|Warning: mysql_query(): Can't connect to local MySQL server through socket '/var/lib/mysql/mysql.sock' (2) in /www/dreamersi/seattlerobotics.org/public_html/encoder/header-article.inc on line 45 Warning: mysql_query(): A link to the server could not be established in /www/dreamersi/seattlerobotics.org/public_html/encoder/header-article.inc on line 45 Warning: mysql_close(): no MySQL-Link resource supplied in /www/dreamersi/seattlerobotics.org/public_html/encoder/header-article.inc on line 63|
MAKING PROTOTYPE PRINTED CIRCUIT ASSEMBLIES
This article is aimed at a hobby/home based electronics enthusiast. I am assuming that you have developed a schematic and tested it with a bread board or plug board and want to take the next step to incorporate it into a design of some sort. You believe that the next process involves designing and assembling a "professional" printed circuit board, but aren't really sure how to go about this process. I address the following issues:
- Do I really want to design, produce, and assemble a pcboard?
- What tools are available for designing?
- How do I buy a pcb?
- Can I make my own pcb?
- How do I put it together?
Should I really do this?
One of the first questions you must ask yourself is whether you really need or want to make your own printed circuit board. The usual reasons for and against this process are listed in the chart below. Listed below the chart are explanations for each item. Check off the ones that apply to you and decide for yourself.
|It’s high tech||It’s expensive|
|It needs to be smaller||It’s messy-and hazardous|
|It needs to be pretty||It will take several tries|
|The parts only come in surface mount||It ties you down|
|It interfaces with a commercial board||It takes time|
|You’re going to make a million of them||It already exists|
|You just want to|
Reasons for building a printed circuit board:
It's high tech.
You've come up with a really cool robot and you want a cool control system to direct it. All those wires hanging off the perf board are ugly and get hung up in things. A "professional" design requires a "professional" control system. I will grant that laying the input wire for a 14 bit A/D converter across a switching power supply will definitely affect you're A/D accuracy. On the other hand, you can route your wires carefully, and I've seen some extremely durable and powerful systems built up with wire wrap technology. High tech is as high tech does. I'm not sure high tech is a valid argument either way.
It needs to be smaller.
This reason is among the best for developing a pcb design. Imagine, if you can, a cell phone built with proto board and wire jumpers. More to the point, a mini or micro sumo robot doesn't have a lot of space or weight allowance for controls, so a pcb control is ideal. Shown below are Dave Hylands Minisumo control and Bob Cooks Microsumo control.
It needs to be pretty.
This argument is like the first one. Put your ugly design in a small enclosure and it becomes instantly pretty. On the other hand, if you'd like the audience to appreciate your effort, this may be valid point. Functionally pretty is another good reason. Cleaning up your design compacts the components and simplifies installation.
The parts only come in surface mount versions.
This statement is only true for a few components today, but the electronics industry is moving in that direction. It is difficult to solder SMT parts to a Vector board.
It interfaces with a commercial printed circuit board.
You may, for example, want to use a PC104 configuration to interface with a single board PC to input and analyze your video signals. If you use the standard interconnections and footprints, you can add a modem, GPS, or a lot of other features all tied to the main bus.
You're going to make a million of them.
This reason is why printed circuit boards have succeeded in industry. A good design is easily replicated. You may have the next, best design for a highly accurate compass-or whatever-and are almost on the way to forming your own company. Larry Barello, for example, sells his ARC1.1 board for the SRS workshop robot and to the world at large over the Internet. I don't think he's sold a million of them yet, but I bought one and found it to be a great help as a "brain" platform.
You just want to.
This is probably the most common reason for this development project. As noted below, you can pick and choose how much you want to do. Pride of ownership is very powerful and may overwhelm any and all objections.
Reasons against building a printed circuit board:
There is no doubt that your pcb assembly will cost money. If you elect to have someone else manufacture the pcb for your robot, you should budget $50 just for the basic pcb-not including the parts-and figure on two or three tries to get it right.
