This article has been written by our own Dan Parr, a member of the RDDS Design Team whose experience in the Safety Critical DO-254 Design Workflow makes him perfectly placed to illustrate and explain the fantastic work involved.

Over to you Dan –

Background:

For this article, we will be taking a closer look at challenges faced developing the 24” High Integrity Touchscreen Display for a programme  we worked on. Due to the Display’s unique and safety-critical purpose, we decided, with our customer, to develop the unit against the standards of RTCA DO-254 (Design Assurance Guidance for Airborne Electronic Hardware), to Design Assurance Level B.

Ruggedised displays have been a staple for RDDS since our beginnings in ‘97 and since 2010 our displays have predominantly featured included touchscreen capabilities. Further, our team has developed a robust and risk-mitigated workflow for developing to higher Design Assurance Levels for DO-254 and -178, so we were confident that we could satisfy the requirements.

Due to the operating environments, we elected to develop with resistive touch technology. This decision was made due to the practical touchscreen use-case and need for greater control over potential EM-Emissions, in this instance we sourced a 5-wire touch panel (sidebar: I’m certain we’ll be going into greater detail on the pros and cons of different Touchscreen Tech in a later article). To meet the DAL level and the customers requirement to minimise complex programmable elements, an alternative interface chip was selected to drive the panel, as the one we most commonly use had programmable registers.

Calibration:

To provide some broad background of the mechanics of resistive touch and its calibration – a resistive touchscreen panel uses conductive layers to detect the point that contact was made. Each possible point of contact can be represented as an X,Y coordinate within the “touchspace” of the panel.

For these coordinates to be useful, they must be translated into the corresponding X,Y pixel location of the image displayed beneath (“pixelspace”). Due to electrical noise, manufacturing variations and slight differences in the position, size or resolution of the touchscreen panel and display panel, touchspace and pixelspace don’t always align perfectly and as such, always require a degree of individual calibration. Most commonly, the differences between the two will involve some degree of rotation, scale and offset; this is normally uniform across the whole of the display.

To overcome these alignment errors, it is common practice to calibrate the screen. This is usually done using a 3-point calibration; targets are drawn in pixelspace at 3 specific and independent points and the corresponding coordinates are measured in touchspace when the user presses each one. With these two sets of coordinates, it is possible to produce a transformation matrix to correct the rotation, scale and translation between the two coordinate spaces.

With the high integrity, safety critical requirements of our customer in mind, it was imperative for the team to develop a solution which would operate within a maximum 1% margin of error in touch accuracy.

To aid in development and ensure any hardware and software issues were captured early, a prototype display was produced. With hardware in hand, the engineers got to work verifying that the prototype operated within this 1% margin.

The Challenge:

During testing, however, it was found that the differences between touchspace and pixelspace were in fact, not uniform; our testing and research showed that there was a much more pronounced deviation along the edges of the display that would persist even after several attempts to calibrate using the 3-point method. Furthermore, to compound the challenge, said inconsistency was not consistent across the display, with different corners presenting different (minor) deviations than their opposite partners.

The Team got together and began to discuss the problem, after much review of potential paths to resolution, we decided that the most efficient and effective path to solving this problem required moving to a more complex calibration method: N-Point Calibration.

N-Point calibration can be performed using any number of sample points and uses a mathematical procedure, called Least Squares Fitting, to produce a curve that most closely matches the samples. By producing a curve, rather than a line, this method ensures that the conversion will be more consistently accurate across all parts of the screen, rather than just the areas that closely match the linear transformation.

An N-Point calibration algorithm was implemented in the touchscreen driver and testing was carried out to determine the optimum number of sample points. Initially 5 points were tested (shown below), but it was found that the accuracy still slipped outside of the specification as you reached the centre of the edge of the screen (the point furthest from any sample point).

With Least Square Fitting, the more data samples you use, the more accurate the resultant curve will be. With this in mind, testing was carried out using 9 sample points. As shown in the image, the additional points were added in the centre of the screen edges, so as well as giving a larger sample set, it also ensured the areas most affected were close to sample points. These additional points proved to be sufficient to bring the touchscreen performance within the required accuracy. Whilst more points may have further refined the accuracy, it was decided that the diminishing return along with the added time required to perform the calibration meant that more than 9 points were not necessary. Additionally, there is a case to be made that with increased calibration points, the potential for operator error grows further too.

Conclusion:

While a comparatively small challenge on a much larger programme, the high integrity requirements of DO-254 design meant this required a robust and provable solution, something which the team delivered in spades. We at RDDS pride ourselves on producing high quality products that meet our customers unique requirements. To that end, we recognise the importance of early prototyping wherever possible, to allow us to carry out thorough testing and de-risking and to ensure that we deliver an end product to the high standards we, and our customers expect.

Speaking generally of the industry and ourselves, it seems we’re often not eager to share the challenges our teams face on programmes and instead focus our outward messaging on the perfected end-result. While from a Marketing Background I can appreciate the logic, this strikes me as a shame. Some of the most incredible work our Engineering Departments perform is in identifying, quantifying and ultimately resolving the myriad challenges of high integrity design, often having to come up with new and novel solutions for unforeseen challenges. Further, in the grand scheme of such long term programmes, challenges and resolutions like the one we’ve described today often get lost as a smaller part of a big picture.

To learn more about RDDS’ high integrity design services

Speak to our team