Amiga 600 schematics nearly put into Kicad

[Stedy]

Originally Posted by rvctech

I'm not sure if that is correct.

From my reading of the schematic, _RST is generated by GAYLE, and this depends on the signal _KB_RESET which is an input to GAYLE. The _RST signal is the system-wide reset and also keeps the 6570 in reset.

The machine begins in a reset state (Q511 is ON) and _KB_RESET is de-asserted by the 555 (U14) output when it reaches its timeout (0.1uF capacitor charging via a 47K resistor to reach the threshold voltage).

GAYLE de-asserts its output _RST once its input _KB_RESET (driven by the 555 and output BJT Q511) is de-asserted.

This then brings the 6570, CPU and the rest of the machine out of reset.

The 6570 can generate a new reset at any time by re-triggering the 555. However the 6570 is not required to bring the machine out of reset and the machine should function correctly without it.

Of course if the 6570 is not loaded, then Q622 should not be loaded either, otherwise it will hold the machine in reset. R624 is also not meant to be loaded.

I missed the fact that GAYLE drives the _RST signal, as I assumed all signals on the left of a schematic symbol were inputs, unless labelled otherwise. I found the document from GAYLE and what rvctech was correct, I was wrong.

I spent a while comparing the schematics on Amigawiki to an original Commodore document and found quite a few discrepancies. Will any of these cause major problems?

Annotating your list below:

E353R 47 not 68 Noise suppression/EMC compliance, tuned to each PCB variant.
E354R 47 not 68 See above
E363R 47 not 68 See above
E364R 47 not 68 See above
C321 100nF not 3900pF Part of audio filter circuit, value changes affect the filter response of the audio circuit.
R321 360 not 1.5K See above, part of audio circuit.
C331 100nF not 3900pF See above, part of audio circuit.
R331 360 not 1.5K See above, part of audio circuit.
R325 390 not 2K See above, part of audio circuit.
R335 390 not 2K See above, part of audio circuit.
C26 0.01uF not 0.047uF The capacitors below are power supply decoupling. Values would be tuned to PCB and chipset variants
C27 0.01uF not 0.047uF
C33 0.01uF not 0.047uF
C34 0.01uF not 0.047uF
C7 0.01uF not 0.047uF
C8 0.01uF not 0.047uF
C29 0.01uF not 0.047uF
C28A 0.01uF not 0.047uF
C28B 0.01uF not 0.047uF
C36 0.01uF not 0.047uF

These were mistakes by me I think:
E513R 68pF not 47pF
E514R 68pF not 47pF
E515R 68pF not 47pF
E516R 68pF not 47pF
All Noise suppression/EMC compliance, tuned to each PCB variant.
Most changes are due to the AC termination used on signals, a quick primer: https://www.diodes.com/assets/App-Note-Files/AB023.pdf

The ones on the parallel port, E513-E5167 are part of a low pass filter, with a cutoff frequency of 23 MHz and current limiting in the event of a short circuit.

I'm not sure about these:
On the internal floppy connector all 27 ohm resistors should be 68?

Are there any original Commodore schematics for rev. 2 so I can check the floppy section? or can anyone with an actual board handy check them please? the ones in question are E586R, E596R, E580R, E595R, E599R, E593R, E592R, E591R, E590R, E589R, E588R, E582R, E587R, E581R, E594R, E583R, E584R, E585R. I want to know if they're 27 or 68. I've looked at a photo of a 2B board and they don't have any markings. Actually they look a lot more like ferrite beads used elsewhere.

See my comments above on parts tuned to PCB variants.

Back in the 80's and 90's, EMC compliance was a black art and the causes and effects not as well known as nowadays. Also, issues due to signal integrity, were just starting to appear as logic got faster edge rates. Stories around CMOS PALS breaking a design that used bipolar PALS are legendary where I work.

