Iron - Description of a block of memory from the printer Robotron CM 6329.01 M. Part 2.

<b>Iron</b> - Description of a block of memory from the printer Robotron CM 6329.01 M. Part 2.
  About printer malfunction

   The printer does not respond to external data (excluding 
normal rarely missteps by pressing its buttons) and produces 3 
"banning" of the signal: passive signal of choice - not ready 
printer to work, nevybor printer, or even its absence 
(obtaining this signal, the modern computer displays a message 
that there is no printer), an active busy signal - not ready to 
accept the information (automatically generated when nevybore, 
so everything here logical), the active error signal - or open 
the paper (if the sensor unlocked and the paper was 
previously), or taking an illegal instruction (code ESC), or 
mechanical failure - jammed feeder head (it is equipped with a 
stroboscopic detector). 

   Interestingly, when foreign "not ready" and "error" the 
printer itself after the test and goes into standby mode 
"ONLINE", without giving an error signal. It turns out that 
CENTRONICS or produces the wrong data or wrong programmed by

printer from the printer to malfunction. Of course, both events
not unlikely, but there are other interesting facts.

   Speccy often uses a simplified protocol in parallel
exchange for the LPT: besides the common wire (and its the same 
screen as if it a), of course, are 8-bit data bus, and only 2 
of Governors Signal: STROBE to the printer, to send a byte of 
data, and BUSY from printer to record his employment, to 
suspend or stop printing. In practice, BUSY can not be used if

data transfer is relatively slow (eg BASIC), then the printer 
will have time to unload your clipboard. According to 
statistics 90% of the standard dot matrix printers with LPT 
port, excellent work on the simplified scheme (a 80% slow, 
without BUSY), and of the remaining 10% of the additional 2% 
can be made to work without a change in the scheme, trickier 
software algorithms, although in this case there inappropriate 
moments (the program can start typing, even if printer not 
included), and must observe certain rules of with the printer. 
In fact, nearly all of the standard signals to LPT port Speccy 
is not even connected, and physically impossible to use. 

   Tales From the above it is clear that the test of the printer
computer system Speccy can get direct access to LPT by
simplified algorithm, since almost all the signals the port 
"are ignored by hardware, signal unavailability (at fault 
printer) are simply not visible and can be printed, not paying 
attention to him (or rather, try to transfer data as if the 
printer is selected and fully functional, in spite of its 
signals). Although we make of it a busy signal, we can also 
ignore it (This is our business as we respond to external 
signals), because employment in some (very rare) printers can 
be denoted testing of graphics mode and receiving the data 
block for print (Probably, and going on my printer DAEWOO).


   So I did - try typing not paying attention, because
the printer itself logically prepared to receive data after 
power - stood in the mode of "ONLINE". But nothing happened! 
For byte did not react, that inclines us to the fault 
interface, but I do not rule out hardware and software 
configuration, lock ... 

   Rummaged in the instructions did not find anything forbidden 
in the position jumpers, checked the lock on the sensor open 
lid and found the sensor - a reed switch, allow sound jumper, 
but the printer does not works bytes BELL, and I heard nothing. 

   Just in case, decided perevklyuchit button mode "ONLINE", and
when you hear the signal. Repeated several times - when off.
"ONLINE" heard the sound. My first thought: unlock bridge,
and the printer is always beeping at these pressings 
(programmed). 

   Again tried to pass code BELL, but the printer just in case 
molchal.Na again perenazhal "ONLINE" and to my surprise, the 
sound have not heard, but the printer with the paper advanced a 
little forward. 

   At first thought it was a glitch printer, because I do not 
sound cut off, but then I remembered that I gave last code 10, 
ie, translate stroki.Neskolko by pressing "ONLINE", I was 
convinced that it works off the printer, and transferring BELL, 
by pressing the same button I heard a sound. It turns out an 
interesting thing: The printer produces prohibiting job signals 
that he is ready to work on input information does not react, 
but when you click "ONLINE" (And when you turn off this option) 
fulfills the last byte received (and as it turned out, stamps, 
if the character). This suggests at least partially by the 
interface because data and Gate (specifically tested the 
fixation of the next byte strobe) pass through it. Perhaps it 
is programmed incorrectly, if some level or control signals do 
not come out of the printer? 