If you produce your own pcb at home as explained below, you'll have to deal with Ferric Chloride, Acetone, and possibly other messy, hazardous chemicals. You also have to decide what to do with these products when you have finished your production run. Trying to dispose of these products is getting more difficult by the hour.
It will take several tries.
I've met few designs that were right the first time. A good designer can get in one or two tries, but often it takes 3 or more "spins" to finalize the product. I've seen "blue jumpers" (indicating fixes) on commercial products, so don't let it get you down.
It ties you down.
Once you finish your design, you tend to get "locked into it" due to the time and effort invested. This is the downside of pride of ownership.Revisions are a normal and healthy part of product development.
It takes time.
There will be the time to design, the time to produce (yours or theirs), the time to assemble, and the time to troubleshoot. Maybe that plug board in a box doesn't look so bad after all.
It already exists.
There are many commercial designs for all sorts of pcb based functions available on the internet from places like Polulu, Sparkfun, etc. You may find that exactly what you need is commercially available. A commercial board will save you time and money.
How do I design the PCB?
OK, so you've analyzed all the pros and cons and decided to go ahead and design and build a custom pcb. You accomplish the design function using any one of a number of Computer Aided Design programs available on the Internet in free (sometimes limited) versions, reduced price versions, student versions, or full blown commercial packages. A few free programs are:
|ExpressPCB||www.pcbexpress.com||Ties only to their production website|
|EAGLE||www.cadsoft.com||Limited to 100 points|
Most free or low cost programs allow you to create "decals" (accurate pictures of component layout), assemble schematics, and generate pcb layouts. Spending money (or more money) opens access to decal libraries, interference checking, auto-routing, signal integrity checking, SPICE models, and other features. If you're doing a basic control board, the "extras" are probably not required. They are nice to have, however. The SRS has offered an EAGLE training class in the past and will probably do so again as demand requires. Although it is beyond the scope of this document to teach you how to use one of these programs to design a pcb, many of the programs have tutorials to assist you in the basics of design. The process consists of:
- Defining part geometry (creating decals) and associating schematic symbols with this geometry.
- Entering a schematic.
- Defining a board outline (and enlarging it as necessary).
- Creating a netlist (a list of components and connections) from the schematic and placing the associated parts and connections (components and "wires") onto the board.
- Locating the components (often following the schematic) to minimize trace length and maximize signal integrity and function.
- Routing the "wires" into traces (auto-routers may help).
- Writing Gerber and Excellon files (defining the geometry, traces, and drilled holes).
I've used both EAGLE (Cadsoft) ($695 commercial package) and PADS (Mentor Graphics) ($10k commercial package) and each has features and limitations. The differences between these two programs are mostly at the convenience end. The best way to learn how to implement the design process is to locate a program with a good tutorial and jump in.
An additional tool that is a "must have" if you intend to design your own pcb's is the GCPrevue program offered free at www.graphicode.com. This program is invaluable in reviewing your Gerber files prior to-or during-production in order to spot problem areas.
How do I create the board?
With your Gerber and Excellon files on hand, you are now ready to fabricate your pcb. You can buy from a commercial source or produce your own in your kitchen or garage
Buying a PCB
Before you buy your pcb, you need to define (for yourself) the feature level your pcb requires. In review, the features of a pcb are:
- The substrate (fiberglass reinforced plastic) which is commonly .062" thick.
- The copper pads, traces, and vias (plated through holes).
- The plating (solder, tin, nickel, or even gold).
- The solder mask (screened clear layer designed to exclude solder).
- The silk screen.
PCB's are commonly defined as 1 layer (copper pads and traces on top only), 2 layer (copper pads and traces on top and bottom with vias), 4 layer (two .030 boards with both top and bottom layers glued), or more using more sophisticated production methods. The usual Personal Computer Main Board has about 18 layers. The usual "amateur" pcb is 2 layers, although 4 layers are common when high accuracy components (that 14 bit A/D), communication (serial), or switching power supplies are involved. Usually the "inner" layers are power and/or ground planes to suppress random noise signals on the communication and control signal lines.
