Amiga 600 schematics nearly put into Kicad

[Mick]

Ok cheers. If I find a problem with one PCB will it be the same across the batch or are the defects more random?

[SteveMoody]

Originally Posted by Mick

Ok cheers. If I find a problem with one PCB will it be the same across the batch or are the defects more random?

I would expect that if there was a flaw in the board then it would be the same across the entire batch rather than just one. I haven't seen any random PCB manufactuting errors for quite some time

[Mick]

Ok thanks. I'll see what Acill's investigation finds first and then I might pick up a multimeter.

I've done an updated board with the CXA1145/Modulator, Z221/Z222 (heard they degrade/leak anyway) dropped and just the CXA2075M and S-video connector instead of the modulator but I'm at a stalemate unless I know what's wrong.

BTW if anyone downloads the A1K files there's an error in the A1200 schematics, the CC_A[16] global labels on the PCMCIA page should be CC_A[0].

[Stedy]

Mick,

Where can I download a copy of the PCB and associated schematics that match the PCB you had fabricated?

I'll take a look through for any issues, as best I can.

Can you or Acill tell me what does not work, it it the black screen of nothing on power up?

[Mick]

The files I used to manufacture the PCB's are on the zone from 15 days ago it includes the Gerber's I think, I've already noted a few mistakes I've since found but none of them should prevent it from working I don't think. If you use the latest KICad it says 3 things are not connected to ground but they are as I've tested a physical board, it's probably because it's not a full connection and KiCad changed something in the recent update to flag up partial connections to VIA's.

Also remember in those files U12_1 is ground and U13_1 is VCC, I hadn't figured out how to sort that at the time. I sorted it by replacing the ground symbol with a "GND" net label and the VCC symbol with a "+5V" net label.

Acill hasn't gone into much detail about it, I think he said something like he had a cycle but otherwise it's not working. He's been busy though what with Christmas and everything so I don't want to bother him about it, I appreciate him building it. I'm still hoping it's just a faulty chip as I sent a donor board I bought off Ebay for measurements, it looked like the owner had been trying to replace the ROM chip and ripped up some pads but who knows what else may have been wrong with it.

If Acill is too busy with work etc going forward I have a couple more PCB's if anyone else wants to build a now knowingly BETA board and try to diagnose it.

[katarakt]

If you're going to look for a multimeter, make sure it uses 2x 1,5V and not a 9V batterie! The recommended PeakTech 2025 uses 2x 1,5V.

[Mick]

I've ordered one, thank you.

[Mick]

FWIW here are the updated boards. We just need to try to figure out what is wrong. I don't want to waste any more money.

[Stedy]

Hi,

Starting to review the files now, will take a few evenings to complete.
What values parts are you using around E120-E129 (EC120-EC129) in your schematics?
They should be 22pF not 22nF.

As you mentioned the current board did not complete a cycle, I'm checking everything around the MC68000, Agnus and Gayle devices first.

[Mick]

I just copied the schematics, they say simply "22" no mention of units, or are you talking to Acill? I'll label them pF just to be clear.

Is E120C right as 0.22UF? what about E112C is it 22pF?

[Mick]

Without more information it's only guesswork but I'm wondering if the CXA2075M has anything to do with it? the website which I saw the mod on spoke as if was a straight swap but they had done it on A1200 which doesn't utilise the RGB output. The RGB output in the datasheet for the CXA2075M looks a lot simpler than what the A600 has.

edit.. then again the CXA1145 datasheet looks pretty much the same as 2075 for RGB output so Commodore must have just done something a bit different?

[SteveMoody]

Usually with values on schematics i tend to assume that if there's no units then take the lowest value which would be pF in this case.


Looking at the schematics i would have thought that for EC120 0.22uF would be too large a value. I'll have to have a look at the original schematics but would have expected them all on the clock to be the same value.


I've also have a look at the CXA2075M and nothing jumps out as being wrong, matches up to what i've got on a computer i'm building.

[torsti76]

Originally Posted by SteveMoody

Usually with values on schematics i tend to assume that if there's no units then take the lowest value which would be pF in this case.


Looking at the schematics i would have thought that for EC120 0.22uF would be too large a value. I'll have to have a look at the original schematics but would have expected them all on the clock to be the same value.


I've also have a look at the CXA2075M and nothing jumps out as being wrong, matches up to what i've got on a computer i'm building.

Well, 220nF (or 0.22uF) is nothing special in Amigaland. The 1200 has at least 20 of those.

[Mick]

Ok thanks. Steve since you're expert can you look at this and tell me if anything is wrong or could be improved, does the composite output need changing?

https://i.postimg.cc/gk327LZs/schematic.png

[hooverphonique]

If they're connected between a crystal pin and GND, they're likely to be 22pF. Like shown here: https://www.electronics-tutorials.ws...r/crystal.html

[Mick]

E120C and E121C don't seem to be listed in the service manual I used to get component sizes, E121C isn't installed anyway but I've looked at a few photos and E120C is 1210 size, I'm wondering if it it could be 0.33uF since I've not seen 0.22uF used at all on the A600, only 0.33uF. 0.22uF seems to be used instead of 0.33uF for decoupling on the A1200 though. It only really connects to the expansion port anyway.

As an aside if you sent the 14mhz signal to the CPU clock input would it run at run at 14mhz without screwing up anything else?

[Mick]

E199R is a bit mysterious as well, should it have a value or can you just install any ferrite bead? it looks like I put a * before I'll change it to FB like the schematics as I know now * generally means not populated.

[alenppc]

Originally Posted by Mick

Ok cheers. If I find a problem with one PCB will it be the same across the batch or are the defects more random?

Sorry for the late reply, I was away for the holidays. The fabrication errors were random, say out of 100 boards ordered 20-25 would be unusable due to various issues. Some even had VCC shorted to the ground. Mind you, this was on the 328 which is a very simple board, I can only imagine that the error rate would be different on a complex board like this.

Even JLCPCB had some fab issues but the error rate was way lower, usually below 1%, I would rarely get 2 unusable boards in a single batch although they could not produce good boards for the free 501 expansion for me but elecrow come through on that one.

This might not apply to this project as they may have different check procedures for more complex boards, but you never know.

[Mick]

I'll have to see what Acill finds and maybe buy the board off him if it's a write off, spend a while checking everything. As if designing it wasn't hard enough you have to also navigate a minefield of dodgy PCB manufacturers.

[SteveMoody]

Originally Posted by Mick

Ok thanks. Steve since you're expert can you look at this and tell me if anything is wrong or could be improved, does the composite output need changing?

https://i.postimg.cc/gk327LZs/schematic.png

Not sure i would call myself an expert on it but here's what I have as as my output stage from the CXA2057M on my board. The design for this is taken directly from the reference design on the datasheet.


I'm not really using the CVOUT signal here but when checking the output on an oscilliscope it looks like it should be correct.