Tell me how many Euro Dimes it takes to push a key!

[webwit]

Whatever.

[ripster]

Acer Blacks - 7.5gx 8 (1 Euros-7.5g)+4g (1nickel = 3.92g)= 64g. Measured by Runeazn.

[ripster]

Am now looking for Green ALPS numbers.

My bet is it should be around 45g.

[ripster]

I'm back!

[xbb]

I'm back!

till when?

[webwit]

I have a theory about this.

[Jim66]

I have a theory about this.

Does it involve nickels?

[ripster]

Here's a clever variation.

Swedish Krone. Wrapped in tape.

IBM N2 - 78g


I always assumed Sweden is part of the Monetary EU. It's like collecting stamps, I learn about keyboards and world affairs all in one hobby.

[ripster]

Got any Euro 20cent coins?

[kbdfr]

Got any Euro 20cent coins?

No, but I can PayPal you 1€ if it helps

[ripster]

Shipping would be high for a .2Kg roll of them.

Mug an American Tourist. They are the ones with shorts and white socks.

[yakill]

Best thing ever were the Deutsche Mark coins
weighted
2g -> 1 Pfennig
3g -> 5 Pfennig
3.5g -> 50 Pfennig
4g -> 10 Pfennig
5.5g -> 1 Deutsche Mark
7g -> 2 Deutsche Mark
10g -> 5 Deutsche Mark

[ripster]

Am now looking for Green ALPS numbers.

My bet is it should be around 45g.

It ended up being 50g.

[webwit]

It is not.

[ripster]

It is not.

You sure?

[webwit]

Which green Alps version was measured, complicated or XM?

[7bit]

Best thing ever were the Deutsche Mark coins
weighted
2g -> 1 Pfennig
3g -> 5 Pfennig
3.5g -> 50 Pfennig
4g -> 10 Pfennig
5.5g -> 1 Deutsche Mark
7g -> 2 Deutsche Mark
10g -> 5 Deutsche Mark

What about 2 Pfennig?

[The Solutor]

Here's a clever variation.

Swedish Krone.

Have you tried with a travellers cheque ?

[ripster]

ALPS Complicated.

The RipOmeter never lies.

Green Alps Complicated Force Graph


Chart and pic from Silencium

[webwit]

The ripometer is +- 2 coins, therefore sometimes lies 2 coins, sometimes 0, and you don't know unless you can compare with 3rd party measurements. Why again isn't your method redundant by definition? By the way, if it would be like your graphs (just over 50), one would need an extra coin, and the measurement would have been 55g. That suggests that results are manipulated to match with existing, precise measurements, or that people are dropping the coins, causing inertia, and therefore not only the force from the weight is measured. If you got 50 while it's over 50, the result is invalid and should be tossed away.

[ripster]

More datapoints = MOAR ACCURACY!

Somebody check their NMB Whites or Purples. These are linear so you have to stack slowly.

British 1Lb (or whatever you write) is 9.5g but you have to shim or remove the switch because of their diameter. This dude got 30g which I suspect is too low since he got a 85g bottom out force.

[webwit]

More datapoints = MOAR ACCURACY!

No matter how hard you shout it or how often you shout it, that is not correct. I'd advise bringing it to GH or OCN, where there might be support for not using a brain.
Of course you can prove me wrong by posting your mathematical proof below that "dithering" can be applied to your method to increase accuracy:

[7bit]

Hey Ripster!

For a dime, I push ANY KEY and pretend it's whatever grams you like![1]

----------------------------------------
[1] 1 dime per key

[webwit]

I've actually found one use for the ripometer. As discussed earlier, its accuracy is +- 2 coins. It's +- 1 coin because of the interval. For example, with 5g coins, to measure a switch of 50.1g you need 55g in coins, which is one coin inaccurate. On top of that, +- 1 coin because of other inaccuracies, such as the person measuring it slightly dropping the last coin or not perfectly stacking the coins.