The solder mask and silk screen features are added to make the pcb easier to assemble. As I mentioned above, the solder mask layer assists in excluding solder bridges (shorts) between adjacent pads, especially when the design includes SMT components. The silk screen layer shows part outlines and reference designators (identifying part locators) to make the assembly process easier. These two features are "nice to haves" which are not strictly required for our purposes-although they surely do help-and they add cost and time to the fabrication process.
Listed in the table below are a number of sources I have found which serve the needs of our level of production (~2 boards) and complexity (2 or 4 layers(Lyrs)), the features/service they offer (basic, soldermask, silkscreen(SM,SS)), the days to produce(DTP), and the approximate cost for 2 of a 3" x 4" pcb. The ET column refers to whether the design is electrically tested for unintended shorts after fabrication.
|Use their software|
We often buy a 2 layer prototype style board with no solder mask or silk screen for the first "spin" of a board design to see if our concept works. We "graduate" to the full scale version for the final product. This approach is not always the simplest-or even the least expensive-as noted below. The decision to include SM, SS, and ET is often based on the complexity of your design. For example:
1. If you include a ground or power plane, using the prototype level board invites shorts to that plane at many locations. In this case, solder mask is highly recommended.
2. If you design a 4 layer board, electrical testing for unintended connections is almost a must as the inner layers are inaccessible.
Several sites offer an alternative to the traditional (chemical) production sites using a machining process that mills the un-needed copper from a double sided pcb blank (see www.pcbprototyping.com or www.lpkfusa.com ). Machines are available from $6000 up (more than we like to spend). Their huge advantage for small companies is completed pc boards in hours rather than days. Home built versions are available as noted below. A tricky part of this process is making the vias from the top to the bottom of the board. Rivets and electroplating methods are both used.
Produce your own
Rather than buying your new design from a commercial source you may decide to produce it at home in your kitchen or garage. There are several great websites that focus on the materials required and the methods involved (http://www.fullnet.com/u/tomg/gooteepc.htm ). The process involved has 5 steps.
Print the layer traces and pads on high quality paper or film.
Transfer the image(s) to a single or double sided copper clad board. This may be as simple as ironing a laser toned image onto the cleaned surface.
Place the pcb in an etching solution to remove the un-required copper. Heat and agitate as required. A $40 commercial etching tank is shown at left (Webtronics).
Drill the appropriate holes required.
- Plate or rivet in vias.
There are several points to remember.
- The chemicals involved (Ferric Chloride, Acetone) are hard on your skin and on the environment. Treat them with special care and dispose of them responsibly.
- The etching process is very sensitive. It requires the right temperature and constant agitation to insure that all parts of the pcb are exposed to a constant level of fresh etching solution. Many sites sell tanks with heaters and agitators built in to assist at this level.
- Copper etches at varying rates. The effort to remove all the copper from the vacant areas often results in removing fine traces as well. It is best to keep your traces as wide as possible.
- 2 layer boards must be carefully aligned from top to bottom layer. Often this process involves drilling alignment holes for both sides.
As I noted above for the commercial process, you can also machine your two sided board at home with a CNC router or mill. At www.hobbycnc.com you can find plans to make your own cnc router. I have been immersed in that project for about a year and there are many unanticipated challenges to the process. The appeal, however, of having a prototype in hours rather than days is great. This unit can also be used for plastic part creation, engraved signs or symbols, or almost anything flat that can be milled. Figure on spending $600 or so for a basic model-and more if you wish to simplify the process.
PUTTING IT ALL TOGETHER (ASSEMBLY)
The final step (or almost anyway) in the pcb production process is assembling the board. Depending on whether you use through hole components or SMT components, you have several alternative methods:
|Hand soldering||Low Watt soldering iron, wire solder||
|Solder paste, hand reflow||Solder paste syringe, small heat gun||
|Solder paste, oven reflow||Solder paste syringe, toaster oven||
|Stencil paste, oven reflow||Solder paste, stencil, toaster oven||
This is the old standby method that applies to everything from 0201 SMT parts (ground pepper size) to 4/0 wire (Welding Cable). The key here is to make sure that you have a soldering iron small enough to suit the process desired. Your local electronic hobby shop can help you with that. SMT parts are often heat sensitive and so less than 30 watts is an ideal iron size. If you've never soldered, take the SRS level one robot class. You'll get a great robot and learn soldering with patient and focused coaching.
