Have you always wanted your very own Z-Wave widget-thing-a-ma-bob-doohickey? Silicon Labs recently released the Thunderboard Z-Wave (TBZ) which is an ideal platform for building your own Z-Wave device. Officially known as the ZGM230-DK2603A, the TBZ has sensors galore, expansion headers to connect even more stuff, comes with a built-in debugger via USB-C and can be powered with a single coin cell. Totally cool! I am working on a github repo for the TBZ but right now there are three simple sample apps in Simplicity Studio to get started.
ZGM230 Z-Wave Long Range Module – +14dBm radio – 1mi LOS RF range
ARM Cortex-M33, 512/64K FLASH/RAM, UARTs, I2C, SPI, Timers, DAC/ADC and more
Built-in Segger J-Link debugger
USB-C connectivity for SerialAPI and/or debugging
RGB LED, 2 yellow LEDs, 2 pushbuttons
Hall Effect sensor
Ambient Light sensor
6-Axis Inertial sensor
1Mbyte SPI FLASH
Qwiic I2C connector
SMA connector for antenna
Coin cell, USB or external power
Firmware development support via Simplicity Studio
There are three sample applications in Simplicity Studio at the time of this writing (Aug 2022 – SDK 7.18.1);
The TBZ ships with the SerialAPI pre-programmed into it so you can use it as a Z-Wave controller right out of the box. Connect the TBZ to a Raspberry Pi or other computer to build a Z-Wave network. Use the Unify SDK to get a host controller up and running quickly or use the PC-Controller tool within Simplicity Studio for development and testing. The SwitchOnOff sample app as the name implies simply turns an LED on/off on the board via Z-Wave. This is the best application to get started as the ZGM230 chip is always awake and is easy to debug and try out. The SensorMultilevel sounds like a great app as it returns a temperature and humidity but at the moment it does not use the sensor on the TBZ and simply always returns a fixed value. SensorMultilevel shows how to develop a coin-cell powered device. Additional sample apps are expected to be available in future SDK releases but I am working on a github repo with a lot of sensor support.
Naturally a single Z-Wave Node doesn’t do much without a network. You’ll need some sort of a hub to connect to. Most of the common hubs (SmartThings, Hubitat, Home Assistant, etc) will at least let you join your widget to the network and do some basic control or status reporting. You need either a pair of TBZs or perhaps purchase the even cheaper UZB7 for the controller side and then the TBZ for the end-device. Then you have a network and can build your doohickey and talk to it over the Z-Wave radio.
Plug in the TBZ to your computer and open Simplicity Studio which will give you a list of applicable documents including the TBZ User Guide. Writing code for the TBZ definitely requires strong C programming skills. This is not a kit for an average Z-Wave user without strong programming skills. There is a steep learning curve to learn how to use the Z-Wave Application Firmware (ZAF) so only experienced programmers should take this on. I would recommend watching the Unboxing the 800 series video on the silabs web site to get started using Simplicity Studio. I hope to make a new video on the TBZ and publish the github repo so stay tuned.
Have you created a Thing-a-ma-bob using the TBZ? Let me know in the comments below!
The two Z-Wave 800 series chips from Silicon Labs have flexible GPIOs but figuring out which one is the best for which function can be challenging. There are a number of restrictions based on the function and the energy (sleep) mode you need the GPIO to operate in. Similar to my posting on the 700 series, this post will guide you to make wise decisions on which pin to use for which function.
The tables below are a compilation of several reference documents but all of the data here was manually copied out of the documents and I could have made a mistake or two. Please post a comment if you see something wrong and I’ll fix it right away.
The table below lists the pins from the most flexible to the most fixed function. There are more alternate functions than the ones listed in this table. The most commonly used alternate functions are listed here to keep the table readable. Refer to the schematics and datasheets for more details.
Port A and B are operational down to EM2, other GPIOs will retain their state but will not switch or pass inputs. Thus, use port A and B for anything special and use C and D for simple things not needed when sleeping (LEDs, enables, etc).
Only the ZG23 QFN48 pin numbers are listed in the table. The QFN48 is expected to be pin compatible with future version of the ZG23 with additional Flash/RAM so I recommend using it over the QFN40. The WSTK2 is the Pro DevKit board with the LCD on it which comes as part of the PK800 kit. There are two sets of holes labeled with Pxx numbers on them which are handy to probe with an oscilloscope. The Thunderboard Z-Wave (TBZ) also has 2 rows of holes which are ideal for probing or connecting to external devices for rapid prototyping.
