PDA

View Full Version : How can I diagnose the keyboard port of a XT Clone board?



nestor
April 20th, 2012, 11:24 AM
I have a bondwell BW38 motherboard that always shows a "KEYBOARD ERROR. PRESS F1" message in bios, no matter if the keyboard is plugged or not. I can see a 74LS322 next to the 8255 so I assume it uses an XT keyboard. How can I find where is the damaged component?

Stone
April 20th, 2012, 12:03 PM
Have you tried another keyboard? Have you wiggled the keyboard connector? Have you used this keyboard in another machine? Have you tried an AT keyboard? If you have answered yes to all these questions, you're screwed. :-) Or at least the board is.

Chuck(G)
April 20th, 2012, 01:33 PM
Very very often, the 5-pin DIN connector on the board comes loose compromising the soldered connections or worse, a trace is broken or lifted.

But basically, all of the XT motherboards use nearly identical keyboard circuits.

nestor
April 20th, 2012, 03:05 PM
Have you tried another keyboard? Have you wiggled the keyboard connector? Have you used this keyboard in another machine? Have you tried an AT keyboard? If you have answered yes to all these questions, you're screwed. :-) Or at least the board is.

Yes, the board is screwed :( I tried with a keyboard and with the AT2XTKBD adapter that I use with other XT machines, but nothing, the board doesn't receive key presses. I installed the generic XT bios from Plasma and managed to boot from a floppy disk, launching in autoexec.bat Prince of Persia in demo mode for 30 minutes, so the board seems to be working.
This board doesn't have a DIN5 conector, only a 5 pin header. GND and VCC are easy to discover, DATA is tied to 74LS322, CLOCK comes from a 74LS125 and /RESET is the last one.

Any idea before unsolder ICs one by one?

slincolne
April 20th, 2012, 03:14 PM
The first thing I would suggest is that you get your multimeter out and check to ensure that the keyboard socket is providing a +5VDC supply to the keyboard (if you can make a pass-through adapter see if it's still 5VDC with the keyboard plugged in).

If the keyboard has power, I'd suggest then getting a logic probe (or oscilloscope) and check the keyboard clock and data signals to see if the bios boot diagnostics are working. When the XT boots it sends a command to the keyboard, and if it does not get the correct response it puts up the 'F1 Error' message.

You should see a series of short pulses on the clock and data lines, and they should be 'nice' 5VDC TTL signals.

Looking at the schematics for the IBM XT Keyboard (Sheet 9 of 10) there are some capacitors on the lines for signal conditioning. It's a good bet that one or more may have failed and shorted out.

You could also just try pulling the 'LS322 in case it has failed.

Have fun :-)

Sean

Floppies_only
April 22nd, 2012, 05:43 AM
The first thing I would suggest is that you get your multimeter out and check to ensure that the keyboard socket is providing a +5VDC supply to the keyboard (if you can make a pass-through adapter see if it's still 5VDC with the keyboard plugged in).

Sean

Sean et. al. Multimeters put up with a lot of abuse but I was taught that the way to measure voltage was to put the red positive lead on the positive supply of voltage and the black negative lead on the ground or negative voltage line. A voltmeter is a high-resistance circuit that shunts a little bit of current away from the device under test. To measure current, you have to have all the current go through the meter, unless it is an AC clamp on style meter that measures the magnetic field surrounding the conductor.

I might be a good idea to tap the ends of the test leads onto a grounded piece of metal before touching the circuit under test, to dissipate any static charge it might have picked up.

Sean (also)

slincolne
April 24th, 2012, 01:57 AM
Thanks floppies_only - but I've over 35 years experience with electronics so I think I can stumble my way through just fine :-)

BTW - for the purposes of this thread, the resistance of the multimeter is insignificant. Your 'bargain basement' multimeter will typically have a resistance of 1 meg - resulting in a parasitic draw of 5 micro-amps assuming a 5VDC supply. A better meter will have a higher resistance. For the purposes of testing a keyboard to see if there is power being delivered (i.e. testing for a broken socket) or if there is a short circuit, or if the voltage is significantly different from the 5VDC (i.e. below the LS322 Vmin of 4.5 or above Vmax of 5.5) such an infinitesimal load won't affect any decision (and probably is beyond the precision of the meter).