25
\$\begingroup\$

Ten years ago, I worked at a small engineering department within Philips Semiconductors that prototyped all designs before getting them professionally fabricated outside. The method was pretty standard:

  • Print artwork using a cheap laser printer onto semi-transparent film
  • Tape down top artwork onto bottom artwork and visually align them.
  • Cut a PCB to approx size, drill a hole at edge where a wire can be inserted to hold the PCB between insertions in chemicals.
  • Insert PCB between artworks, and expose using a homemade UV light box.
  • Hook wire into hole on PCB and dunk in a vertical tank containing Sodium Hydroxide and wait until developed
  • Put in spray tank and spray with clean water.
  • put in vertical tank containing Ferric Chloride etch and leave until etched.
  • put in spray tank and spray with clean water
  • put in etch remover until etch removed
  • Put in vertical tin tank for tinning.

The results were beautiful looking PCBs that were very easy to solder and professional looking, despite a lack of solder mask and screen.

Is this the standard method still used by professional engineering departments when making prototype PCBs today, or are there other methods that give similar great results?

\$\endgroup\$
  • 2
    \$\begingroup\$ I don't know of a single company in this area that has their own PCB prototype shop. It just doesn't make economic sense. \$\endgroup\$ – Olin Lathrop Jun 11 '13 at 13:20
27
\$\begingroup\$

None that I know of. 15+ years ago I also worked for a small engineering firm that did what you described above. Today, in my basement, I can layout PC boards with free or low cost software, generate a Gerber file, email those files to a circuit board house and have boards made for less than $3/sq in with better quality that I can dream of doing. Mask and silkscreen are typically included for the same price. Turnaround is between 5-15 working days depending on how much you want to pay.

It's no longer reasonable to make PC boards yourself unless you need turnaround time measured in minutes. It's so cheap that if you place a reasonable value on your time, you lose money doing it onsite.

\$\endgroup\$
  • \$\begingroup\$ Just to clarify, that $3/sq inch excludes a fixed tooling charge, right? I have yet to find a place that would make me a single 6 sq inch board for $18 ... \$\endgroup\$ – us2012 Jun 11 '13 at 16:31
  • 1
    \$\begingroup\$ I was approximating the specs of a number of places. For example, at oshpark.com you'll pay $5/sq. in. for 3 copies of pretty good quality boards. So you would get three of your boards for $30, or $10/each \$\endgroup\$ – lyndon Jun 12 '13 at 0:37
19
\$\begingroup\$

About 15 years ago I hired an EE straight out of U.C. Santa Barbra. His first week on the job he asked me, "How do you prototype PCB's? Wire wrap, breadboards, or what?" My answer then is just as valid now. We design a normal PCB, have it professionally made, and if they work we ship the product. If it doesn't work, we fix it, build new ones, and try again.

There are PCB shops and contract manufacturers that specialize in making prototype quantities of bare PCB's.

The vast majority of companies make their prototypes in the same way. There are exceptions, of course. Some companies etch their own PCB's. Some use something like a milling machine for PCB's. Some will do wire wrap. And some will use breadboards or perf. board. But for anything but the simplest of boards probably 95+% of companies just get prototype PCB's made from a professional PCB fab shop.

While this method seems to cost more, in the end it is cheaper. We spend less time debugging, building, and futzing around with potentially low-quality and unreliable prototypes. Plus, we are able to use 4+ layer PCB's with plated holes-- which is HUGE. In the past 15 years I've designed maybe FOUR 2-layer boards and 40+ more complex boards.

\$\endgroup\$
  • 3
    \$\begingroup\$ Totally agree, +1. I design electronics professionally. Trying to make may own boards just doesn't make sense. For less than one hour of my time, I can get quality boards with silkscreen, solder mask, plated thru holes, and more than 2 layers for the equivalent of a extra hour or two. Turnaround is 1 to 1 1/2 weeks from internet order to FedEx delivery. I use that time to get the BOM in shape and have parts ordered. Usually boards and parts show up at about the same time. While waiting on that, I work on some other project. \$\endgroup\$ – Olin Lathrop Jun 11 '13 at 13:19
11
\$\begingroup\$

I feel like someone should mention PCB milling as a prototyping method. Quicker and less messy than a chemical process, and the same machine that does the milling can also drill holes and rout the the board edges.

Two-sided boards are accomplished by drilling a few registration holes in the board that drop on to pins on the milling machine. Mill one side, flip, mill other side.

Still probably not very common in professional environments since prototype runs are so cheap, but for times when faster turnaround than a prototype run can provide is needed, milling is pretty great.

\$\endgroup\$
  • \$\begingroup\$ +1 excellent suggestion, but expensive I guess. \$\endgroup\$ – Larry Harson Jun 6 '13 at 18:07
  • 2
    \$\begingroup\$ This works great for big analog stuff: wide traces, easy to solder and fast turnaround. Anything smaller pitch than SOIC and I'd rather pay for a quick turn (with soldermask) or design and wait for a cheaper board. \$\endgroup\$ – Chris Gammell Jun 11 '13 at 20:32
7
\$\begingroup\$

Time has definitely moved on from those days and I endorse what the other answers say/imply about the cost of getting proper multilayer PCBs made externally but I'll add something else.

In my experience the use of circuit simulation tools means that you can reduce the number of PCB iterations to get a design correct. Add this to the almost insignificant cost of multilayer boards and there is no real need for in-house prototyping any more.