   To verify this was to transfer to the printer consistently
Bytes 32 ... 255. Below are control codes, and it makes no sense
them once more to touch, t.k.diapazonom 32 ... 63 5 LSBs, and so
will provereny.Nelegkaya it was a venture, because after each 
byte is transmitted it had to "see" the printer, which pressed 
button "ONLINE", and after printing the symbol should have been 
pressed again "ONLINE" and send the next byte. The printer is 
correctly printed. 

   Printed tables, I surprisingly discovered that she was 
essentially repeated, this suggests ignoring the 7-th (highest) 
bit, if it ever passed through interfeys.K addition, the 
passport symbol table was actually 7-bit. But perhaps this 
"Evil" phrase: the unused bits in the cliff and not affiliated 
with printerom.Togda can assume that the interface port is 
programmed incorrectly (7-bit can mean the necessary modes of 
operation) and he does not accept the data stream, but because 
of an erroneous mode transmits only the last (although this 
depends on the button which is not connected with the interface 
directly, the program queried separately, perhaps, the logic 
signals erroneously matches - but 3 matches highly unlikely). 
You can check this by using graphics mode. But in the statement 
was a warning about printers with 7 needles, it just drops the 
high bit - and do not check its passage (t.k.sisteme he did not 
need one). Were doubts about the inclusion of graphics mode. 
Normally the printer as it freezes, taking the entire set of 
image data, and then all at once their prints, but in our case, 
with such "violent" byte-castings, as if not bump into the 
system A printer error (such as: interruption of the flow of 
image data, output out of print mode on the button "ONLINE"). 
But all decided to experiment. 

   I passed the inclusion of graphics mode (bytes 27,75,11,0) 
with Listing 11 bytes of graphics on the printer, as usual, did 
not react. Then he began to pass themselves graphic bytes 
(decided pass "ladder" - a serial search of all bits, and full

line and at the beginning and end of an empty space, t.e.bayty 
0,1,2,4,8,16,32, 64,128,255,0). Printer to print them, as 
always, one by one, except first and last zeros, normally 
worked graphics mode and came out of him (I sent the code for 
the sound check). But on a ladder, because of the dried paint, 
the points were very vague, and although all 8 were visible, I 
decided to play safe and give a further 7 graphics bytes in the 
same mode: bytes 27,75,7,0,0,% 01010101.0, 10101010.0%, 255.0 
(symbol% denotes the entry in the binary form, then go% bits 
from older to younger). The result was that each needle printed 
by one, with an intermediate distance, and you print 4 needles 
from the bottom and then the other, counting from the top, and 
then all 8. Due to the distance between the print needles and 
shift bits of the press (and the distance between the positions 
of print) was clearly shows the imprint, and one could say - 
all 8 data bits caught in the printer, there is no break, and 
the interface to the data bus is not defective. 

   Pleased that the printer is able to print it all: the text 
and graphics, but how to do it, upset, because almost every 
byte (Except managers) simulated a few seconds, and print

would solve only a big fan or "tourists" out of interest
"Not just buttons ponazhimat. Of course, quite possible to 
gather a small shemku that after taking the byte "clicks" a 
button, but it should push it again only after working the 
current byte, and the time of processing depends on the byte, 
and from the print position, you can enter the standard delay, 
but 5 seconds per byte (If not more) - not too impressive speed.


   Disassemble the printer is not wanted (ROM is not excluded), 
and the interface checked three times - all connections are 
normal. It remains only to analyze and predpolagat.Itak, for 
example, the interface is faulty and gives false signals, and 
the printer is ready to print, but due to the failure not 
receiving data. Then there are the options: fault only the main 
chip interface (presumably, this is not only I / O port, but 
the shaper signal interruption); More and more defective chips 
(through which the signals pass to a computer). I must say at 
that moment I was not sure that the printer has a Z80, but the 
analogy signals clearly seen. 

