On Sat, 23 Nov 2002, Constant Brouerius van Nidek wrote:
> With my XT which is still in daily use for emailing, I have
> startup problems.
> At the memory count, the count stops at 512, requires a F1
> and enters then in the IBM personal computer basic, a
> start where I never got acquainted with.
I tend to agree with the suggestion that the power supply may be getting
weak. If you have an accurate voltmeter, check that the voltages between
the power connector at the board and the case of the power supply (Ground)
are accurate within 5 % of specification. If you have an oscilloscope,
you can use the single sweep trigger to verify that the combined AC plus
DC errors are still within that 5 % specification. If not, you can use
an AC voltmeter in series with a relatively large electrolytic capacitor
to get an idea how bad the AC ripple is.
Another simple test is to swap with a known good power supply. Make sure
that you have the right pins connected in the right order. The two black
wires at the ends of the two connectors should normally be together in the
middle. If you are careful, you can even use a 350 W AT power supply to
provide +/-5 and 12 VDC. (The excess wattage capability just isn't used.)
Wrong connections and short-circuits tend to be spectacular -
engineers with a ring on their little finger should avoid the
"red-hot ring" syndrome.
The big electrolytic capacitors in a power supply tend to lose capacitance
with age, by leaking or drying out. If that is the case, new ones of the
same rating (both working voltage and capacitance) and able to fit in the
same location (not too fat, say) are usually fairly cheap at a surplus
electronics supply store.
CAUTION! I once found about 350 VDC on a large capacitor that had been
removed from an industrial-sized power supply 3/4 days earlier. The
"Electret Effect" can cause electrolytic capacitors to recover part of
the voltage normally across them some time after they were disconnected
and short-circuited with a big resistor. If you aren't used to working
on switching power supplies, get someone who works around high voltages
(like a TV repairman) to to do it for you. I know at least four people
who have been "bitten" by high voltage on a big electrolytic capacitor.
Fortunately, none of them needed artificial respiration or CPR, as they
all made the mistake of working alone. Three were incapacitated for at
least 20 minutes, and one suffered occasional heart arrythmias for a
week or so. Using two hands (connects across the heart) and hurrying
are other "No-Nos".
> Up to know it sufficed to press the memory chips in the
> four available banks but this time I have no luck.
> Before I start working on the banks, trying to find
> replacements for the chips (if they still can be found new
> or in my supply of old chips) I wanted to know if somebody
> on the list remembers if the faulty chip can somehow be
> found.
There are several software memory tests available - some are even free to
download. In a pinch, you can even use DEBUG (available with DOS) to move
around blocks of "00"s and "FF"s to look for stuck bits. The only hitch
is that these require a working system - or at least one that the lower
part of memory works.
> According to my memory (also sometime faulty due to
> age ;-)) the faulty chip should be somewhere in the fourth
> bank. Is that right?
Seems like a good guess. Depending on how the 640 K is made up of RAM
chips - in any case, it is probably the row farthest from the CPU. Low,
or fluctuating voltage could cause random memory errors, or cause a
weak bit to err consistently, too.
> Whether that is in the rows 0 to 7 or
> in the row P (forgot what that one was called is something
> I cannot remember ever to have know.
The RAM on an XT may be made up of 640 KBytes (1 KByte = 1024 Bytes,
and each Byte has eight bits numbered from 0 to 7 plus a nineth Parity
bit which is used to check the other eight for errors.
If the memory chips were 64 K x one bit, that would be nine rows
(0 to 7 plus Parity) to hold any byte; by ten columns (10 x 64 K = 640 K).
If the memory chips were 256 K x 4 bits, that would take two rows plus
another row of 256 K x 1 bit (Parity) for any byte. Some XTs only had
two columns of 256 K x ? bits for a total of 512 KBytes to start with,
however, you could add an extra 128 K - but you had to reset some jumpers
accordingly. (Check your jumpers don't have a bad or incorrect
connection, too.)
The memory is addressed from 0000:0000 to F000:FFFF in Hexadecimal, which
counts from 0 to 9, then A to F - so A is ten, F is fifteen, 10 is sixteen
in decimal. FF is 255 decimal, 100 is 256 decimal, 400 is 1,024 decimal
(or 1 KByte), and FFFF is 65,535 decimal (64 KByte).
The first four digits (before the colon) are the segment, which is shifted
and added, so that 8000:5D03 and 78AB:D253
8000 78AB
5D03 D253
----- -----
85D03 85D03
are two ways of addressing the same byte, number 548,099 decimal (counting
from 0, so it is the 548,100th).
Boyd Ramsay
[EMAIL PROTECTED]
To unsubscribe from SURVPC send a message to [EMAIL PROTECTED] with
unsubscribe SURVPC in the body of the message.
Also, trim this footer from any quoted replies.
More info can be found at;
http://www.softcon.com/archives/SURVPC.html