Now ripster keeps yelling his method is accurate because of dithering. This is not correct, firstly because of the interval. Same 50.1g switch, it would dither to 55g, thus providing a false sense of accuracy. Furthermore, the secondary inaccuracies might discriminate, and this is where it gets interesting. It could mean the result is dithered to 56g, providing more false sense of accuracy, the more results get in.

+- 2 coins might mean it's more often +2 coins than -2 coins, for example. It could discriminate because for instance dropping the last coin would have that effect.

And this is exactly what multiple ripometer results are dithering. The results don't dither to an accurate result. The results dither to an average error.

So there you have it - the ripometer is an excellent tool, when used by a lot of people and when you have accurate results from another method to compare to, to find out how much the ripometer sucks.

[ripster]

This picture brings tears to my eyes.

A Fujitsu Peerless - 80g.

[Konrad]

Just curious, but what tolerances are permitted in the switch specifications themselves? What if the design or manufacturing produces something like 45±5g? After a threshold higher precision does not yield higher accuracy. I suppose calculating an average from measures across a larger number of keys would lead to better accuracy, plus it could demonstrate any switches which fall outside the norm.

Also curious, why is the nickle-weighting measure condemned? I see that US coinage is indeed manufactured to exacting (and surprisingly metric) mass specifications, I doubt many other objects massing 5.000g could be purchased for 5 cents. If I'm not understanding something about switch operation then please explain; if the reason is the patent-pending Rip-O-Meter-2000™ then no explanation necessary, I already understand.

Also available are specifications for Euro coinage (which differ somewhat from the above reported values) and Canadian coinage.

[The Solutor]

Just curious, but what tolerances are permitted in the switch specifications themselves?

Cherry is pretty vague about anything on their switches, no word about the plastic used, no word about the durability testing methods, no word about the switch tolerances.

I suppose calculating an average from measures across a larger number of keys would lead to better accuracy

Obviously, after someone told him even Ripster started to measure more than one key.

I suppose calculating an average from measures across a larger number of keys would lead to better accuracy, plus it could demonstrate any switches which fall outside the norm.

Doing large tests with this method is stupid and boring, you can easily test the evenness of a keyboard using a spare reference switch upside down, what i call the switch-o-meter.

[webwit]

Also curious, why is the nickle-weighting measure condemned?

You can read all about it... in this topic.

It's too inaccurate to be useful, in fact it does more harm than good when the data cannot be trusted. It's like measuring people's body size by using sticks measuring 30cm in length, or better, let people measure their own length this way and have them report it, and then actually start using the results. It would be an exercise in futility. Either measure it well, or don't measure.

[webwit]

It is all basic highschool physics and maths. I doubt this is unique for Dutch high schools, so people must have not been paying attention.

You know, you needed to answer questions like, to give a very basic example: if you have a thermometer of +- 3% accuracy and you measure the temperature of two liquids, first 100 Celcius, the other 98, can you conclude the second is less hot than the first? Or do you only know they're both pretty hot? Which you already knew.

Likewise you can toss ripometer results. Don't be trapped into a false sense of accuracy. You only get to know if a switch is heavy or light. Which you already knew by operating the switch.

[The Solutor]

You know, you needed to answer questions like, to give a very basic example: if you have a thermometer of +- 3% accuracy and you measure the temperature of two liquids, first 100 Celcius, the other 98, can you conclude the second is less hot than the first? Or do you only know they're both pretty hot? Which you already knew.

Use the correct terms, you are speaking about precision, which is what is missing on the ripometer method.

With an inaccurate thermometer you don't know if the water is 100 Celsius or 97, but you will find for sure if the sample A is hotter than the sample B.

With an imprecise thermometer you don't know if the sample A is hotter than the sample B.

In the ripometer test the accuracy can be easily improved using tinier coins, but the precision will still be pretty low because the tolerances on the switch itself, especially when the switch is tactile.