1 of 1 people found this helpful
I think they are two sides of the same coin. There will be requirements driven by electrical design and restrictions placed by fabrication/manufacturing, and I think it is a process that goes hand in hand - You have to keep in mind one when thinking of the other, you cannot isolate them.
By and large, your manufacturing capabilities, be they on site, or sub-contracted, will be defined. You cannot introduce a new process without first researching the implications of doing so.
For example, let's say you work for a company that is yet to have a BGA/HDI requirement. If that company does some form of control circuitry, they are going to come into contact with it at some point, perhaps a component that has been identified for a new product only has a BGA option. At this point, one needs to investigate the implications of introducing such a part.
What are implications of using microvia?
What are the capabilities of your fabricator?
How is the part going to be placed?
What are the rework requirements?
Can current equipment cope? (X Ray inspection etc)
You could be introducing a part that will give a small saving in terms of component cost, but also have a massive overhead for new equipment and training.
There will be times when electrical needs will have to bow to what is physically possible. It might be desirable from an electrical point of view to have back to back BGAs, but in manufacturing (not to mention tracking terms) an absolute nightmare.
Of course, the primary purpose of the PCB is to do a job, and it is no good if that job becomes impossible if design constraints make it so. To this end, the manufacturing side also has to bend a little. And we have seen elements of this with things like boundary scan; it was recognised that having full nodal access for a 1000+ pin device simply wasn't feasible, so an alternative solution was provided.
With regards to fabrication, the main driver is cost (a fabricator will do pretty much anything if you pay them enough ). To this end there are things you can do - Perhaps the biggest of these is panel optimisation. Reduce the amount of raw material you are using. As a crude example, rather than having "L" shaped boards placed adjacent to each other, have them "interlocked" and reduce the scrap area by 30% or thereabouts. So even the Mechanical designer needs to be aware if your company has a standard panel size of how that board will sit in a panel; Has the space he has allowed for a given board mean that you can only fit one on a standard panel and there will be 40% scrap?
However, obviously introducing new technologies and processes into fabrication is going to have a cost impact. Engineers need to be aware when conceptualising their product that a 10 layer, blind/buried/stacked via board is going to have a cost premium over a 4 layer PTH board.
In summary, if your product is going into any form of mass production, then, I don't think you can prioritise one aspect and then the other. For a design to be successful you need to think about both from the outset, that means having a Electrical designer and a production engineer working side by side at the conceptual stage, or an Electrical engineer having in-depth knowledge of current manufacturing/fabrication processes.
With regard to your panel optimization comment, do you typically design your PCB’s for a panel or do you let the fabricator take care of the whole panelization process?
For instance, do you create the panel design on company standard panel templates?
Either step and repeating the same PCB a number of times or even by mixing multiple PCB’s in the same project so that you build a system panel?
Another aspect to designing your PCB’s for a panel is that manufacturing constraints may exist at the panel level that restricts the placement and routing of the boards. E.g. for specific technologies a supporting wire touches the panel for mechanical relieve during soldering of the components on the panel.
This requires a placement keep-out in the middle of the panel. If a board is placed on the panel this keep-out should be reflected on the board. Stiffening bar locations, reference holes and clearances at the edge of the panel may also have to be taken into account.
Designing the PCB(s) to fit the panel presents challenges that many customers may never experience if they are not mass producing boards for consumer products?
We have our own assembly line on-site, so product is made here rather than being subcontracted out. To this end we design panels designed to run down our lines. This places a restriction on PCB panel width, which for us, is fixed. The reason for this is so that the line does not have to be stopped, and the track width adjusted everytime you run a different product down the line. (You could use pallettes, but this can be expensive). The length of the panel can be pretty much what it needs, but obviously you need to be aware of panel strength etc.
So I have a standard width of 236.2mm (9.3in) to design in. And the single/multi board approach changes dependant on product. We do run several panels that are a made up of all the required boards for a single product. Of course there are limitations, it obviously means that your copper weight and layer count for all boards is going to be the same, and you need to be aware that a panel with a lot of routing is prone to "sag" when it goes over solderwave/reflow oven processes. Quite a lot of work it put into panelisation, in as much making sure that the breakout process is relatively easy, but the panel remains robust when being processed.
We also produce panels that have several instance of the same board, and typically the same issues apply; get as many as you can on a panel, whilst keeping the panel strong.
Panelisation decisions are usually taken at the very outset of a project, not at the end when it comes to gettign the board fabricated, as is perhaps usually the case. Because we have a fixed width of 236.2, a board of ,say, 120mm width would be disastrous, as the would lead to a significant amount of waste (And real estate is the single biggest PCB cost factor).
