Discover ways to optimize Z-Wave IoT units for 10+ yr battery life with sensible steering on sleep modes, energy consumption, troubleshooting, and measurement instruments for Z-Wave builders.
Introduction
Wi-fi IoT units are usually powered with batteries. Thus, a key design criterion is: how lengthy will the batteries final? Z-Wave builders should steadiness battery lifetime, battery kind and performance. Easy Z-Wave sensors can run on a single coin cell for 5 to as many as ten years. Door locks want 4 AA batteries to mechanically throw the locking bolt, however the electronics nonetheless contribute considerably to battery life. Most units want at the least a two-year battery life in any other case prospects really feel they’re always altering batteries leading to a one-star assessment.
There are three parts of battery life calculation:
- Common present whereas sleeping
- Common present whereas awake
- Awake period
Battery lifetime calculation:
For instance: A tool with 3 AAA batteries has a capability of 1000mAhr, a median awake present (Iawake) of 12mA for 0.2 seconds (Tawake) with 6 wakeups per hour (Nwakeups) and a median sleep present (Isleep) of 0.030mA, you get 1000/(12*(0.2/3600)*6+0.03=29,000 hours or 3.3 years. Thus, the important thing for lengthy battery life is to attenuate the common sleep present and restrict the variety of instances the gadget wakes up. Clearly matching the dimensions of battery to the duty at hand is essential. Too small of a battery and the consumer is irritated with changing it too typically, too giant of a battery will increase the bodily measurement of the product and the associated fee.
Are You Sleeping?
Identical to any new father or mother, the important thing for an IoT gadget is to sleep each time potential. Each millisecond of “awake” time drains the battery. How does the IoT firmware developer know when the gadget is awake and when it’s asleep? Observing the gadget state generally is a problem as typically the gadget doesn’t give any indication of being in both state or worse, it is going to give the mistaken indication. One other problem is that energy consumption in a lab surroundings is likely to be fully totally different than in the true world. Each firmware engineer wants entry to high quality present measurement instruments. I’ve an inventory of inexpensive choices later on this publish.
Z-Wave Function Sorts
Function Sorts are outlined within the Z-Wave specification part 8. Z-Wave units have two Function Sorts for battery powered units:
- Listening Sleeping Finish Node – LSEN (aka FLiRS)
- Reporting Sleeping Finish Node – RSEN
LSEN units get up as soon as per second and briefly pay attention for a Wakeup Beam. The LSEN function kind is usually used for door locks and thermostats the place the consumer needs to work together with the gadget, unlock the door or change the temperature, however the change doesn’t have to occur immediately like turning on a light-weight. A one or two second latency is suitable. This mode makes use of a bit extra common energy however the capability to speak with the gadget at any time is a requirement. Sensors who ship readings greater than as soon as per hour might wish to use LSEN mode as an alternative of RSEN as a result of quicker wakeup time and the power to verify the studying at any time as an alternative of doubtless ready for hours.
RSEN mode is the bottom energy consumption mode the place the gadget will usually stay in a really low energy mode for hours. Exterior wake-up sources equivalent to a door opening or a water leak will instantly wake the gadget and start speaking, however more often than not the gadget is in deep sleep. The gateway should watch for the gadget to wake-up to verify the battery stage or change configuration which is usually greater than an hour and could possibly be 24 hours or extra.
Z-Wave has a brand new Wake On Occasion Finish Node (WOEEN) function kind which is an power harvesting gadget and not using a battery. These units have much more stringent energy necessities as they need to talk utilizing just some millijoules. Growth of this Function Kind requires quite a lot of energy evaluation and optimization past the methods I’m offering right here.
