Placing components is usually just a matter of trial and error. First you start with the things that have a fixed location, and then put the rest where you think it is easiest to route. Don't forget to put the bypass caps close to the power pins. Do a quick and dirty layout, to see if it will work, and adjust/erase if necessary.

Speaking about power, you don't seem to have a power connector, or are you using the AVR ISP connector ?

For as much as possible, try to put your pins on a 100 mil grid.

Are you having a PCB manufactured, or are you going to etch it yourself ? Make sure your trace width, drill size, and via sizes are all set appropriately before you start.

Forgot about that. I’ll take the power from an ATX power supply, I’m only going to need a connector, an on/off switch and a fuse.

I’ll try to etch the board myself and try a trace width of 0.01”. There are some limitations, especially with the through holes.

_________________
Marco

Careful with PC power supplies. They can deliver a lot of current, so one mistake on the board is enough to vaporize traces and components. I always recommend using a power supply with current limit. Or at least put a resistive element in series, like a small bulb from a bike light.

See my brief write-up on custom PCBs at http://wilsonminesco.com/6502primer/CustomPCB.html, and the links there.

If you're sticking to the slower parts, you can get away with nearly anything (like the huge 2-layer boards of early home computers did with no ground plane) but I still like to encourage good habits and neat work that will pay off when you get into the faster stuff.

Again, remember you can mix up the data lines to RAM, and you can mix up the address lines to SRAM, if it makes the layout easier. Like ElEctric_Eye said, there's really no reason for them to tell which data lines are which, or which address lines are which. The fact that CADs don't gracefully let you change them is one of several reasons I do not use schematic capture software and ratsnesting (although I had to get very proficient at it at my last place of work). In spite of the schematic capture, we still got PCB routing errors, unlike the method I use now when I do the schematics by hand and check the PCB layouts by hand. (I use CAD only for PCB layout, not the schematic.) Make sure you use the sections of a quad NAND for example that reach the best for the layout. If you use schematic capture, you'll have to do some back-annotating after you see which sections reach the best.

I have not done very much digital-only PCB layout, but I've done a lot of analog, up to 500 parts on a board; and packing it in by hand and routing it while I'm placing it, I get things so dense that basically you can't see the top surface of the board except at the edges. That's not possible if you just sprinkle the parts in and then set the autorouter to work. A human can always do a much better job than an autorouter. As for trace width, most PCB manufacturers don't start charging extra until you get down below .006". Even at .010", you can get two traces on the same layer between pads of DIP ICs.

_________________
http://WilsonMinesCo.com/ lots of 6502 resources
The "second front page" is http://wilsonminesco.com/links.html .
What's an additional VIA among friends, anyhow?

I can confirm that! They can deliver quite some fireworks


I totally agree, you have to set a standard as high as possible.
I would really love to make Daryl’s SBC3 or similar. But frankly that’s too complex for me to start with.

You have to put yourself in amateurish beginners shoes. Those blogs swirling around the net show little steps with results. That’s what makes it attractive. It’s a big leap from having a simple breadboard setup running and your workbench computer, POC1.1 or SBC3.

Yes, I learned about that possibility on building a VIC-20 memory expansion.

that’s an ongoing process, I use “delete” and “invoke” to select the right section. I discovered that not so long ago. It’s all learning by doing.

Checking everything by hand and even make your own parts library is indeed the best thing to do. For instance I used a VIA symbol I found on the net and luckily I discovered that VSS and VCC were swapped . Even on EAGLE’s standard libraries a silly PNP transistor had E and C swapped.



That’s years of experience of course. And certainly of great value that you share that knowledge. But one can only show or tell about a skill, not hand it over to someone. And yes, autorouter sucks. But it’s tempting and sometimes useful for small sections I believe.

_________________
Marco

Of my two last previous efforts, long before I finished the first one I could see that my growing ideas for it were not going to fit in the case, so I started over. Then on that second one, I bit off more than I could chew, trying to do too much at once before anything at all was working, and I got discouraged and quit working on it. The workbench computer came a couple of years later, and in its orginal form it was simpler but left room for expansion. It only took me a week to build in its original form, and had a big blank space on the 4.5"x6.5" board. Later, over the years, a lot of things were added, including the mezzanine board since the main board ran out of space. The original approach however made it possible for everything else to just be additions, without starting over. The processor's own buses, BTW, do not go up on the mezzanine or to the board-edge connector.