I think you are quite right that many aspects of panel design may never be encountered if you sub contract out all assembly work, you may pass on a single up image for the sub-contrctor to do with as he pleases, however, I do think that panelisation can be as critical consideration as any other part of the design process; if you get it wrong then it can lead to all sorts of problems. After all a subcontractor is going to have all the constraint requirements you talk of - Central support rails/Panel tooling holes/Stiffening bars and jigs etc.
I would like to participate in this discussion, but it looks as if it was branched off from a previous discussion, which I can't seem to find. (Which I wanted to read before blurting in this one...)
When I look at the "Board Fabrication, Assembly and Manufacturing" list, I only see four discussions (none seem to be what this one is "continuing...").
Am I lost?
Jack (aka "the new guy")
The original discussion was a blog post from Steve, which you can find here:
1 of 1 people found this helpful
Thanks for the link, Pete. (I got a little confused not being able to find it, and then wrote to an administrator that the blog wasn't being reported in the blog list, and now have just come to the realization that maybe you were abandoning the blog to turn this into a discussion. If so, I apologize for interfering and please tell Ed to change it back to the way you want it, ok?)
Anyway, you asked some great questions about a subject that I'm interested in. I can't let this one go by...
steve_hughes wrote:All, > >
How much time and effort (resource) are dedicated, during the PCB design process, to preparing the board for fabrication and manufacturing?
Without meaning to sound sarcastic, my answer would have to be 100%
Being a designer, I see my role as "converting ideas into reality". The engineer has an idea for something he wants to try, and he is depending on me to interpret it correctly and transform it into something he can use, something he can hold in his hand and work with. So my answer to your question is that I spend ALL my time preparing the design for fabrication and manufacturing (are those two different things?)
but I know what you really meant... a certain amount of time has to be spent on packaging, placing routing, and an additional amount of time has to be spent modifying the database with downstream processes in mind. For the record, I'm a designer for Caterpillar, and our products aren't high-volume consumer products. Most of our products are robust small-volume designs that are intended to last a few decades in harsh conditions. That shift of emphasis only means we have different considerations, but we are still highly involved in the best DFM practices we can develop.
steve_hughes wrote:All, > >
Are the electrical constraints met first then fabrication and manufacturing requirements addressed later?
How much are you prepared to compromise yield to ensure that electrical constraints/performance are met?
My view is that we never choose one over the other. Of course, the primary consideration for a designer is to incorporate everything related to the engineer's idea to give it the best chance of success. enough copper for current, enough clearance for voltage, noise and timing and filtering and etc etc etc but at the SAME TIME we are forming this matrix of constraints into an actual product, that's what a designer does. So I really see myself as a bridge between theory and the real world, and I feel a huge responsibilty to incorporate the needs of BOTH SIDES into the design. Granted, the engineer probably knows more about what he wants than I do, but an equally valid statemnt would be that I probably know more product development in the real world than he does. So yes, we make compromises. If he really needs to approach a specific impedance, we will give a certain amount of weight to that. If the signals travel over a flexible section that has a very thin dielectric, which forces traces that are so thin that the yield may go down, then we have to discuss the need and determine how valid it is. So yeahyeah yeah you've probably heard a million times how good design is a series of trade-offs, but a lot of it is just education and communication. I have recieved some relatively absurd requests in the guise of "constraints" in my life, but if a way can be found to gently question the origin and science behind these "constraints", real design work can be done that results in a well-designed circuit that can be made with reduced waste, time and error.
gotta go for now,
Welcome & many thanks for persevering through these teething issues. I was asking the same question, where is my blog?
It seems that we have a relatively lively & interesting debate going here now. Thank you for your input.
So I have a related question, concerning CAM/DFF which I will respond in the original blog where you also posted a reply, or maybe even start a new thread.
It would also be great to hear other’s views on these subjects.
Well, your blog is now listed! I saw it earlier today....
I personally prefer you posting it here or as a new thread, or as a new blog entry. Posting it as a comment makes it less visible to new people, in my opinion,
Besides, you're a man with his own blog! Try it as a new blog entry... (can you remove your own comments?)
Also, if I could ask a question, I noticed at the top of this thread a yellow "star" that says something like "one answer is worth 5 points" or something like that. What the heck IS that?
p.s. I'll compile that article into a PDF and try to figure out the best way to post it, ok? These discussion threads/blogs don't allow attachments, do they?