ASCI u68 (SystemX)

Brent Hilpert hilpert at cs.ubc.ca
Sat Oct 17 00:33:12 CDT 2015


My guess is the RS232 is not involved in the initial-boot monitor, I wouldn't necessarily be expecting anything from it at reset.

At this point I'd be aiming to figure out the keypad/display operation. The firmware EPROM likely maps to the upper 2K of address space as the reset vector is at FFFE,FFFF.
EXAMining the contents of those two bytes (big-endian order) may produce an address in the range F8xx-FFxx, the entry point to the monitor.
If that pans out and the memory contents in that range look suitably random or differentiated then it would affirm the ROM is in that range.
Or poke around with EXAMine to try to locate the RAM, then figure out the command to load RAM, and EXAMine it to see if you can get back what you entered.

There was a bunch of writing on a cover plate which could be seen but not discerned in a photo, was that perhaps a command structure for the keypad?

If you can load RAM, you can enter and execute a program that modifies memory, but you'll need to know the memory map in more detail to do IO.
Could figure out the address mapping for the IO by experimentation and inference from the 18S22, some reverse engineering around the 18S22 would help.

Or there's dumping and disassembling the monitor EPROM which might turn up IO addresses, perhaps routines for character IO, maybe a console monitor or a serial-line download routine, etc.


On 2015-Oct-16, at 9:59 PM, Brad wrote:

> Thanks so much Brent!  Your insight on the 6821 proved to make this thing
> run way better!  Swapping 29 and 30, the system 'boots up' consistently and
> properly.  Before, you'd hit reset and then any key and it might come up or
> it might not, or it might generate weird messages and then go dark.  Now it
> comes up every time.
> 
> I connected the RS232 again and tried bauds between 110 and 9600 but no
> dice.  I'm assuming bits, parity etc are all standard.. hrmmm.  Curious now.
> Wish a manual still existed somewhere.
> 
> -----Original Message-----
> From: cctalk [mailto:cctalk-bounces at classiccmp.org] On Behalf Of Brent
> Hilpert
> Sent: Friday, October 16, 2015 9:46 PM
> To: General Discussion: On-Topic and Off-Topic Posts <cctalk at classiccmp.org>
> Subject: Re: ASCI u68 (SystemX)
> 
> On 2015-Oct-16, at 7:30 PM, Brad wrote:
> 
>> Here's some detail shots of the unit (including two of the front and 
>> back of the mainboard).  Note the 6821 in the middle that has 'BAD' 
>> scratched into it.
>> 
>> Front:
>> http://s1381.photobucket.com/user/unclefalter/media/20151016_191325_zp
>> scaec0dg8.jpg.html 
>> http://s1381.photobucket.com/user/unclefalter/media/20151016_191426_zp
>> smagz2dcc.jpg.html
>> 
>> Back:
>> http://s1381.photobucket.com/user/unclefalter/media/20151016_192343_zp
>> sdgtagrl1.jpg.html
>> 
>> Keypad:
>> http://s1381.photobucket.com/user/unclefalter/media/20151016_192440_zp
>> sioikuzdu.jpg.html
> 
> 
>> From what can be readily discerned, with some presumption, in functional not
> numeric order:
> 
> IC26 6875		clock generator
> 
> IC22 68B00		processor
> 	. . obviously
> 
> IC7 TBP18S22	PROM 256*8
> 	probably for address mapping of the RAM / ROM / IO.
> 
> IC19 2716		 EPROM 2K*8	 (or 2708 1K*8, note jumpers)
> 	firmware/monitor ROM
> 
> IC18 2716		 EPROM 2K*8	 (or 2708 1K*8, note jumpers)
> 	optional ROM
> 
> IC16 6514		SRAM 1024*4
> IC17 6514		SRAM 1024*4
> 	1 KBytes of RAM
> 
> IC30 6821		IO 2*8+4
> IC31 6821		IO 2*8+4
> 	IO to the keypad & LED display.
> 	It looks like the hex-to-7-seg decoding is done in software,
> 	over on the kbd/display PCB is a CA3082 7*NPN-tran, presumably
> 	used as LED drivers. One of the chips is masked by the photoflash.
> 
> IC29 6821		IO 2*8+4
> 	Goes to connector P1, just GPIO pins to play with.
> 
> IC15 6850		UART
> IC13 1489		RS232 receiver
> IC28 75150		line driver
> IC25 4702		bit-rate generator
> 	Comprise a serial line interface.
> 	Not clear how the bit-rate is set, I'm guessing the rate selection
> pins of the 4702
> 	are fed by IC14 74175 latch, fed in turn from IO pins on IC30, so
> the baud rate would be
> 	be under programmatic control.
> 
> IC27 555		timer
> 	Not clear what that's for, might be an oscillator/timer for
> kbd/display scanning,
> 	feeding into IC30 or 31.
> 
> The remaining chips are bus drivers and a little glue logic.
> 
> --
> 
> It's pretty straightforward for what it is.
> Date codes of 1981, but the support for 2708 and 2716 EPROMs suggest the
> design is a few years earlier.
> 
> As IC30, the bad 6821, is involved in the keyboard/display you might try
> pulling it and replacing it with IC29.
> If you want to be particular, pull it and check for bad voltages and current
> limits on the IO pins before plugging in a sub.
> 
> The expectation if it's running properly is that you'd press a key for a
> command followed by hex address or data, e.g. EXAMine, enter the address
> XXXX, displayed as entered, memory contents displayed on entering 4th
> address digit.
> 
> 
> -----
> No virus found in this message.
> Checked by AVG - www.avg.com
> Version: 2015.0.6172 / Virus Database: 4447/10805 - Release Date: 10/12/15



More information about the cctalk mailing list