Use the pins at the top of this list first as they are the most flexible
TBZ Qwiic I2C_SDA
TBZ Qwiic I2C_SCL
PC and PD are static in EM2/3
EM4WUx pins can wake up from EM4 sleep mode on a transition of the GPIO
BRD4210 and TBZ have 32KHz crystal mounted
Accurate timing while sleeping – Time CC
Trace pins for debug & code coverage
Trace is configurable for 4, 2 or 1 data pin
JTAG data in Trace Clock out
Pins below here should be used primarily for debug
Packet Trace Interface (PTI) data
RTT UART printf and Trace D0
These two SWD pins should ONLY be used for debug and programming
SWD debug clock
Pins below here are fixed function only
Not used by Z-Wave
Not used by Z-Wave
RFIO on ZGM230
Matching network to SMA
Push buttons on DevKit boards
1.0uF X8L cap (unconnected on ZGM230)
Inductor to DVDD for DCDC – 3.3V
3.3V In/Out based on mode
VDCDC on ZGM230
Highest voltage – typically battery voltage
3.3V for +20, 1.8V for +14dBm
1.8V or 3.3V but less than PAVDD
Power Supply Pins
Obviously the power supply pins are fixed function pins. The only really configurable parts to this set of pins is the voltage to apply to the IOVDD, AVDD and whether to use the on-chip DC to DC converter or not. If your device is battery powered, AVDD should be the battery voltage assuming the battery is nominally 3V (coin cells or CR123A). AVDD can be measured by the IADC in a divide by 4 mode to give an accurate voltage reading of the battery. This avoids using GPIOs and resistor dividers to measure the battery level thereby freeing up GPIOs and reducing battery drain. IOVDD should be set to whatever voltage needed by other chips on the board. Typically either 1.8 or 3.3V. The DCDC should be used in most battery powered applications unless a larger DCDC is present on the board already to power other chips.
The other configurable voltage is the RFVDD and PAVDD and the choice there depends on the radio Transmit Power you wish to use. For +14dBm PA an RF VDD are typically 1.8V. For +20dBm PAVDD must be 3.3V.
Every product has unique requirements and sources of power so I can’t enumerate all possible combinations here but follow the recommendations in the datasheets carefully. Copy the radio board or Thunderboard example schematics for most typical applications.
Debug, PTI and Trace Pins
The two Serial Wire Debug (SWD) pins (SWCLK and SWDIO) are necessary to program the chip FLASH and are the minimum required to be able to debug firmware. While it is possible to use these pins for other simple purposes like LEDs, it is best if they are used exclusively for programming/debug. These should be connected to a MiniSimplicity or other debug header.
The SWO debug pin is the next most valuable pin which can be used for debug printfs in the firmware and output to a debugging terminal. Alternatively, the UART TX and RX pins can also be used for debugging with both simple printfs and able to control the firmware using the receive side of the UART to send commands.
The two Packet Trace Interface (PTI) pins provide a “sniffer” feature for the radio. These pins are read by Simplicity Studios Network Analyzer to give a detailed view of all traffic both out of and into the radio. The main advantage of these pins is that they are exactly the received data by the radio. The Z-Wave Zniffer can also be used as a standalone sniffer thereby freeing these pins for any use. The standalone Zniffer however does not show you exactly the same traffic that the PTI pins do especially in noisy or marginal RF conditions. Thus, the PTI pins on the device provide a more accurate view of the traffic to the device under test.
The Trace pins provide additional levels of debug using the Segger J-Trace tool. These pins output compressed data that the debugger can interpret to track the exact program flow of a running program in real time. This level of debug is invaluable for debugging exceptions, interrupts, multi-tasking RTOS threads as well as tracking code coverage to ensure all firmware has been tested. Often these pins are used for other purposes that would not be necessary during firmware debug and testing. Typically LEDs or push buttons can be bypassed during trace debug. There are options to use either 4, 2 or even 1 trace data pin but each reduction in pins cuts the bandwidth and make debugging less reliable.
LFXO and EM4WU Pins
The Low Frequency Crystal Oscillator (LFXO) pins are typically connected to a 32KHz crystal to enable accurate time keeping within several seconds per day. If supporting the Time Command Class, I strongly suggest adding the 32KHz crystal. While you can rely on the LFRCO for time keeping, it can drift by as much as a minute per hour. While you can constantly get updated accurate time from the Hub every now and then, that wastes Z-Wave bandwidth and battery power. Both the Thunderboard and BRD4210 include a 32KHz crystal so you can easily compare the accuracy of each method.
Reserve the EM4WU pins for functions that need to wake the EFR32 from EM4 sleep mode. These are the ONLY pins that can wake from EM4! Note that ports PC and PD are NOT able to switch or input from peripherals while in EM2. See the datasheet and reference manual for more details.
Many of the remaining GPIOs have alternate functions too numerous for me to mention here. Refer to the datasheet for more details. Most GPIOs can have any of the digital functions routed to them via the PRS. Thus, I2C, SPI, UARTs, Timers and Counters can generally be connected to almost any GPIO but there are some limitations. Analog functions have some flexibility via the ABUS but certain pins are reserved for special functions. Hopefully these tables help you make wise choices about which pin to use for which function on your next Z-Wave product.
