Re: pendant
Posted by
Ray Henry
on 1999-09-05 08:09:15 UTC
List
I'm doing this from memory and will look up the details asap -- I'm also
posting before coffee. I really hope that you will be able to figure out
work arounds for all of the problems that I encountered in working on a
keyboard emulator approach to a machine control operator interface.
The connector will be one of the following:
5pin 180 deg DIN 6pin MINIDIN
---------------- ------------
_
(2) 1 CLOCK (5) (6) 1 DATA
(5) (4) 2 DATA | 2
(3) (1) 3 (3) (4) 3 GND
4 GND 4 +5V
5 5V (1) (2) 5 CLOCK
6
The signals are open drain and TTL compatable.
This operator interface idea using keyboard signals got my interest a
couple months ago. I ripped apart a cheap keyboard and mapped all the
signals onto the chip. Then I began to discover the real problems inherent
with it.
The keyboard port is bi-directional.
Each key press sends a signal. (eleven bit serial)
Each key release sends another signal. (eleven bit serial)
The shift, alt, control, and windows keys work the same.
So to send a "<" signal you have four 11 bit serial codes.
[shift] - [<] - [release <] - [release shift]
To make matters worse the codes are sorta asc-2 but not really cause asc-2
runs out at 64 characters and runs from most sig bit to least sig bit. Kbd
code contains a parity bit for checking but runs from least sig to most sig.
I even went so far as to wire up and program a stamp to send a complete set
of codes each time a switch was closed. That way I wouldn't have to press
shift, control, or alt and a switch. I also had it decoding quadrature and
sending the + or - to increment axis position for a handwheel. But it was
deathly slow and erratic. And two buttons pressed at the same time made a
real mess.
At this point I'd prefer an I/O system that is more reliable than the mess
I cooked up on my workbench -- perhaps one that used something like the NML
message system that NIST has developed.
If you want, I can dig out the stuff I collected.
Ray
-----OM-----
Date: Sun, 05 Sep 1999 09:13:54 GMT
From: john@... (John Stevenson)
I'm doing this from memory and will look up the details asap -- I'm also
posting before coffee. I really hope that you will be able to figure out
work arounds for all of the problems that I encountered in working on a
keyboard emulator approach to a machine control operator interface.
The connector will be one of the following:
5pin 180 deg DIN 6pin MINIDIN
---------------- ------------
_
(2) 1 CLOCK (5) (6) 1 DATA
(5) (4) 2 DATA | 2
(3) (1) 3 (3) (4) 3 GND
4 GND 4 +5V
5 5V (1) (2) 5 CLOCK
6
The signals are open drain and TTL compatable.
This operator interface idea using keyboard signals got my interest a
couple months ago. I ripped apart a cheap keyboard and mapped all the
signals onto the chip. Then I began to discover the real problems inherent
with it.
The keyboard port is bi-directional.
Each key press sends a signal. (eleven bit serial)
Each key release sends another signal. (eleven bit serial)
The shift, alt, control, and windows keys work the same.
So to send a "<" signal you have four 11 bit serial codes.
[shift] - [<] - [release <] - [release shift]
To make matters worse the codes are sorta asc-2 but not really cause asc-2
runs out at 64 characters and runs from most sig bit to least sig bit. Kbd
code contains a parity bit for checking but runs from least sig to most sig.
I even went so far as to wire up and program a stamp to send a complete set
of codes each time a switch was closed. That way I wouldn't have to press
shift, control, or alt and a switch. I also had it decoding quadrature and
sending the + or - to increment axis position for a handwheel. But it was
deathly slow and erratic. And two buttons pressed at the same time made a
real mess.
At this point I'd prefer an I/O system that is more reliable than the mess
I cooked up on my workbench -- perhaps one that used something like the NML
message system that NIST has developed.
If you want, I can dig out the stuff I collected.
Ray
-----OM-----
Date: Sun, 05 Sep 1999 09:13:54 GMT
From: john@... (John Stevenson)
>Subject: Pendanton my
>
>On 4 Sep 1999 10:16:49 -0000, you wrote:
>
>>I have started a web page on the pendent project as conceived by Mo. It has
>>a circuit diagram, brief explanation, and picture of the project to date.
>>You can check it out on my web site http://www.ktmarketing.com go to the
>>Home Machining page and select the link titled "Info and pictures of the
>>manual pendent I am building". It is the last one in the top group.
>>
>>BTW, the same idea will work with any brand of PC based controller software.
>>The only change is to what keys control what function.
>
>
>Thanks for the info above. This is high [ read within a couple of weeks ]
>to do list.lead
>I was actually going to go that way and have already perloined a card and
>from an old keyboard.change over
>I hadn't thought about a change over switch though and that has given me some
>ideas. Also I didn't realise that you only need to change over 2 wires.
>What I propose to do is build an interface junction box as in the drawing and
>fit this on the side of the machine, next to the monitor but fit the
>switch in this box not the pendant. I will then connect the switch to aspring
>loaded hook arrangement so that when the pendant is on the hook the keyboardthe
>works, when you take it off the hook to use it the switch changes over and
>pendant is activeALT,
>I run a Beaver CNC with the AHHA setup and a pendant is very usefull when
>setting up.
>I am still working on the layout of keys to suit as the AHHA makes use of
>CTL, and Shift a lot.rate in
>Ahha also use the plus and minus keys on the numerical pad to alter feed
>a move.
>
>Info time now. On the board I have there are 5 wires
>Black goes to ground on the board and the other 4 go to a socket.
>I have tried tracing them but they go to either small surface components or a
>small IC and not straight to the main chip.
>They are coloured Yellow, White, Red, Black any clues which they are ?
>This is a PS2 keyboard and I will be fitting it into a 5 pin din set up.
>Will this fit ??
>--
>Regards,
>John Stevenson
>Nottingham, England