N-key rollover test.

[iMav]

Ok, use THIS LINK to test your keyboard's rollover capabilities.

Reply in this thread with your results and with as much info about the keyboard as possible (including the interface, USB, PS/2, etc).


Enjoy!

[iMav]

Happy Hacking Keyboard Professional 2
Interface: USB
Operating System: Mac OS X 10.5.1 (Leopard)
Max keys accepted: 7

[zillidot]

There's a problem with the way the testing is done. On non n-key-capable keyboards, the results depend on which keys you press.

Please include in the replies, the set of keys used in the testing. And please post the *minimum* number of keys on which the keyboard fails (i.e. not all of the keys pressed are registered). I'll post an example below.

To find this minimum set of keys, try to press keys on adjacent rows simultaneously. (eg QWE and ASD)

(iMav can you add this to your first post? Just so that people know what to do.)

[zillidot]

Apple keyboard (old squishy white version)
Interface: USB
OS: Mac OS X 10.5.1 Leopard
Minimum number of keys on failure: 4 - tested with: QWAS


(Note that it happily accepts 6 keys if I hold down the home row)

[iMav]

@zilldot

Would it be possible (or feasible) to come up with a specific set of tests to use in conjunction with the rollover test page?  i.e., Press letters x-y-z...press e-r-x-c...etc  and standardize it?  Or is that ultimately flawed due to the various in keyboard configurations?

Basically, what I would like you to do (if you are willing, of course), is to write up a set of instructions for everyone to follow regarding this.  I will add it to this thread as well as to the rollover test page.

[zillidot]

Would it be possible (or feasible) to come up with a specific set of tests to use in conjunction with the rollover test page?  i.e., Press letters x-y-z...press e-r-x-c...etc  and standardize it?  Or is that ultimately flawed due to the various in keyboard configurations?

I think keyboard designs are similar enough to have a standardised test. I suggest the following:
1. Q-W-A-S
2. Q-W-E-A-S-D
3. Q-W-E-R-A-S-D-F
(and so on...)
When a failure is encountered (i.e. not all the keys register), try to reduce the number of keys to find the minimum at which the failure occurs.

Most non n-key capable keyboards would fail on 1. My guess is that keyboards limited by the USB protocol would pass 1 and 2 but fail on 3 (but I don't have one with me to test right now).

We can transpose these tests to different locations on the keyboard, eg a few keys to the right, or up or down a row. But I suspect that wouldn't make a difference, unless it's an unusual keyboard design. If we encounter those we can always add more tests to the list.

[xsphat]

MacBook internal keyboard
Keys: qwe
max:3

[pex]

I think keyboard designs are similar enough to have a standardised test.

I can't agree with that from my travels.

[puntium]

pex: I think zilldot is referring to the keyboard wiring and not necessarily the language or the layout.

Most foreign keyboards are variations of US keyboards with different keycaps and a few extra keys. It takes a long time to develop a good wiring for a keyboard that prevents ghosting for commonly used key combinations. It usually has to do more with human factors (i.e. in what ways are you unlikely to contort your hands.. for ex, no one would ever press qwas together naturally). Because people's hands tends to be the same, these wiring schematics become rather universal.

[xsphat]

Model M 1391401, 1984,
4 keys
qwer

[zillidot]

Filco Majestouch FKBN104M/EB (n-key rollover model)
Interface: USB
OS: Mac OS X 10.5.1 Leopard
Max keys: 6 (tested with QWEASD, fails with QWERASD)

(Note that the documentation says that it has true n-key rollover only if PS/2 connection is used. On USB it is limited to 6 keys.)

[xsphat]

Weird, the HHKB Pro 2 gets more than the Filco.

[pex]

Model M 1391401, 1984,
4 keys
qwer

Model M 1391401 wasn't made in 1984.  You're saying a 1391401 doesn't get n-key rollover? WHAT.

pex: I think zilldot is referring to the keyboard wiring and not necessarily the language or the layout.

Most foreign keyboards are variations of US keyboards with different keycaps and a few extra keys. It takes a long time to develop a good wiring for a keyboard that prevents ghosting for commonly used key combinations. It usually has to do more with human factors (i.e. in what ways are you unlikely to contort your hands.. for ex, no one would ever press qwas together naturally). Because people's hands tends to be the same, these wiring schematics become rather universal.

If only it were that simple.  It's surprising the number of judgment calls we see on keyboards as far as what blocks what.

[iMav]

Model M 1391401 wasn't made in 1984.

