Microcontroller Tips

Microcontroller engineering resources, new microcontroller products and electronics engineering news

basics

Microcontroller power source considerations for Arduino

By 1 Comment

Powering an Arduino has a little magic to it. It wasn’t obvious to me when I first started working with them, but Arduinos have on-board regulation. Taking advantage of this allows for longer power supply wiring runs by using a higher voltage power supply than the nominal 5V or 3.3V needed by the microcontroller (MCU) for logic levels. Some Arduinos accept 6Vdc to 16 Vdc of input voltage, which is well above the maximum rating of the MCU, but the Arduino board regulates the supply voltage accurately as well as additional power for Arduino peripherals.  I’ve seen veteran engineers baffled as to the reasoning behind powering a 3.3V Arduino with a 9Vdc power supply until it’s been explained to them.

Microcontroller power requirements

Selecting an appropriate power source for microcontroller-based designs is often overlooked.  While concentrated effort and forethought may be in the details of the design itself, many performance and reliability problems can be traced back to the selection and connectivity of the power source.  The Arduino family of development boards provides solutions for these problems, but it’s easy to get it wrong without a good understanding of what options are available when you’re designing it.  It’s not as simple as saying that a 5V Arduino uses a 5Vdc power supply and a 3.3V Arduino uses a 3.3 Vdc power supply.

Arduino power requirements

Many Arduinos use the ATmega328P microcontroller.  Microchip’s ATmega328 has a wide range of acceptable Vcc voltages. (Vcc is the regulated DC supply voltage needed to operate an IC and is often referred to as a supply voltage for ICs.) Most commonly, Arduinos are designed to work at either a 3.3 V level logic for low power consumption or 5 V logic to be compatible with legacy TTL logic devices.  The examples provided below refer to 3.3 Vdc devices where the power source considerations are more critical. However, the same principles apply to 5 Vdc devices.

Figure 1:  Arduino board specifications. (Source: www.arduino.cc)

Example: Arduino Pro Mini

First, let’s assume a circuit design that uses something like the Arduino Pro Mini. The maximum current draw of the Arduino is 200mA.  It’s unlikely that the Arduino itself will draw 200mA, but let’s assume that between the Arduino and the other devices connected to it, they draw a total of 200mA. The ATmega328p datasheet shows that the minimum voltage of a logic high at a pin is 90% of Vcc. Thus,  if Vcc is 3.3 Vdc, the minimum voltage at a pin that will be considered a logic high is 0.9 * 3.3 Vdc = 2.97 Vdc.  Any value seen at a digital pin below 2.97 V is in an indeterminate range and will cause unpredictable results from the Arduino.

There is always some distance between a power source and an Arduino. The greater the distance, the greater the voltage loss across the power supply’s wiring. But how much is lost? Since 26 AWG is a common choice for low power circuit wiring and it’s on the smaller end of the range of wire gauges, there is less copper. Less copper means lower cost. Stranded 26 AWG is a good choice due to the flexibility of routing the wires. The 26 AWG is large enough such that it’s rated for carrying as much a 2.2Amps for chassis wiring, which is more than ten times the 200mA current draw we’ve specified for the Arduino’s maximum current draw for our design.  A 3.3 Vdc power supply and 26 AWG seem like a great choice, but let’s take a closer look.

Power source line loss

Quality 26 AWG wire has a resistance of 40.81 ohms per 1000 feet or 40.81 milliohms per foot. With 200mA flowing through the power source’s wires, we’ll have a voltage drop per wire run as investigated below. Keep in mind that we need to run the wire from the power source to the Arduino and then back again the to the negative side of the power source. We can see that at ten feet, we’ve lost 5 percent of our 3.3 Vdc source. At 20 feet, we’ve lost almost 10 percent. This action reduces the voltage applied to the Arduino to just 4.5V; the low limit of our guaranteed maximum digital logic high voltage.

Table 1: Line Loss. Distance indicates the physical distance between the board and its power supply. (Source: Author)

Twenty feet seems like a reasonable distance for most applications.  However, we’ve only considered the resistance of the wire itself to this point.

Contact resistance

Contact resistance is often not considered or even understood. The resistance of the 26 AWG wire is 40.81 ohms per 1000 foot based on the cross-section diameter of the wire. However, every point in the wiring where we’ve placed a connection in the wiring, we have created a point where the cross section of the current path decreases, and therefore have a higher resistance point.

Mating a circular connector will only make the pin contact the barrel at a tangential point. Blade connectors produce the same reduced area across the surface. Even a screw terminal can’t match the cross-section resistance of the wire itself. Consider that any termination is susceptible to oxidation over time and increased resistance by connecting and disconnecting the wires multiple times over the life of the system. Each of these points can easily have a contact resistance of 40 milliohms. That’s right; each connection point can add the equivalent resistance of 1 foot of 26 AWG wire. With two connections landed on the Arduino and two connections at the power source, any system will have a minimum of 4 terminations. We now have a voltage source loss of 5 percent at 8 feet and 10 percent at 18 feet between the Arduino and its power source.

Table 2: Line Loss and Contact Resistance

Single supply choice – varied distances?

Therefore, with a typical power source wiring setup, we’re losing five percent of the 3.3V source voltage between the power supply and Arduino at eight feet and 10 percent at 18 feet.  In simple terms, if we were using an adjustable dc power source, we could increase the voltage to compensate for the line loss and contact resistance. However, power supplies are expensive and take up space. Typically, in embedded systems, designers try to have a common power source for multiple embedded controllers. If one controller is one foot away from the supply and the last one is 20 feet away from the power source, the designer has a delicate balancing act to keep each embedded controller within the proper range.

Arduino power source options

Arduino designs give you a way to work around the line loss and contact resistance power source problems by providing on-board regulation. However, there are several ways to power the Arduino and not all of them provide the benefit of on-board regulation:

USB Power – A USB cable is commonly used to program an Arduino through the Arduino Integrated Development Environment (IDE).  Not only does the USB cable provide diagnostics through the IDE serial monitor, but it also offers 5Vdc power to the Arduino through the USB Vcc pin. The 5V USB power is used to power a 5V Arduino directly, or it’s regulated down if it’s a 3.3Volt Arduino.

5V or 3.3V Power – The designer can apply the appropriate voltage to the 5V or 3.3V power pins of the Arduino. These pins are tied directly to the power pins of the MCU on the Arduino board. However, applying power to these pins will cause the MCU of the Arduino to be susceptible to the line loss and contact resistance loss from the power source mentioned earlier.

Vin or Raw –   Arduino may have this pin labeled either “Vin” or “RAW” depending on the Arduino Variant being used. A common mistake is to apply a 5V or 3.3V power source to this pin. The problem with this is that not only do you have the line loss and contact resistance loss mentioned earlier but that this pin is the input to the on-board regulation circuit. Like any voltage regulator, you need to supply slightly more voltage into the device than you expect to get out of it. If we applied 3.3 Vdc to Vin, we would lose about 0.5 volts through the regulator. That means the microprocessor and connected peripheral devices would only be running on 2.8 Vdc, at best. Combined with our noted line loss and contact resistance loss, we can be running well below our required voltage level.

