Weird Video issue CPC464 "Solved"

[imperious]

Hi Everyone,

1st post here from a new CPC464 owner.  I have an Australian CPC464, Board revision MC0001A that I got off a friend just Yesterday, who got 2 of these
from His Uncle who wanted them to go to a good home. Comes up AWA on the screen. One has a Monochrome monitor and the other has the Colour monitor. 
As You would expect my friend has the Colour monitor and the good working unit. He also has the Monochrome monitor as I don't want that.

This unit was tried on both monitors first and gives the symptom pictured. I got it working on my GBS8200 easily enough. I am hoping someone doesn't
tell me it's a ULA fault as I'll likely give it back to Him as a spare if that's the case. So far I have reseated the Z80 cpu and pushed down on the ULA chip,
also tried a spare 4164 over each Ram chip but that made no difference to the screen. If this problem was on a c64 I'd suspect one of the 3 ROM chips, but
don't really have much of an idea with this. No dry joints that I can see. I do have a Programmer and 27c256 eproms if it miraculously turned out to be a dodgy FW/basic Rom.
One other thing, the Yellow colour doesn't change, but the colour of the text can be the red/pink colour pictured, or dark or light blue at turn on.

Any help much appreciated.



     

[CPC_Fan]

Hi and welcome to the forum imperious. Unfortunately I am not a hardware expert, so I cannot help you with your problem. However I am sure that one of the forums resident experts (probably Bryce), will post here soon enough with some helpful and friendly advice 

[gerald]

If you have the pasic promt working, Z80/ROM/RAM are OK.
Looks like the gate array is not working properly. Ink color and mode does not seem right.
The gate array being on socket, you can confirm by swapping it with an other one (well same ref, 40007 on a MC0001A)

[imperious]

Thanks for the information, it's what I feared unfortunately. The motherboard can take a 40010 also, but I won't be paying $50 Aus to get one
here. A couple of Years ago the Exchange rate was a lot friendlier and $30 would have been a lot more acceptable.

The computer is in pretty good nick, so I might hang on to it and see if any cheap ULA's show up.

[Bryce]

I'd guess the Gate-array too. However, it may not be actually damaged. Remove it from the socket and clean the pins of the IC and the socket with isopropanol and it might solve the problem.

Bryce.

[qbert]

As You would expect my friend has the Colour monitor and the good working unit. He also has the Monochrome monitor as I don't want that.   

Too bad for you you did'nt take the monitors :

- the colour one provide much better image for vintage games graphics than any LCD display currently available, unless you add some expensive video pre-processing hardware. The GBS 8220 is definitely not of these ones. Also remember all vintage graphics were designed for CRT screens, not for the now "pixel perfect" LCDs.

- the green monitor is really good for any serious activity on the CPC. The monochrome image is crispier than on the colour output and, if your are in any serious programming, provide less eye-strain.

[CPC_Fan]

Thanks for the information, it's what I feared unfortunately. The motherboard can take a 40010 also, but I won't be paying $50 Aus to get one
here. A couple of Years ago the Exchange rate was a lot friendlier and $30 would have been a lot more acceptable.

The computer is in pretty good nick, so I might hang on to it and see if any cheap ULA's show up.

Have you tried Tot0's site imperious? I have just had a quick look and have worked out that either a 4007 or a 40010 from him should work out at about $32 Aus including tracked and insured postage.

Link here CPC Circuits - Cent Pour Cent

[TFM]

Could be an error in the ROM too, so mode 3 gets used instead of 1.

[Bryce]

Could be an error in the ROM too, so mode 3 gets used instead of 1.

That wouldn't cause the dodgy text font though.

Bryce.

[TFM]

That wouldn't cause the dodgy text font though.

Bryce.

Save a MODE 1 switch-on message and reload it in mode 0 (or 3) and you will see just this effect. Of course pen and paper are switched here too. An additional ROM error.

[Bryce]

Really? Must try that out.

Bryce.

[MaV]

Yep, as TFM said, that's the MODE 1 startup message in MODE 0 and different colours.

[Bryce]

But why aren't the characters mode 0, they look corrupted?

Bryce.

[MaV]

The letters are correct for MODE 1. You can cheat the firmware into thinking you're still in one MODE, while you've switched the hardware to another MODE. The result for MODE 1 letters shown in MODE 0 looks exactly like the screenshot.

So, either - as TFM surmises - the MODE is set in ROM to 0 (with all other parameters still set for MODE 1), or the hardware has some kind of strange and singular glitch and doesn't switch to MODE 1 correctly.

The colour changes irritate me. It looks like it's a hardware problem after all.

[Bryce]

Possibly a data bit not making it to the CRTC?

Bryce.

[gerald]

Possibly a data bit not making it to the CRTC?

It could be the bit 0 stuck at 0.
- Mode is on bit[1:0] of the mode/rom register. if bit 0 is stuck at 0, only mode 0 and mode 2 are possible
- To set the color you have to select the ink register via an index.
    - border is OK (index is 16, bit 0 = 0)
    - ink 0 is wrong (index is 0), but has color of ink 1
    - ink 1 is random (index is 1), looks not initialised

So a check/cleanup on GA IC and socket may help. IC115 (74LS244) may also be the problem.

[Bryce]

Thanks for doing the analysis, saves me from doing it

Bryce.

[MaV]

@bryce: You can do the MODE 0/MODE 1 effect even in BASIC:

Code Select Expand


10 out &7f00,&84
20 goto 10

Execute this on the start screen and it'll look quite like on the picture (colours are different obviously). It won't be perfect because interrupts get in the way (which set colours and mode back to the current values of the firmware variables).

[Bryce]

Cool, I'll try it tonight

Bryce.

[arnoldemu]

Possibly a data bit not making it to the CRTC?

Bryce.

The CRTC generates addresses, so more likely is the Gate-Array.

[arnoldemu]

    - border is OK (index is 16, bit 0 = 0)

border is selected if bit 4 is set to 1, bits 3..0 can be any value. So bit 0 could still be 1.

16-31 will also select border.

[arnoldemu]

at reset time, I believe only border is set to black. So clearly it's initialised here and initialised correctly too.

All other colours could be random.

Border is good colour, so the hardware colour number used will be one of:

4 or 16.

At reset time mode will be set to 0, lower and upper rom will be enabled. Screen will be written to ok because you can write to ram under the rom.

The colour for pen 0 shown here is probably hw colour 3, but may be coincidence.

Bit 0 could be stuck at 0.

What happens if you try to change the mode from basic?

[Bryce]

The CRTC generates addresses, so more likely is the Gate-Array.

Yes, but it receives data from the databus.

Bryce.

[gerald]

border is selected if bit 4 is set to 1, bits 3..0 can be any value. So bit 0 could still be 1.

16-31 will also select border.

I was wondering, but did not make any test. Good to know 

[TFM]

Either a problem in the GA or in the ROM. Check / clean them first.