Bring your product to market on time and within budget – join the Eurocircuits Community
This is the fifth in a series of articles where we discuss important topics in PCB design and manufacturing. Please follow us on social media and sign up to our mailing list to get our publications.
Fiducials are markers that are used as a reference for placement or size; their name, interestingly, is derived from the Latin word for ‘trust’. Theoretically these markers can take any shape as long as they can be recognised with certainty. In PCB design we use fiducials to help machines and the materials they’re working with to be as accurately aligned as possible. Therefore, the markers’ geometry and placement are crucial for the reliability of what we’re manufacturing. As usual with the topics we’ve covered so far what seems quite simple — just add three round things on the board — is more complex than it first seems.
Fiducials are made of exposed plated copper with a soldermask clearing, usually with a minimum of a 1:2 and a maximum of a 1:3 ratio.
Why do we need fiducials?
When we submit a board to be manufactured we send the physical representation of the board to the fab, usually as Gerber files. If we want the board to be assembled with components we also need to communicate their coordinates and rotation. For this we also send a Bill of Materials (BOM) and Component Placement List (CPL), or use formats like Gerber X3 that already have that information embedded within them. The machines that we use to automatically place components take that information and then need to find the actual physical spot on a panel that has one or more boards on it.
We can help the placement machines do this in two ways. We can try to very accurately align the panel of PCBs on the machine, although mechanical and manufacturing registration tolerances normally means that this doesn’t work reliably. So instead we use markers on the boards or panel so that the machine aligns itself to what’s in front of it; this is much more robust because it doesn’t depend on mechanical tolerances or operator skill.
How do we use fiducials?
Here we see a panel with fiducials containing two PCB instances with three fiducials in each; those are circled in blue. The fiducials are at the ‘corners’ of the PCB, as far away from each other but without interfering with the functionality, like being inside the antenna keepout area.
The most common PCB fiducials are made of a round bit of plated exposed copper and a centred soldermask clearing, with a ratio of 1:2 or 1:3 (this is a range specified by the IPC). The exposed copper’s diameter is usually somewhere between 1mm and 2mm. Since we’ll be using the fiducials primarily for component placement, we create them in the same process as the landing pads (it must be done this way according to IPC-7351B). That gives us the best registration alignment; we wouldn’t want to use silkscreen, as its registration tolerance is normally much more relaxed. Then, we remove the soldermask for increased contrast and reduction of glare, and make sure that they won’t be ‘solderpasted’ because that will dramatically reduce their effectiveness.
To save space it’s helpful to place fiducials on the same place on both sides of the board, but this is not a manufacturing requirement. A fiducial on one side prevents a via or a through-hole component from being placed at that spot, so placing fiducials ‘on top’ of each other is more space efficient.
Type of fiducials
Global fiducials are those that provide alignment for the entire board. Normally we’d put three of them somewhere near the corners of the board, and as far apart from each other. We also need to avoid identical fiducials next to each other; not only does that defeat the purpose, it can confuse the assembly machine. One round fiducial doesn’t give us much information; two is better but can still cause confusion about the orientation of the board. Three fiducials allow the machine to be certain about orientation and also to be able to compensate for any stretching, scaling, twisting, or warping that may have happened in the manufacturing and assembly process. By placing the fiducials near the edge of the board and far apart we maximise our use of them. Four fiducials can cause confusion again if the machine stops looking after it found three and gets the orientation wrong.
Local fiducials are ones that help the placement machine even further by giving it markers close to fine-pitch components that need better alignment than other components require. Here we use the fiducials for alignment although rotation information can also be extracted depending on the pattern. The IPC recommends two local fiducials located diagonally from each other, across the component; local fiducials can be shared between close components if space is tight. While global fiducials are usually required, the need for local ones really depends on the capability of the placement machines and on the size of the board: some machines won’t need them and some may only need them for larger boards. Local fiducials can be annoying to place because at that small of a scale they take a lot of space, usually right where we’d like to fan out routes away from the component or place a bypass capacitor. It’s probably best to talk to the manufacturer about whether local fiducials are helping at all, or how small you can make them.
Panel fiducials are a form of global fiducials that are placed on the ‘waste’ material of the panel.
Various panel fiducials that Eurocircuits uses on the edges of its panels.
When Eurocircuits assembles your PCB it does not, in fact, need fiducials on your board: we add our own fiducials on the panel your board is part of, and we use features on the board (like pads) as local fiducials if that’s necessary. We do, however, recommend that you add three fiducials if the board might be manufactured elsewhere in the future.
Using a 1.5mm copper diameter fiducial is ideal for most cases, but you can go down to 0.6mm diameter if needed. You don’t need to tell us which features are fiducials; our process will find them either way, just don’t place them too close together. And, as above, if you’re out of space on the board, leave them out, we’ll find other features to rely on. Our machines can handle more than three fiducials on the board, but there’s no point in having more than three.
Here are our resources for the specs that are relevant to what we’ve discussed so far:
I was wondering about why we normally use round fiducials. I’ve asked Twitter and others and there are many theories:
Round objects are easier for machines to locate.
For HAL finishes the convex shape on a round fiducial will still be round while in a square one, for example, it might not be quite square any longer.
It’s easier for machines to find the centre of a round shape.
Round shapes have the least surface area.
Equal etching round the shape.
It encourages the use of multiple fiducials instead of less effective oddly-shaped ones that could in theory include rotational information, but is hard to process.
It’s a feature that’s most distinct from what a traditional board might have, which is mostly rectangles.
Round mounting holes can double as cheap fiducial.
We always did it that way.
Some of those are probably more correct than others. The machine vision needs to accurately find the fiducial and then estimate its exact centre, so a round shape seems to maximise this outcome the most.
Be the first to read all about our news and information. Follow us on