Table 3:  Line Loss, Contact Resistance, and Regulator Loss

VIN pin, properly used

Despite the problem described above, using the VIN or RAW pins are the solution to the source voltage loss. On the Arduino board, the  VIN or RAW pins are the input to the voltage regulator on the Arduino board.  All we need to do is apply a voltage in the specified range to get the desired regulated output to the Arduino. A supply voltage of 6 Vdc to 12 Vdc applied to Vin, or RAW will power the Arduino’s microcontroller, overcome any line or contact resistance voltage loss, and provide power output to the 5V and 3.3V pins of the Arduino to power peripheral components. The range of the Arduino’s input voltage is based on the voltage requirements on the entire board, including the energy required for the MCU to powering peripherals.

Conclusion

Off-the-shelf power supplies within the range of 7 Vdc to 12 Vdc are not as common as 3.3Vdc or 5Vdc power supplies, but they are available. It’s tempting to use the more common 5 Vdc and 3.3 Vdc power supplies for Arduino boards, but from the facts presented above, it’s necessary to use the less common alternatives for the best regulation and microcontroller performance.

8-bits and counting: 8-bit MCUs are still going strong

By Leave a Comment

Microcontroller units (MCUs) never stop evolving; a more functional, featured MCU follows every model. Embedded systems have grown with continually growing MCU technology, yet there are many traditional MCUs that are still selling in high volumes. Atmel 8051 is one of those legendary MCUs which have developed a considerable demand in the market.

The Atmel 8051 is one of the bestselling descendants of the 8-bit Intel MCS-51.  Four decades ago, Intel introduced an 8-bit MCU with limited features, the MCS-51. In just a few years after launch, the 8051 design was leading the embedded market and multiple manufacturers introduced their own versions of the 8051. For several years the 8051 MCUs offered appealing functions and features that increased growth for this architecture as more designs piled up for a multitude of end products.

Back in the 1980s, there were very few MCUs with comparable features to the 8051 architecture as the best option for embedded products. Decades later, what’s keeping the 8051 alive?

8-bit MCUs
Figure 1: Intel® P8051 microcontroller. (Source: en.wikipedia.org)

Today’s 8051 MCU applications

Continual improvements to 8-bit MCU architectures help them retain MCU unit sales volumes, even with cost-competitive 32-bit MCUs. Today’s 8-bit MCUs are still landing new and legacy design-in sockets that require lower prices, ultra-low power consumption, and small physical form factors. For example, an 8-bit (one byte) wide processor consumes less power than 16- or 32-bit MCUs. The latest Cortex M7 based 32-bit MCUs can process at speeds of hundreds of MHz, yet the use of traditional 8-bit MCUs are not declining in the market.

Many commercial product vendors still trust the 8051 MCU for various reasons, the most prominent being low cost and low power consumption. Despite the more limited features and old architecture, the 8051 is still used in some wireless communication products. The latest HC-10 Bluetooth modules are based on 8051 core. Although 16-bit or 32-bit MCUs can replace the functions offered by 8051s, the 8-bit MCU is sufficient for many product developers. Another reason for retaining 8-bit MCUs is that 8-bit MCUs are easier to program and troubleshoot than 16- and 32-bit MCUs because there’s physically fewer bits to address, debug, and investigate.

Examples of 8051 MCU-based devices include Bluetooth gadgets,[i] wireless transceivers, used inside the communication system of wireless power meters, and more. [ii] No doubt the modern transceivers based on 32-bit cores are exceedingly effective, but 8051 MCUs are suitable for low-cost products, especially for products where one cent in savings per unit makes a huge difference at high volumes.[iii]

Apart from commercial products, 8051 MCUs are still actively used in academic activities, as well.  Many academic institutions still prefer the architecture and programming of the 8051 for teaching basic embedded systems and relevant topics. For example, it’s much easier to wrap your brain around 8-bit constructs than 16- or 32-bits in terms of detailed study of operations, memory, addressing, and so forth. Project development boards are widely used in coaching activities, which is why 8051-based development kits are still in demand today.

Aside from 8051 MCUs, many other 8-bit MCUs have made their way into commercial products. Microchip is undeniably the largest supplier of 8-bit MCUs. Microchip’s low-cost 8-bit chips are accompanied by free tools, which also buoys 8-bit MCUs as a low-cost option. An 8-bit MCU can accomplish more than would be expected. For example, in 2014 OPU SAT was launched into space with a Microchip’s 8-bit PIC16F877 MCU controlling its communication system.

8-bit MCUs used in satellites

Figure 2: System block diagram of OPU Sat. (Source: directory.eoportal.org)

Even though 8-bit MCUs have made their way into space, wider-bit MCUs are preferable for areas demanding a higher level of processing. The OPU SAT also contained a Microchip 16-bit PIC24FJ256GA110 MCU in its command and data handling unit.

The PIC24FJ256GA110 MCU is somewhat famous in satellites as it has been used in other satellites as well. The PIC24FJ256GA110 MCU was also used in the Aeneas 3u Satellite as a flight processor. [iv]

8-bit MCUs can clock at decently high frequencies considering that many applications don’t require massive amounts of rapid calculations. Some applications only need to check a status once a minute, for instance. Thus, 8-bit MCUs are more than enough for many applications. The central processing unit, in many cases, is a minuscule part of most devices with multiple peripherals. The Atmel 8051 MCU has certain limitations for its peripherals as compared to 32-bit MCUs. However, there are many applications for MCUs where the peripherals are enough for developing products that do not involve intense, high-speed processing.

Several general-purpose input/output (GPIO) lines, Pulse Width Modulation (PWM), and some flash memory can be suitable for small-scale projects. However, as a significant drawback, the 8051 MCU does not support Analog to Digital Conversion (ADC), and an external ADC has to be connected to it for reading analog signals. Nevertheless, there are many 8-bit MCUs with an integrated ADC, such as Microchip’s ATtiny202/402 AVR MCUs. However, 32-bit MCUs with ADCs typically have multiple ADC channels on the chip. Some 8-bit MCUs have Digital to Analog Converters (DACs) on-chip. Nevertheless, the cost-cut with 8051 MCU as compared to 32-bit MCUs enables low-cost production.

Memory is also an area of difference between 8-bit and higher-bit MCUs. For instance, 64 KB of flash memory in an 8051 MCU is insufficient for detailed instructions whereas 32-bit MCUs like the ATSAMS70Q20B MCU have 2 MB of flash memory. Not every MCU project requires flash memory as high as 2 MB.

In the end, it’s true that most 8-bit MCUs are no match to 32-bit MCUs in terms of speed, processing capability, support for numerous peripherals, and memory, but 8-bit MCUs are still more suitable for devices where extensive features and high processing speed are not essential.

Communication protocols

Traditional MCUs have limited communication protocols, but over time, MCUs have evolved. Multiple communication protocols are necessary for extensive circuits where an MCU has to communicate with other devices. For example, the Microchip ATSAMS70Q20B MCU supports Master Controller Area Network (MCAN), Serial Peripheral Interface (SPI), I²C (Inter-Integrated Circuit), Ethernet, Universal Serial Bus (USB) and the Universal Asynchronous Receiver-transmitter (UART). However, the 8051 MCU supports limited communication protocols which restrains its use in areas where multiple communication protocols are required. Thus, for limited communication protocols, 8051 MCU can serve the purpose.