   If you look at the circuit interface CENTRONICS, it is 
evident that through the auxiliary circuit is only an exchange 
of signals (Signals, taken from the main chip and past shallow 
logic back into the main chip is not returned). This means that 
the failure of simple chips will not change the logic of the 
byte general way, in addition the fact that any byte will be 
transmitted or not, and passed with an error either true 
(despite what the logic elements will be denied), t.e.printer 
software can not "know", "check" and "use" in their work that 
other interface circuits are faulty. He takes data from the 
main chip, and only it can control. A Here are the data 
received is not true (because of malfunction of the shallow 
logic), he sees, but decides for the correct data. 

   Now we shall understand these might be "incorrect" data (we 
assume that just because their system is not working as it 
should). There 4 signals and data bus. STROBE signal and SM are 
fully on the basis of published experiments (gate goes to only 
one leg of the main chip, and hence exactly comes inside - in 
the the main chip - it can not work properly, but this is not a 
malfunction "walk-through circuits"). The remaining 3 signals 
in question as my system in general are not issued and hang 
(default) inactive state. But all of these signals, whatever 
they are (at fault "walk-through" mikruh) in the main chip is 
not received, we can block the bridges and the printer will not 
respond to them, ie, the state has no meaning (simplified 
protocol). Well, with the output (the printer) signals much 
simpler: having passed "checkpoints" mikruhi, a printer, they 
do not come back, and he "still" that with them, but I just 
ignore their condition, giving data on printer as if it were 
fully functional. Hence the conclusion: even If all other 
Defective mikruhi interface and, therefore, give false signals, 
the printer must print normally received data t.k.shina data 
and strobe are working, and other signals Printer does not see, 
and ignores the other on the basis of jumpers. But once he did 
not prints, we can assume: faulty chip main interface

or the printer itself is defective.

   Now consider the alleged logic of the printer, as well as 
the reception and testing of bayt.Itak, turn on the printer. 
Interface port programmed, and part of its "external" inputs 
runs on input (Data bus and 3 control signals), as part of the 
withdrawal, except He is programmed to logic "capture" of bytes 
and the formation of Interrupt on any event - which actually is

reception of the next byte. This assumption is based on the 
signals arriving at interfeys.Analogichno is programming the 
internal printer ports, responsible for the display / speaker 
(O), buttons / jumpers (input), engines and needle (pin), but 
with its own logic of work (compared with the interface 
CENTRONICS). 

   As a rule, such systems often use interrupts.
Interrupt - this is a reaction to an external processor 
signal.Pri performing basic tasks the processor can receive an 
interrupt signal (There is a special entrance) and interrupt 
the main program (Certainly not in the middle of the team, and 
between them), go to the program interrupt handling, and after 
the return from it and continue to perform basic tasks. There 
are different modes and different interrupt how they 
obrabotki.Po special team interrupts can be disabled, and the 
processor will not pay attention to them (to distract from the 
main task), the other team can resolve them. In Depending on 
the mode, interrupt signals may lead to the implementation of 
various actions in the program. With them, perhaps a 
multi-(almost parallel) execution. It is only logical to use an 
interrupt in this printer. 

   The logic of action printer is approximately as follows: 
test the memory and mechanism (if that is defective, given the 
error signal and the system hangs) after checking the key, and 
if any of them pressed, it is system testing, to check on

keys to stop him, or programmable ports scanned key jumper, and 
given the display, as well as going working out the changes on 
the buttons / jumpers, t.e.protsessor turns to the core problem 
- the handling of its input interface (keyboard and jumpers). 
It is logical, because not known when exactly the user will 
transmit data for printing, and there is no sense in vain to 
check the interface to unknown data (including interference 
connection). When the interface comes bytes transferred (ie, 
not just received some new information, but also sampled -

