On-Line #07 22 октября 1995

Profi     Доработки     Железо

Iron - Finalization of reference sync computer "Profi" v5.03.

                 IRON


             Done!


   As I already wrote earlier, I had work to change the work of 
reference sync computer "profiles. So, please hurry - work a 
success! Now the details. 


   Initially there was a sync versions boards 3.xx, it
Perhaps the darkest creation and sharpen the focus on it I do 
not I will. Let me just say that you can recognize this version 
of the black frame around the screen SPECTRUM, and the program 
does not run letters QARX on top of the curb, but somewhere in 
the middle of the screen. Then the specialists of "Condor" was 
written version of "SAMX6" which was implemented SPECTRUM full 
screen, plus the correct Signal INT. And although, like 
everything was normal, I was not satisfied with the position of 
the upper curb effects. 


   The idea that there is "guilty" INT, proved to be incorrect, 
the case was in the conversion of strings. As we know, right 
"deploy" 312 rows SPECTRUM raster (screen + border). In many 
schemes, such as the "Pentagon 128", is computed 320 lines - It 
leads to an apparent violation of sync. The most noticeable if 
the BASICa give the command "SAVE" - instead of good 
distinguishable, rising to the curb lanes will be visible to 
the chaotic jumble. In "Profiles" things were almost as 
expected, only counting eight cycles (16 * 19 + 8 = 312) 
occurred in personnel momentum, rather than before them. This 
led to the fact that the formation of the screen starts at 8 
lines before, and looked as if the image on the curb "launched" 
on one of familiarity. Although "SHOCK MEGADEMO" went fairly 
well, the sprites do not disappear. So the first problem point 
was the position calculation. Here is the first moment I 
removed. 


   And then turned a funny thing: the picture on the screen / 
curb were poured as the "Pentagon 128" with modified INTom that

led me to believe that the "PROS" versions 3.xx - 4.xx is not 
that otherwise, as an improved and feature-rich "Pentagon"! 
However, only on the bus and makeshift. I'll tell you that 
there is important signal WAIT - the "Pentagon" and the old 
"PROS" he simply was not used. So get a version of "SAMX6M". 
Well going SHOCKom, unfortunately, had to sacrifice. But, 
finally fell into place upper curb the effects and went 
"Multicolor" in corporate programs. 


   Another thing that got me depressed was the fact that
QARXe was evident progress of the ray, which forms a border at 
the top right corner. Practice has shown that a key role played 
duration of the signal INT. And, again, that's funny, duration, 
chosen exactly as in the corporate SPECCY, there was not 
optimal. Apparently, due to many parameters of the circuit 
computer shorter INT processed in exactly the same as strictly

Fixed INT in the company SPECCY. By the way, then that INT is 
formed at the beginning of the frame (ie the trailing edge of 
an inverted human clock) - not quite true. In fact, he formed a 
little earlier. This conclusion I did, based on the fact that 
the circuits that implement the principle described above, give 
the program the following effect ACADEMY - curbs the image is 
shifted relative to the screen in the upper left corner of 
exactly one pixel. This time I did not clean up, though 
theoretically it is done possible. But the apparent progress of 
the beam in QARXe I almost eliminated by choosing the 
experimental duration INTa. Now the duration INTa somewhere 
around 8 - 8.6 ms. While in theory "it is not quite well, in 
practice no problem as "critical" software was not. 


   At this point, two revised and modified version of the 
reference sync are called "SAM7CS" - for versions boards 3.xx - 
4.xx and "SAM14CS" - for versions boards 5.xx. Both the driver 
passed all measurements at the frequency and duration of sync, 
human frame, etc., the results were good, and now all the 
boards will be equipped with data sync. 


   "Fifth" version of the card gave me a very pleasant 
surprise. For by the fact that in the scheme of the processor 
signals are formed from WAIT video controller, get the perfect 
timing. Such I have not yet seen! In all the "critical" the 
program is completely coincide ALL curb the effects, work all 
"MULTICOLOR" s, and the notorious SHOCK MEGADEMO works so 
perfectly that I was sitting spellbound for five minutes, not 
believing what was happening. Here, apparently impact that 
application of the latest version of the asynchronous circuitry 
in the SPECTRUM mode has achieved full emulation of "opaque" 
tire brand ZX SPECTRUM. All this gives me the right to say that 
today the "PROS" version 5.03 of the most successful and 
compatible models commercially available computer. Naturally, 
the owners have purchased cards for all versions can contact me 
for modifications of the computer. 


   In principle, the old "TRAC" you can try cross-connected 
WAIT as the "fifth" version. This, I believe, will undoubtedly 
polozhitelny result, but due to the large number of new 
compounds that will be a very difficult task. While the 
enthusiasts can do this, at the same time transferring the 
experience to "The Pentagon" has made happy owners of the most 
common 128Kb computer in our country. 


   The main technical summary of the above is as follows: 
duration of INT in the SPECTRUM-compatible computers are fixed 
no-one and is chosen experimentally the stability of the 
"critical" programs, no doubt also that the second significance 
of the signal, which is responsible for compliance - is a 
signal WAIT, and there is a desire to consider how to properly 
cross-connected. The second conclusion is: it is impossible to 
achieve 100% compatibility, and proprietary chip ULA can 
proemulirovat a maximum of 99%, and then only with great 
expenditure of time and technical means that it is unlikely the 
main mass of users and producers seem to fit. 

P.S. For even more compatibility, on top of the board, perform 
the the following:


All versions: look for the wiring diagram muz.protsessora chip, 
where the mixed A13 and A15, remove the leg from the chip A13, 
and connect it to the A15. You will go to digitized melody (Eg 
BLAVA 128). You can also find free element "AND" (LI1) instead 
of the A14 and submit to the inverter mixture of A14 and M1. 
This is a safety net to avoid clogging the data bus 
muz.protsessorom. 

For version 5.03: Vypayayte resistor R15 on the top board, 
connect the 2 and 3 findings m / DD45 circuit with a signal 
CSAP6 (3 foot m / s circuits DD39). After that dramatically 
improves synchronization. 

            Kirill Gromov (C) CHUNG Software (FFC)