Hi there!

Over the past few months I've had the opportunity to run the self-test on three different HP 28Ss.

Prior to displaying the "OK-28S" message, the units appear to run some sort of "screen test" (I'm probably not describing it properly, but hopefully you know what I mean).

The interesting thing is; each unit's screen test was different!! No two units showed the same designs during this test.

I would have expected that the self-test would run in exactly the same way on all 28Ss; can anyone explain the differences?

At first I thought that maybe this was an indication of different ROM versions; but two of the units were manufactured within weeks of one another!

Any insight would be greatly appreciated!
Thanks!
B.

Unlike the 42S, which has some built-in constants for this purpose,
it seems that the 28S uses random patterns for the display tests.

Raymond

Hi;

this is just a guess: as you are running RAM tests, have you cleared RAM contents prior to run them? (e.g. Memory Clear procedure) Maybe, just maybe, RAM contents are used to generate the display patterns (I had one HP28S, Brazilian made, recently stolen, and I bought an used, USA-made; did not have the chance to compare these patterns).

Cheers.

Actually, I have cleared the memory in all 3 units, and each unit's test is identical.

Although the test results vary from unit to unit, the test result on each unit is always the same, regardless of how many times I run it...I've tested it at different times of the day, just in case the clock had something to do with it. I've also removed the batteries and reinstalled them, and once again, they are still the same.

Although I'm not much of a computer expert, logic tells me that either (a)the ROMs are different or (b)there is some piece of data that the unit uses to randomize the test. Based upon my test results, I can only assume that (a) is correct.

Does anyone else have any other ideas?
B.

There is only one HP-28S ROM version: 2BB.

You can check it by keying: #10d SYSEVAL.

Regards.

J-F

If you think there are different versions of the 2BB ROM, check the CRC correction words of the two 64KB pages of the ROM. You can find the two CRC words (two CRC's per 64KB page) at #1FFF8-1FFFF and #3FFF8-#3FFFF. To read them you need a PEEK program.

But that isn't necessary, the garbage you see in display during the self test on the HP28S is a remainder of 32KB user RAM test. The used RAM test destroys the RAM content, so the old content must be saved first. The display RAM is a good place for this. The different display output during the self test is a function of the different content of the user RAM at testing time.

If you don't believe this, check the following code.

Regards,

Christoph

Disassembled HP28S RAM test code:

* RAM test

018DF   P=      0

018E1   LC(5)   #C0000			start of user RAM

018E8   D=C     A

018EA   GOSUB   [01941]			save #400 nibs of RAM into disp area

018EE   LC(5)   #00043

018F5   P=      4

018F7   GOSUB   [015D0]			user RAM test of area with pattern (destroys content)

018FB   GONC    [0191A]			Ram OK -> NC -> jump

* error condition

018FE   CD0EX

01901   B=C     A

01903   GOSUB   [01972]			restore #400 nibs of RAM

01907   C=B     A

01909   D0=C

0190C   LC(1)   4

0190F   ST=1    9

01912   GOSUB   +0008A                  # Address: 0199C

01916   GOTO    +00008                  # Address: 0191E

0191A   GOSUB   [01972]			restore #400 nibs of RAM

0191E   P=      0

01920   LC(5)   #00400			next page of RAM

01927   D=D+C   A

01929   LC(5)   #D0000			reached last address?

01930   ?D#C    A, GOYES [018EA]

01935   ?ST=1   9, RTNYES		return on error

0193A   LC(1)   4

...



* save #400 nibbles of RAM at D[A] to display

01941   C=D     A			D[A] addr. of RAM page

01943   D0=C

01946   D1=(5)  #FF840			start of display (slave)

0194D   LC(5)   #00220			no. of nibs to copy

01954   GOSBVL  04FF2			copy!

0195B   D1=(5)  #FFC00			start of display (master)

01962   LC(5)   #001E0			no. of nibs to copy (#220+#1E0=#400)

01969   GOSBVL  04FF2			copy!

01970   RTNCC

* restore #400 nibbles of RAM at C[A] from display

01974   D1=C

01977   D0=(5)  #FF840

0197E   LC(5)   #00220

01985   GOSBVL  04FF2

0198C   D0=(5)  #FFC00

01993   GOTO    -00031                  # Address: 01962

01962   LC(5)   #001E0

01969   GOSBVL  04FF2

01970   RTNCC

* copy routine

04FF2   C=C-1   A

04FF4   GOC     +0004A                  # Address: 0503E

04FF7   P=C     0

04FFB   CSR     A

04FFD   C=C-1   A

04FFF   GOC     +00025                  # Address: 05024

05002   A=DAT0  W

05006   DAT1=A  W

0500A   D0=D0+  16

0500D   D1=D1+  16

05010   C=C-1   B

05013   GONC    -00011                  # Address: 05002

05016   C=C-1   XS

05019   GONC    -00017                  # Address: 05002

0501C   C=C+1   X

0501F   C=C-1   A

05021   GONC    -0001F                  # Address: 05002

05024   A=DAT0  WP

05028   DAT1=A  WP

0502C   CD0EX

0502F   C+P+1

05032   D0=C

05035   CD1EX

05038   C+P+1

0503B   D1=C

0503E   P=      0

05040   RTNCC