that it is received), the port stores it in its register
(It is a fact based on the tests) and forms a signal 
(confirmation of the byte) of the transmitting device and 
activates busy signal (after all, bytes, even though he 
entered, but not yet processed). However, the port generates an 
interrupt processor. Processor (if not too busy with higher 
priority) drew attention to this fact and proceeds to interrupt 
processing routine, which first checks the jumper settings and 
system variables, and on this basis takes the byte, putting it 
in buffer, and can, once processing, and then removes the busy 
signal (to an external device can send the next byte) and out 
of the interrupt handler, returning to the main problem. The 
reliability of such actions is very high, at least, the fact of 
using interrupts. 

   Later, inspired by a relatively simple repair such
printer, I made out that, properly inspected, cleaned, alcohol 
and checked. Interestingly, when the outer dust Printer board 
was almost in perfect condition (it was a weak blue plaque next 
to the connector interface, blackened by the flying feet of one 
resistor, slightly oxidized one fuse, but still serviceable, 
and like all specialties). In general, the "repair" has given 
nothing, everything was in order, and the printer then does not 
become work better. 

   That's when I found 2 more similar to the port (as in the 
interface) and 2 other specialized chip, which, presumably, are 
programmable timers - breakers and can generate signals for DMA 
mode and multi-tasking interrupt for working press, including 
time slots for the mechanism. But to say exactly what these 
chips can not, t.k.nikakoy documentation to them I do not have, 
it is clear only that they use the above signals and is likely 
to have contact with the microprocessor Z80 (microprocessor in 
one package). But in our country as we know, Z80 secretly 
brought, on the other did not think (there were not needed), ie 
they are not well known. 

   The result: the scheme nalazilsya, natestilsya ... In fact, 
fully confirmed his theory of using interrupts. Almost all of 
the interface signals, with the same name, go to the Z80, and 
come (no breaks). Thus, the theory of fault next (possible fault

simple chip interface for the statistics will not exclude, but
Even in this case they are not alone to blame, and not primarily
Although almost certain that they work): Failure of the main 
chip Interface; fault similar chip printer chip fault of 
another printer other malfunction of the system as Total 
(detail, including small chips, memory unit, and also the 
processor - in short, no rule - a kind of TOTAL case), the 
fault memory (almost exclusively the loss of some data - a lot 
and do not need a little, but aptly ...). How to transfer is 
not a priority! You should also remember the past this printer, 
and exactly what he originally was not defined as working 
(though the test is always held) t.e.nelzya exclude failure 
from the outset - a marriage (I'll take it to the GENERAL 
case). But I would prefer it to be defective or faulty memory. 
Although it is easier: to resolder a small chip or reflash Rom? 
The first is easier, but if the chip is not small, then get one 
where it - Nothing is known about him, do not simulate - just 
me! 

   In favor of the main chip interface CENTRONICS says his
ability to capture and transmit data byte to the processor, and
as a reaction to most of the input signals from the printer 
(although not all of them can be checked). Consider these 
signals. 

   Addresses A0 and A1, input is the choice of port registers -
half of their options fully operable, because and allows 
programming port (albeit incorrectly), and read from it accept 
information, but there is a dark place in the options of access 
to registers port. A6 to the input CS; had he not been working 
with the Port nothing to read. The signal RD - it was he who 
reads the incoming bytes, without nikak.Signal IORQ responsible 
for access to the port (as opposed to memory) and duplicates 
the signal of CS (allowing you to control the choice of ports 
in different locations, without the CS would have to mix the 
two Signal shallow logic, and serve as a signal IORQ choice), 
and hence input rabotaet.Signal CP works too, t.k.blagodarya 
input clock frequency and is an autonomous capture of incoming 
bytes, the signal STROBE. 

