We all know that breadboarding has a very limited bandwidth potential when it comes to clocking computers. Despite a few people pushing 10 MHz+, I don't believe the experts recommend going much above 1 MHz for computers on a breadboard.

My questions are about performance with breadboard wire (thick, insulated) that is short and roughly the same length for buses. Wire that has been soldered onto these cheap perf-boards. I don't imagine the soldering has any benefit other than preventing the wires from coming unplugged.

Inductance, stray capacitance, etc. is still an issue...I get that.

So, is that really any better than plain breadboards?

The reason I ask is that I just finished my first "apartment" design. Right now, it's just the CPU, clock and power. But the module is on a small perf-board, the wires are short, soldered, has decoupling caps, etc. I only plan on clocking it at 1.79 MHz but was wondering if it could handle the full 3.58 MHz (my design requires a 3.58 MHz clock for some audio).

Next, I plan on building a ROM module that plugs into the top...just like an Arduino shield with long pin headers.

This is not going to be my final design. I just want a way to rapidly prototype different designs without keeping VIA's, 65C02's, etc. on breadboards. I constantly find myself knocking wires out, cutting more and more cables to span large areas, etc.

So I guess the underlying question is how fast is too fast for simple soldered designs like that? I'm not trying to build a complete back-plane. Just some stackable modules that will allow me to prototype my final design easier.

Thanks for any pointers.

_________________
Cat; the other white meat.

There are so many variables, even from the way one person builds to how another person builds in the same size, that it would be hard to even give rules of thumb as far as numerical limits go. Just keep the parts close together so the wires can be short, and keep them direct and avoid the big hoops of wire, and I think you'll be fine at 3.58MHz. The bigger issue is really not so much the clock speed itself, but the slew rates you get from the faster parts, although a slower clock speed does help to some extent by giving more time for the resulting ringing to die out before the next clock edge. If you can only concentrate your effort in this regard on one thing, make it the clock lines, because a dirty clock's ringing can cause multiple triggers on what's supposed to be a single edge, and crash things.

This is a nice part about wire wrap though. You can put the sockets shoulder to shoulder without needing any room between them for routing traces, and the #30 wires can take a very direct route across a crowded board, going between pins of ICs it does not connect to, without limiting their accessibility. If two pins that need to be connected are, say, 2.7" apart, and at a diagonal, the wire connecting them can be something like 2.8" to 2.9" long (plus the length to wrap), not having to go around the outsides of sockets or ICs. You can even get perfboard with a ground plane, or, if you really want the extreme, planes on both sides so one side can be ground and the other can be Vcc.

_________________
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?

Yes, taking the most care with the clock feels like good advice - also taking care that the power distribution is clean and decoupled.

I suspect there are three potential disadvantages with breadboard designs:
- poor electrical connection: many resistive and friction-only connections
- the internal metals run in parallel on adjacent rows
- the wires could end up longer than they otherwise would
These needn't make every design fail - indeed, most designs work - but would be enough to give a bad reputation.

It would be nice to have some quantitative calculations about stray inductance, capacitance and resistance to see how likely it is for these strays to interfere with operation at MHz speeds.

I've had success with perf board, for a simple system, well past your planned speeds.

The 65C02 on this board is clocked at 16mhz. It's co-processor not unlike the one I pictured in my introductory post.

The only things connected to the 6502 are the rom, 2 ram chips and the memory decode GAL. All communication with the host goes via the ADT7130 dual port ram.

I imaging though that's right on the edge of what's possible for this design. I've done 3 boards of this design, the first of which will "only" clock to 14mhz the later 2 will clock at 16.

I agree what Garth and Ed said, but I don't think it is a show stopper. Considering the clock rate we are discussing for our designs we do not have to worry about wire lengths. My 65xxx test setups with 4 to 5 breadboards and some decent planning where you place each chip all run easily up to 10MHz. Sure wire-wrap will give you the shortest path, but breadboards are not really bad either. Also the free wires on a breadboard have almost no crosstalk, they are too far away from each other for these frequencies. Also the parallel metals are rather short (0.5") and "far" away from each other. The biggest issues I have with breadboards is the contact quality and that wires accidentally come off when doing changes. Therefore I use jumpers for the power-rail to the power pins of the chips. K&H makes very nice sets with the jumpers having the matching color to the number of 0.1" units they span. Also I place many (5-10) decoupling capacitors along each power-rail. To make the design smaller I put together the breadboards in a way so I only have one power-rail between the chip area blocks the power-rails are attached with tongue and groove to the central block and can be removed. Also you need to make sure you have a clean power source and it is a very bad idea to run 20cm dupont wires from the power-supply to the breadboard. I always take care of the power distribution as a first step in the layout to make sure the contacts are good and short. You can measure the voltage drop from one end to the other when drawing 300-500mA.

In fact all these considerations are valid for soldered and wire-wrapped designs as well. Paying attention to the most basic requirements avoids having basic problems and helps debugging strange effects.

In other words, you should be able to run your setup with 3.58Mhz without any issues, provided the access time of your RAM and ROM matches the CPU speed and your peripherals are of matching rating.

P.S. I am currently playing with a DCJ11-AE design on a breadboard running at 18Mhz and it works like a charm.