Hmmm, I guess someone is going to have to post a pic of their "birth certificate".  

I've got a Model M 1393278 made in 2007.  

If only it were that simple.  It's surprising the number of judgment calls we see on keyboards as far as what blocks what.

I don't think anyone is, intentionally, trying to pass misinformation.  I'm asking anyone that is willing to draft one up, to come up with a comprehensive, idiot-proof, rollover test plan that we can post here so that everyone is testing the same things.  Once that is done and we have a general consensus, I will clear out this thread and allow everyone to resubmit using the standardized test.

[pex]

Hmmm, I guess someone is going to have to post a pic of their "birth certificate".  

I've got a Model M 1393278 made in 2007.  

Sometimes people misread the copyright/patent/whatever line on the keyboards as year of mfg.  However, the 1984 line was on 1391401 over a decade's-worth of models.  I think I read that November '85 was the earliest 1391401 anyone could come up with.

I don't think anyone is, intentionally, trying to pass misinformation.  I'm asking anyone that is willing to draft one up, to come up with a comprehensive, idiot-proof, rollover test plan that we can post here so that everyone is testing the same things.  Once that is done and we have a general consensus, I will clear out this thread and allow everyone to resubmit using the standardized test.

If we're going to be true to the cause, I just don't want any keyboard getting away under- or over-credited.  I don't have enough keyboards lying around to find a semblance of proof for my baseless, anecdotal claims on keyboard electronics layouts.  However, until someone does some specific and somewhat scientific testing, I don't think any of us, with our anecdotal evidence, should make final decisions on what constitutes a test!

I mean, we can find out n-key rollover if we take a standard school ruler and run it across any line of a keyboard.  That'll get at least 10 keys, probably many more.  But what if it passes on the top line, but we do it to the next line, and it fails?  What if it passes on the top line but the numpad isn't n-key rollover?  It's certainly not n-key rollover, then.

If we're not doing testing only for n-key rollover and looking to describe any-key rollover, well, the testing gets that much harder.

I just suggest we're going to have to put a bit of work into this (unfortunately) to get things right.  I mean, why does it make sense I can type "i4lw,z" but not "flip"?

[zillidot]

If we're going to be true to the cause, I just don't want any keyboard getting away under- or over-credited.  I don't have enough keyboards lying around to find a semblance of proof for my baseless, anecdotal claims on keyboard electronics layouts.  However, until someone does some specific and somewhat scientific testing, I don't think any of us, with our anecdotal evidence, should make final decisions on what constitutes a test!

I mean, we can find out n-key rollover if we take a standard school ruler and run it across any line of a keyboard.  That'll get at least 10 keys, probably many more.  But what if it passes on the top line, but we do it to the next line, and it fails?  What if it passes on the top line but the numpad isn't n-key rollover?  It's certainly not n-key rollover, then.

I think what we're trying to do (or at least, what I'm proposing) is to find key combinations which cause keyboards to fail (i.e. not register all the keys pressed). You're right in saying that it's practically impossible to test for true n-key rollover. But it is easy to test if a keyboard does NOT have n-key rollover - you just have to find a key combination on which it fails!

So to simplify the search for this key combination, what I'm suggesting is a list of key combinations on which keyboards commonly fail. We can then test keyboards against these key combinations, and record which combinations it fails on. Note that we're mainly interested in the SMALLEST key combination on which it fails.

It's not possible for the list to be comprehensive, but if we come across an unusual keyboard which fails on an unusual key combination, we can always add it to the list.

[xsphat]

Hmmm, I guess someone is going to have to post a pic of their "birth certificate".

Yeah, 84 would be the copyright date. My 1391401 was made 05 June 1988. You buckling spring buckeroos are sticklers.

[xsphat]

Since no member here has tried this on a PS/2 computer, I'm guessing none of us have one that's worth the time, so really what is the point of n-key rollover? I think the vast majority of humanity can live just fine with the ability to use only three alphanumeric characters in conjunction with the existing modifier keys.

We all use USB to connect our peripherals, so who cares? I just don't understand this.

[pex]

We all use USB to connect our peripherals, so who cares? I just don't understand this.

No.  I also still use Betamax and Laserdisc.

I'm trying to kick things off here with some empirical study.  I've dissected my broken Model M.

http://solutions.sveit.com/modelm.html is where I'll be keeping notes on it; it is partially relevant to this thread because for my keyboard I may be able to /prove/ what combinations issues I could expect for at least one keyboard: mine.  We can use this, in part, to develop our strategy.