To achieve the EMC compliance and signal integrity fixes, various high speed timing fixes (yes those infamous A1200 ones) were applied to fast signals. The resistor/capacitor values were determined by empirical analysis (a nice way of saying trial and error. The values used would be in part determined by the unique PCB stackup. The A600 was a two layer board, these is no controlled impedance, which is a great help for achieving EC compliance and signal integrity.

The seemingly random array of E components were form EMC/EMI compliance and the 'icing on the cake' was to seal the mainboard inside a metal faraday cage, the same one most of us have probably removed from our Amigas. I hope this helps explain the reasoning behind all this.

With any non-functioning digital design, I always check the power supplies, reset circuit, configuration and clock circuits first as they account for 90% of the problems I see.

[rvctech]

Originally Posted by Mick

I spent a while comparing the schematics on Amigawiki to an original Commodore document and found quite a few discrepancies. Will any of these cause major problems?

The value of EMI components is generally not super critical and generally you might not even load them to start with unless they were involved in something important (i.e clocks, bus, reset, etc).

Since Acill has not provided any details on the failure, it's probably best to wait for the board to return to the UK for diagnosis. It's probably not worth your time to try and guess what is wrong when you have no information available to you.

[Acill]

Send me a PM and I will get the info. I can pay for it, no worries.

[Mick]

Ok thanks Acill, Stedy is going to have a look at it for me so I'll just let him send you his address.

I've fixed the schematics the best I can, mainly just changing symbol inputs/outputs to the correct type and using those power flags which is an oddity of KiCad. Nothing has changed in terms of what is connected and where but the only warning that shows in ERC now other than global labels not being connected (ones Commodore never connected) is one about pins 16 and 17 of U45 being connected to ground.

[nikoskob]

Hi Mick,

As promised, here is a pic of the floppy area of my Rev 2D A600 motherboard.
Hope it helps deciphering that area.
If you need more pics, just let me know.
Cheers

[Mick]

Thanks nikoskob, it matches the schematics I've copied.

If you can take some photos of the other areas of your board I'll have a look through and see if I can find anything else amiss? appreciate it.

[nikoskob]

Originally Posted by Mick

Thanks nikoskob, it matches the schematics I've copied.

If you can take some photos of the other areas of your board I'll have a look through and see if I can find anything else amiss? appreciate it.

No problem, will do once I get home from work tonight.

[Eriond]

Originally Posted by Mick

I was just thinking because of things like the A314 project it probably would not be compatible in that case.
...
Otherwise, what would be the best way to disable the second megabyte? jumper?

As long as you you make the ROEn line available on the expansion port, and that being connected to all the motherboard RAM chips' OE pins, the A314-600 will be just fine with your board.
You see, once inserted, the A314-600 disables all onboard RAM, and instead serves 2MB of emulated DRAM (chipmem) to FatAgnus from it's own SRAM. Thank you Commodore engineers for this most unexpected feature!

[Mick]

Well I'm just messing a bit tbh. I really need to get a basic no thrills design working first. At the rate I'm going someone else will have finished it for me.

It'd be handy to know exactly what signals your board and stuff like the A604N and other 1MB expansions use though.

Do you know what the _X1M signal is for Eriond?

[Mick]

If we can't find what's wrong I'm thinking of getting some more PCB made with manually routed power and maybe separate the shield stuff (external connectors etc) from the ground plane more. I think the way I've done it a lot of external noise will be polluting the ground plane and maybe with the power being a plane decoupling is not as effective? got to find someone to build it though. I might have to just leave it and hope someone else finishes it.

[Stedy]

Originally Posted by Mick

Well I'm just messing a bit tbh. I really need to get a basic no thrills design working first. At the rate I'm going someone else will have finished it for me.

It'd be handy to know exactly what signals your board and stuff like the A604N and other 1MB expansions use though.

Do you know what the _X1M signal is for Eriond?

Looks like the _X1M signal is not used. It connects R162 but is not driven by anything.
Don't panic yet. I'm great at fixing impossible problems.