SMT assemblers often view the hand soldering method negatively because of the difficulty of holding the part, the wire, and the solder. Gripping the wire in your teeth is an unhealthy-though widely used-method. A simpler method is simply to tin one of the pads lightly, locate the part with tweezers, heat the tinned lead to anchor the part, and then solder the leads starting at the opposite end. By tinning only one lead you will insure that the part is flat on the pads after reflowing that first lead. Fine solder (.015") is available for SMT soldering.
Through hole parts are much simpler to assemble and solder. My favorite method is to place the pcb on ¼" thick black anti-static foam glued to a flat board. The parts are then bent (if required) and trimmed to show about 0.2" lead below the body. They can then be placed into the board with the leads projecting into the foam. After all parts are in place, simply tack one end of each component from the top, remove the board from the foam, turn it over, and finish solder all parts on the back side. I suggest you use water soluble flux and-finally-clean the pcb thoroughly with a brush and very hot (>150 deg. F) water. A common cause of trouble is neglecting to clean your boards properly.
Paste and hand reflow
This method is the standard method for SMT prototype production. Paste is applied to all pads using a hand or air operated syringe. Use less paste than you think you need! The parts are then placed in position and carefully located. When placing the parts, try not to push the multi-leaded parts around too much, as the solder will smear and tend to bridge. It is better to lift them slightly with a dental pick and reposition.
A small heat gun is then moved over the parts to reflow the solder. It is great fun to watch the parts square themselves up as the solder surface tension draws them into position. Carefully apply only enough heat to each component to see the solder melt and then quickly move on. Overheating parts is one common problem with this method.
Another common problem is applying too much solder applied to the pads. The solder may spread out (especially on boards without solder mask) and form bridges that are not visible to you because they are under the SMT part. Oops! I built one board using this method in about 6 hours, and then spent another 6 hours pulling the parts back up again to remove the bridges. That was definitely bad form.
Paste and Oven ReflowIn this method, the paste and parts are applied as above and then the pcb is placed in a reflow (toaster!) oven to reflow the paste. Several sites address the issue of conversion, temperature control, and time (www.sparkfun.com and www.pcbexpress.com ) and our own Kenneth Maxon has written a great article in the Encoder on this subject. He also has great pictures regarding the art of solder application. Since I've been burned by bridges so many times, I'm not sure I'd do it his way, but practicing the "art" will make you (or at least your solder joints) shine!
Stencil Paste and Oven Reflow
This method use a stencil screen (plastic or stainless .006 sheet with holes cut over the pads) and a common putty knife to apply the solder paste. Up until recently it would cost you about $125 for a stencil from a place like www.pcbexpress.com or similar. Polulu has lately been offering a plastic stencil for much less that appears functional. My personal experience with these "hand held" stencils is that the resulting set up lacks rigidity, and rigidity is required to insure that the paste ends up at the right spot on the board-especially on fine pitch parts. Maybe I didn't use enough tape?
When ordering a stencil, get the thinnest (commercial = .005") you can get and ask for a 10-20% reduction on the pad sizes. Ask your stencil provider for assistance in the recommended reduction amount. The point of the reduction is to reduce the chances of bridging. I vote for that!
Clean, Clean, Clean!
As I stated above, use water soluble flux (SMT too!) and wash with lots of hot water. If you don't, you'll get odd and hard to reproduce results. Mostly that is a bad thing.
After cleaning, check out your work. A good solder joint should be shiny and have smooth fillets from the part to the pad on all the visible sides. No lumpy or dull work allowed here-it doesn't hold or conduct well. Again, the web has lots of information on what that joint should look like. Type in "IPC-610" (the standard of the soldering inspection world) on Google and find the home sites and other sites with explanations referring to it.
At the end of the day (or days) after you have designed, built-or had built-your pcb, and assembled it, you'll have a thing of beauty-at least for you. Who knows, you might have the next Ipod!