The new Z-Wave 800 Series silicon is now shipping in volume and fully supported by the Silicon Labs tools so it’s time to get to work designing new products! In this post I’ll describe the main advantages and the difference between the chip version (SoC) and the module. But first I want to invite everyone to watch the Tech Talk on using the new 800 series developers kit: ZWAVE-PK800A.
Unlike the 700 series, either the SiP module or the SoC can be used for either controllers or end devices. In the 700 series the EFR32ZG14 SoC is only usable on gateways and only runs the SerialAPI. The ZGM130 module is used for all end devices and can be used on gateways. Thus, the 800 series gives you more choices for which chip/module to use that best matches your product needs.
What’s the difference between 800 series Module vs. SoC?
Here’s the short list of differences:
ZGM230S SiP Module – easier to use
Integrated crystal, RF match, decoupling
Factory calibrated CTUNE
34 GPIO – 44 pin SiP 6.5×6.5mm
+14dBm Max RF Transmit power (lower battery current targeting End Devices)
More expensive unit cost but just add antenna and a few passives
23/31 GPIO – 40/48 QFN 5×5/6x6mm (48 pin compatible with a future larger flash/ram device)
+14dBm or +20dBm Z-Wave Long Range RF Tx power
Line powered devices should use +20 for additional RF Range
Lower unit cost but more companion parts, antenna and crystal calibration required
Both require an external antenna and require regulatory (FCC/CE) testing
The ZGM230S System-in-Package (SiP) Module is a superset of the EFR32ZG23 System-on-Chip (SoC). The module adds a handful of inductors and capacitors for the DC-to-DC regulator and RF matching and the 39MHz crystal which is pre-calibrated at the Silicon Labs factory. The module is easier to manufacture (fewer components and no calibration) but is limited to +14dBm transmit power in Z-Wave Long Range. Modules are more expensive due to the integration but the cost crossover is at pretty high volumes.
The ZG23 SoC is the chip inside the module. The main advantage of using the SoC is that at high volumes, it is cheaper. The SoC supports +20dBm Z-Wave Long Range transmit power which can nearly double the radio range over the module. But +20dBm demands a lot of battery power so it typically cannot be powered with coin cells but must use a CR123A or AA batteries. Getting FCC to pass at +20dBm can also be a challenge and careful matching of the antenna is required. On the factory test floor, every unit manufactured must have the 39MHz crystal calibrated. Details of the calibration process are described in User Guide 522. The crystal calibration is necessary to ensure reliable radio communication and is a process that requires a spectrum analyzer and several seconds of testing. Your manufacturing partner has to be equipped and knowledgeable to properly calibrate each unit efficiently.
500 vs. 700 vs. 800 Series Comparison
Are you still working with the Z-Wave 500 series and need more details on which series to upgrade to? Fortunately we’ve put together this comparison table to answer those questions. I have to say that once you’ve written and debugged code for a modern ARM processor, you will NEVER want to use the 500 series 8051 8-bit CPU ever again!
Which Z-Wave Series to use?
In these times of long lead times and limited silicon availability, the main question of which Z-Wave chip/module to use may come down to which ones you can get! Silicon Labs keeps some inventory of all of our chips available thru our distributors Digikey, Mouser and Arrow. Each day a few hundred chips of all types are placed into inventory so anyone can buy enough to build prototypes. If there are zero available today, try again tomorrow or the next day. At this time (end of Q1 2022), we are able to supply the 500 series pretty well but the supply outlook for 2023 is uncertain. The 700 series has limited availability so if you already have orders placed and have been given allocation, you are OK. The 800 series is our most advanced process which Silicon Labs and our fabrication partners are investing in upgrading capacity so availability will improve late in 2022 and into 2023. Any new product development or upgrading of 500 series products should use the 800 series. This outlook changesliterally daily so contact your Silicon Labs sales person for the latest recommendation.
The choice of 800 series is easy – do it! The improvements and availability over the 500 and 700 series makes using the 800 series a no-brainer. So the next question is Module or SoC? That decision has to be done on a case-by-case basis as there are a lot of factors to be weighed. The first hurdle is the total unit volume you expect to purchase. If you’re not in the 100,000+ per year stage, then the recommendation is to use the module as it is simply easier to manufacture. The crystal calibration requirement for the SoC is non-trivial and demands expertise and equipment to do it properly. If your target market is not the US, then the module is also the way to go as the additional RF power isn’t available except in the US region as Z-Wave Long Range is only available in North America. I recommend you contact your local FAE to discuss your needs and we’ll help guide to the appropriate solution that balances cost vs. complexity.