   Remain the two dark horses: M1 and the IEO. M1 has (in this
case) to reset signal INT, the processor does not react to
it twice (the processor, receiving INT, did not immediately 
react, but INT mozho reset only when the react, or processor 
that INT miss), to confirm their response processor generates

combination IORQ and M1, which recognizes the port, dropping
INT). The idea is that this entry should be defective, because 
participates in After the formation of the interrupt, but his 
passive state allocated a combination of entries in the port 
(by process of elimination, WR t.k.signal the port is not 
supplied), and if the record does not work as expected, the 
port is either not zaprogrammiruesh, or you can do it wrong 
(both options are extremely improbable - although the fact : 
port is programmable, but the effect is not true), and another 
thing, which is not decoding the signal M1 (at fault chip port) 
can constantly reset INT so that the processor did not notice 
(too short INT). IEO signal comes, of course, if its level 
corresponds to a resolution granting termination, you may need 
not simple level, and the momentum is IEO in question, as well 
as similar chip in the printer, with which here comes IEO 
(perhaps wrongly). 

   Actually, by studying the circuit can be very significantly 
to declare: no printing, t.k.ego processor is not interrupted. 
This was confirmed by a eksperiment.Poskolku during the 
CENTRONICS interface is located inside the printer can not 
verify the acting in the signals. I dismantled the interface 
and brought two Posting: signals INT (comes with chips) and IEO 
(comes on a chip). During the tests, I gave the bytes to the 
printer and took it for these signals. It turned out that the 
transfer byte status signal is absolutely no change, but when 
you click "ONLINE" chip generates INT, and B worked through the 
printer (this can be seen in the weak change in signal voltage, 
short pulse t.k.zafiksirovat tester is impossible). Another 
signal is also changed, with one these changes more evident, 
and the other just barely (can not remember any how, but the 
fact that they vary). This completely confirms my the theory of 
using interrupts in this case. Everything is logical: if 
Pressing formed an interrupt and the processor draws the 
attention of the incoming bytes. Rather it is logical only 
recently, but relationship pressing with the emergence of the 
interrupt signal, the chip interface - a very strange fact. In 
this case, INT button can be produced for two reasons: 1. the 
chip enters signal from the printer indicating the interrupt 
enable, and chip fulfills pulse interrupt 2. exit from the 
regime of "ONLINE" CPU accesses to the port interface to 
program it, because the printer does not have to take the bytes 
outside the regime "ONLINE", and thus generates INT. Note that 
in any event, this event is bizarre - print byte on exit from 
the print mode. Of course, we remember that the majority of 
printers, taking line of text, do not move, and the imprint it, 
only by adopting the code newline or carriage return code that 
appraises as adopted by the end of the line or string is 
printed at withdrawal from the "ONLINE" (as many bytes as the 
unit has managed to take), though perhaps it happens in the new 
entrance to the regime "ONLINE" (as after filling the paper if 
it's over, and the text nedopechatan). From this position it is 
possible (though questionable) say that our printer tries to 
print a line that he did not passed and adopted at the same 
time last byte in the register zafiksirovany port - that he 
still can. 

   It turns out that the chip interface CENTRONICS at least 
partially defective because even generates an interrupt, but 
not always, when it should be. Of course, there remains the 
likelihood of its partial failure, and failure of such a chip 
in the printer itself, with which this signal goes IEO, as well 
as others who form such a signal initially. It can be assumed, 
and such malfunction: the inner port can not completely see the 
condition of bridges. Those that block the unused interface 
signals are not checked and the printer thinks there is no 
lock, and waits for the appropriate signals to the computer. 
But this contradicts the fact that the necessary signals 
received from a modern computer and did not yield results, as 
well as the fact that since entry is blocked (due to lack of 
signals), what then ... are printing at least one byte from the 
input, let and last ... All this suggests a false programmed 
port (not excluding his fault), because of what the printer is 
not working as it should. Given the total number of ports - 
their 5, is not enough any program when you switch to

there were surprises. Although the two are mostly programmed
true, otherwise the motion mechanism as well as display, 
keyboard and a jumper, normally would not be controlled. 
Another fact, equally telling, and the incorrect programming 
interface, and possible about his failure, concluded that the 
port not only does not generate an interrupt signal (which may 
be a hardware reason - not received the proper signal from the 
printer IEO), but gives false information that somehow come 
with a normal data bus. Question: where they come from? Or they 
are distorted in most ports because of its failure, or true 
programmable port - that is, wrong. It can be assumed and then 
the firmware is not suitable for this interface - interface 
specified IFSP, and we CENTRONICS, and therefore the port comes 
incorrect information (for another interface, it is true). 

