Amiga 600 schematics nearly put into Kicad

[rvctech]

Originally Posted by Mick

I don't really have much choice unfortunately, manual routing just blows my mind and I've tried to look for alternative autorouters that are supposedly much better but they're hundreds if not thousands of pounds.

My ECAD package costs thousands to tens of thousands and it has a very average autorouter, a good autorouter would be at least thousands.

Manual routing is a bit like soldering, it's all a matter of practice

[Mick]

We will have to see what happens with Acill after the new year, if it still doesn't work with the above things tested then I have no idea why it doesn't work and so far nobody has found anything.

Does length of traces and via's impact the value of resistors that are needed?

[rvctech]

Originally Posted by Mick

Does length of traces and via's impact the value of resistors that are needed?

What do you mean by value of resistors?

[Mick]

The Ohms value, as I've said I'm not really electrically minded. I just wondered if longer/thinner traces and via's where there weren't any before meant more resistance along the route and therefore it might mean a lower value resistor is needed in place of the originals. In other words is it a bad idea to not just copy the way Commodore routed? in the schematics they mention a few times that various components are for EMI control so I'm guessing not copying the routing exactly changes conditions like the amount of EMI?

[rvctech]

Originally Posted by Mick

The Ohms value, as I've said I'm not really electrically minded. I just wondered if longer/thinner traces and via's where there weren't any before meant more resistance along the route and therefore it might mean a lower value resistor is needed in place of the originals. In other words is it a bad idea to not just copy the way Commodore routed? in the schematics they mention a few times that various components are for EMI control so I'm guessing not copying the routing exactly changes conditions like the amount of EMI?

What you're describing only matters in applications where extremely high accuracy is required (for example in analog instrumentation). This should not be an issue for an A600 motherboard.

The EMI stuff is mainly for compliance with mandated emission requirements that existed at the time (FCC). This design would probably still work even with the EMI components removed altogether (although in the case of ferrite beads you need to replace them with a 0R link).

In a high speed digital design, the lengths of traces on a parallel data or address bus do need to be length tuned to ensure that they are close to the same length (transmission lines). But the A600 is not a "high speed design" (14MHz?) and so I would not expect this to have any impact.

[Mick]

Okay thanks for explaining, I've learned a lot along the way with help from people on here.

I'll pin my hopes on it being R161 then. If not I'll blame manufacturing unless someone can find something.

[Stedy]

@Acill

Do you get any colours on startup or just a black screen?

@Mick

Board bringup/testing is the fun part, don't panic yet.

I did review the original layout/schematic you shared, whilst the autorouter was not as neat as me it did a reasonable job. I would have separated some critical tracks from others (clocks, interrupts and analogue signals) a bit better but for 1990's tech it was Ok. It'll probably be something simple.

Links to a few faultfinding guide on my website, first the A500/A2000 troubleshooter:
http://www.ianstedman.co.uk/downloads/a5kts1.jpg
http://www.ianstedman.co.uk/downloads/a5kts2.gif

I have some more generic processor stuff too if needed.

[rvctech]

Originally Posted by Mick

Okay thanks for explaining, I've learned a lot along the way with help from people on here.

I'll pin my hopes on it being R161 then. If not I'll blame manufacturing unless someone can find something.

No worries mate. Just to make sure, your PCB manufacturer included electrical testing?

When performing bring-up of a new board for the first time, it's often a good idea to install a set group of parts at a time, verify, and then continue. For example, power supply first, then CPU, etc. This can make it easier to fault find if something doesn't work, compared to loading everything in one hit.

I just checked the schematics, if R161 (_OE on DRAM) is not connected to ground, it will definitely cause the machine to not work.

If you have a spare PCB from the same batch, have you tried to perform a continuity check between pad 2 of R161 and GND to see if they're connected? This would eliminate all uncertainty for that part.

[Mick]

Originally Posted by rvctech

No worries mate. Just to make sure, your PCB manufacturer included electrical testing?

I'm not sure I used PCBWay? they ought to have done for the price.

Originally Posted by rvctech

I just checked the schematics, if R161 (_OE on DRAM) is not connected to ground, it will definitely cause the machine to not work.

If you have a spare PCB from the same batch, have you tried to perform a continuity check between pad 2 of R161 and GND to see if they're connected? This would eliminate all uncertainty for that part.

Yeah I've told Acill to try it I'm just not sure if it is or not, KiCad says it isn't connected but looking at the copper fill there does seem to be some touching.

