Bernkastel7734 wrote on 2022-10-25, 19:49:

I do not have much experience with that, but maybe it's RAM related issue. In Olivetti M20 hardware manual I have found such Video logic diagram.

The Video BW pin should always be driven either high or low by the "video control" circuit (assuming it's a TTL level signal, as in MDA). Even if the video RAM failed, you should be seeing clear edges and defined levels. Looking at the DC offset in the MDA image: Do you connect scope ground anywhere? The MDA signal shouldn't be going below zero if DC coupled (the screenshot says it is DC coupled), and not connecting ground on a handheld battery-operated scope can result in "false AC coupling", as well as degrade the signal overall.

I recommend to check whether the BW video pin is connected to anything by trying to inject negative voltage into it and check whether it's clamped. With most IC technologies, injecting negative voltage at an output pin of an unpowered chip is clamped to -500mV. Use a meter in diode check mode, red wire to GND, black wire to BW Video. If the meter indicates "open circuit", the video output trace is likely interrupted somewhere.

It shows open circut only on VSync, and that's weird because VSync seems to be working

I guess that there is problem with video signal generation. Its CRTC is HD46505.
https://www.silicon-ark.co.uk/datasheets/hd46 … eet-hitachi.pdf
I've checked what does output on it's pins looks like and:
-all data lines (D0-D7) are floating,
-refresh memory adress lines MA0-MA9 are fine ( regular high/low transitions), but MA10-MA13 are floaring,
-raster adress lines from RA0 to RA3 are fine, but RA4 is floating,
-both chip select and register select lines are floating.
By floating I mean they are around 1 volt, with is not a correct TTL logic state.
So by that I guess that there are some broken traces/bad IC around the adress buss.

Bernkastel7734 wrote on 2022-10-29, 20:00:

-all data lines (D0-D7) are floating, -refresh memory adress lines MA0-MA9 are fine ( regular high/low transitions), but MA10-MA […]
Show full quote

-all data lines (D0-D7) are floating,
-refresh memory adress lines MA0-MA9 are fine ( regular high/low transitions), but MA10-MA13 are floaring,
-raster adress lines from RA0 to RA3 are fine, but RA4 is floating,
-both chip select and register select lines are floating.

Please check your probing setup. It is fine that the lines you describe as "floating" are at constant level while your PC is idling. D0 to D7 are only used when the CPU comminucates with the CRTC. MA10 to MA13 are expected to be constantly low when a single 40x25 screen (1000 characters) is displayed. Chip Select should be constantly high. It is an input to the CRTC, a bad level on chip select would indicate a fault in the address decoder, not the CRTC. For text modes with an 8x9 to 8x16 font, RA4 is constantly low. I expect you checked with an oscilloscope, so make sure that grounding is correct and that you are not operating in AC coupling mode.

You measured the horizontal and vertical frequencies, and they were fine. This basically proves that D0-D7, chip select and register select work. Otherwise, the BIOS could not have properly set up the video mode. Also, the hsync and vsync outputs work. What does not seem to work is the video output. That one is not generated by the CRTC, but by external logic that fetches a character from the video RAM at the address given by MA0..MA13, then looks up the character shape using the character code and the row address bits in the character generator ROM and finally shifts out the individual pixels as given by the character ROM.

I am clueless but am wondering why the B/W signal hit a 50v peak but you picture shows a 100mV scale. Could be bad ground and picking up external radiation but not able to show the true signal levels ?
The B/W is coax from the website and maybe bad core, bad ground ? Just a guess. Seems odd.

Horun wrote on 2022-10-30, 02:58:

I am clueless but am wondering why the B/W signal hit a 50v peak but you picture shows a 100mV scale.

You are looking at the "max PkPk voltage" (that has ever been encountered since resetting the stats), not the "current PkPk voltage", which is 552mV, and coarsely matches the image on the screen.

By replacing, one 7400 IC, I believe to be bad, I got something like that on screen. Which looks to be a ' non bootable disk' error ( or some sort of such error) because when I press any key on the keyboard screen goes blank for a second and it comes back to that state. But as You can see there is some incompatibility down there...

And I have just fried character generator ROM by inserting it backwards ffs... I have been tring to find it's dump online but no success. So I guess that the end of the journey..

Now why would you go and fry that poor old rom.

Anyway,
It's just character data and should be easy enough to recreate.

fyi, the Olivetti M20 hardware manual indicates that it also supports color.? Page 116 - 3.3 Video Interface
edit character size is mentioned on page 86 - 2.2 CRT display

Yes it does, but for color it needs special RAM expansion board I do not have.

weedeewee wrote on 2022-11-02, 14:34:

Anyway,
It's just character data and should be easy enough to recreate.

But how to recreate such eeprom?
I wonder if it is using a clone of MC6845 will MDA character generator ROM work?...

Bernkastel7734 wrote on 2022-11-02, 14:40:

weedeewee wrote on 2022-11-02, 14:34:

Anyway,
It's just character data and should be easy enough to recreate.

But how to recreate such eeprom?
I wonder if it is using a clone of MC6845 will MDA character generator ROM work?...

I was looking for m20 rom and found a m20.zip file in relation to mame emulation.
There's a 2KB chargen.bin file inside of the zip. It looks like the data you need.
Guess you'll only need to download the file, source a compatible chip, program it and place it on the mainboard.

Damn, that actually can do the trick. It seems I was too angry at myself to properly look for that dumps.. Thank You a lot!

Found a chart of compatible chips, but since they are all fused PROMs I have no way of programming them, so I need to find suitable EPROM and make an adapter, I guess. That MMI 6381-2J is 55ns one, so I need to find fast EPROM.

I've made a post about that broken PROM on other forum, and someone said it's RAM mapping not character generator, so I guess I have a big problem

Annoying indeed, and no use in buying a used one, since it seems to be a OTP ROM, unless you know it came out of the same machine as you have.
I've been looking at the manual, and while there are some schematics & board layouts in there. it's hard to find which chip it is that got fried. (especially since the pdf isn't searchable.)

Oh yeah, the schematics are bad indeed.
Well,
one said that it is not character generator because characters are stored in RAM,
another said that they're in BIOS,
and that PROM datasheet claims it's used for character generator.
Sooo.
And yeah, I won't buy used one

Well, I just read your post on vcfed.

Seems like tpaxia is willing to program an equivalent one, like the am27s181 that was also suggested there.
Guess that only leaves one to find the chip & the data to be programmed into it.

Well, I'm not sure if he would ship it to Europe

Why not? If you're willing to pay for the shipping & chip cost.

oh, I think i finally found which chip it is... indicated on the boardlayout as PAG3
edit, it's likely PA63 :-p
dow.