Power consumption

When choosing an MCU, special attention is most often paid towards the power budget and requirements. MCUs consume considerable power during operation, so several levels of idle and sleep states can make a difference during inactivity. The latest MCUs can sleep during inactivity which saves considerable power in the long term. Traditional 8-bit MCUs such as 8051 may not have any such features. Sleep states during inactivity can make a large impact on power consumption, so older MCUs consume more power than modern MCUs with otherwise equivalent features.

8-bit MCUs still going strong at Microchip

Microchip is still producing new 8-bit MCUs, generating a significant amount of revenue from 8-bit MCUs. Silicon Labs also releases new 8-bit MCUs. 8-bit MCUs are still used in plenty of products in legacy products and in new designs. 8-bit MCUs tend to be easier to program and understand on a deep level compared to 32-bit MCUs and are not likely to go away as long as an 8-bit MCU costs less than an equivalent 32-bit MCU.

[i] https://www.alibaba.com/product-detail/High-performance-and-low-power-MCU_60391542233.html?spm=a2700.7724838.2017115.93.106b4e3dyptiH4

[ii] http://www.ti.com/product/CC2511

[iii] https://www.alibaba.com/product-detail/Good-Quality-Integrated-Power-meter-with_60770229988.html?spm=a2700.galleryofferlist.normalList.129.60741e813LizGS

[iv] https://earth.esa.int/web/eoportal/satellite-missions/a/aeneas

The complexities of cache

By 1 Comment

Traditionally a cache is a safe place to keep things hidden until you can retrieve them. However, a cache within computer terminology is any storage that is managed to take advantage of location for rapid access. Cache is local memory that exists in the data path between the processor and main memory. A cache will hold a collection of data that has been recently referenced and has a high probability of being requested by the processor again. Nearly every modern general-use processor, from desktop computers to servers and embedded processors, makes use of a cache. Cache is local to the processor, i.e., it is located on the same chip as the processor. Cache is used to speed up processing because off-chip memory takes more processor cycles to access than does cache. But in the race for ever-faster computing, a cache isn’t as simple as a memory bank for stashing data the processor might need next.

In order to understand cache better, it’s important to understand the term “memory hierarchy.” Memory hierarchy differentiates computing storage based on the amount of time it takes to interact with a type of storage.

cache
Figure 1: The memory/storage hierarchy of a computer. Cache is shown as one layer, but modern caches may have several levels of hierarchy within the cache itself.

A direct-mapped cache maps each memory location to one location in the cache. Each cache memory is tagged with address information that corresponds to the address of the data in main memory. If a cache location is not loaded with good data, perhaps just after boot up, it will have a “valid bit” that shows whether or not the cache entry contains a valid (working) address. A valid bit is a field in the tables of the memory hierarchy that indicates that the associated block in the hierarchy contains valid data. Therefore, a cache location may be occupied with nothing or junk, but the valid bit shows whether or not the cache can be overwritten with the valid data. Reading from a cache is faster than writing to the cache, as with any type of memory, because reads do not change the contents of a memory location. Very large caches can create their own problem of increased latency, defeating the purpose of the cache to begin with. To defeat this issue, caches can have several levels, called a multi-level cache. In a multi-level cache, the most expedient data is stored in the cache at the top of the cache hierarchy, lower-level data, in terms of relevancy to the processor’s immediate task, is stored in the second level, and so forth. It’s common for caches to have up to four levels (i.e., an L4 cache), but of course, more or fewer levels can be used. The highest level of cache, before accessing external memory lower on the hierarchy, is called the last level cache (LLC). A question arises about what happens with caches for multi-core chips. Multi-core chips can have a common shared cache for all cores where global variables might be stored at the highest level (L3 or L4, for example) but a dedicated L1 cache is local to each processor since sharing an L1 cache across several processor cores would just add to the complexity and latency.

What happens if the cache does not have the required data? When the cache does not include required data in cache memory, it’s termed a ”cache miss,” and the data must be fetched from lower down on the hierarchy, perhaps from a much slower hard disk drive, which slows things down. If the processor encounters a cache miss, extra steps are required. If everything was going along swimmingly and the processor has been serviced by a full pipeline, the cache miss creates a stall in the pipeline that’s servicing the processor, and the processor experiences a stall. Whereas a processor interrupt requires all working memory registers to be saved while the processor executes a new task, the cache miss simply freezes everything, waiting for the memory fetch. Some processors get around a cache miss by processing out of order execution (OoOE). With OoOE, instructions are executed in order of data availability rather than in the order established by the software program, thus avoiding idling of the processor.

Caching data is faster because of it’s physical location and the physical nature of the media used for a cache. However, caches also employ predictive mechanisms to keep data that is most likely to be needed in the faster cache memory, versus fetching data from farther away and slower memory media. Caches are generally right 95% of the time, and yet when data is not present in the cache, the option to retrieve the data from lower down the memory hierarchy is always present. However, the prediction miss rate gets higher as larger blocks of data are required. In other words, if a cache is too small to handle the average block of data that’s require to complete a set of related tasks, then the benefits of having a cache reduce, and the processor (computer) gets slower. In the case of a desktop computer, one can increase the size of the cache manually, but this is more difficult in an embedded processor since embedded processors are typically intended to operate without human intervention. Some elaborate schemes have been created to regain cache performance, but are something of a bandage to the problem.

These methods are called early restart and requested word first (or critical word first), and are, in simple terms, involve resuming execution before the transfer process is complete and depend upon favorable race conditions between the processor resumption and completion of memory fetch. In this case, if memory fetch is too slow to complete, then the processor has to wait, defeating the purpose of the cache.

 

LoRaWAN security design tips: Designing with a stand-alone security chip

By Leave a Comment

With the growth of the Internet of Things, cyber attacks are increasing at an alarming rate (Figure 1). Key provisioning, a process of key generation and device authentication, is a critical part of establishing security. Whoever holds the key will be able to access the payloads, i.e. the valuable data stored in the application server(s). While the LoRaWAN specification defines the process of key provisioning, it is up to developers to ensure that the key provisioning steps are securely carried out.

During the manufacturing process, the end devices are assigned two root keys as specified by the LoRaWAN specification v1.1 (NwkKey and AppKey ). When the end devices attempt to connect with the application server via the network server, by sending the root key related information over, there’s no guarantee that this process is secure. In some cases, the information is just written down by the contract manufacturer’s employees on a piece of paper. The real risk is that if a hacker gets ahold of the root key by either copying or re-keying, the system security will be compromised.

stand-alone security chip
Figure 1: Cyber attacks are increasing at an alarming rate. (Image source: Microchip)

A security chip comes to the rescue

Last February, Microchip Technology introduced a single-chip (ATECC608A) solution where a separately downloaded, one-of-a-kind code is paired with a chip that has a unique ID. The security chip is an 8-pin, low-power IC which draws 150nA in sleep mode. As shown in Figure 2, ATECC608A is assigned a unique ID when it is shipped to the customer/manufacturer. Additionally, a manifest file can be downloaded by the customer. To further increase security, the customer obtains the chip and the manifest file via different channels.