I don't have a multimeter if Acill can get it working I might buy one but I'm a bit demoralised at this point.

I can send a PCB if anyone wants to put the time in to investigate.

[deimos]

Originally Posted by Mick

I don't have a multimeter

If it's just this one check, perhaps you can MacGyver a continuity check with a bulb or LED and a battery?

[Mick]

Originally Posted by Stedy

I did review the original layout/schematic you shared, whilst the autorouter was not as neat as me it did a reasonable job. I would have separated some critical tracks from others (clocks, interrupts and analogue signals) a bit better but for 1990's tech it was Ok. It'll probably be something simple.

It was the best I could do without freerouting getting stuck or crashing, I've started using another 1.3 version that seems a lot more stable but without understanding what is wrong for it not to work it's all a bit up in the air. About separating critical traces I don't think you get that depth of control with it unless it's possible to set trace clearance on a net by net basis. I'll have to investigate.

Ideally we could probably do with someone manual routing the files on A1K, maybe even get the A1200 board down to 4 layers which I'm unable to do with freerouting.

[Mick]

Originally Posted by deimos

If it's just this one check, perhaps you can MacGyver a continuity check with a bulb or LED and a battery?

I've been thinking what I could use but it'd probably mean dismantling something I'll be wanting to use a few months down the line, I'd might as well just wait for Acill to check it.

[Mick]

I just put together a circuit with a PC fan and R161 is definitely connected to ground, so I'm back to thinking that it must be a dodgy chip on the donor board I gave Acill? I think Acill is using a Kyocetera oscillator from the same board so pin 1 being connected to ground shouldn't be an issue? I don't see what else it could be?

[Mick]

Alright, I checked the other things that KiCad is claiming are unconnected and they're connected so it looks to me like something changed with the recent update, I didn't think I would have ordered PCB's without checking everything.

So that just leaves the following mistakes I've since found:

Originally Posted by Mick

Problems I've already found (not sure if he's tried yet):

Major stuff
Oscillator pin 1 connected to ground (disabled oscillator output?)

Minor stuff
Parts on the back for the A500 keyboard connector is unnecessary (plus SOT23 pads wrong as I copied A500 schematics)
U49 pads wrong
Some components placed sightly wrong

Not sure
Missing junction on PCMCIA programming section
I forgot to do the controller port swap on pins 2 and 3

So assuming Acill left U49 and the SOT23 components on the back unpopulated like I said what else could it be? could the missing junction I've found or forgetting to swap the pins on the controller section cause any serious issues?

[alenppc]

Originally Posted by Mick

I'm not sure I used PCBWay? they ought to have done for the price.

Let me just chime in, I would recommend against using PCBWay, as during TF328 production days they were known to deliver between 25-30% defective PCBs in any given batch.


From my experiments, Elecrow seems to do the best job followed by JLCPCB.

[Mick]

Well that'll be a kick in the teeth if true although at least I'll have an answer as to why it's not working, I'll just have to wait for Acill to test the chipset and see. If it just turns out to be a faulty chip I'll order some more boards with all of the fixes.

[rvctech]

Originally Posted by alenppc

Let me just chime in, I would recommend against using PCBWay, as during TF328 production days they were known to deliver between 25-30% defective PCBs in any given batch.

Almost all manufacturers offer electrical testing, but on rare occasions it is necessary to request this as part of the order process.

According to the PCBWay website they perform electrical testing of "all multi-layer boards". Sounds like they do this by default. I don't use this company though and cannot speak "from experience" .

[alenppc]

Originally Posted by rvctech

Almost all manufacturers offer electrical testing, but on rare occasions it is necessary to request this as part of the order process.

According to the PCBWay website they perform electrical testing of "all multi-layer boards". Sounds like they do this by default. I don't use this company though and cannot speak "from experience" .

These companies say a lot of things that they don't follow through. When confronted about the quality issues, PCBWay simply said "we will make sure the next batch is okay", but then it came with the same issues.

Meanwhile, JLCPCB won't bevel their boards no matter how hard you try. Same excuse, we will do it next time, but the next batch is the same as before.

[Mick]

I'm considering getting a multimeter to spend 15 million years checking connections one by one if Acill has no luck, I'll be damned if I don't succeed after everything I've already put into it. What's a good model?

[torsti76]

I like my Peaktech 2025. It's got a very good price-value.

It even measures frequencies up to 20MHz, which is kinda perfect for Amiga use.