Contact him and see if he has an SBC-3 board to sell you.

_________________
http://WilsonMinesCo.com/ lots of 6502 resources
The "second front page" is http://wilsonminesco.com/links.html .
What's an additional VIA among friends, anyhow?

If Daryl can't provide you with an SBC-3, I have enough parts here to build two more POC V1.1 units, with or without the SCSI host adapter. Lemme know via PM if you are interested.

_________________
x86?  We ain't got no x86.  We don't NEED no stinking x86!

The incandescent lamp as a ballast suggestion is a good one if you're at all concerned about a major blowup at power-on. Automotive tail-light lamps are a good choice for that sort of thing.

Incidentally, any reasonably good PC power supply will crowbar if subjected to a direct short. However, it's not something that should be relied upon in powering up a brand new design.

_________________
x86?  We ain't got no x86.  We don't NEED no stinking x86!

But if there's only 20A running through your chip, the PC PSU may not consider that a short

You may need a relatively substantial load, probably on the 5V to keep it in regulation.

Arlet is right, I’ve had vaporized traces in the past. (even a coil on a C64 board ), then I started to use fuses. They saved the day a quite few times.

I know what you mean, it’s a dedicated power supply with an additional 9V~ transformer and worked without, but I have used a few LED’s and resistors for bling bling anyway.





That’s a very nice offer, and really tempting.


I should rephrase that statement, I meant I would really love to design and build a SBC like those.

If doing it myself isn’t going to happen, I definitely would like to make use of that offer. For now that feels a little bit like cheating. And I won’t like to lose the drive designing it myself.

As a side project, I am close to programming my first CPLD. I have to wait for some capacitors. The software is installed and I have made a small sample project. It’s ready to roll...
Daryl kindly assured me that the programming cable and test board I’m building should work for all XC95xx 44-PIN CPLD’s.

For designing a ‘816 based SBC my greatest challenge is routing a board of that size. I probably will make use of the ExpressPCB’s board manufacturing service. I have relatively good confidence in designing the glue logic with a CPLD.

It’s really exciting!

Edit
PS.
do I *need* a special tool to swap those PLCC44 IC's?

_________________
Marco

Since you live in The Netherlands, make sure you compare prices with Eurocircuits. http://www.eurocircuits.com. They'll let you upload Eagle .brd files directly, so it's really convenient.


You could use a small screwdriver, but the socket is easily damaged. If you plan on removing the PLCC more than once, I'd recommend getting one of those tools.

This is worth drawing attention to. And, the warning applies to some IC voltage regulators as well as the regulator in a PC power supply. Although some regulators do maintain their specified output voltage even in the absence of any load, others respond badly when the load current is too low. For an IC regulator, check the data sheet. If there's a minimum load current specified, and that value is non-zero, well, you need to observe that spec! I just quickly checked the 7805 and LM317 data sheets; the former has no minimum load current spec but the latter lists a minimum load current of 5 mA.

As noted, for a PC power supply the minimum load could be very substantial. But detailed spec's on such supplies are hard to come by, meaning you'd probably have to resort to an experiment. ( For such an experiment, use resistors as a load --not your precious project! )

We know you're a stickler for proper terminology, BDD, so I hope you'll welcome a reminder. In a another thread PaulF very amiably corrected you in regard to the term crowbar:cheers
Jeff

_________________
In 1988 my 65C02 got six new registers and 44 new full-speed instructions!
https://laughtonelectronics.com/Arcana/ ... mmary.html

I use a small 200 watt power supply to run POC. I determined by testing that the crowbar point was about 12 amps on the 5 volt output and about 15 amps on the 12 volt output. That would be sufficient to turn some chips into smoke generators.

Since much of what we build is powered by 5 volts, inserting a 7805 regulator into the circuit would have the effect of limiting current to a bit over one amp into a short. That much current probably won't damage the average PCB or wire-wrap assembly.

_________________
x86?  We ain't got no x86.  We don't NEED no stinking x86!

Yes, using the +12V rail and a 7805 is a better plan than using the +5V rail directly. The 7805's overcurrent protection kicks in a lot quicker.