Construction techniques for prototypes / one-offs

[baljemmett]

Hi all,

Hoping I can canvas some opinions on methods of building one-off or prototype designs; I'm trying to decide how to put together my next project, and I get the feeling there are better ways to do it than sticking to stripboard (as I have with all my other contraptions).

The project is a homebrew CPU built from 74-series chips -- think Magic 1 (and plenty of other examples) but smaller and simpler! -- so it's going to consist of a fair number of logic chips running at maybe a couple of megahertz (if I'm lucky).  The various ways to build things that I've seen around are:

• Stripboard -- I think this might prove to be more hinderance than help, since most interconnections are unlikely to line up conveniently so there's a lot of track cutting going on (unless using the more expensive DIP or tripad varieties) -- and of course the issues Dave's discussed in the past, with maintaining signal quality etc.
• Matrixboard -- using jumper wires for nearly all interconnections -- strikes me as a lot more flexible and probably not much more fiddly than stripboard overall?
• Home-made PCBs -- I've never etched and drilled my own before, but I suspect it'd probably work out about the same as strip/matrix board in terms of effort (drilling vs. making jumper wires?)
• Professionally-made PCBs -- I think this will be prohibitively expensive in terms of setup costs, since there are likely to be several reasonably-large boards and only one small one will be reused in any quantity -- but it'd look good!
• Wirewrap -- this sounds expensive, fiddly and frustrating, but very flexible
• Wiring pen(cil)s -- I'd forgotten these until recently, but the technique looks like a cross between wirewrapping and matrixboard construction

So it looks to me like there's going to be some tooling cost whichever route I go, apart from simple strip/matrix board techniques; PCBs would probably look the best, but also require a lot of confidence in the design before committing anything to copper or it'll get expensive to respin them.  Surface-mount PCBs would significantly reduce the amount of drilling I'd need to do, but I've not had much practice at SMT soldering (could always construct some other bits and pieces this way first to get my hand in, of course!)  I'm currently leaning towards the wiring pen, since it seems to offer plenty of flexibility, and possibly some PCBs for one particular board that'll be useful in quantity.

Does anybody have any thoughts or experiences that might be helpful?  Perhaps I should just try a few things out and see which I'm most comfortable with, but since I've little experience of this sort of thing I'm hoping those of you with more might know if any of them are always going to turn out badly or suchlike.

Thanks for any words of wisdom!

[Bored@Work]

I know, I should not answer this, because this is one of those questions where a second after I answer this new requirements are presented with ifs and buts, but anyhow, in decreasing order of preference:

1. Professional made PCB

2. Professional made PCB

3. Professional made PCB

then

4. Pre-made, off-the-shelf prototype board, Eurocard format, pattern depending on chosen IC package format. E.g. for THT with DIP patterns. And what you call wiring pencil, but without any of those far too expensive and relatively useless plastic combs.

5. Perfboard and wiring pencil (see above).

6. One of the Manhattan construction techniques with dead bugs.

[Psi]

It really depends how much free time you have and how quickly you want a working prototype.

It's true that its quite hard work to etch your own pcbs, however once you've done it a few times you get fast enough to go from a pcb design to an assembled board in a day.  Now true, you could just sit down at your computer for 15min and email off the gerber files to get the board made professionally and you get all the benifits of that. A much better quality board and no problems doing 2-layers but you wont have the board at the end of the day or even the next day you have to wait anywhere from 3 days to 2 weeks for them to make and post it to you.

When i have a circuit i want to build/test I start using throughhole components whereever possible and just solder their legs together in a grid pattern to test out various individual small sections of the circuit. I find this a really quick way to get a small circuit built, changing values is fast too.
Once i've tested all the sections and i'm happen with the way they are working i will either...

A) Built the whole circuit out on breadboard (can bit tricky if you want/need to use smt components or ICs)
or
B) Create an inital pcb design and print it out on press-n-peel-blue pcb paper, iron it on a blank pcb and etch. Then drill holes and solder the components on.

After i've tested the board and probably changed or added a few components and/or added wires for extra tracks i will fix up the pcb design with these changes and either make a new final board using press-n-peel blue (if its just a one off hobby project and i have time) or send the pcb layout to ourpcb.com and get some professional boards made up.