[xsphat]

OK. What this study needs is someone with a true n-key rollover keyboard to test this out with iMav's test on a PS/2 connector-baring computer. What would you say if it was still only 7?

[iMav]

Since no member here has tried this on a PS/2 computer, I'm guessing none of us have one that's worth the time, so really what is the point of n-key rollover?

The last computer I had that had PS/2 connections was passed on to a friend a couple years ago.  

n-key rollover may not be important to you (and, really, it is not all that important to me either), but there are several members here who are quite interested.  I'm all for defining some standardized testing and allowing the membership to help properly identify rollover functionality of the various keyboards we have available to us.


Actually, now that I think about it...I DO have a PS/2-equipped computer out in the workshop.  I'll have to go dust it off and see if she'll fire up.  

[zillidot]

Since no member here has tried this on a PS/2 computer, I'm guessing none of us have one that's worth the time, so really what is the point of n-key rollover? I think the vast majority of humanity can live just fine with the ability to use only three alphanumeric characters in conjunction with the existing modifier keys.

I would have tested it on PS/2, but have been a bit lazy. I have seriously considered using a PS/2 keyboard on my gaming PC. While I agree that you only need 3-key rollover for normal typing, it is quite easy to end up pressing 4 or more keys simultaneously while gaming.

To be honest, I don't consider n-key rollover to be that important either (except for gaming). But it's nice to have. And since there's so much disinformation out there, it would be good to have some reliable data about it.

I'm trying to kick things off here with some empirical study.  I've dissected my broken Model M.

Interesting pictures! I see no signs of diodes in the keyboard matrix that are apparently required for n-key rollover...

[zillidot]

OK. What this study needs is someone with a true n-key rollover keyboard to test this out with iMav's test on a PS/2 connector-baring computer. What would you say if it was still only 7?

Ok I did it!

Filco Majestouch FKBN104M/EB (n-key rollover model)
Interface: PS/2
OS: Ubuntu 7.10
Keys: more than I can press

I tested the key combos I listed above: QW-AS, QWE-ASD, QWER-ASDF, etc. It passed all of them, at least until I ran out of fingers (up to 20 keys, pressing 2 keys at once with each finger).

Also tried pressing many keys at once using a book. The results were a bit variable, possibly because I'm not depressing all the keys correctly. A few times I had problems with "stuck" keys where the OS thought the key was held down when it wasn't. But that only happened when I tried to press around 30 keys at once. I conjecture that a buffer is overflowing somewhere, either in the keyboard or in the OS, causing key release events to be lost.

I gave up testing at this point as I don't have 30 fingers.

[iMav]

Keys: more than I can press

Maybe we should have a contest.  Who can post a screen capture from the rollover test page with the most keys recognized?  

[Mikecase00]

I've tried testing several key combinations on the n-key rollover test page using my 1391401 through a PS/2 KVM switch.  It's interesting really, I've had as many as 8 keys register (the most I've tried) and as few as 0.  Certain key combinations seem more likely to register higher counts than others.  I suspect that the KVM switch might be hindering performance a bit, so I'll tray again w/o it later.

The Model M keyboard actually has some on-board memory to buffer characters in case the computer your typing on is too busy to accept input (this was a more important consideration back in the M's heyday).  I don't recall what the actual buffer amount was, I think it might have been in the neighborhood of 13 chars.

[Mikecase00]

Here's my best so far

[Nonmouse]

Well, on my MX-8100 (G80-8113HRBUS-2, PS/2 connection), here's the best I've got a screencap of:


The only limit I've found is physically being able to press the keys down.  These were pressed both on the main keys and on the number pad.

I suppose I should see if I can rig a way to press all 120 keys at once, to see if it really is true n-key roll-over.  Being as that I already know that I can press more than twice as many keys as I have fingers, I may not be too hot to find out...

[xsphat]

Six keys at a time with this:

http://geekhack.org/showthread.php?t=288

[fkeidjn]

The keyboard I used for this test:

-IBM 84-key PC AT buckling spring keyboard similar to this keyboard
-This is a capacitor buckling spring keyboard as opposed to the model M, which is a membrane keyboard with buckling springs.
-I used a AT to ps/2 adapter and connected the keyboard through ps/2 connection.

The result, after mushing as many keys as I could, is here, 39 keys!?

[xsphat]

7 keys at a time with my brand new shiny Filco (model number in sig).