Figure 2: The authentication process using stand-alone security and third-party Join Server. (Image source: Microchip).

How does the authentication process work?

As shown in Figure 2, there are multiple steps involved. Step 1 is self-explanatory. Steps 2 and 3 show the two important components that the customer will obtain to achieve security; the chip and the manifest file, as explained earlier. Step 3 is the most important. At this point, the customer will upload the manifest file information to the Join Server. (See footnote 1). In this case, the Join Server is supported by a third party, The Things Industries (TTI). Note that the manifest file contains all the needed information, including the unique ID of the device. Prior to this step, TTI has already received the chip information from Microchip. Any information uploaded by the customer can be verified by TTI. Once verified, true authentication is complete, as shown in Step 4.

What if the chip is stolen?

There’s always the possibility that the chip could be stolen and fake devices built. However, TTI has the chip identity on record, so if a manifest file is missing or a fake ID is created, the authentication will fail and the request to join will be rejected.

Using the combination of a security IC chip and a manifest file protects the true identity of the end device. The four-step process allows the customers/users to achieve a high level of security. Even if the security IC chip is stolen, hackers still cannot connect with the application server or join server to steal data. Finally, one other advantage of using the manifest file is its scalability. Tens of thousands of security ID’s can be efficiently uploaded to the join servers all at once. 

Footnote 1: The LoRaWAN specification calls out the application, network, and join servers. The network server is operated by the network provider. The application is operated by the end user. Alternately, the end user may choose not to own but to lease the application server from the network provider. Finally, the join server is typically operated by a third party independent of the network provider to increase security. In this case, the network provider will not have access to the security keys.

When asked the impossible, ask questions before proceeding

By Leave a Comment

Sometimes a client will ask for the impossible and expect it at low cost. Or worse yet, the sales engineer in your company will win a project on a low bid because he/she thought it could be accomplished in a “creative” (but impossible) way.

What if you were tasked with a wirelessly communicating embedded project whose electronics, circuit, components, PCB, and all, were required to fit inside a furnace? Typically, a thermocouple or RTD would be placed inside a furnace and wiring would take the sensor signal into a box that holds the electronics. But what if you were asked to create a project where the sensing devices, a 7-segment LED display, and wireless communication were all placed inside the furnace?

Figure 1: The planet Venus has been too hot to explore, because devices would burn up upon approach. (Source: nasa.gov)

Yeah, sometimes engineers get asked to do the atypical. In some cases, the challenge is met with creativity and innovation. And sometimes you must admit that the request can’t be done with today’s technology. With decades of experience, this would be my train of thought if asked to complete such a project.

Think at a higher level: what’s the real goal?

What is the objective here? Is to measure furnace temperature or build an embedded design that can withstand 125°C continuously? So many thoughts come to mind when I read this headline beyond just finding a technical solution for the chosen path. The first engineer I ever worked for gave me several pieces of advice I still practice today. Much of the advice can apply to the headline above.

Like me, you might have heard, “There must be a dozen ways to accomplish the same goal and more than a thousand ways to get it wrong.” Measuring and data logging temperatures well above the operating temperature of the electronic control devices is not a unique problem. Metal smelting furnaces and glass kilns operate between 1,000°C to 1,500°C. The material and walls of these ovens appear bright red at these temperatures! Why limit the suggested design to just 125°C? Why not make it 1,500°C? Obviously, because the design would melt in second; like Arnold Schwarzenegger in the finality of “Terminator 2”.

MIL-SPEC components are available at 125°C. These are considered high-reliability parts since they are designed, tested, and certified to operate continuously at 125°C. Component qualifications are submitted to life tests that subject sample lots to full power at a 125°C ambient temperature for 10,000 hours, perhaps cycling 1 ½ hours on and ½ hour off. The power cycling is supposed to further stress the expansion and contraction of the internal component materials. However, MIL-SPEC components are not actually used at 125°C continuously in practical designs. Rather, the fact that the components will operate at an acceptable failure rate at the extended temperatures, they are even more reliable at lower ambient temperatures.

Just because some electrical components are available at a 125°C rating doesn’t mean that placing them inside the oven they intended to measure is the only solution. Obviously, problems quickly arise trying to source all the components that can withstand the continuous temperature.

If someone asked me to do this project, I would take a step back and look at how others have accomplished this goal. What’s wrong with that? A more traditional approach would be to mount the thermocouples inside the furnace and wire them to the outside of the furnace. Depending on the design requirements, the thermocouple might be just insulted with high-temperature insulation or enclosed in a metal sheath. The thermocouple wiring would be long enough to reach the embedded control or monitoring electronics that are mounted in a more reasonable temperate environment.

The engineer must also consider that there is not always a direct commercial-to-military equivalent for all components. In fact, due to the time it takes to qualify high-reliability components for life testing, MIL-SPEC components often lag commercially available components by years. Many commercial components just don’t have an equivalent due to the demand and cost that it takes to qualify these components. In military, aerospace, and avionics applications, the reliability factor is more important than the latest technology.

What about the mechanics that are related to an embedded design?

Modifying the mechanics of the furnace for wiring access or mounting sheathed temperature probes might quickly get away from the skills and abilities of the electrical engineer who is designing the embedded system. People don’t want to be wrong, especially engineers, and tend to stick with the familiar. An electrical engineer will tend to provide an electrical solution. A mechanical engineer will tend to offer a mechanical solution. A software engineer will give you a software solution but blame the electrical engineer or mechanical engineer if their solution doesn’t work as expected (insert lopsided grin; no one is perfect).

However, the tongue-in-cheek statements above don’t mean that the electrical engineer shouldn’t seek the advice of someone qualified to change or modify the mechanics of the system. Another piece of advice from my first boss was, “You don’t have to know everything, you just need to know where to find the answer.” You got me…. these were in the days before the modern internet, so finding the answer meant more than just googling for answers. There is no substitute for talking to people and asking questions. (For which forums can be a good enough substitute, but not as good as whiteboarding with some experienced engineers.)

Although I’ve seen engineers in some companies hold tightly to their disciplines, the best designs are a balance of mechanics, electronics, and software. There are many advantages to seeking out a multidisciplined balance if you are ever presented with a tuck-it-inside-the-furnace problem. (This is not to say that I haven’t run across mechanical engineers who provide wiring access holes large enough for the wire bundle, but not the terminated connectors, too!) The approach needs to be a cooperative give-and-take of knowledge and ideas.

Who is asking for a narrow solution?

However, exchanging ideas with other engineers isn’t the only problem with the design objective. Is the client asking for the control and datalogger to be installed internal to the furnace without understanding the technical challenges and limitations? Maybe the client isn’t knowledgeable enough about what they are asking for, and the engineer isn’t confident enough to educate them and diplomatically suggest alternatives. There is a difference between doing what the client says and providing what the client needs. Again, if the realistic objective is to control and monitor furnace temperature, why is the engineer trying to mount the embedded controller inside the furnace?

