On the MFM Emulator hardware: How annoying is it to assemble one of them? It's time consuming to do it right, and I would say that a moderate level of soldering is desired. However, I believe that David even will sell a fully-build version now, sothat may be your best option.
I see there's been a good deal of subsequent technical discussion between you and DoN, so I hope you get that to work. It will take me more time to review all that has been said, and provide any meaningful comment, but just by skimming it, it allappears to be good progress. Getting your original keyboard to work is by far the most ideal solution, in my opinion.
As far as the keyboard manufacturer, when I tore one of mine apart, I see that it was built by CORTRON, but you may have seen this already while inspecting your keyboard. See a picture at http://bit.ly/1NDkg32 (hosted on Google Drive, sorry, DoN)
Given the world's shortage of these proprietary keyboards and mice for this machine, I still would like to pursue the "building the adapter" idea at some point. It wouldn't be as elegant, but it could at least be functional, and allow many differentoptions for hobbyists like us who like these machines.
On Wednesday, December 23, 2015 at 7:31:20 PM UTC-6, Convergent MightyFrame wrote:
Well, the good news is that I can receive signals from the keyboard
and mouse with a standard TTL serial hookup. I can power them with a 5v supply, and nothing explodes. This is good, but I haven't figured out
what they're saying yet. :) No work on it over the last week or so
either.
As far as the keyboard manufacturer, when I tore one of mine apart, I
see that it was built by CORTRON, but you may have seen this already
while inspecting your keyboard. See a picture at http://bit.ly/1NDkg32
(hosted on Google Drive, sorry, DoN)
Yep. I haven't seen any indication that these things are available
elsewhere either, although I have a slight suspicion that some 3b2
keyboards may have been very similarly designed -- perhaps to the point
of being internally compatible with a different plug.
On 2015-12-23, Chris Smith <protheus@byteorder.net> wrote:
could find out. There are two ICs of interest here. First, there's a Signetics 8450. I have no idea about this and Google doesn't know
either, but I wonder whether it's a slightly modified 2650. In that
case, we've got a general purpose CPU there. It's in a huge 40 pin DIP package, so that could certainly be the case. I can't really find
anything connected to the keyboard plug that far in -- at least not on
the pins I'd expect if this were pin-compatible with the 2650 -- but there's a good chance I wouldn't find it by then, so...
It may be a keyboard encoder chip -- which does thinks like
drive signals to the horizontal rows of keys and look for which vertical
row sees the pulse. Then, inside, it is programmed to generate a
specific code for each unique keypress.
Worked fine until it was quoted. AT least we are both using
fixed pitch fonts for our newsreaders. One trick to help avoid problems
is to never use the TAB key, but instead enter just the number of spaces needed.
The shield connects to chassis ground, which is a safety trace
around the board edge, and is grounded to the metal chassis by all the
screws which mount the board (except perhaps ones in the middle).
Note that the pins may be high for zero, so you ground them to
send a signal. (If there is a bar over the signal name, it is zero when high, and true when low.)
IIRC, the keyboard cable plugs into the connector to the right
of the keyboard shelf on the computer, and to the connector on the left
of the keyboard, so the coiled cable is hidden under the keyboard when
it is at rest. This leaves the right-hand connector on the keyboard for
the mouse.
On Tuesday, December 22, 2015 at 9:42:06 PM UTC-6, DoN. Nichols wrote:
Anyway -- the MC6850 means that the longest data string would be
one start bit, eight data bits, and two stop bits.
Which is interestingly close to what the PC/AT or PS/2 keyboard does
anyway, now that you mention it.
The signals which seem to be connected externally -- based on the
schematic, are:
KBRST (from the RST pin on the CPU -- probably used to reset the peripheral at boot.)
KBTXD (from the TXD line on the 6850)
KBRXD (to the RXD line on the 6850)
O.K. Reasonable.
There's a signal called 19.2Kbd going into the send and receive clocks
of the 6850, so it's probably a good guess that the peripherals operate
at 19.2kbps. Page 260 also seems to indicate that VCC is +5v.
The 19.2k clock might be also sent to the keyboard so it does
not have to be generated on the keyboard.
I'm not sure how this works out at the keyboard port and a quick
glance at the manual doesn't seem to show this. I suspect the keyboard
will need Vcc and ground signals on top of everything else.
Yes -- and maybe the clock, too. (Are there five pins on the
connector?)
So perhaps
Vcc/Gnd/RX/TX/not connected -- in some order or another.
I would expect the VCC and Gnd to be two ends of the connector
row of pins. Just a matter of finding out which is which.
