🚀 go-pugleaf

In article <2eb4edac@ralf>, Ralf Brown <ralf@telerama.lm.com> says:
>
>I just tried it on my AnyKey, and found:
>     <left-shift> <f> <t>    fails on the 't'
>     <left-shift> <g> <t>    fails
>     <f> <left-shift> <t>    fails
>     <g> <left-shift> <t>    fails
>     <left-control> <f> <t>  fails on the 't'
>     <f> <left-control> <t>  fails
>     <left-control> <g> <t>  fails
>     <g> <left-control> <t>  fails
>but
>     <right-shift> <f> <t>   works
>     <f> <right-shift> <t>   works
>     <right-ctrl> <f> <t>    works
>     etc.

The AnyKey keyboard is a hard contact keyboard.  This means it is layed out in a 
matrix, with each key at the junction of an x and y matrix line.  When you press 
a key, the keyboard firmware sees one x line and one y line as being active, so
it can tell what key is pressed.  When you press more than one key, it can usually 
tell which key is pressed, except for one special circumstance: when you press 
three keys and the keys form three corners of a rectangle in the matrix.  When 
this happens, the firmware can't tell if those three were pressed, or if the 
key at the forth corner was pressed.  This is called a phantom, and most keyboards 
will send an 'ff' code to signal that condition.  To get around this situation, 
most keyboard manufacturers take great pains to lay out the matrix so that a 
phantom will occur for only the most unusual of key combinations.

If the AnyKey phantoms on such simple combinations as left shift-f-t, then the matrix 
is very poorly layed out.  Thanks for the ammunition.<G>

By the way, a capacitive keyboard does not phantom at all.  You can press up to 
eight keys at once, and the computer will see all eight scan codes.

        Mike Burton
        Senior Software Engineer
        Key Tronic Corp. (one guess what we make<BG>)

In article <39loq8$7e6@oldklatu.interserv.net>, mburton@mortimer.keytronic.com (Michael Burton) says:

>If the AnyKey phantoms on such simple combinations as left shift-f-t, then the matrix 
>is very poorly layed out.  Thanks for the ammunition.<G>

Is my face red!  I tried the left shift-f-t combination on our 122-key keyboard and 
it worked fine, so I assumed that this combination would work for most of them. Wrong! We manufacture the Microsoft Natural Keyboard, and it exhibits this problem, too.
Take Gateway's advice and don't press all three keys at once.

	Mike

In article <39loq8$7e6@oldklatu.interserv.net>
           mburton@mortimer.keytronic.com "Michael Burton" writes:

[Interesting stuff, but why would you need to press eight keys at
once (unless you were a mac user :-)...]
> 
>         Mike Burton
>         Senior Software Engineer
>         Key Tronic Corp. (one guess what we make<BG>)
> 
Tronics?
-- 
Stuart Baldwin - at home - email: stub@boxatrix.demon.co.uk
West Didsbury, Manchester, UK
(Sorry, thought I was in demon.local)

In article <39opu8$mk7@oldklatu.interserv.net>
           mburton@mortimer.keytronic.com "Michael Burton" writes:

> Is my face red!  I tried the left shift-f-t combination on our 122-key keyboard and
> it worked fine, so I assumed that this combination would work for most of them. Wrong! We
>  manufacture the Microsoft Natural Keyboard, and it exhibits this problem, too.
> Take Gateway's advice and don't press all three keys at once.

This must be a case of cheese-paring by the manufacturers. AFAIK, to get a
matrix with full n-key roll-over you need a diode in series with every
switch ... that's 102 diodes, at maybe 0.2p each in the quantities they use,
but by the time everyone has added their markup it would probably add about
5 quid to the final price.

--
Tim                                       [this sig intentionally left blank]

In article <784329927snz@boxatrix.demon.co.uk>, stub@boxatrix.demon.co.uk (Stuart Baldwin) says:
>
>Tronics?
>-- 

Key Tronic.  Despite the best attempts of the entire rest of the 
world to add one, there is no 's' in the name.  Never has been.
	Mike

In article <784476806snz@bierman.demon.co.uk>, tim@bierman.demon.co.uk (Tim Bierman) says:
>
>In article <39opu8$mk7@oldklatu.interserv.net>
>           mburton@mortimer.keytronic.com "Michael Burton" writes:
>
>> Is my face red!  I tried the left shift-f-t combination on our 122-key keyboard and
>> it worked fine, so I assumed that this combination would work for most of them. Wrong! We
>>  manufacture the Microsoft Natural Keyboard, and it exhibits this problem, too.
>> Take Gateway's advice and don't press all three keys at once.
>
>This must be a case of cheese-paring by the manufacturers. AFAIK, to get a
>matrix with full n-key roll-over you need a diode in series with every
>switch ... that's 102 diodes, at maybe 0.2p each in the quantities they use,
>but by the time everyone has added their markup it would probably add about
>5 quid to the final price.

It's a little hard to pare .5 oz of cheese off of a piece that weighs .3 oz. 
In U.S. terms, if the diodes cost .02 cents a piece, then that's 2.04 added 
to our costs. Since our margin (profit) on most keyboards is less than 2 
dollars, we sould be selling the things at a loss.  Maybe we could make up 
the difference in volume, eh?<G>

If we attempted to pass the cost of the diodes along to our (OEM) customers, 
they would take their business to NMB or whoever, who would sell them a 
keyboard WITHOUT THE DIODES.

