When I was in school components fit on solderless breadboards and we made circuits using breadboards, power supplies, meters and oscilloscopes. Many of today's components don't appear to be breadboard friendly, so how is it done today?
Is circuit design software and simulation good enough to go straight to a PC board? Or do you use surface mount to breadboard adapters? Do you still use a soldering Iron to solder or paste solder and an oven?
I'm wanting to tinker with some circuits but some chips I'm interested in only comes in MSOP or other packages that look intimidating to attempt to solder.
You can stock up on a LOT of tiny SMT to through-hole boards and just solder the SMT onto those and drop them into your solderless breadboard or else wire-wrap sockets. Perhaps someone on the web sells these in batches or else has a kind of nifty board with a bunch of SMT outlines to through-holes on them and you can use that? Some companies sell these expensive adapter boards, too. But they usually aren't very cheap. Dead-bug point-to-point is another method. But some of these things have really fine spacing -- usually you can find a package with wide enough spacing to get by with.
For fast circuits, where cross-talk and transmission line effect can be important, there isn't much point in building a prototype on a breadboard - the printed circuit layout is a crucial part of the design.
This doesn't mean that the printed-circuit base prototype is going to work first time, even with the best circuit design software and simulation. Something always seems to go wrong, though with careful and expert engineers it is rare that you can't get the circuit working with a little bit of cut and link.
For lower-frequency work, turn the MSOP packages up-side done and glue them - dead-bug style - to your prototyping board, and solder wire- wrap wire or enamelled transformer wire (both around 30 SWG - 0.3mm OD) onto the leads.
The last time I did this, I ended up using a prototyping board with a "collander ground plane" on one side, and cut up the rings of copper around the lead-holes on the other to get enough small pads close together to keep the wiring compact. It worked quite well.
My motto: If it works on a breadboard, it's not worth producing.
On my current project, I have to feed the simulator pcb parasitics and component parasistics to get accurate simulations. I've had to bench test to get some parasitics. Once parasitics are included, scope results and simulation results get close.
If all looks good on sim, I make a pcb, etch it and bench test it.
I trashed all my breadboards years ago and only work with smd parts. I do hot plate soldering.
It's that or .... I put the micro leprechauns to work with tiny soldering pens.
In our case, "assembly" is fully automated pick-n-place on the same line as the manufactured board. For a new board manufacturing will use it for the temperature profile too. We have a prototype machine for raw boards but it's never worked. If they can get it to work I may use it for a few test circuits.
One problem is that device models often aren't available for fast parts, or all you get are S-params when you need large-signal time-domain stuff. So sometimes you can learn a lot by hacking some FR4 and testing parts.
I never breadboard entire products, or even complex circuits... just enough to characterize parts or simple subcircuits.
I tend to design a PCB with CAD software then have prototype PCBs made. There are several companies out there who do 'pooling', i.e. they amalgamate many designs onto one PCB, that way you end up only paying a small fraction of the tooling cost of the PCB. Some companies can handle 6 layer boards with this process. Example in the UK is PCB Snap from Spirit Circuits
formatting link
This can be quite cost effectve for producing protptypes that are as close to the final product as practicable.
Another way is to design with CAD but make your own PCBs. One way that works reasonably well is to produce gerbers and print them in mirrored form on a laser printer with high contrast (i.e. lines are as black as possible) onto glossy paper, then use an clothes iron to transfer the image directly to a sheet of copper laminate, and then etch. The reason this works is the ink is somewhat resistant to the etching chemicals. There is a product that is specifically designed for this (can't remember its name, it uses blue paper to ease the transfer process), do some Googling for homemade PCBs.
As for soldering, with a fine tip soldering iron and a little practice it's not that difficult to solder a MSOP or TQFP to a PCB. Get yourself a good magnifying glass on a stand, lots of no-clean flux, some silver laden solder, and the most important part small thin solder braid for tidying up the inevitable shorts between pins and your away.
I use ExpressPCB to make boards for parts of the design. Things like power supplies are made as a PCB with connectors that are wired to the other boards. I use 1206 parts and put in some extra parts and layout such that cuts and jumps are easy(ish) to do.
Think. Read data sheets. Think. CAD. Make on production line.
Pcb design rarely takes longer than a lash up on strip board or whatever and always works better.
Very occasionally I have lashed up a small part of a design but I almost always regret the waste of time.
I like to get things to production as soon as possible. As many issues get highlighted in production that alter the design as get found during debugging.
The new reflow oven gets commissioned on Monday which I hope will reduce some of these production snags. Like LEDs die but inductors don't solder.
We sell about 80% of our rev A boards, with no prototypes. Assuming the first unit won't work is self-fulfilling, and a good way to make sure the third iteration won't work either.
And my cabin automation hardware worked absolutely unaltered the first time!
Well, I would if I were selling them. I'm a consultant designer, so generally I deliver working protoypes on the undestanding they are just prototypes and the customers make the first small batch. Given that the actual design information will probably need to be tweeked slightly anyway to match any mods I make it's not worth me spending more money than necessary. For a hobbyist making one or two off it's really not worth it.
I have to say most of my designs usually work with only a slight modification (sometimes none, sometimes component changes, sometimes the odd wire or track cut).
Sometimes though it's easy to assume something works when actually it's close to the edge, but you still need to thoroughly examine the product anyway. We all know that sometimes there are real pressures to ship, in some respects a board that doesn't work first time can be beneficial in focusing attention to smaller details. Of course the design has to be tested thoroughly anyway, what I'm saying is it can speed that process up if the thing doesn't work first time and you've got real pressure to get it out the door :)
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.