And I remember once finding where the connector is shown in the
schematics. But they are upstairs, and I am downstairs, nad it is
getting close to time for my P.T. exercises, so I'll leave it to you to
find that.
I may take another look later for the connector information. I went
ahead and hit the keyboard with a continuity tester just to see what I
could find out. There are two ICs of interest here. First, there's a Signetics 8450. I have no idea about this and Google doesn't know
either, but I wonder whether it's a slightly modified 2650. In that
case, we've got a general purpose CPU there. It's in a huge 40 pin DIP package, so that could certainly be the case. I can't really find
anything connected to the keyboard plug that far in -- at least not on
the pins I'd expect if this were pin-compatible with the 2650 -- but
there's a good chance I wouldn't find it by then, so...
There's also an SN7442AN, which is a pretty common BCD to decimal
decoder, with a known pinout. Some of the pins here are continuous to
the port on the side of the keyboard.
When you look at the back of the keyboard, the connectors look --
excuse the bad ASCII art --like this:
__________
| ------- | <- Cable shield contact.
1| . . . . |4
5| . . . . |8
| _ |
-
More wires than I remember. If you count the shielding plate, you
have 9, but it doesn't seem to really use nearly that. Both pins on the
left side are continuous with Vcc on the decoder chip. All the rest of
the pins on the top are continuous with ground on the decoder chip. I
don't believe the cable shield _was_ continuous with ground on the chip,
but it connects to a large trace that runs the perimeter of the board,
so it goes somewhere.
What we have then is something like:
__________
| ------- | <- Cable shield contact.
1| + - - - |4
5| + x y z |8
| _ |
-
Just enough space left over to hold rx, tx, and rst. If I'm right,
rst is what will make the keyboard flash its caps lock LED, so I may be
able to locate that one by simply powering it at 5v and applying a
signal to each pin in turn. The logic level should pretty much be
equivalent to vcc anyway, so there's not really much chance of breaking anything that way. That would just leave me with two options for a
complete pinout.
mean that the keyboard would need to handle the mouse as well and
somehow send the data through.
Yes -- it does. I think that takes us up to six pins on the
keyboard connector. One of the data runs allows the computer to command
the state of case lock and such, IIRC.
Note that the mouse plugs into the keyboard, and is routed
through to the computer.
Right, that. Given everything else that's going on here, I wonder
whether the mouse isn't just another set of simple switches and some
pulse encoders, more or less, that are handled by the keyboard directly.
This seems feasible, but I'm not sure how that would work, since I
thought that the mouse could be attached to either side of the keyboard, which would mean the Vcc and ground would be fixed in both spots, and
there would only be three other pins available. Not quite as many as
you want. If I'm wrong though, and the mouse port is different than the
one I was just poking at, we have more or less exactly the number of
pins we need. 1 signal each for five switches or encoders (x,y,1,2,3),
Vcc, ground, and that would leave us with one left over.
The mouse likely generated the same codes as the cursor motion
keys, plus three for the buttons. Since there are 256 possible codes
which could be sent by the MC6850, there should be plenty -- especially
since it does not have as many special keys as the Sun does. :-)
Now that you mention it, that seems a reasonable assumption to start with, anyway.
On 2016-01-05, Chris Smith <protheus@byteorder.net> wrote:
The 3B2 *may* have actually been made by AT&T -- or some other
contractor, and there is not much chance that the keyboards would be >interchangeable.
On 2016-01-05, Chris Smith <protheus@byteorder.net> wrote:
On Wednesday, December 23, 2015 at 7:31:20 PM UTC-6, Convergent MightyFrame wrote:
[ ... ]
Well, the good news is that I can receive signals from the keyboard
and mouse with a standard TTL serial hookup. I can power them with a 5v supply, and nothing explodes. This is good, but I haven't figured out
what they're saying yet. :) No work on it over the last week or so
either.
No oscilloscope to watch it with? You need to set it up for single-shot triggered by the beginning of the output pulses.
For the keyboard problem, I think a clever thing to do would be to get a $5 "orange PI", plug a USB keyboard into that, and then have it drive the 3b1 connector.
I'm hoping to use a serial port out of the BeagleBone on the MFM drive emulator to plug into the serial port on the back of the machine -- then I can ssh to the 'bone, and talk to the 3b1 as a serial TTY over the network rather than staring at itphysically. The display was never a strong point for the machine, which I blame on the non-square pixels.
In article <slrnn8ooo2.g19.BPdnicholsBP@Katana.d-and-d.com>,
DoN. Nichols <BPdnicholsBP@d-and-d.com> wrote:
On 2016-01-05, Chris Smith <protheus@byteorder.net> wrote:
The 3B2 *may* have actually been made by AT&T -- or some other >>contractor, and there is not much chance that the keyboards would be >>interchangeable.
