Amstrad Plus and IM2 bug

[dragon]

Im stunding my gx4000 revc




With the 74ls27


First nor:


Pin 9+10+11=8


Pin 9,10 wait
Pin 11 a13


Output to pin 8-> pin 5


Second nor


Pin 3+4+5=6


Pin 3 iorq/clk6


Pin 4 unkown, the polimeter not show me continuity between z80 reset pin 26 and pin 4 i can't folow it disapear under the labels of board.


Third nor:
1+2+13=12


1,2,13=pin 6


Output clk6 niorq pobably.

[dragon]

None

[PulkoMandy]

About the possibilities to "fix" the hardware:


The problem is different to 128K support. Fixing this would change the interrupt timing, which means existing software (both written for CPC and Plus) which relies on precise timings would stop working correctly. This makes the fix more annoying than the bug.


Also, the bug is well known and has several software work-arounds (not using IM2, putting the code only where A13=1, etc). So fixing it makes writing code for the Plus a little less annoying, but does not suddenly open many new possibilities (it's not like we can suddenly have twice as more DMA channels or something).


The compromise seems not worth it.


Nonetheless, now I'm curious what the original bug which led to LK106 was?

[dragon]

All that was done by amstrad. Even the amstrad fix change, between board revisions.


Looking track with pictures in web appears pin4 in 464+ is connected to the reset of the fdd (asic pin 52).


That in theory is not connected in the console.

Edit: appears,They have  made a little change between revision b and  the other.


In the original fix(hc chip)they take nreset from z80 line.
In the 74ls27n  they use reset in oposite way. So they take the other reset line of the asic in the computers(used by the fdd ic).. As the gx4000 don't have these pin enabled they remade the tracks to take it from asic pin 52 fdc  but for the inside part. So at normal look it appears not connected, but is connected to pin 4 ic 116 internally.

I don,t think amstrad have timmings in mind when made the fix, i bet was made by sone chinese guy in orion factory when they view something they not like in the first computer out of factory chain.


Anyway as i say fix it or not is only a option. Don't like, not fix it, like, fix it thats all. In the same manner, the sprectrum guys chosse fix the sound bug in +3 or not.


If amstrad continue selling the plus more son or later they have fix it in new motherboard revision, as he do in spectrum. Thats the true.


Probably was simply as add m1 to the ecuation. When m1 is low if  a13 is low wait is low  iorq is low and reset was high niorq=0


When m1 is high if a13=0 wait=0  iorq=0 and reset=1  niorq=1

[dragon]

I  found a very fun thing about the fix.


Remember the review of gx4000 in 1990 lets take a look! Rewind!


One of the first gx4000 sending for the press to review number "14".





Take a look at ic 116. Reset signal is taked from the reset singnal of the ay sound chip. But the most extrange thing was they have connected pin2 of the ic 116 to the ay bc2 port!

[arnoldemu]

bc2 on the ay is wired to 5v.

I remember reading that issue of aa. I read it over and over a few times

Is the ASIC the same version everyone else has? It looks like it is.

So, I wonder what is behind the claims that earlier ASICs are a little bit different? One day we'll find out

I was also thinking - on broken Pluses, what if IC16 is the one that breaks and the ASIC is fine? It would be interesting to know.

[roudoudou]

Kevin, will Arnold mimic the bug? 

[arnoldemu]

Kevin, will Arnold mimic the bug? 

yes

[dragon]

bc2 on the ay is wired to 5v.

I remember reading that issue of aa. I read it over and over a few times

Is the ASIC the same version everyone else has? It looks like it is.

So, I wonder what is behind the claims that earlier ASICs are a little bit different? One day we'll find out

I was also thinking - on broken Pluses, what if IC16 is the one that breaks and the ASIC is fine? It would be interesting to know.

Who said the asic was little diferent?

Is dificult to tell the gx4000 rev a motherboard in cpcwiki have number 4, the aa motherboard have number 3 and are little diferent, So probably somewhere is 1 and 2 early boards.

Another strange issue is the rev a of the computer not exist in the wiki. ¿Anyone have view it outside the service manual?.There is a posibility that are broken and amstrad retire it from sell.Also there is a posibility, that the cpc works nomal with lk6, And the fix is only to maintaining  compatibiliy with early boards that maybe don't have lk6.

The ic is a normal ls chip, he can die of course cost less <1e Maybe bryce can have someone broken.


edit:

This  ***In theory*** should fix the bug with systems using 74ls27n, i know  nobody want it, but its funny do it.

The principle is easy, original amstrad circuit have a13,wait,iorq,reset,This is the same except they add m1 to the equation. So if cycle m1 is on(low), Iorq is not changed with a13=0 wait=0, until m1 finished,The rest works as amstrad original circuit. reset is active high(pin 52 asic) with  74ls27n in boards. (is active low in hc chip)