of much wider wire than these keyswitch springs.

[jacobolus]

this is starting to get a bit off-topic, but i'm a bit fan of measuring things in-situ. things like case and pcb and plate rigidity all make a difference w/rt how the switch feels in practice.

For sure, however if the purpose is demonstrating differences between switch types, then every extra difference between keyboards makes that comparison more difficult. Ideally we can measure *both* some examples in a keyboard, and *also* some comparative examples where the different types of switches are sitting in the same plate.

i'd rather see us design a device that is precise but not accurate but that can be assembled at low cost by anyone who can build a phantom/gh60/ergodox. i think this is particularly important because it allows the user to see, along a given metric, how their modifications to their boards affect that metric.

That sounds great. Do you have an idea about how to build one of those?

[Melvang]

The problem isn't in actually making springs.  It is in keeping the consistency between individual springs to not notice a difference.  At the scale that these would be it wouldn't take much of a difference to notice in a switch.

[mkawa]

here is a contact:

gardner spring
18003313263

note they work in imperial units

[nubbinator]

the other thing to remember about these things is that no matter how accurate or precise your measurement device is, it's completely a statistical property that needs to be modeled as a distribution over a large number of samples.

Absolutely. My plan, if HaaTa can get me set up measuring, is to try to measure at least 10 of each type of switch, and ideally more from several different keyboards/batches.

For better comparisons among Cherry MX and among Alps-mount switches, I’d like to pull switches out and put them just in a bare plate clamped down to something, and make sure to use the same type of keycap to test different switches. Then test each switch at least 10-20 times, pressing it at different places on the key top, and different speeds, if possible. I dunno.. we’ll see what’s possible in a few weeks.

After a large amount of data has been collected, we can try to figure out what the best way is to graph it, to avoid measurement noise and effectively demonstrate the properties of the switch, but also show variance from switch to switch and press to press. Maybe I can make some kind of javascript driven interactive graphics on a webpage somewhere.

Even more fun would be to capture a video and audio of the key being pressed, and try to sync those to an animated bar moving across the force graph.

If that works out well, would be nice to add various switch modifications like orings, landing pads, lubricant, replacement springs, etc., and try to show what effect those have.

Need a minimum of 30 switches, preferably more statistical significance.  When you account for age and use variance, well, the numbers needed would grow even larger.

[jacobolus]

Absolutely. My plan, if HaaTa can get me set up measuring, is to try to measure at least 10 of each type of switch, and ideally more from several different keyboards/batches.

Need a minimum of 30 switches, preferably more statistical significance.  When you account for age and use variance, well, the numbers needed would grow even larger.

Sure, whatever. Fortunately, I’m not aiming for perfect oracular insight into all switches in the world, nor is that necessary to be a useful resource. Measuring 10 springs each from 3-4 batches of each type of spring (or switch, or whatever) wouldn’t be impossibly difficult, assuming access to the measurement equipment and enough sample switches.

[Techno Trousers]

I'd be in for probably 300 of the light clear, AKA 'true ergo clear' springs. I think those would be perfect for all my planned upcoming MX modding projects.

[korrelate]

All righty. Sorry I've been away folks. I started a new job recently and I've been trying to get myself up to speed.
Anyway, I have read through all of the posts put up since my last visit and I've got a couple of updates.

I still won't have a ton of time to spend on this, but I will spend what I can. Mkawa, you've touched on a bunch of points that I have to thank you for.

So here's what's on the agenda for the next couple of weeks:

1. I want to create a short list of manufacturers.
2. Contact them and see what's needed as far as specs go.
3. Post the results of the conversations in (2.) above on the forum.

Mkawa: ygpm.

Cheers & thanks for the patience!

[Ishamael]

I'd be interested in the lighter clears discusses here as well.

[Techno Trousers]

Bumper for updates? I'm in on the stainless steel ANSI 125 plate buy, so some of these springs will let me complete a couple of TKL projects.

[mkawa]

i'm not looking to add to my crushing set of responsibilities at the moment. i'm afraid you guys are on your own.