Troubleshooting
The traditional downside when debugging a battery powered gadget is isolating a excessive present draw. When debugging a brand new design, the standard downside is that the board is drawing a whole bunch of microamps (and even many milliamps!) when it ought to be beneath a couple of tens of microamps. The apparent very first thing to verify is that if the CPU is sleeping. Load the firmware right into a devkit so it’s the solely gadget drawing energy. This isolates all the opposite sensors or circuits on the board so at the least you understand the CPU is asleep. Right here’s a quick guidelines for prime present debug:
- Is the CPU in Sleep mode?
- Is the high-speed oscillator off?
- Are GPIOs floating? Unused inputs would possibly oscillate which may draw quite a lot of energy
- Are GPIOs combating pullups/pulldowns? Strive disabling pullups/pulldowns within the debugger
- Are different circuits drawing energy?
- Minimize energy traces, elevate energy pins, elevate circuit pins, take away circuits from the board
- Debug GPIOs inflicting phantom energy draw?
- Disconnect ALL debug connections – SWD/JTAG and even floor
The massive downside comes when prospects begin returning the product due to brief battery life. You could have optimized the battery life within the lab, however did you check it in the true world? One thermostat I labored on would always ship a humidity report as a result of the humidity studying would bounce backwards and forwards by one p.c with every studying. By no means noticed this within the lab nevertheless it turned fairly frequent within the area. This was simply solved by including hysteresis to the readings and requiring the humidity to alter in a single course by greater than three p.c earlier than sending a report. Fortuitously, we have been capable of OTA the firmware and clear up the issue earlier than it turned widespread.
A major problem with early FLiRS (now referred to as LSEN) door locks didn’t develop into an issue till many locks on many Z-Wave networks have been put in in a single house constructing. The unique WakeUp beam solely had the NodeID of the door lock to get up. The issue was that with many Z-Wave networks in shut proximity, each door lock having NodeID 02 (which was the case on all of the networks within the constructing) when the controller awoke their door lock, all of the door locks on all of the adjoining networks would get up as a result of they have been all NodeID 02. Battery life in lots of circumstances was just a few months as a result of gateway polling every door lock each evening which meant each door lock awoke a whole bunch of instances each evening. This was solved years in the past by including a hash of the HomeID to the WakeUp beam which just about assures solely the specified node will get up.
A very tough battery life downside occurred with a door/window sensor that in uncommon circumstances would stay awake. We needed to arrange a dozen models and watch the ability on all of them individually and when one stayed awake for greater than 60 seconds then we captured knowledge and stopped the check so we may debug the issue. Turned out the stack was overflowing, inflicting inconceivable situations as I discussed in Half 5 of the Builders Journey. Typically the gadget continued to operate, nevertheless it simply didn’t sleep inflicting the battery to die in a couple of days. Turned out the trigger was a fancy interplay of interrupts and the Z-Wave mesh rerouting on the identical time which is why we didn’t see it in lab testing. Masking all interrupts whereas servicing the sensor and a few cautious changes to the stack sizes solved the issue which was OTAed to all models within the area.
Keep in mind that every radio transmission makes use of a major quantity of energy. Hold all communication to a minimal particularly when powered with coin cells. Be sure that all potential values of configuration parameters can’t by accident trigger fast transmissions.
Instruments For Measuring Energy Consumption
Each Z-Wave firmware developer wants entry to energy measurement instruments. These instruments can vary from Silicon Labs AEM which is a part of the event package to costly laboratory instruments. Measuring the present draw versus time and some GPIOs is vital for quickly optimizing battery life. Invoice Hewlett, co-founder of Hewlett Packard, has a preferred quote: “You can not handle what you can’t measure… and what will get measured will get completed.” The record under offers some instruments I’ve used for optimizing battery life – let me know what you’ve had success with!
5-1/2 digit Digital Multi-Meter (DMM) or Bench Energy Provide
DMMs with at the least 5-1/2 digits have sufficient dynamic vary to measure milliamps to nanoamps. Fashionable DMMs include USB and will be remotely managed and document waveforms for voltage and present however usually lack extra channels for GPIOs or UARTs. DMMs are tremendous correct for very low currents which is why I usually use one to exactly measure deep sleep present. A very good DMM prices $250+ however there are multi-channel energy provides which have related capabilities for about the identical value. I typically use an affordable DRoK energy provide with an OLED show to know when the gadget is sleeping or not, however I don’t want nanoAmp accuracy.
