Accurate CNC - Myths and Truths In PCB Prototyping
The intention of this publication is to clarify some technical aspects of PCB prototyping machines, specifications and designs that are part of our knowledge and expertise. It will serve you well in comparing CNC (Computer Numeric Control) systems.
In this article we will use the following measurement units;
- Mils, 1mil = 0.001 inch, for imperial measurement system
- Inches, 1inch = 1000 mil
- Microns (µm) 1 µm = 0.001 mm, for metric measurement system
- Millimeters (mm) 1mm = 1000 Microns (µm)
Metric units are in brackets.
1) Machine resolution
First we would like to remind you that the positioning resolution and positioning repeatability are not related to the absolute (compare to a reference point) positioning accuracy. Machine resolution represents only the smallest movement that can be made by the positioning system. Due to the nature of mechanical engraving a resolution in the range of 0.1 to 0.3 mil (2.5 to 7.5 µm) is needed for high quality milling process. The upper limit is a bit rough for the most precise operations possible and the lower is too precise to have a real impact to the quality of the final product. Lately we are seeing technical specifications that point to 0.005 mil (0.12 µm) or even less. This is a statement that can’t be supported by actual results. Here are a few, of many, factors that make these specifications suspicious.
Screws and nuts that have less than 0.005 mil (0.12 µm) backlash and nonlinearity in a single turn in a similar range are available; unfortunately the cost of one pair of 15 inch screws and nuts with this specifications is staggering $10K. Obviously these screws and nuts are not used in a machine that cost $10 to 25K. Typical screws and nuts used in high end machines have 0.05 mil (1.2 µm) backlash and similar nonlinearity per turn. This is still 10 times more than the posted resolution. One can see that when the direction is changed you will have 10 steps error if your resolution is 0.005 mil (0.12 µm).
Stepper motors nonlinearity
Any stepper motor (despite of the design) has larger nonlinearity then the resolution declared above. If the system is servo based your rotary encoder will still have a similar nonlinearity and it still will depend on the lead screw nonlinearity and the nut backlash. Even the best case (servo system with linear scales) will only improve your absolute accuracy, but linear scales with 15-20 inches (350 -500 mm) length and a resolution less than 0.02 mil (0.5 µm) are also out of the price range that can be used in the desk-top prototyping machines.
Positioning system rigidity
The next point of our argument is that even the best designs in positioning systems are not rigid enough to maintain their physical position when changing cross forces are applied. A good design will have an elastic movement of less than 0.05 mil (1.27 µm), at any position, when a cross force of 10 oz (2.8 N) occur (a typical value for PCB machining). Now this is also 10 times more than the proposed resolution of 0.005 mil (0.12 µm). More rigidity can be achieved with heavy, strong components, but it isn’t practical to have prototyping machine on your desk that weighs 300-400 Lbs.
Spindle run out
Spindle run out is an extremely important parameter in the specifications of a PCB prototyping system and quite relevant to the resolution of the positioning system. Usually it is measured at a certain distance from the end of the tool holding collet, and represents the total tool movement in the plane that is perpendicular to the tool axis. Typical values of this distance are 3/8 inches (or 10 mm for the metric measurements). The Highest quality spindles have Spindle run out in the range of 0.1 – 0.2 mil (2.5 – 5 µm). Run out bigger than 1 mil (25 µm) is not acceptable for PCB prototyping systems. It is very interesting to note, that the same manufacturers, that specify resolutions of partial micron do not have listed any specifications of their spindles run out.
As a bottom line - to create a design that has a theoretical resolution of 0.005 mil (0.12 µm) without any practical results is driven by commercial advertisement forces - to catch the eye of a less informed person. We definitely think that this does not offer any advantages to the entire business of mechanical PCB prototyping.
We are ready to discuss this issue with anyone who is interested in more facts and details.