[Mick]

Yeah it just connects to ground, a lot of things don't make sense to me. I wondered with it being the same value as ROE and in the same place on the schematics if it was something like a circuit on memory expansions that detect if R162 is there (therefore connected to ground) and disable the memory if it isn't, like the ROE but just for the expansion.

I'd try to check the custom chips if Acill didn't get around to it, the board I sent as a donor which I used to check measurements was sold as faulty, it looked like the guy had tried to replace the ROM socket and pulled up some pads but you don't know if the reason he was messing was because something on the board was faulty to begin with though.

[Mick]

I just had a look at some photos nikoskob sent me of his rev 2d board, there's a couple of differences.

E126R - 47 not 27 (CCK-A)
E127R - 47 not 27 (CCK)

The CCK signals seem important?

edit: another E199 is a 27 ohm resistor, not a ferrite bead?

[Mick]

I had a look through all of nikoskob's Rev 2D board as he was kind enough to take some photos.

I found the following differences compared to the Wiki schematics I copied:

E353R 47 not 68
E354R 47 not 68
E363R 47 not 68
E364R 47 not 68
R321 - 680 not 1.5K
R331 - 680 not 1.5K
R325 - 360 not 2K
R335 - 360 not 2K
R345 - 680 not 22K
E199 - 27 not FB
E126R - 47 not 27
E127R - 47 not 27

They're mainly audio/input again but also some near Agnus.

I know there may have been differences ordinarily but it might be useful information later on.

His board seems to be a lot closer to the original Commodore schematics rather than the Wiki ones.

Commodore schematics
E353R 47 not 68
E354R 47 not 68
E363R 47 not 68
E364R 47 not 68
R321 360 not 1.5K
R331 360 not 1.5K
R325 390 not 2K
R335 390 not 2K

[nikoskob]

Originally Posted by Mick

I had a look through all of nikoskob's Rev 2D board as he was kind enough to take some photos.

I found the following differences compared to the Wiki schematics I copied:

E353R 47 not 68
E354R 47 not 68
E363R 47 not 68
E364R 47 not 68
R321 - 680 not 1.5K
R331 - 680 not 1.5K
R325 - 360 not 2K
R335 - 360 not 2K
R345 - 680 not 22K
E199 - 27 not FB
E126R - 47 not 27
E127R - 47 not 27

They're mainly audio/input again but also some near Agnus.

I know there may have been differences ordinarily but it might be useful information later on.

His board seems to be a lot closer to the original Commodore schematics rather than the Wiki ones.

Hopefully this helps towards getting your board working.

If you need anything more, give us a buzz. More than happy to help

[Mick]

Will do thanks, I thought it was worth checking in case Wiki made some mistakes in drawing up their schematics.

[Acill]

Ian has reached out to me. I've pulled the chips and tested them off the first board I did and started working on a 2nd one. I am sending him a known good set of chips and the rest of the parts off the donor boards. The parts included are from my working 600 I decided to strip for this project to be sure I was working with good parts.

[Mick]

So the custom chips you used were all alright then? that doesn't bode well then (it was my last hope ) but again thanks for the help.

If I knew what the problem was and it was something simple I would order more updated PCB's with all of the stupid problems fixed to save you guys messing around with beta boards but without knowing it just seems like a waste of money. If stedy can get it working I'll send you both a couple of new PCB's as a thanks.

[Mick]

Apparently the basic version of ToPoR costs just shy of $2000 so I think the best course of action is probably to find whatever the hell is wrong, check stuff like video output is working okay and double check everything else and then find someone to manually route it professionally.

[Mick]

I am going to manually route it similar to how Commodore did. I was close to giving up but I realised earlier that freerouting is not just an autorouter but a complete routing package and it's about 1000 times better/easier than KiCad for routing and produces much better results.

Stedy you are best off just using sockets for the custom chips because if/when this is done I'm going to get new boards made because I'm about 0.5% done and the routing already looks like a work of art, I'll do away with the split in the ground plane as well and if it still doesn't work after that it will have to go down as one life's great mysteries.