NOTE: When making your own pcbs drilling the holes for throughhole components is the most annoying part. It takes time and effort to dent all the pads so the drill will be centered and even then the hole will often go off-centre anyway. So it's good to use surface mount as much as you can to reduce the number of holes you need to drill.

[baljemmett]

I know, I should not answer this, because this is one of those questions where a second after I answer this new requirements are presented with ifs and buts

Don't worry, if I was confident enough in my requirements to state any more I'd have included them   Or maybe they'd have answered my own question -- at the moment I'm too inexperienced in the practical side to have nailed anything down firmly yet.

but anyhow, in decreasing order of preference:

1,2,3. Professional made PCB

Yes, I suspected that would be the right/ideal answer; what worries me is cocking something up and having to go through hideously-expensive numbers of iterations.  I might play with some smaller parts of the design and see if I can get the hang of it, and maybe prototype the system using another method and transfer it to PCB for a 'finished product' if it goes well!

4. Pre-made, off-the-shelf prototype board, Eurocard format, pattern depending on chosen IC package format. E.g. for THT with DIP patterns. And what you call wiring pencil, but without any of those far too expensive and relatively useless plastic combs.

5. Perfboard and wiring pencil (see above).

Ah, now that's interesting; I had been eyeing up Eurocard format boards, especially since there are plenty of varieties with spaces for DIN41612 backplane connectors.  Seemed like a sensible way to spread a system across boards in a nice orderly fashion, as opposed to using ad-hoc headers and cables.  Two I had been looking at were the Roth RE-315 and RE-060; I think these are the sort of thing you mean for 4 (with DIP packages) and 5, respectively?

Thanks for the advice; I think I'll order a wiring pencil (the combs do indeed look rather expensive so I'll skip them if they're of questionable utility!) and some boards over the weekend and see how I get on.  If nothing else, it'll give me an order reference with which to enter Farnell UK's current prize draw

[RayJones]

LOL, I was thinking of starting a post just yesterday bemoaning about the difficulty of using veroboard (stripboard)

I just lashed up a real simple one off job yesterday using this devil spawn, and spent about 1/2 hour debugging all the fine solder bridges I left between tracks.
Should have worn the reading glasses methinks

Being a simple design, the fixed nature of the board was not a great hindrance, but yeah it pays to be careful how you run the tracks, and making sure you do cut where required.

In the end I should have used the simple matrix board without copper that I have heaps of and would have done a better, neater, and probably more reliable job - the veroboard tracks are easy to unintentionally break too.

My conclusion is veroboard is most definitely devil spawn. It seems tantalising at first glance but no end of pain.

[Bored@Work]

Two I had been looking at were the Roth RE-315 and RE-060; I think these are the sort of thing you mean for 4 (with DIP packages) and 5, respectively?

Yes, this is the stuff I am talking about. And I would recommend to do some planning upfront, like parts placement, instead of ad-hoc placement and wiring.

I think I'll order a wiring pencil

Consider making your own. I have seen commercial wiring pencils with plastic tips. They don't last the first few wires, especially if you are inexperienced (hot wire from or during soldering, a slight pull, and the wire melts the tip).

(the combs do indeed look rather expensive so I'll skip them if they're of questionable utility!)

They help to keep order, but introduce at least two problems: Crosstalk, because signals run in parallel in close proximity, and difficult to replace wires.

If you start wiring, start with the power rails, and use hookup wire, not the pencil. Then use the pencil to wire from inside (middle of the PCB) out. Do the wires with with short length first.  Crossing out all connections that you have done on a copy of the  schematic.

You will find it takes time and passion. After you have done your first board you will think that a professional PCB isn't such a bad idea.

Regarding the time it takes to get a professional PCB. It matters if you are a business and time is money. If you are a hobbyist it doesn't matter. Just have a few projects running in parallel. While you wait for the PCB for one project you work on another one.

[DrGeoff]

Wire-wrapped all my computers back in the 70's. Fun Fun Fun!

[Neilm]

I always take the view that it depends what you want to do. You mention that you want it to run at a couple of MHz,so I would recommend a proper PCB, and a professionally made one will be easiest if you want to do multiple layers.

Wirewrap will certainly be a problem with SI. As the wires endup going everywhere so you'll endup with crosstalk on pretty much everything, veroboard will introduce a lot of stray capacitance that also can cause SI issues.