2) Automatic tool change
Automatic Tool Change (ATC) is a factor that contributes to better productivity and convenience for the operator. At the same time it makes the prototyping system more expensive and in most other systems less accurate. This is not true with our ATC systems however. Our opinion is that ATC is third in the row of important factors; 1- precision and 2- affordability. It is not our business to decide how a customer will select the machine model and parameters, but it is our business to ensure accuracy of our products does not suffer with added features. We just feel it is necessary to comment on the hidden accuracy problem related to ATC in most other systems. This being tool penetration accuracy.
Tool penetration accuracy is a factor that varies in different operations used to create PCB prototypes. Here is a review of needed tool penetration accuracy related to the machining processes:
Drilling: an accuracy of ± 10.0 mil is completely enough (except for drilling blind holes in multilayer boards, very rare application in PCB prototyping)
Cutout: same as drilling (± 10.0 mil)
Insulation and copper rubout using stub end mills: ± 1.0 mil
Insulation of designs with fine features using V shaped tools: ± 0.2 mil
We will call the last process Primary Insulation (PI). It is the most demanding for the accuracy of the tool penetration accuracy and in the same time one of the most used in PCB prototyping.
Now, what is the relation between ATC and tool penetration accuracy?
Some ATC systems use the plastic rings to set the tool position inside the spindle collet.
The plastic rings have been in the PCB industry for a long time, long before the technology of mechanical PCB prototyping. Usually they are factory installed at a distance (of upper surface of the ring) to the tool tip = 800 mil. With its popularity and long history this has became a world wide standard. The problem with this is the fact that this distance is not from the tip but from the shank end and it is calculated based on tool overall length that varies by 2 mils or more. The variation is actually transferred to the position of the ring relative to the tool tip. This problem does not exist in the classic PCB production because there are no processes of PI and copper rubout and the drilling and cutout were getting sufficient accuracy using the plastic rings.
Lately we found that some manufacturers offer extra precision ring setting systems that utilize an “in advance measurement” of tool length and appropriate ring setting. Also now are available ring setting systems that utilize electronic measurements. Due to the relatively low proliferation, these kinds of ring settlers are relatively expensive units.
For the readers with less experience in the tools – the ring setting problem comes from the fact that the rings are installed on the tools using significant force, that can’t be applied to the tip of the tool. The conclusion of this potential problem is that you have to know what is available from the supplier of your ATC prototyping system to have this problem solved in advance.
The second most important problem is the shank diameter tolerance. As you probably know all prototyping systems are using tools with shank diameter of 125.0 mil (3,175 µm). This shank diameter appears to be world wide standard too. Typical tolerance is +0.0/- 0.2 mil (+ 0/- 5 µm).
The problem here is the geometry of the spindle collet. In order to get maximum clamping force from a limited pull force (usually spring generated) the spindles designed for ATC are using collets with a small cone angle. This multiplies the shank tolerance by 4 to 20 times. In other words even your ring is set at perfectly accurate position the collet will close (clamp) your tool at different position related to the spindle rotor, depending on the actual shank diameter. The error goes in the range of 0.8 to 4 mil (20 to 100 µm). In most cases these errors are limiting the performance of the PI process.
The only available solution seems to be individual tool calibration procedure after each tool change as performed by our ATC systems. We don’t know of any other ATC, PCB prototyping system that is designed to do this accurately. The leading company in this business suggested manual tool change and adjustments for the usage of V tools keep in mind, that the insulation milling using V tools is one of the most common operations in PCB prototyping. Our ATC models satisfiy all the requirements for the accuracy needed in PI.
Now we have made one more significant step ahead offering models with calibrated screws and models with complete servo feedback control based on precise linear encoders. In order to achieve the maximum accuracy, all new models are equipped with thermal compensation for the axes and material. We don't know any manufacturer (LPKF, MITS, T-TECH, ...) that have published specifications of the real physical/absolute accuracy of the systems offered by them. With the invention of the new models we got an absolute accuracy that we are proud to present to the public. All results and measurements are backed by NIST certified tools and technologies. New Models.
The new technologies in the models also completely eliminated some of the factors discussed above, such as Positioning backlash and Stepper motors nonlinearity New Models - More Information .
NOTE. We will add more information to this article on a periodic basis in order to cover more aspects on the PCB prototyping technology.