| • |
The IBM 5150's POST will go into MANUFACTURING TEST mode if it finds that no keyboard is attached and that the motherboard's keyboard data line ('KBD DATA') is at a TTL LOW state (or is grounded).
An example of how to do that is shown at here.
 |
| |
|
| |
|
| • |
In MANUFACTURING TEST mode, the POST's keyboard test (301 test) is skipped.
 |
| |
|
| |
|
| • |
In MANUFACTURING TEST mode, when the POST puts an error message on-screen, the speaker is beeped as well.
 |
| |
|
| |
|
| • |
In MANUFACTURING TEST mode, the POST gets channel #0 on the 8253 timer chip to trigger a "BLINK LED" operation. Earlier, channel 0 of the 8253 timer was set to output a square wave of about 18.2 Hz, so, at about 18.2 Hz, a subroutine ("BLINK_INT") is run that toggles the PB6 pin (bit 6 of port B) of the 8255 chip. Per here, the PB6 pin going LOW pulls the keyboard clock line ('KBD CLK') on the motherboard's keyboard DIN connector to LOW. The result is a square wave of about 9.1 Hz on the keyboard clock line on the motherboard's keyboard DIN connector.
So, presumably, for MANUFACTURING TEST mode, IBM plugged a custom device into the motherboard's keyboard DIN connector.
The device:
- Grounds the keyboard data line ('KBD DATA'), so as to put the POST into MANUFACTURING TEST mode; and
- Has a LED connected between ground and the keyboard clock line ('KBD CLK').
My homemade device, which includes the LED, is shown at here.
It has been observed that blinking of the LED will not happen if the POST declares a video card error (1 long beep then 2 short beeps when there is nothing plugged into the motherboard's keyboard DIN connector). If you are in that situation, set the two video switches on the motherboard to the EGA setting.
 |
| |
|
| |
|
| |
|
| • |
In MANUFACTURING TEST mode, if switch 1 on the motherboard's switch block SW1 is found to be ON (i.e. no floppy drive/s present), then restart the POST.
On the screen, you can tell when that restart happens: the blinking underline cursor disappears for a few seconds.
 |