The 3B2 was made by AT&T. It definitely had a Western Electric
processor in it. It was one of the first true 32 bit microprocessor
systems.
But I agree that it's doubtful that keyboards for it would work
in a 3B1 / 7300.
Not much news on this recently, but I did finally manage to borrow the other keyboard so that I could attempt to recover the machine. Reformatted the drive -- had to adjust the geometry slightly, but it seems to have formatted and tested ok.Installed the OS again, which went reasonably well except that my dev kit seems to be completely messed up. Still, dd is working and while this may not recover the floppies for me, it may allow me to download images from another dev kit and install them.
Chris
I have no idea what people pay for that kind of thing myself, (I can tell you that people seem to pay about the same for the floppies themselves as they always have... not that I don't have a few laying around, but I've been looking for clean ones tomake new 3b1 media recently) but if my experience with the disk is any indication you might actually be able to press the thing back into service, depending. I had forgotten how resilient those old drives actually are.
Chris
Based on some more work, I've decided the following:
I may have not had my power wired in properly and the data I've had so
far may have been noise. I'll give this another shot in a bit and maybe
get more interesting data.
Now that the machine is running, I can tell you that the port does
generate 5v where I expect it to, so that's at least something.
The Burroughs/Unisys keyboard, once open, starts looking a _whole_ lot
like the AT&T one. Based on continuity tests at the plugs (which are
also 8 pins) I find that three of them are connected together and two of
the others are connected together, leaving three other lines. It would
make a whole lot of sense for these to be negative, +5v, and (rx, tx,
rst) respectively, exactly as in the other keyboard.
I'm kind of
wondering how difficult it would be to just wire it in and try it.
Perhaps not much.
Interestingly, when the bad keyboard is plugged into the machine with
a mouse on the other side, the mouse works fine. This suggests to me
that there must be some logic in there that's working properly. Perhaps whatever runs the keyboard matrix is the only thing that's not
functional.
I'll poke around a bit at it next week hopefully. This is true of the
Unix PC keyboard. I can't plug the other one in since I don't currently have the computer for it. :)
O.K. I've gone to _The Reference Manual_, and found that the
serial chip for the keyboard is on sheet 14 of the system board
schematics. It gives the following pin functions:
KBDTXD --> JK-7
GROUND --> JK-1,2,3
+5V --> JK-4,8
KBDRST --> JK-5
KBDRD <-- JK-6
Chassis Ground to JK-shield
with "JK" being the jack on the system board into which the keyboard
plugs.
At a guess it is something like this:
5 6 7 8
: : : :
1 2 3 4
Making the GROUNDs a row along one side (bottom as I show it here, but
check which three are shorts together -- perhaps examine the underside
of the keyboard PC board under the connectors) and the +5V is the two together at the end past the GROUND pins. That would make the KBDRST
the one adjacent to the first ground pin, the KBDRD the middle of the
three, and the KBDTXD adjacent to one of the +5V pins.
That's the plan. I believe the connector is something called sdl.
The old pc keyboard used a six pin connector to hook the cable up to the keyboard. This one uses an 8 pin one. It's something like a modular
plug with contact edges along the top, but also a shielded shell. It's flatter and wider than an RJ, with catches on the sides instead of on
the bottom. I'm not absolutely sure of this but I can order a couple of the connectors and try it. Not yet sure how I'll get them crimped on to
a cable, though. I don't have the original cable that went to the computer, but in pictures they look D-shaped on the system end.
So -- this is a female connector on the keyboard itself, instead
of on a cable.
I'm thinking that the transmit bursts are short enough (high
baud rate) so the chances of a collision with the mouse signals is
minimal. Perhaps the mouse signal is sent through a gate which is
turned off when the keyboard is about to transmit a keystroke. (Likely turned off just long enough before so a mouse signal could complete or
time out before the keyboard character or characters were sent.)
Anyway -- you should expect the keyboard end of the 3B1 cable to
be the same as the computer end -- no crossing of pins between the ends.
On Friday, March 4, 2016 at 9:17:12 PM UTC-6, DoN. Nichols wrote:
On 2016-03-05, Chris Smith <protheus@byteorder.net> wrote:
The Burroughs/Unisys keyboard, once open, starts looking a _whole_ lot
like the AT&T one. Based on continuity tests at the plugs (which are
also 8 pins) I find that three of them are connected together and two of >> > the others are connected together, leaving three other lines. It would
make a whole lot of sense for these to be negative, +5v, and (rx, tx,
rst) respectively, exactly as in the other keyboard.