If you look at the requested design, it’s asking for the entire embedded project, including the readout, to be placed inside the furnace. The question should be why you would try to source a seven segment LED display rated for 125°C? You will likely not be able to find one, because a human cannot survive reading an LED display while trapped in a 125°C oven. There is no such Commercial Off the Shelf (COTS) device; it would need to be custom made at a very high price. One application in need of such a rugged LED display might be on the planet Venus.

Better designs exist for the more fixable problems

If mounting several thermocouples externally on a furnace makes the wiring runs too long to multiplex (MUX) them at a single controller, then don’t take that approach. One possible solution could be that separate controllers are needed to read each thermocouple and then upload the data to a central controller. (Definitely cheaper than a custom 125°C tolerant display.) How you solve the problem all depends on the size and layout of the furnace that the engineer is trying to measure.

You don’t have to take the initial approach that’s requested and plow through with a suggested solution when so many others seem possible, which leads me to my final piece of advice from my old boss, “Why is there never enough time to do things right, but always enough to do things over?”Yes, the engineer who tries to put the whole project, electronics and all inside the furnace can finish a functional design, but it will never be reliable.

Taking a step back to seek more traditional solutions would not be a waste of time. Sometimes you have to take the right road and reject the unreasonable request….unless you have massive funds to commission custom devices or decades to do the R&D necessary to create a currently unavailable device.

This FAQ was inspired by a conversation on EDAboard.com. Check it out here.

Understanding single-ended and differential communication systems

By Leave a Comment

Do you know the difference between single-ended and differential communication schemes? And when should you use them, which one you should use, and why?

Single-ended communications set ups are meant to be used only for short runs or to provide communication between devices on the same Printed Circuit Board (PCB). Single-ended communications will reference the signals with respect to ground. However, as the designer attempts to move devices farther apart from each other, the signals in a single-ended communications bus become susceptible to Electromagnetic Interference (EMI). EMI causes the ground and signal levels to shift out of spec such that bit levels cannot be distinguished. However, a differential communication (or signaling) system can significantly extend the distance between wired devices.

TTL logic levels

Electronic designs rely on the interconnectivity of the components to function. When components are in close proximity of each other, such as residing on a common PCB, the designer typically connects components directly with respect to a common digital ground plane on the board. The length of the board traces is short enough where signal loss from the resistance of the traces is negligible. Digital components can be easily connected as long as the components use the same logic levels.

While there is little concern about line loss, direct connections of discrete logic components can take an excessive amount of board space to route all the necessary traces for complex devices. Consider the memory circuit shown in Figure 1. Using discrete memory integrated circuits (ICs), there are more than 120 connection points, just so we can access a measly 4 Kilobytes of memory. Imagine how many connection points there would be and the size of the PCB required to support 4 Megabytes or even 4 Gigabytes of memory using this scheme!

 

single-ended and differential communication
Figure 1: Discrete Component Memory Design. (Image:  docplayer.br)

Modern memory

The ability to multiply the memory circuit above in Figure 1 comes from improved technology and manufacturing methods that move thousands of circuit connections on board a single chip. But with all the connections moved on board a single IC, how do you access the memory? Again, improvements in manufacturing and technology allow designers to interface with the device over a serial communication system. An SD Card is just one example of a serial memory device. While the latest cards can use a more advanced serial protocol, all SD Cards support the legacy SPI interface.

Figure 2:  SD Card (Image: sandisk.com)

SPI interface

SPI stands for Serial Peripheral Interface. SPI (sometimes called “spy”) is a synchronous serial communication interface that uses only four wires for its connection to other devices. Data transfer takes place between a master device like a microcontroller and a slave device like an SD card using a serial clock (SCLK), two data lines (MOSI and MISO), and a Slave or Chip Select (SS).

Figure 3: SPI Interface connections.  (Image:  sohu.com)

The detail of the SPI protocol is beyond the scope of this article. However, the SPI example is mentioned because the communication signals are referenced from ground. Referencing communication signals from ground is also called “single-ended communication.” The characteristics of single-ended or logic level-based communications are that the connection needs to be short enough to prevent line loss and provide noise immunity from electromagnetic noise. This is why you see SD Cards used in cell phones, computers, and embedded controllers where they are designed on the same printed circuit board as the microcontroller or microprocessor they serve. The SPI interface does not lend itself to being connected remotely over a long distance from the microprocessor it supports.

I2C interface

I2C (Inter-Integrated Circuit) is another common serial communication protocol. Like SPI, I2C is a synchronous serial communication interface. “Synchronous” means that it transfers data a bit at a time along the Serial Data Line (SDA).  However, I2C requires only two wires for communication: The Serial Data Line (SDA) and the Serial Clock Line (SCL). While only two wires are needed for I2C compared to the four wires of SPI, I2C has a slightly lower transmission rate than SPI. One example of a memory chip that uses I2C is Microchip’s 24CW64. With only two pins for power and two pins for the I2C interface, this device can store 64 Kilobytes of data for a microprocessor. Just like the SPI example, the I2C EEPROM uses single ended, ground-to-Vcc digital logic levels for the I2C lines. Since memory chips are typically mounted near the microprocessor, concerns about line loss and noise immunity are low.

Figure 4: I2C Interface. (Image: https://nguyentiensk.wordpress.com)

The I2C communications bus has expanded well beyond merely serial access to memory chips. Hundreds of peripheral devices exist that use I2C for easy integration to microcontrollers:

  • Temperature Sensors
  • Accelerometers
  • Proximity Sensors
  • GPIO Expanders
  • Touch Sensors
  • Color Sensors
  • Distance Sensors
  • Driver Control
  • OLED Displays
  • Other microcontrollers
  • A-to-D Converter
  • Light Sensors

Indirectly, interfacing to these devices is just like accessing the external memory locations of the microcontroller. If it makes sense for the peripheral to be mounted on the same PCB as the microcontroller, the designer still has little concern about line loss and electromagnetic noise immunity.

Electromagnetic interference: the problem with distance

What if you need a sensor in a location that’s remote from the microcontroller? Suppose you want to measure the temperature of something that exceeds the operating temperature of the microcontroller? You would need the thermocouple and sensor to be placed at a distance from the microcontroller. In another application, suppose you want to provide feedback to the user on a remote OLED display? Even passing messages between two microcontrollers in a distributed design would require communicating over a longer distance than the restraints of a single printed circuit board.

As the distances increase between I2C devices, the communication lines become more susceptible to EMI again. The effects of EMI may not appear obvious at first, but these problems are all around us in the physical environment. One basic electromagnetic principle is that current flowing through a conductor produces a magnetic field along the axis of the conductor. In turn, a conductor in a magnetic field will produce current in the conductor (Faraday’s Law). These two principles are how transformers, motors, and generators work.