\$\endgroup\$
5
\$\begingroup\$

I work at a small laboratory at colege. Here you have to make a gerber file (usually using Altium), and then send to the person responsible for prototyping (me), who will use a LPKF S42 to make the board. Sadly it looks pretty homemade.

I attached this picture from one of our prototypes. The left one (multilevel green) is made in some Board House; the right one (kind of brown) I made by myself.

enter image description here

\$\endgroup\$
  • 2
    \$\begingroup\$ sigh that almost looks like a bitcoin mining rig. \$\endgroup\$ – Bryan Boettcher Jun 26 '13 at 20:21
3
\$\begingroup\$

So this is what we follow in our Lab. Only for a two layer board. More than two layer boards end up going to the vendor.

  1. Take a print out of the board image (original size) and make sure only the bottom, drills, pads, via and dimension layers are on the image. We use Eagle for this. Use photographic pages of GSM less than 140.

  2. Tear the page suitably to place over a PCB of certain size and iron (yes using an Electric Iron) the board keeping the printed side on the copper. Do this for atleast 15 minutes till you are sure that the tracks would have transferred onto the board.

  3. Put the board in water and remove the excess paper and etch the board using Ferric Chloride as you mentioned. Once etched see the original design and using a compass make holes on all places where drills need to be done. Use appropriate bits for drilling (depending on the size of the component you wish to fit). Also, drill all the vias and add jumpers on the top layer (if applicable).

So, This is the final result. :)Final PCB after drilling and soldering..

Edit: A few parts, which i thought were self explanatory, added now.

\$\endgroup\$
  • 1
    \$\begingroup\$ That can't be the final result from the steps you mention. I don't see any of the print material on the tracks, and the process you describe doesn't create holes. There is obviously a lot more work to get to even a mediocre board than you list. You should also point out that this method doesn't yield plated thru holes, soldermask, or silkscreen, all of which make assembly and debugging easier and more reliable. Add up the time per step and loss of resulting productivity, and I doubt this makes economic sense unless you get paid very little. \$\endgroup\$ – Olin Lathrop Jun 25 '13 at 19:29
  • 1
    \$\begingroup\$ "our lab in Delhi", do you mean your bedroom? ;) \$\endgroup\$ – Larry Harson Jun 26 '13 at 19:09
2
\$\begingroup\$

I'm not sure how professional we'd count as being, as we're a small design house rather than a huge multinational electronics corp...

Generally, very basic prototypes are done with a scalpel & copper-clad board if needed, or if more complex, we just send it out to Seeedstudio / Iteadstudio for $20, it's a no-brainer. If we need something done quick, we pay a more local "proper" PCB house to do a one-off or very short run. It costs, but that's business.

\$\endgroup\$
  • \$\begingroup\$ $20 for a single PCB? \$\endgroup\$ – Larry Harson Jun 6 '13 at 14:41
  • 3
    \$\begingroup\$ Nope, $20 for 10 PCB's, delivered. It's insane, and opens up all sorts of possibilities. Unless you're in a rush it's not worth your hours whittling something from copper-clad when you can e-mail a CAD file and have a pack of 10 turn up a fortnight later. \$\endgroup\$ – John U Jun 6 '13 at 14:48
  • 1
    \$\begingroup\$ It's $50 if you want the 10 boards delivered using DHL, which is 3 days from China, plus about 4 days of factory time. Yes, $50, for 10 boards, with a one-week turn. There are size limitations, and design rules are 8/8 or bigger, but it's fast and cheap. \$\endgroup\$ – Jon Watte Jun 6 '13 at 19:15
1
\$\begingroup\$

Typically, we send our stuff out to be printed for us, even for prototypes. Sometimes this can get expensive - 10 revisions of our work and there's over $10k. But for an initial investment of $7k, you can buy a CNC mill that will cut out dual layer PCBs very nicely. This isn't bad for prototyping, but not the best answer for mass production. We tend to steer away from chemical etch because it has a tendency to etch unevenly, and when your PCB has over a pound of copper on a small board (power supplies) you can create hotspots. For LV projects, chemical etch isn't too bad.

\$\endgroup\$
  • \$\begingroup\$ Unless I'm mistaken, PCB houses still use chemical etching. \$\endgroup\$ – Larry Harson Jun 26 '13 at 19:11
0
\$\begingroup\$

We scarcely need yet another go-around of this question, but since it is here, for completeness the clothes iron & junk mail paper laser toner transfer method is a useful and faster alternative to the photo etch method for those same-day one-off experiments.

\$\endgroup\$
  • \$\begingroup\$ Can you give me a link to where this question has been asked before? Thanks. \$\endgroup\$ – Larry Harson Jun 6 '13 at 17:50
  • \$\begingroup\$ 5 examples on the first page of the search results... \$\endgroup\$ – Chris Stratton Jun 6 '13 at 17:56
  • \$\begingroup\$ I looked through all the pages with the tag " pcb-fabrication", and didn't see anything discussing alternative ways of prototyping PCBs, apart from the obvious - get professionals to do it. The clothes iron etc method looks interesting. \$\endgroup\$ – Larry Harson Jun 6 '13 at 18:05
  • \$\begingroup\$ There is not a 1:1 correspondence between tags and subjects, use the search box on the upper right. \$\endgroup\$ – Chris Stratton Jun 6 '13 at 18:12

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.