When working with high-precision sensors, it is crucial to consider the housing. In this article, we discuss the mechanical design considerations at play in a custom board design for an inclinometer subsystem.
There are many considerations for mechanical design when designing a system with a high level of precision. Recently, I designed an inclinometer-based custom PCB with such precision that I need to design a heavy and stable case to protect the integrity of the data it collects.
As I mentioned in my summary of the inclinometer subsystem, the general project should allow a resolution of 0.001 °, which means that the sensor will detect it if the plate moves even 1 µm. This means that you will have to digitally recalibrate the board each time it snaps into its mount. To make things easier, I decided to fix my card to a holder for less need for constant digital recalibration.
The picture above shows the assembled board that is kept in the aluminum tester during the preliminary test.
See the links below for more information on the Precision Inclinometer Project Series. This article specifically refers to the PCB layout article.
Manual plate adjustment with differential drive screws
Since I’m Mark “Hard-way” Hughes, I decided to add a manual adjustment system. At one end of the bracket, I chose a differential drive screw mechanism. On the other hand, I added two additional polished M3 acorn nuts on the opposite end of the table mount. These two non-adjustable feet bolt to the bottom of the dash and become a pivot point opposite the differential screw. Also, this arrangement allows a certain amount of tuning to test the project.
An exploded rendering of the PCB holder, alignment pins, PCB, and PCB clamps
Differential screw mechanisms have two slightly different pitch threads that rotate simultaneously about a central axis. They are arranged so that as one screw advances, the other is withdrawn. The result is a compound screw whose effective pitch is the difference of the pitch of the two original screws.
This video shows a generic differential screw while this video shows the Thorlabs differential screw used in this project.
A render of the differential screw. The thread size has been exaggerated in this illustration. Image by Mark Hughes
This differential screw mechanism can be adjusted in multiple ways. Rotating the thick part (shown in blue above) provides a rough fit. Rotating the fine part (shown in red above) provides fine adjustment. Rotation of the center part (shown in green above) simultaneously affects both roughness and fineness, providing a microfine (differential) fit.
If you have access to the company’s credit card, your friendly local machinist will be able to make custom differential screw mechanisms. Due to budget concerns, I opted for a 25 µm / rev differential set screw manufactured by Thorlabs.
A complete revolution of the mechanism changes the height of the end of the PCB holder by 25 µm, but the screw can be turned by smaller amounts.
In the case of this differential screw, 1 µm corresponds to a rotation of approximately 15 °. While this is not an ideal level of fit, it is at least within the realm of plausibility. And, as a backup, I have the ability to calibrate the device digitally.
PCB Mechanical Design: Heat and Mechanical Stress Mitigation
We incorporate some mechanical design considerations into our PCB design, itself.
Six 3mm notches are placed 1 inch on center along the long edges of the board that will engage the indexing pins on the aluminum bracket. The PCB is held in place by two 3 “long aluminum fasteners that will be held loosely against the PCB with M3x0.5 or # 4-40 hardware.
Rather than using fixed mounting holes that create concentrated stress points, this design allows a small amount of pressure to be applied over a large area of the board, securing it while still allowing some movement parallel to the plane of the board.
4 layer PCB pile shown above. Dark green is the ground grid. Red is the 12V network. Salmon is the 3V3 network. Light green is the 2V5 network. Dark blue is the 5V0 VRef network.
The only component on this board with mechanical considerations specified in the datasheet is the U4 5V voltage reference, Linear’s LT1027LS8.
The datasheet indicates that “power and ground planes should be skipped under the voltage reference IC”, which is relatively easy to achieve with path outputs. Of greater immediate concern is the directive for “[cut slots] Through the PC board on all four sides. “The grooves should be as long as possible, and the corners are big enough to accommodate the tracing of the traces.”
Figure 8 of Datasheet showing suggest board cutting patterns to use.
This is done to minimize any mechanical stress on the PCB near the voltage reference. Changes in humidity, as well as temperature, can cause the PCB to contract or stretch, which, in turn, can put stress on the LT1027LS8 package. These proposed cutouts reduce any thermal conduction of heat from nearby components, in addition to providing mechanical stress relief.
The width of the cutout, sometimes called “kerf”, is determined by the capabilities of the PCB manufacturer, which in my case is a lower limit of 1mm. I chose to extend each cut with a small angled relief to extend the groove and further isolate the components.
It might be possible to alleviate stress even more with more elaborate designs. If you have any advice on this, please let me know in the comments below as I did not consult with a mechanical engineer for this design.
Cutting power and ground planes can be dangerous during PCB design. All signals will find a return route and a disrupted ground plane can make that path unpredictable.
And according to the “Hard-way Hughes” paradigm, I should take this opportunity to observe that the sensor provides a radiometric output and both the ADC reference input and the sensor power input are fed from the output of this reference. voltage. Ratiometric means that the output is scaled linearly with the reference voltage, so a change in the reference level should affect the input and the output equally.
Therefore, the voltage reference could deviate substantially before it began to affect data conversion, so the additional work required to incorporate the recommendations for the datasheet may be largely unnecessary for this design.
I’m sorry I didn’t incorporate a voltage accumulator / follower between the voltage reference and the ADC, especially since I had one on the board with two available channels. I will have to test the board to determine whether or not the design works acceptably without it.
An artist’s impression of thermal / stress finite element analysis.
The above artist’s impression of a combined thermal / stress finite element analysis illustrates how the plate cutouts serve to mechanically and thermally insulate the inner section of the plate from the heat generated by the ICs, as well as some theoretical pressures applied to the outer sides of the joint. Board cutouts reduce the overall voltages applied to the area of the board occupied by the voltage reference IC.
There are no calibration switches on the PCB, as pressing them could disturb the board alignment. Instead, a 20-pin 20-pin FFC carries several capacitive touch lines from the microcontroller to an exterior control panel that is not yet designed. The data will be taken from the board through the USB to UART interface.
When creating high-precision analog circuits, some devices require significant consideration for maximum performance. Although I could have easily ignored best practices or omitted some of the manufacturer’s recommendations on this design, I made every attempt to integrate them for educational purposes.