Filco: NKRO over USB for REALZ!

[laffindude]

Bride of Helvetica?
[ Attachment Invalid Or Does Not Exist ] 48614[/ATTACH]

[laffindude]

a) The Windows driver issue does not affect my converter - NKRO avoids the problem.
b) It's fixed in Windows 7 anyway.

http://geekhack.org/showthread.php?18674-Filco-Stuck-Modifiers-and-Keys <== this issue is related to that issue fixed here?

[tp4tissue]

Heh, that's not how USB works. The polling isn't done by the CPU, it's done by the USB chips. Only IF there has been a packet received does it bother the CPU.... and that's an interrupt, baby! And packets are only sent when there is some change in key state.

So, polling keyboards at 1000Hz doesn't load up the CPU any more than at 100Hz.

Mice are different - they will load the CPU up more if you poll them more, because they can usually find something to report (they just report smaller movement increments each time).


USB devices reserve bandwidth on the bus, so it's only if you've overloaded the bus that that can happen. And that's pretty unusual these days, with most motherboards having multiple USB hubs.


Bear in mind - that sort of stuff is written by a marketdroid, not a techie. Even the techies working on a keyboard are unlikely to know much about USB, and even less about OS drivers, since they tend to just buy in pre-programmed chips from Holtek or whoever.


a) The Windows driver issue does not affect my converter - NKRO avoids the problem.
b) It's fixed in Windows 7 anyway.


No... I am Helvetica!

It is totally not fixed in Windows 7. is this a recent patch?

[Soarer]

http://geekhack.org/showthread.php?18674-Filco-Stuck-Modifiers-and-Keys <== this issue is related to that issue fixed here?

It is totally not fixed in Windows 7. is this a recent patch?

Oh wait, this is a different issue then. Thought you were talking about the 6-keys-plus-modifiers bug in Windows' USB drivers.

Any PS/2 keyboard has a queue of output events - maybe only 16 bytes or so. It can only send one byte a millisecond, roughly. Since each key release results in at least two bytes, it doesn't take much to fill the queue. And it's much easier to simultaneously release many keys at once, than press them at once (unless you use something flat to help you).

Some keyboards might have bigger queues than others, but the fundamental problem remains.

There's nothing any converter can do to fix it.

[laffindude]

Gotcha. I don't know what issue TP4 is talking about though, since his BW triggers whatever it is.

[tp4tissue]

Gotcha. I don't know what issue TP4 is talking about though, since his BW triggers whatever it is.

I hate the black widow with a passion. Damn that keyboard suck so hard,,, thank god for amazon return policy.. K60 and k90 were downers too.

[litster]

How do you encounter it on the Blackwidow when it is 2kro?

Gotcha. I don't know what issue TP4 is talking about though, since his BW triggers whatever it is.

I hate the black widow with a passion. Damn that keyboard suck so hard,,, thank god for amazon return policy.. K60 and k90 were downers too.

Your black widow hatred still doesn't explain how you experienced the problem with black widow, a 2KRO keyboard.

[Soarer]

I hate the black widow with a passion. Damn that keyboard suck so hard,,, thank god for amazon return policy.. K60 and k90 were downers too.

Nearly all keyboards are... 'barely sufficient', LOL. Of course, most of them are fine for typing, but beyond that there really aren't many (any?) brilliant ones.

Replacement controllers is the way to go for the ultimate board currently - Full-speed USB, 1000Hz polling, fast scanning, etc. is what you really want. You'd never touch PS/2 again, I assure you.

[laffindude]

Hate it as much as you want, but which bug did you actually trigger?

Edit: I really should get into the habit of replying immediately after opening a new tab. Ninjaed by 10 minutes is just embarrassing.

[Soarer]

And, tp4tissue, how about some comment on the myth-busting parts of my post?

[laffindude]

I wish Teensy would get a firmware/software combo that is as easy to reprogram as Aikon or the Chameleon.

[tp4tissue]

And, tp4tissue, how about some comment on the myth-busting parts of my post?

I'm having trouble finding specific literature on the i8042 chip,, but you did not do any mythbusting, you've merely added more fluff without any substantiation or sources.

It's like calling into question what the plus sign means, and what does 2 really mean, when asked 2+2,, you can do this indefinitly,

[Soarer]

I'm having trouble finding specific literature on the i8042 chip,, but you did not do any mythbusting, you've merely added more fluff without any substantiation or sources.

It's like calling into question what the plus sign means, and what does 2 really mean, when asked 2+2,, you can do this indefinitly,

Oh, you're funny. 2+2. Hahahaha.

Myth #1: The CPU polls the keyboard. PS/2 is better because it uses an interrupt.

False. It's the USB chip that polls the keyboard. Read about USB drivers.

Myth #2: Keyboards will get swamped by other devices on USB. PS/2 is better because it has no contention.

False. USB devices reserve bandwidth, and are guaranteed to get it, unless you've put too many devices onto the same controller. Just don't do that! Read the USB specs.

Where is your substantiation and your sources? Please don't quote any more marketing material... that really is 'fluff'.

[Soarer]

I'm having trouble finding specific literature on the i8042 chip

Try 8242BB: "Intel 8242BB Technical Reference Manual Version 1.0 August 15,1994". Since...

The Intel 8242BB is programmed with IBM's keyboard
controller firmware. The 8242BB provides an
off the shelf keyboard and auxiliary device controller
for AT, PS/2, EISA, and PCI architectures.

[limmy]

I saw custom PS/2 boards that has super fast repeat rate. Can a USB board do that? By repeat rate I mean number of key signals per second when a key is pressed down.

Here is a youtube video of 356CL (OTD project custom keyboard) testing different repeat rates.
[video=youtube;I7puVuMktzk]http://www.youtube.com/watch?v=I7puVuMktzk[/video]

I set repeat rate to the maximum on my windows keyboard settings, but I want faster rate.

[breakaway]

You can get a faster rate with a autohotkey script or go linux whichever suits you.

[bhtooefr]

USB is a far more elegant approach, here.

Rather than spam the computer with keystrokes, it just sends one make code, one break code. If the OS wants fast repeat, it can do fast repeat. If the OS wants slow repeat, it can do slow repeat. If the OS wants no repeat at all, the OS can do no repeat at all.