Neil

[Zad]

Get a load of wire-wrap sockets and wire wrap it, at least in the prototype stage. This is even how the original small mainframes and minis were prototyped. Several even stuck with the wire-wrapped backplane for production. There is plenty of speed in wire-wrap, and will easily hit the 74LS speed limit specs.

[DrGeoff]

Did a complete 68000 @ 8MHz based proto with wirewrap back in the 80's with no problems.

[saturation]

Dead bug whenever possible for one-offs, fast to make, great for all frequencies.

[baljemmett]

Thanks for all the input, folks; certainly some food for thought!

I think what I'll do is try a few things out first; I've got a couple of small projects I could undertake with different methods, to see how I get on and give me some idea of the relative effort and time involved.  It'll probably be the wiring pencil first, especially if the postman rings my doorbell in the morning -- I ordered a steel-tipped one with a spool of solderable enamel wire included, as I know the enamel wire I have to hand doesn't take well to soldering!  Thanks for the tips on getting started, BoredAtWork; I'm sure you're right about it convincing me to invest in a proper PCB, and maybe by then I'll be more comfortable with the cost involved

From some quick research it looks like etching a surface-mount PCB would probably be the next thing to try out, and then perhaps wirewrap after that (mostly based on cost of entry; the cheapest wirewrapping tool I've seen in the usual places is a bit pricier than an 'all you need' PCB starter kit, even before the cost of the sockets etc., so might as well do it in that order.)  That'd also give me some experience at PCB design which could come in handy if I decide to put one particular repeated sub-component on 'mass'-produced boards instead of wiring each one individually.

Time-wise, well, let's put it this way; the last revision I made to one of the schematics is still on the plotter upstairs, and it's dated sometime in August, so I'm in no rush.  This project is mostly just to keep me amused so waiting a couple of weeks for PCBs to arrive or spending ages wiring a board won't be a problem; I just want to eventually have something I can be proud of and say "I made that"

Thanks again for the advice, everyone

[Simon]

I've not read the whole thread but me personally I like to make my own PCB's, bit daunting at first but not so hard. In fact soon as I get this building site of a house of mine plastered (and not me plastered) and the carpets in I was going to setup a hobbiest PCB service to offer to the forum members for basic projects as there does not seem to be this intermediate service available, either you get all the gear and make your own or you pay over the odds for one from china. Of course there is batch pcb but I've never used them.

I'm also fast coverting to doing designs in SMD as is saves having to drill holes and of course it is all smaller and so cheaper on the PCB plus i have the possibility of useing both sides

[Chasm]

Prototyping in China is a very viable option today.
Even for one offs as long as you have time to wait for the slow s&h and a board size that fits into the cheap offers. The ability to go for very fine structures, have solder stop and silk print also helps a lot. Fine structures without solder stop is just a pain.

I have almost all the gear to etch my own prototypes only a nice way to drill holes into them is missing. As I don't do much at the moment and have a high confidentially that I get my layouts right (Yeah, right.) it is cheaper to order into China than to buy the last tool.
Less fiddling with chemicals and way more lazy too.

That will probably change when I'll do some power electronics again. The discrete stuff I have to mind tends do be large in board size and should not have any structures under 100 thou or so. 

[Jon Chandler]

Since my boards finally came after a six week delay which seems to have been a one-time problem with China Post...

As Chasm said, Chinese fab houses are making some decent boards.  ITead Studio will make 10 boards up to 5cm x 5 cm for US$12 + $4 shipping or up to 10 cm x 10 cm for $28 + $4 shipping with reasonably quick turn-around time, so making a board isn't quite as big a risk as it used to be.  Not quite instant gratification but soldering parts on a "real" PCB with silk screen and solder mask is pretty nice.

I posted a picture yesterday in the thread on PCBs here.

When it comes to making PCBs and being sure everything is right, Dave's PCB Design Tutorial is the best reference I have seen!


Jon

[Chasm]

$5 for s&h for the 10x10cm version.

There is also the option to ship with DHL Express instead which costs about $25 for 10 boards. But then it gets into regions where looking at other options becomes prudent again. Both local to cut on S&H and also on fabs that have shorter guaranteed turn around times.

There are also various eBay vendors, but that is usually something for larger boards or actual production runs. For the latter it may make sense to spend more money on a known vendor if you have critical layouts (read close to the minimum specs of the fab). Production run after all implies that you need the boards and don't necessarily have time for a re-run of them.