Agreed -- with likely "negative" ('ground') as the three wires,
and the +5V being the two together. But verify this with the original
keyboard plugged into the computer -- don't go by what I guess is
correct. :-)
I'll poke around a bit at it next week hopefully. This is true of the
Unix PC keyboard. I can't plug the other one in since I don't currently
have the computer for it. :)
wondering how difficult it would be to just wire it in and try it.
Perhaps not much.
What are the connectors like on the new keyboard? If something
other than the one for the 3B1, I would suggest making an adaptor rather
than modifying the cable. This could make it easier to swap wires
around at need.)
That's the plan. I believe the connector is something called sdl.
The old pc keyboard used a six pin connector to hook the cable up to the keyboard. This one uses an 8 pin one. It's something like a modular
plug with contact edges along the top, but also a shielded shell. It's flatter and wider than an RJ, with catches on the sides instead of on
the bottom. I'm not absolutely sure of this but I can order a couple of
the connectors and try it. Not yet sure how I'll get them crimped on to
a cable, though. I don't have the original cable that went to the
computer, but in pictures they look D-shaped on the system end.
Interestingly, when the bad keyboard is plugged into the machine with
a mouse on the other side, the mouse works fine. This suggests to me
that there must be some logic in there that's working properly. Perhaps >> > whatever runs the keyboard matrix is the only thing that's not
functional.
I *think* that 5V, ground, and the txd are simply routed through
the keyboard. Maybe the reset, too. I've never tried holding down a
key for repeats while moving the mouse to see whether they interfere
with each other.
How would that work if there's only one data line going out of the
keyboard? I mean, wouldn't there be some multiplexing going on in
there, or am I missing something? Or maybe you mean tx from the
computer into the keyboard...? That would make some sense.
On 2016-03-05, Chris Smith <protheus@byteorder.net> wrote:
The Burroughs/Unisys keyboard, once open, starts looking a _whole_ lot
like the AT&T one. Based on continuity tests at the plugs (which are
also 8 pins) I find that three of them are connected together and two of the others are connected together, leaving three other lines. It would make a whole lot of sense for these to be negative, +5v, and (rx, tx,
rst) respectively, exactly as in the other keyboard.
Agreed -- with likely "negative" ('ground') as the three wires,
and the +5V being the two together. But verify this with the original keyboard plugged into the computer -- don't go by what I guess is
correct. :-)
wondering how difficult it would be to just wire it in and try it.
Perhaps not much.
What are the connectors like on the new keyboard? If something
other than the one for the 3B1, I would suggest making an adaptor rather
than modifying the cable. This could make it easier to swap wires
around at need.)
Interestingly, when the bad keyboard is plugged into the machine with
a mouse on the other side, the mouse works fine. This suggests to me
that there must be some logic in there that's working properly. Perhaps whatever runs the keyboard matrix is the only thing that's not
functional.
I *think* that 5V, ground, and the txd are simply routed through
the keyboard. Maybe the reset, too. I've never tried holding down a
key for repeats while moving the mouse to see whether they interfere
with each other.
KBDTXD --> JK-7
GROUND --> JK-1,2,3
+5V --> JK-4,8
KBDRST --> JK-5
KBDRD <-- JK-6
Chris, or all:
Thank you for your contribution here!
KBDTXD --> JK-7 # keyboard transmit data (from kdb to comp)
GROUND --> JK-1,2,3 # Ground
+5V --> JK-4,8 # and Power
KBDRST --> JK-5 # Keyboard reset
KBDRD <-- JK-6 # Keyboard read data (from computer to keyboard)
Now, on an older-skool PS2 keyboard, there is a clock and a data pin.
I'm wondering if you might have some insight onto how these translate:
I'm guessing that KBDTXT is equivalent to the "Data" pin on the PS2.
But what about KBDRST and KBDRD?
Would either of those be equivalent
to the PS2's clock signal?
I'm just starting to dive into "mapping" what one of these keyboards
sends on each keypress, with the hope of emulating one of these for a
UNIX PC if one gets a UNIX PC without the more rare factory keyboard.
http://unixpc.blogspot.com/2017/03/at-unix-pc-keyboard-signal-decoding.html
As soon as I can get some experiments of my own going, I'll report
back here also.
| Sysop: | Keyop |
|---|---|
| Location: | Huddersfield, West Yorkshire, UK |
| Users: | 293 |
| Nodes: | 16 (2 / 14) |
| Uptime: | 210:40:35 |
| Calls: | 6,619 |
| Calls today: | 1 |
| Files: | 12,168 |
| Messages: | 5,317,248 |