Figure 5 Current Induces Magnetic Field (Image: Wikipedia)

But how do these principles affect the digital communication systems discussed above? As the designer starts to move peripheral devices off of the PCB of the microcontroller or attempts to talk between microcontrollers separated by larger distances, the communication lines become more susceptible to the inductive coupling of other devices in the environment. The communication lines are susceptible to having undesired current coupled into them by stray magnetic fields from transformers, motors, and other electro-mechanical devices. Relays and solenoids can be particularly troublesome using digital communications, due to the high in-rush of current when they engage and disengage. The current coupled into a single-ended digital communication system can drive the digital logic levels well above and below the accepted digital logic levels of the system. Influencing the logic levels to an unknown state between a digital high and low is just as troublesome for the data integrity of the digital communication system. While you should take care to try to prevent such interference, it cannot always be avoided.

Differential communication

At the cost of some more complex circuity and design considerations, the engineer can minimize the effects of electromagnetic interference. The most significant improvement can be the utilization of differential communication drivers to replace single-ended communication over long distances.

For differential I2C, our system goes from 2 wires to 4 wires. We still only have the data line (SDL) and clock line (SCL), but we will use two wires for each signal. As the name “differential” communication suggests, the digital value of a bit is the difference in the voltage level between the two lines for each signal. The precise voltage levels of each line become less important. Rather, it is which line is more positive than the other that determines whether a digital 1 or 0 is received. The differential transceivers on each end of the communication line will convert the differential signals to the digital voltage logic levels of the microcontroller(s).

When you choose wiring for differential communication, consider keeping both wires for a single data signal as close as possible to each other. A multiconductor cable or “twisted pair” wiring is excellent for maintaining that closeness. The reason you want to choose wiring like this is so that any influence from an external electromagnetic field will influence the voltage induced in both wires, and in the same way. If the electromagnetic field creates a one-volt drop in the differential communication lines, both conductors see the same voltage drop. Even though the absolute voltage level has changed, the difference between the two wires remains the same. In this way, the integrity of the data being transmitted over the differential line is maintained.

Figure 6: Differential I2C Transmission. (Image: Wikipedia)

While you are trying to keep your single ended I2C solution within the size of a single circuit board, the differential solution can support distances of up to 10 meters at a 10K bits per second.

An example of an I2C differential driver is the PCA9615 by NXP Semiconductors uses this chip in a breakout board for prototyping and bench testing, too.

Figure 7: Spark Fun PCA9615 Break Out Board. (Image:  Sparkfun.com)

The PROs and CONs

Utilizing serial communication protocols can greatly reduce the number of connection points or pins as well as the size of your PCB design. If the design can be kept to a single PCB, single-ended communication between micro-controllers and peripherals is the easiest and most cost-effective solution. However, care must be taken if the peripherals need to be off-board or you need to communicate between microcontrollers over a significant distance. In these cases, spending the extra time and money to integrate a differential communication system is well worth the increased performance and reliability.

Cybersecurity basics: Server and end device relationship to LoRaWAN network, Part 1

By 1 Comment

Providing end-to-end security is the ultimate security goal of every wireless network. Each reported successful cyber attack serves as a reminder that achieving security is easier said than done. Let’s first examine the process.

Most of us know about using strong passwords and encryption. The most vulnerable security component is key management during device activation and authentication. As with physical properties, if you hold the key to the front door, you can get in. The same concept applies to IoT applications. It’s vital to manage the keys to ensure no intruders have access to your key copying or rekeying it. A smart city may have 20,000 smart street lights (end devices) that need to be connected to the server (controller) for the first time. And connecting 20,000 end devices through a process called device activation and authentication is no trivial task.

Simply put, authentication is a process to ensure “You are who you say you are.” A device can easily present an ID that looks real to the server. But is it? If a fake ID is accepted, the whole network is compromised and hackers can steal data information without the knowledge of the end device owners.

cybersecurity
Figure 1: The LoRaWAN block diagram shows how end devices (left) connect to the network servers (right) via gateway hosts (middle). (Image courtesy of LoRa Alliance)

How LoRaWAN network provides end-to-end security

A long-range wide area network, commonly known as LoRaWAN, is an open, standards-based and fast-growing network platform. In just a few years, it has grown from a few network providers to 100 worldwide. Like most long-range networks such as Sigfox and NB-IoT, LoRaWAN connects devices and gateways to servers. It is capable of bi-directional communication with end-to-end security, while some applications or networks only offer unidirectional communication. Typically, these low-power networks (LPWA) including Sigfox, NB-IoT, LoRaWAN, and others support devices with 5-10 years of battery life, making LPWA solutions very attractive to developers. The broad-based applications include smart homes, smart grids, infrastructure, smart farming, industrial IoT, smart cities and smart manufacturing (also known as Industry 4.0).

As shown in Figure 1, the blocks on the left represent end devices, which can be sensors or edge devices. In the middle are blocks representing a gateway and on the right, the blocks represent customer owned application servers or network servers owned by network providers. These servers may be located remotely. Other terms used by LoRaWAN include an application server, network server, and joint server.

  • An application server hosts the application software, which controls or communicates with the end devices.
  • A network server is owned and used by a network provider to forward data traffic between the application server and the end devices.
  • A joint server is an independent server owned by a third party other than the network provider. (Sometimes, the join server can be an independent server owned by the same network service provider.) This is used to keep the root key. Some users prefer this method to increase network security. With the key kept by the joint server, even though the network provider’s platform is hacked, the user’s information is secure because the hacker does not have access to the key kept outside the network provider’s domain.

In part 2 of the series, we will go over how authentication works to achieve end-to-end network security between the end devices and the application servers.

Protecting an MCU: Build your own optocoupler

By Leave a Comment

Microcontrollers are widely used for controlling transistors and relays. Transistors and relays are used for actuating or (energizing) equipment with a higher voltage than an MCU can handle. For example, fans, lights, television, door locks, furnaces, and appliances can be automated using an MCU, however, most operate at higher voltages compared to the voltage tolerances as designated in an MCU’s Input/Output (I/O) specification. Microcontrollers used in Arduinos, Microchip’s PICs, and STMicroelectronics’ MCUs usually operate in the range of 3 – 5 VDC. When high voltage equipment is controlled by (and electrically connected to) an MCU, the high voltage equipment produces back EMF headed towards the MCU’s circuit and adds noise to the signals. These noisy signals can cause circuits to malfunction or cause permanent damage.

Another major challenge is protecting sensitive circuits from the risk of receiving electrical surges. High voltage equipment, powered by a high voltage source, can initiate an electrical surge. If a sensitive circuit is exposed to an electrical surge, the components can burn up immediately. In engineering slang, this means an electrical surge can “let out the magic smoke.”

To reject back EMF, noise, and electrical surges from entering an MCU circuit, optocouplers are often used. Optocouplers create a safe connection between high voltage equipment and microcontrollers with a means of complete electrical insulation. In case the high voltage circuit induces an electrical surge, the surge remains only on the output side of the optocoupler, and the circuit at input side remains safe and unaffected, as both sides are electrically isolated. Optocouplers are also called photodiodes, optoisolators, photocouplers, photo relays, and optical isolators.