   Generally, this is strange, because the interface itself - 
separate, removable module that connects to a standard signal, 
and a general method of data exchange is used for all models of 
interfaces, but they can be programmed, of course, in different 
ways, despite what the logic of the transfer, but once a 
removable module, then by necessity it can be replaced (no 
matter what insert the plug), and print from another source, a 
different protocol. The idea is that when such simple 
replacement should not get into the printer and change 
proshivku.Proshivka itself must determine the type of interface 
(once it is provided with jumpers) because of the socket 
interface a lot of empty contacts, which can be used for 
auto-configuration of the plug. However, if we compare the 
interface circuits, then both Other contacts are not used, and 
thus autoconfiguration impossible, but initialization of the 
ports - a survey immediately after their inclusion - will give 
different results in different interfaces, allowing to 
distinguish them. True, it can be assumed that such a 
convenient - autoconfiguration - the firmware just is not 
enough space, or information, or simplified. Otherwise, get 
silly: the interface must be changed together with the 
firmware, so casually, and a table with Cyrillic change 
possible, or barter ... 

   Take the scheme both interfaces and checks on the basis of 
logic of signals for data transmission and pertinence of the

bits of the port. Conclusion: yes, very much like the truth. If 
our program CENTRONICS a willingness to accept data protocol 
IFSP, then the port generates the signal nevybora (in the IFSP 
is just readiness), and the error signal from CENTRONICS again 
normal - "Infallible" - for the IFSP. About the busy tone: it 
is not cleared because the programs the output port of 
discharge of employment as input (for IFSP), and a passive one 
input to zero, of course, not become. In this case, the bytes 
are not accepted, because all control bits do not enter the 
port (these inputs are connected to the inputs shallow logic, 
which themselves are waiting for a signal from the port, and 
eventually such "Mnogovhodnye" points scheme hangs in the 
passive one (the level is not TTL input connected - a unit, but 
it is sensitive to interference). In general, it seems that 
blame for firmware and program the printer as our CENTRONICS 
IFSP, and therefore it does not work properly. At issue is that 
the printer can not print one (last) bytes. This can be 
explained as interference in the free entry (as if not 
connected, according to our theory with the firmware) causing 
false positive combination of control signals and yet what a 
system failure or incorrect programming features of the port, 
its distinctive logic work at the same time.


   The processor is clearly defective, t.k.delaet their 
business. Of course, not to exclude it, but too well the state 
board and the conditions of his operation. In my experience, no 
one has not yet dead, although there were circuit outputs. They 
say someone burned, but such trials, which this printer, and 
never dreamed of. 

