Related Content
Contact Information
Viewpoint Systems, Inc.
800 West Metro Park
Rochester, NY 14623
Phone: 585.475.9555
Fax: 585.475.9645
Viewpoint Data Management, LLC.
800 West Metro Park
Rochester, NY 14623
Phone: 585.475.9555
Fax: 585.475.9645
Viewpoint News, April 2011
Part 1: Embedded Systems for New Product Prototyping
by James Campbell - jac@viewpointusa.com
Suppose you have a new idea for a product or even an incremental improvement to an existing one. This product needs to measure and control and possibly have a user interface. You also need to put this new product through some paces to see if the concept works and to work out the kinks. You need prototyping!
How do you decide whether to spin your own custom board or use components off-the-shelf (build versus buy)? What about packaging, communications, and user interface hardware? In our experience, there are many options to consider.
Also, you should also know that NIST sponsors local Manufacturing Extension Programs (MEPs) to provide assistance on Product Prototyping. In the Rochester area, High Tech Rochester (HTR) is the local MEP. Check out http://www.htr.org/resources_mep.asp. Viewpoint works with Ed Sullivan or Robert Hutchison at HTR.
Considerations for Product Prototyping
Products use mechanical parts, electrical parts, sensors, batteries, user interfaces, and many other components. For the purpose of prototyping a new product, we just need a functional equivalent. Usually that means we can take components off-the-shelf (COTS) and add the proprietary part. I want to focus on the measurement and control components in this newsletter.
In case you didn’t know, Viewpoint has helped many companies convert ideas to prototypes and these prototypes into platforms for initial product runs. We usually use a small form-factor controller with appropriate I/O. Also, our custom electronics and instrumentation team helps by including packaging, sensor/actuator conditioning, communications connectivity, and user interface.
Here’s a typical flow (at a super high level) which will help define some terms.
Number of Units versus Cost of Development
For prototypes, you should consider avoiding the design and build of a custom printed circuit board (PCB) for the controller, signal I/O, communications, and so on. That route could take a lot of effort for just the few units needed for proving your idea. Also, you might want to be able to sell the first few tens of units with minimal development investment. It makes sense to do some cost analysis to locate the breakpoint between COTS and custom. NI discusses this “buy versus build” point as well at http://zone.ni.com/devzone/cda/tut/p/id/6083. In our experience, if you are expecting to build 1000s of units per year, a custom PCB is likely the better choice. Even then, do the proof of concept and the first few prototypes with COTS. You might find that you will stay with COTS past 1000s of units per year because of the flexibility and rapid development.
Prototype Controller Choice
Ultimately, the product will have a microcontroller (or microprocessor). But, depending on how early you are in your design and proof of concept phase, you might consider using a desktop PC with LabVIEW or a PXI real-time solution with LabVIEW RT to act as the microcontroller. Some points to consider at this stage are size constraints, which might require a smaller controller, or the signal-to-noise ratio, which might require the controller to be closer to the sensors. Let’s focus on the options for a small COTS controller, especially one with signal conditioning options.
NI RIO Platform
The RIO platform from NI comes in two basic flavors: CompactRIO (or cRIO) and sbRIO. Each has a controller and I/O module selections based on the C-series modules. The sbRIO is a bare PCB into which the I/O modules connect. The ‘sb’ stands for Single Board. Check out www.ni.com/embedded and look for the cRIO and sbRIO icons.
For software development tools, these platforms need LabVIEW Real-Time (LVRT) and possibly LabVIEW FPGA (LVFPGA). Viewpoint’s experience shows that it is not as simple to program an application as using LVRT and/or LVFPGA LV on a PC-based platform. The old (20 year?) adage about trading off memory versus speed is back when using these non-PC platforms. What goes around comes around, I suppose.
Microchip Platform
When custom PCBs are the best option, we typically use traditional microcontroller chips. The microcontroller market is very mature, but also mysterious for many people interested in the stages of proof of concept and prototyping. That’s why we typically start off considering the NI way. But, Viewpoint also has capabilities with the traditional microcontroller tools.
For example, we use PIC chips, made by Microchip, a leader in embedded control. They have a broad range of products from tiny, very simple controllers to 32-bit controllers that are fast and full of features. They further subdivide their product line by power requirements from very low, such as the new XLP line that can run on batteries for up to 20 years, to full featured controllers that are fast and integrate on-chip peripherals. The on-chip peripherals include memory (EEPROM, RAM, and Flash), communications (CAN, Ethernet, RF, USB, I2C, SPI, and UART), timer/counters, motor control PWM channals, encoder interfaces, A/D converters and graphic/touchscreen interfaces. Microchip has many pre-assembled development kits that have a number of features already implemented, making prototyping streamlined for a number of applications.
For software development tools, the toolsets available with microchip are their in-house C and assembler compilers along with many third-party vender solutions. They have a very broad library of pre-coded tools simplifying the use of their on-chip periphals plus many algorithms for their DSP based products, software stacks for TCP/IP and Bluetooth, graphics libraries for displays, plus many more.
Conclusion
Finding controllers for proof of concept and prototyping is a lot easier than even 10 years ago. It’s easier than ever to try a product idea in a form-factor that is close to a final product and with appropriate signal conditioning. Plus, it can make good sense to use this platform for the first few tens or even hundreds of unit sales before considering the custom route. Next time we’ll do a rough cost analysis to see if we can find the point where the “buy versus build” breakpoint lies.