Internal view of a phototransistor-based optocoupler

Phototransistor-based optocouplers contain an  Infrared Emitting Diodes (IRED) on the input side and a phototransistor on the output side. Both sides are optically connected, and there is no electrical connection between the input and output side. When a voltage signal is passed on the input side, the IRED turns on, and the light that the IRED emits then falls on the receiving phototransistor, resulting in current flow through the phototransistor’s collector and emitter. [i]

Figure 1: Internal connection diagram of a phototransistor based optocoupler consisting of an IRED at left and a phototransistor on the right. (Source: sharp-world.com)

How to make optocouplers at home

Although many optocouplers are fairly cheap and readily available online, they can be developed at home using an LED, a light dependent resistor, and an opaque tube, perhaps made with duct tape.

Figure 2: Basic components for making an optocoupler at home. (Source: Author)

The phototransistor and LED are inserted inside the tube, facing towards each other. Tightly enclose the tube so that the phototransistor cannot be affected by an external light source. It is recommended to use an internally reflective tube, however, as a non-reflective tube might absorb LED light, affecting the efficiency of the optocoupler. This simple design can be used in hobby projects, functioning quite well as a standard optocoupler. [ii]

 

[i] http://www.cel.com/pdf/appnotes/an3020.pdf

[ii] https://www.autodesk.com/products/eagle/blog/how-an-optocoupler-works/

 

What’s new with Wi-Fi 6? Focus on high-efficiency

By 1 Comment

The first wide-spread home electrical appliance was the toaster, and 90% of homes today still have one.[i] Today, wireless connectivity to the internet is as ubiquitous as the electricity was a hundred years ago. Smartphones, tablets, wearables, laptops, televisions, Nest thermostats, Alexa, Google Home, Ring video doorbell sensors, surveillance cameras door locks, smart lighting, and anything else that qualifies as part of the Internet of Things (IoT) use Wi-Fi. Internet service via Wi-Fi is the new utility and serves an abundance of bandwidth-hungry applications. Older homes have fewer electrical outlets, and too many appliances can blow a fuse—reminiscent of our present-day bandwidth issues.

The development of the new 6th generation wireless standard, 802.11ax, is also called High-Efficiency Wireless (HEW). Wi-Fi 6, or 802.11ax, aspires to quadruple the average throughput per user over the latest 802.11ac specification. Wi-Fi 6 will expand wireless connectivity to provide a consistent and reliable data stream, or average throughput, for every user regardless of whether environments are jam-packed with users. Also, Augmented and Virtual reality, 4K video, and 8K video involve more substantial amounts of streaming data than earlier technologies expected.

Blowing breakers is a thing of the past, whereas high congestion, slow data streams, and narrow capacity are a thing of the present that will continue to cause issues for Wi-Fi dependent applications. The first Wi-Fi standard’s (802.11b) top link speed was 11 Mbps. Released in 1999, 802.11b was slower than a wired connection at the time. The IEEE 802.11ac standard was released in 2013 and finally offered wireless connectivity at a blazing (albeit theoretical) 7 Gbps.

However, now users often experience frustratingly slow data traffic in spite of faster 802.11ac connections, which is like driving a Hennessey Venom F5 (which can top 300 MPH) on congested L.A. highways at 10 MPH. And wireless signal congestion will only get worse since the average household has about ten devices connected to the Internet today and is forecasted at 50 by 2022.[ii] Indeed, few devices in homes today are connected by ethernet cable; almost everything uses a wireless connection.

However, IEEE 802.11ax is supposed to fix the congestion problem. Planned for formal release in 2019, manufacturers have already jumped the starting gun and released 802.11ax routers (which will need firmware updates if the standards change upon official release).

The 802.11ax standard enhances performance in both the 2.4 GHz and 5.0 GHz spectrum bands, theoretically delivering several gigabits per second of throughput to many more devices, all at once, over a single channel. Some of the key vehicles by which IEEE 802.11ax betters 802.11ac include multi-user Multiple Input Multiple Output (MU-MIMO), multiple access modulation, multi-frequency accommodation, Basic Service Set (BSS) coloring, and Target Wake Time (TWT). Many 802.11ax features take a page from modern cellular communications.

Wi-Fi 6, or IEEE 802.11ax, promises a maximum theoretical speed of about 10Gbps, which is 38 percent higher than what you get from 802.11ac. However, although IEEE 802.11ax provides an inconsequential increase in speed, it will be able to handle much more traffic, providing higher throughput. In an analogy to L.A. traffic, 802.11ax, were it a highway, would have 24 lanes on the highway over the present-day six lanes.

IEEE 802.11ax depends on a patchwork of several technologies, for instance, beamforming, MU-MIMO, a multiple access higher-order modulation arrangement, more effective scheduling amongst devices, and a few others.

Key features and applications

The new IEEE 802.11ax/Wi-Fi 6/High Efficiency Wireless standard:

  • is backwards compatible with older versions 802.11a/b/g/n/ac
  • increases by 4X the average throughput per user in highly populated environments such as stadiums, airports, festivals, and train stations.
  • improves traffic flow and channel access
  • improves power management, aiding devices in longer battery life
  • includes multi-user operations for downlinks and uplinks by using Orthogonal Frequency Division Multiple Access (OFDMA) technology and MU-MIMO
Figure 1: A significant change in 802.11ax is that subcarrier spacing is now one-fourth of the subcarriers’ spacing of earlier 802.11 standards, retaining existing channel bandwidths. The IEEE 802.11ax standard defines a larger FFT by four times that of prior FFTs, which multiplies the number of subcarriers. (Image: ni.com)

Multiple access modulation

A significant improvement starting with 802.11ax borrows 4G cellular technology that utilizes multiplexing to fit more users in the bandwidth of the same channel. Orthogonal Frequency Division Multiple Access (OFDMA) modulation is an improvement and an enhanced version of Orthogonal Frequency-Division Multiplexing (OFDM). OFDM was utilized in 802.11ac. OFDMA increases both efficiency and capacity by permitting some devices to inhabit the same channel bandwidth, which helps to sustain several data streams.

Based on the traffic needs of multiple users, the 802.11ax Access Point (AP) determines how to allocate a channel and assigns all available resource units. The AP might allocate the entire channel to just one user just as 802.11ac does it or the 802.11ax access point may partition the channel to provide resources to multiple users at the same time.[iii]

Figure 2: One user per channel (left) vs. multiplexing several users in one channel together using OFDMA. (Image: ni.com)

Multi-user MIMO and beamforming

Multiple Input Multiple Output (MIMO) technology is already an important Wi-Fi technology, beginning with 802.11n. The 802.11ax standard borrows MIMO from 802.11ac and uses beamforming techniques. The access point can aim simultaneous beams at different users, with each beam holding specific packets for the user it’s focussed on. The 802.11ac standard can support sending up to four multi-user MIMO signals at the same time, whereas the 802.11ax increases multi-user MIMO transmissions to eight.

BSS coloring: spatial re-use via color codes

A Basic Service Set (BSS) is a set of wireless clients associated with the same access point. In other words, an access point and devices that are connected to Wi-Fi form a BSS. New in 802.11ax is the ability to remedy the inefficient reuse of channels. Earlier 802.11 protocol used Carrier Sense Multiple Access (CSMA). CSMA is meant to avoid collisions by only transmitting if it senses that the channel isn’t being used. However, in very dense usage cases, CSMA ends up creating interference between users. CSMA does not benefit user density experienced in airports, stadiums, and train stations.

