For 2017 the Embedded Systems Conference in the Minneapolis Convention Center was held on November 8 and 9. This is the first year I’ve gone. I was surprised by how few attendees there were to the sessions I attended the first day. There were in the neighborhood of a dozen attendees in rooms that would hold over 50 people. I attended more of the hardware related sessions. The low attendance may have been somewhat related to this. Colleagues who attended other sessions noted similar attendance levels. There was much stronger attendance on the second day with the 4 hour session I attended on migrating to an RTOS being over full initially then losing about a third of the attendees over the four hours leaving about 40 attendees.
I was also surprised to see no semiconductor vendors in the exhibitor area. As a result I didn’t really get much from the exhibitors. I spent a little time talking to test equipment representatives, PCB fabricators and stuffers and mechanical component prototypers but didn’t take away much I could use immediately.
The first session I attended was titled PCB Assembly in House on the Cheap. The presenter discussed low volume and low-cost ways of building and testing PCB assemblies. To have boards made the domestic example of Oshpark was given and Chinese made boards which can still be delivered quickly can be obtained from Smart Prototyping, Seeed Studio, and Dirty PCBs. Some of the PCB suppliers will also offer a stencil for your board. To populate the boards the presenter said a small SMT machine could be used like those from NeoDen but he had good success having high school students hand place components. He described an optical inspection system he designed using a webcam and open source software to detect missing parts and questionable solder joints. He also used printed jigs for consistent placement of leaded parts. For test, he showed systems that used Arduino base and shield boards for IO and an android mobile device for the UI running an app developed in Python.
I attended several seminars given by Rhode and Schwartz representatives. One focused on power integrity. It was mainly about accurately measuring the ripple on a DC supply rail. The presentation featured the advantages of using their RT-ZPR20 probe. The common method of switching your scope to AC coupling and using a passive probe with a short ground lead has a couple problems. One the AC coupling circuit inside the scope has a lower signal to noise ratio than DC coupling. Another is that passive probes can not accurately pass signals above 100MHz. One way to mitigate this is to solder the coaxial cable from the scope directly to the board. This has the problem of presenting a high load to the supply rail you’re trying no measure. It also does not mitigate the need for AC coupling. The probe being promoted presents a high impedance load and a wide offset range so DC coupling can be used.
I attended another Rohde and Schwartz session on probing methods. The need for active probes was highlighted when frequencies get into the hundreds of Mhz and GHz. Loading over frequency is an important probe characteristic to consider. They showed how their active differential probe with modular tip can mitigate many of the complications associated with signal measurement.
Rohde and Schwarz also offers equipment for IoT device testing. They have a product that will help you determine how well your device can maintain a wireless connection in the presence other wireless signals that are likely to be present.
In their booth, I asked about high voltage signal measurement. They suggested their RTH series scopes with isolated channels. The probes look like traditional passive probes with the exception of having reinforced insulation and high voltage spacing between the ground ring and tip. This enables high bandwidth measurements at high voltage. The scope supports remote user interface in the case that a signal under lockout/
I attended a panel discussion on sensor fusion. This is the idea processing data from multiple sensors can provide a higher value than the value that the sum of the information from the individual sensors could give. Multiple sensors can also allow inferences to be made about an object or process that can’t be directly measured. The discussion also related to the relationship between edge and cloud computing. A helpful analogy was given of how the human spinal column will react to the sensation of burning on a finger by ordering the muscles to withdraw the finger. The brain is not involved. This is analogous edge computing. After the finger is withdrawn a person will examine it and decide how to treat it. This involves the brain and is analogous to cloud computing.
I attended two sessions given by DfR solutions personnel. One was on selecting connectors and the other was common design mistakes. It was interesting that they both mentioned the danger in using connectors with
The Jacob Beningo presentation of transitioning to RTOS based systems was helpful in understanding when an RTOS is appropriate and understanding RTOS concepts. The examples given during the presentation were from FreeRTOS. While he didn’t endorse FreeRTOS this suggests it is worth considering for a project especially considering there are supported and safety variants available and there is some level of integration with the STM32 Cube library.
The presentation on cyber securing medical devices by Battelle seems much like the effort required for UL 61508 certification. There is not a hard requirement for passing certification. It is more about being able to present a credible argument on how your product meets the safety thresholds. For cyber securing devices the FDA has heightened concern over how firmware updates are handled. To model threats, the presenter suggested assembling a data flow diagram and applying a threat assessment score to the points where the data is exposed. OWASP and NIST800-53 are resources for modeling and STRIDE and CVSS are scoring systems.