TinyCurrent
The TinyCurrent is a cute and low cost gadget to transform present into voltage which you then connect with your oscilloscope. It is a easy opamp throughout three change selectable shunt resistors to pick out the vary. The change is the principle limitation of this gadget because the dynamic vary isn’t sufficient to cowl the milliamp awake mode to the sub microamp sleep mode transition. Because it’s a easy analog conversion of present to voltage, it’s simple to connect with a scope together with a number of GPIOs. Measuring the deep sleep present have to be completed with a DMM or different gadget however this works tremendous to shortly see when the gadget is in deep sleep or energetic modes.
Silicon Labs AEM
Silicon Labs Superior Vitality Monitor (AEM) is the first energy measurement device for Z-Wave builders. A WSTK (BRD4002) developer’s package is required to run the AEM which now you can buy by itself as an alternative of a part of a costlier package. The AEM provides 1.8-3.6v as much as 500mA with a 100kHz pattern price and a decision slightly below 1uA. There are 4 digital logic inputs in addition to the VCOM UART. The Vitality Profiler is a device inside Simplicity Studio that gives a graphical interface for the AEM. I discovered the power profiler device suffers from stability points however is succesful.
JouleScope
Joulescope is at present the trade customary for IoT growth. Whereas expensive, it’s the good device for the job. Full disclosure: Joulescope offered me with a JS220 for a earlier undertaking which I now use for each undertaking. Broad voltage vary of +/- 15V, 3A to 0.5nA decision, 300kHz bandwidth and a dependable multiplatform software program utility make the Joulescope indispensable. The JS220 has 4 digital inputs and a couple of outputs which assist correlate present draw with the firmware being executed. The newer JS320 model is now accessible with much more dynamic vary. The screenshot under reveals the dynamic vary from microamps to over 400 milliamps and two digital logic indicators indicating the state of the CPU.
Oscilloscope Present Probe
Historically, an oscilloscope present probe was used for measuring present and concurrently sampling a number of analog or digital channels. The issue with most of those probes is that they have been designed for prime present (60A or extra) and comparatively low bandwidth and are very costly, costing round a thousand {dollars} only for the probe with out the scope. They don’t have the dynamic vary to go from nanoamps to milliamps and customarily don’t have the decision under milliamps. Possibly there’s one in a drawer someplace you need to use however I wouldn’t purchase one in every of these for debugging IoT units.
Subsequent Steps
Half 8 of the Z-Wave Developer’s Journey discusses sub-GHz antennas which may make or break a Z-Wave product. The gap a Z-Wave product will be from the hub is decided by the effectivity of the antenna. I’ll talk about a couple of key components in low-cost antenna design from my expertise of getting deployed many several types of antennas. As we proceed alongside the Z-Wave Developer’s Journey, I welcome your feedback and questions. Please be at liberty to succeed in out to me instantly through e mail.
Concerning the Writer
Eric Ryherd has been on the forefront of Z-Wave innovation since 2003, starting as a guide and later serving as a Area Utility Engineer at Silicon Labs. Over the course of his profession, he has contributed to the design and growth of a variety of Z-Wave merchandise, together with sensors, distant controls, motorized window shades, and in-wall dimmers, a lot of that are in the marketplace right now.
Though he “retired” in 2022, Eric stays deeply engaged in embedded programs and Z-Wave growth via his weblog, DrZWave.weblog, and ongoing IoT consulting initiatives. He’s additionally a well-recognized face at Z-Wave Alliance Unplug Fests, the place he continuously serves because the lead coordinator, supporting interoperability and developer collaboration.