Therefore, 802.11ax improves system-level performance as well as the best use of spectrum resources in densely populated areas with a spatial reuse technique. BSS coloring inserts an identity in a transmitted Wi-Fi signal. Receiving devices see this “colored” signal and know which BSS is sending it, the strength of its signal, and a new logic sequence to optimize the process of deciding when to transmit it and when not.

BSS coloring is in the 802.11ax standard, usable by both access point and client devices, such that the new 802.11 standard can effectively diminish co-channel interference.

Saving battery with improved TWT feature

IEEE 802.11ax has a Target Wake Time (TWT) which improves the battery life of mobile devices. TWT ensures (via precise synchronization) that devices will use full power only when communicating with the access point is necessary.

IEEE 802.11ax reflects a complicated evolution of wireless communication along a growing standard based on 802.11. IEEE 802.11ax will provide 4K/8K streaming video to several users at the same time, significant power savings, and an improved means for AR/VR applications to work better in real-time (e.g., reducing lag between physical actions and a headset’s response, etc.)

It will be a breath of relief to finally have fast, dependable wireless connectivity for everyone in the house for both high definition multi-player gaming and streaming video in one room, 4K streaming movies in two or three other rooms, and twenty or so always-connected “smart home” devices. Perhaps the next problem to solve will be widening the bandwidth and lowering cost for internet pipes serving whole neighborhoods.

To learn more about 802.11ax, look at some of the following resources:

CISCON White Paper on 802.11ax (a PDF), and National Instruments’ Introduction to 802.11ax High-Efficiency Wireless.

[i] https://newatlas.com/go/4824/

[ii] https://cio.economictimes.indiatimes.com/news/internet-of-things/households-have-10-connected-devices-now-will-rise-to-50-by-2020/53765773

[iii] http://www.ni.com/white-paper/53150/en/#toc1

Choosing PCB materials to optimize applications, Part II

By 1 Comment

Have you ever wondered which type of Printed Circuit Board (PCB) materials would be worth investing in to improve signal integrity? Would your high-frequency signaling work better with a tighter, more uniform weave in the core sheet of the PCB? What’s a high-frequency signal, anyway? When should you invest in a rigid, moisture-proof PCB?

Properties of PCB materials

The electrical properties of PCB material directly affect the overall performance and durability of circuitry. Some PCB material properties that may be of concern are the following:

  • The dielectric constant (Dk) of PCB material affects impedance of the circuits. A PCB with a low and stable dielectric constant, even with changing frequencies and temperatures, is suitable for high-frequency [i] Otherwise, if a PCB is made of material with a poor Dk, transmission lines or traces on the PCB will change impedance in unpredictable ways.
  • Electromagnetic waves are created from oscillating electric and magnetic fields. These waves can be absorbed in the dielectric layer of a PCB during signal transmissions, which leads to attenuation. (Attenuation is a measure of the loss of signal strength in a transmission).
  • The measure of signal attenuation with respect to signal propagation is called Dissipation Factor (Df). A PCB material’s Df is a useful parameter to investigate when Signal Integrity (SI) is of concern. Preserving SI is a matter of combating causes of signal distortion, signal loss, crosstalk, ringing, and noise induced by a power supply. Signal strength is a major concern with respect to high-frequency signals and sensitive circuits. In many applications were the electromagnetic signals are feeble, the dissipation factor plays a vital role in preserving signal strength.[ii]
  • The roughness of the copper surface on a PCB can also add resistance to a signal or add resistance in the circuit. Rough copper traces increase the resistance because the topography of the copper surface moves the signal up and down. Surface spikes can create an increase in capacitance, as well. Smooth copper can cost more but may be worth it to preserve signal integrity at RF frequencies.[iii]
Figure 1: The current can tunnel below the surface profile and through the bulk of the conductor (image at left). The current is forced to follow every peak and trough of the surface profile increasing path length and resistance. (Image to the right). (Source: Webinar, isola-group.com)
  • PCBs can have fiberglass sheet cores that are made up of strands of fiberglass yarn and epoxy resin. The dielectric constant differs across the sheet cores and is therefore inhomogeneous. For improved uniformity of dielectrics across materials, a densely woven fiberglass style can be used for the PCB core. The pre-impregnated material around the high-speed signal layers must be used for controlling impedance and signal issues. [iv]

Isola is “a leading global material sciences company that designs, develops, manufactures and markets copper-clad laminates and dielectric prepregs used to fabricate advanced multilayer printed circuit boards (PCBs).”[v] Isola’s expertise includes knowledge on how to mitigate fiber weave effects of PCB sheet cores. Isola’s advice is to use more uniform glass (see Figure 2, below). The uniform glass weaves will have more uconsistent distributions of glass across the entire surface, which reduces variations in impedance for traces. You can use as wide a trace as possible in your layout, route the traces at an angle to the fiber warp and fill, route traces in a zig-zag fashion with respect to the fiber (see figure 2, below), and select materials that have resin and fiber dielectric constants that are as close as possible to each other. High frequency signals (>100MHz) are the most likely to be affected by non-uniform dielectric effects caused by the PCB weave. The fabric weave has a far lower impact as a non-uniform dielectric on a trace that’s carrying a signal at a lower frequency.

Figure 2: It’s possible to mitigate the dielectric effects that PCB fiber weave can produce. The weave on the left (1080) is not as uniform in construction as the weave on the right (3113). The wire shown on top of each weave is an example of how a trace should be imposed against weaves, which also improves the uniformity of the dielectric imposed by the fabric weave.
  • Note that the dielectric of a PCB fabric weave can absorb moisture if it’s exposed to liquids or wicks moisture due to humidity. Moisture absorption affects the overall performance of a PCB. Generally, the moisture absorption value of a PCB dielectric is between 0.01 to 0.2 %. [vi]
  • The ability of a PCB to bear stress is called “flexural strength.” A PCB bend test can be performed to determine a PCB’s ability to resist [vii]

This just scratches the surface of the available choices in PCBs beyond options in traces and the fabric weave that’s sandwiched (laminated) between PCB layers. However, if you are starting to hit a wall on improving a design’s signal integrity, especially with high-frequency signals (greater than 100 MHz or 10 Gbps), then you might want to begin investigating the options mentioned here. Not all PCB options are mentioned in Part I and II of this article, but it’s a start.

[i] http://www.microwavejournal.com/blogs/1-rog-blog/post/23743-selecting-pcb-materials-for-high-speed-digital-circuits
[ii]  https://intel.ly/2SPIRcC
[iii] http://www.microwavejournal.com/ext/resources/Webinars/2014/SLIDES_Isola_26feb14.pdf
[iv]  https://intel.ly/2SPIRcC
[v] https://www.isola-group.com/about-us/
[vi] https://www.mclpcb.com/pcb-material-selection-guide/
[vii] https://www.testresources.net/applications/electronic-printed-circuit-board-pcb-flexural-bend-test/