I don't know if you've noticed, but computer prices NEVER go up.  They always 
go down.  Where do you think that reduction comes from?  Some reduction in 
the cost of some parts, for sure, but mostly they tell their suppliers (like 
keyboard manufacturers) "This time we don't want a $14 keyboard, we want
one for $11.  If you can't give it to us, maybe NMB or Alps can".

	Mike

In article <39tlg7$e40@oldklatu.interserv.net>
           mburton@mortimer.keytronic.com "Michael Burton" writes:

> It's a little hard to pare .5 oz of cheese off of a piece that weighs .3 oz.
> In U.S. terms, if the diodes cost .02 cents a piece, then that's 2.04 added
> to our costs. Since our margin (profit) on most keyboards is less than 2
> dollars, we sould be selling the things at a loss.  Maybe we could make up
> the difference in volume, eh?<G>

Umm ... I wasn't saying it didn't make commercial sense to do this paring
(though I hate to think just how mickey-mouse machines are going to get if
it gets carried to its logical conclusion). Your arithmetic is wrong, though
- I said 0.2 pence, or about $0.003. Multiplied by 102 gives 30.6 cents. I
was guessing at a price, based on the fact that even in quantities of a few
thousand, diodes are less than 0.5p even to people like me. Using surface
mount multiple-diode packs should reduce it even lower than 0.3c.

You have to think of it as a proportion of the cost of the switch, which is
a fairly rugged device rated for enormous numbers of operations. It's
mass-produced in amazing quantities, but then so is a diode.

In the end, it's the customer's option. Personally, I'd pay a few more
pounds/dollars for NKRO. Those who wouldn't pay this have to accept the odd
missed character if they work at touch-type speeds.

--
Tim                                       [                                 ]

In article <784648697snz@bierman.demon.co.uk>, tim@bierman.demon.co.uk (Tim Bierman) says:
>
>Umm ... I wasn't saying it didn't make commercial sense to do this paring
>(though I hate to think just how mickey-mouse machines are going to get if
>it gets carried to its logical conclusion). Your arithmetic is wrong, though
>- I said 0.2 pence, or about $0.003. Multiplied by 102 gives 30.6 cents. I
>was guessing at a price, based on the fact that even in quantities of a few
>thousand, diodes are less than 0.5p even to people like me. Using surface
>mount multiple-diode packs should reduce it even lower than 0.3c.
>
>In the end, it's the customer's option. Personally, I'd pay a few more
>pounds/dollars for NKRO. Those who wouldn't pay this have to accept the odd
>missed character if they work at touch-type speeds.

My guess was a bit high, but you get the idea.  We aren't selling these 
keyboards to the public, we are selling to OEMs and VARs who insist on the 
best price at the expense of better quality.  As long as this is the case, 
and as long as the users who buy from these OEMs and VARs say nothing about 
keyboard quality, you won't see any hard contact keyboards with a diode at 
each switch position. 

My preference is the capacitive keyboard, but that is more expensive to 
manufacture and is being phased out by all keyboard companies.  Our 
capacitive boards were rated at 100 million keystrokes.  The newer hard 
contact boards are rated at a more conservative 10 to 30 million keystrokes. 
Newer ain't necessarily better.<G>

	Mike

In article <39loq8$7e6@oldklatu.interserv.net>
   mburton@mortimer.keytronic.com (Michael Burton) writes:
> 
> The AnyKey keyboard is a hard contact keyboard.  
> (...)

Many thanks to all those who responded to my post. I must admit to being 
quite surprised at the outcome (not least by how many keyboards exhibit 
the problem and how few people have noticed).

Gateway had replaced my keyboard once before I posted this (all at no 
charge). They have since paid for their courier to collect the 
replacement, and I am leaving it at that. I'll just have to get used to 
the fault - I'm not sufficiently bothered to spend time and money finding 
a better keyboard (I needed the macro facility for a specific project 
anyway). 

(The macro facility is great for ftp too - when getting FAQ files from 
rtfm with filenames like Long_group_name_FAQ_(Version_3_1)_(part_1_of_16) 
it can save loads of typing!)

So, thanks again to all those who responded, both by email and posted 
here - especially to those who had to research their answers (books, 
tests on several keyboards, etc). A quick summary: 

Problem: 
My keyboard (Gateway AnyKey) had problems with the key sequence SHIFT F T 
(three keys) pressed and held in that sequence. Some other combinations 
also failed, but most worked okay. 

Gateway said:
Don't type that way!

Usenet / email said:
The problem is a hardware one, common to many keyboards (including many 
genuine IBM keyboards). It appears from IBM references that this problem 
means that such keyboards are not IBM-compatible (ie they don't meet the 
IBM spec) but since many IBM keyboards fail in the same way that argument 
doesn't hold much water :) Apparantly the problem could be fixed by means 
of a reletively cheap design mod, but consumer pressure has driven prices 
below the point at which that becomes reasonable.

I don't think I missed anything - apologies if I have.

Mark.
-- 
int MeaningOfLife (void) { return 42; }    // mcr@holly.demon.co.uk
function MeaningOfLife :integer; begin MeaningOfLife := 6 * 9; end;