   Of course, poor programming and erasing may ROM, because 
then the program will execute in favor of neverno.V This 
version also has some say, is very rare crashes Printer - 
because something is behind them, may, of course, stagnation 
mechanism, or a bad contact when connecting interface, but 
nothing can be ruled out, including damage to the firmware ROM. 
Take, for Had my printer. Initially refused to feed back 
(although mechanically All right, and no matter where 
spinning). Then came the crashes in state lines, and then began 
to fall apart when you turn on the initialization, in the end 
he either did not become tested, no mechanism "Going crazy". I 
was hoping that it's not ROM was a scheme climb that could have 
- checked, but the big chips you can not just check the tester, 
because the state of their signals can be programmed 
incorrectly and should either operand to 100% logic test ROM, 
which I certainly do not, either implicitly trust the contents 
of the firmware, and, among other things, completely aware of 
its logic (to know what chips are programmed when and how, and 
under This check them). As a result of that story, I found 
reference data on the ROM K573RF8 (32 kilobytes) and found that 
it keeps all only 25 thousand hours, and switched off and 
switched on, well, even 3 years, and has been almost 7 years - 
closed to 2 terms. Then I print the contents of this ROM and 
got rid of the file on disk.Ot program, of course, much 
remains, but there was so much every bullshit that of normal 
logic operation can not say (I disassemble the code). Swap the 
bytes have been completely erased, and towards the end of 
memory loss that vozrastali.Seychas printer lies in the closet, 
being completely intact, and there is no firmware him. Although 
the assembler is available to me, I do not know all the 
features of the programming mechanism, so you can spend a lot 
of time wasted, even having a chance to damage the very 
mechanics of the test. In Overall, I would like to get a normal 
ROM with the firmware, and even for 100 rubles (although the 
printer I got for 200, and the money is not sorry something I'm 
still typing on it), but so far there is no place on the 
internet yet have not tried to watch, although the chances very 
high. But in general, willing personally take (within the 
Ryazan) to someone who will be able to use for other purposes. 
Just a pity to throw the system. Someday I will gather the 
programmer and try to revive ... 

   There is another interesting fakt.V my comp without sockets 
soldered ROM K573RF5. It's almost "analogue in our case." Her 
and the volume same pinout and like p ^ 2, differ only in 
electrical parameters (polyphagia). In particular, the storage 
of information it is the same - 12 years, but when engaged, not 
3, and 6 years (twice in p ^ 2). But if we recall how I drove my

Speccy, including around the clock, then my computer worked
2 years continuously (if not more) that reduces the total time
Holds up to 8 years and years have passed, 11 t.e.uzhe 3 years 
as the data can Stretham. However, the tests (there are special 
programs with the checksum) is still talking about repair. But 
even if this Chip erased (I can not ask - desoldering the need 
and the quality of mounting such that the track just 
pootletayut), I do not hurt. Of course, the standard BASIC 
print functions will work badly - buggy, but I do not use them, 
and it does not use a modern program, everywhere has its own 
driver. Perhaps that is why Rom as it is turned off (the CPU to 
it becomes very rare), and this increases the storage time, but 
the term must have already passed, and for about 10 years 
although the information least partly spoiled.


   In favor of serviceability ROM of the printer is the fact 
that to test all the characters are printed normally, and their 
image data loss is not visible. But what are the characters? 
Only ten part of the ROM, and is it in a separate chip, 
possibly sewn later, and, accordingly, with a large reserve 
storage time. In addition, the "Operating System Printer" in 
the work performed constantly all the time until the printer 
simply switched on and the characters extracted only when you 
print, ie rarely used. Therefore, storing their chip is in 
"poluspyaschem" condition, and the data in it lost less kept 
longer ... 

   We arrive at the fact that before actual repairs -
analysis of the chip the printer, it is necessary to ensure 
proper operation of its ROM (correctness of data stored there), 
through an experiment with it replacement, and if ROM in order 
to puzzle over the printer itself (it is possible, comparing a 
working original). 

   In practice, the repair of many Speccy, I've often suspected
loss of data in their main memory (27C512 to import 64 KB)
but not once it was not confirmed! In fact I have no background
Data on these chips. A fault was in the prime prichinah.Prichem 
I have heard that these chips are very robust and in fact not 
"erased" by themselves. But statistics can not exclude it, 
because mikruha erasable. And only a year ago, my suspicions 
about one such import mikruhi confirmed, but only by replacing, 
and as it turned out, her last barely flashed (it was partly to 
marriage). [Ed.: I had 2 of these cases. ] 

   So I'm inclined to believe the wrong programming
interface CENTRONICS - due to internal fault or
data loss (or their incompatibility) in ROM. At least
50% interface working properly (50% sure it completely).



   File is created in the editor ACE0.85, in Pentagon system 
128, the latest edition of the file on Aug. 24, 2004. 

KSA-7G