Smart homes save energy right? Occupancy sensors turn off lights in vacant rooms, scenes and timers activate only the devices that are needed, only when they’re needed, and energy use feedback adjusts our behavior.
It all sounds mostly reasonable, but the additional power consumption of the devices themselves is mostly ignored. While there’s no fundamental reason for them to consume more than a couple of hundred milliwatts, many unfortunately do.
Doing the math
The protocols used (Z-Wave, Zigbee, Thread etc.) are designed to minimize energy usage, with a typical Zigbee routing device IC requiring only around [email protected] in continuous listening mode - i.e. only 25mW.
Devices controlling loads (e.g. light switches or plug-in appliance switches) also need to drive a relay when the load is on, which adds another 150mW or so. This could be eliminated through the use of latching relays, but I haven’t seem them used, presumably because it would add a couple of dollars the BOM for the additional relay cost and driving circuitry (typically an H-bridge).
Add in a 10mW indicator LED and that’s still only 35mW/185mW with the load off/on.
Switchmode power supplies at these power levels will typically only provide ~50-60% efficiency due to the large step down (375V → 4-5V) and very low currents. Assuming a 5V relay and a 55% efficient SMPS followed by an LDO for the IC, that gives net consumption of 96mW when off and 369mW when on.
MCU : [email protected] = 7.5mA@5V = 37.5mW = 68mW @ 55% efficiency
LED : [email protected] = 3.3mA@5V = 15.2mW = 28mW @ 55% efficiency
Relay : = 30.0mA@5V = 150.0mW = 273mW @ 55% efficiency
net = 369mW @ 55% efficiency
Efficiency at such small loads is typically poor
In reality, the efficiency will be lower with the relay off and higher when on. Ideally the SMPS output would also be <5V to reduce the LDO losses.
Let’s compare this with the real world.
Measuring the standby power of various smart switches
Below is a table showing the power consumption of several Z-Wave and Zigbee devices I had access to. Order is alphabetical:
Columns:
W = Watts
PF = Power factor
Rly = # relays on \
Standby power (no relays active)
| Vendor | Model | W | PF | |
|---|---|---|---|---|
Zigbee |
||||
| Aqara | 1G wall switch (E1) | QBKG40LM |
TBD | TBD |
| Aqara | 2G wall switch (E1) | QBKG41LM |
0.13 | 1.0 |
| Aqara | 3G wall switch (E1) | ZNQBKG31LM |
0.15 | 1.0 |
| Aqara | 2G wall switch (H1M) | ZNQBKG25LM |
0.37 | 0.4 |
| ????? | Rotary dimmer | TBD | TBD | |
| Zenismart | Roller blinds | ????? |
0.63 | 0.4 |
Z-Wave |
||||
| ACT | On/off plug-in unit | ZRP210 |
0.70 | 0.4 |
| ACT | 1G dimmer | ZDW230 |
TBD | TBD |
| ACT | 2G wall switch | ZRM230 |
TBD | TBD |
| Astral* | 300W 2G dimmer | LDM32S |
0.45 | 0.5 |
| Astral* | 2G wall switch | LSM12S |
TBD | TBD |
| TKBHome | On/off plug-in unit | TZ68E |
0.62 | 0.3 |
| TKBHome | 2G wall switch | TZ57 |
0.69 | 0.4 |
| Z-VIDAR | 2G in-wall switch | Z-PRL2-V01 |
0.19 | 1.0 |
Standby AND active power (1 or more relays active)
| Vendor | Model | W | PF | Rly | |
|---|---|---|---|---|---|
Zigbee |
|||||
| Aqara | 1G wall switch (E1) | QBKG40LM |
TBD | TBD | 0 |
| TBD | TBD | 1 | |||
| Aqara | 2G wall switch (E1) | QBKG41LM |
0.13 | 1.0 | 0 |
| 0.28 | 0.3 | 1 | |||
| 0.41 | 0.3 | 2 | |||
| Aqara | 3G wall switch (E1) | ZNQBKG31LM |
0.15 | 1.0 | 0 |
| 0.26 | 0.3 | 1 | |||
| 0.37 | 0.3 | 2 | |||
| 0.48 | 0.3 | 3 | |||
| Aqara | 2G wall switch (H1M) | ZNQBKG25LM |
0.37 | 0.4 | 0 |
| 0.71 | 0.4 | 1 | |||
| 1.05 | 0.5 | 2 | |||
| ????? | Rotary dimmer | TBD | TBD | 0 | |
| Zenismart | Roller blinds | 0.63 | 0.4 | 0 | |
Z-Wave |
|||||
| ACT | On/off plug-in unit | ZRP210 |
0.70 | 0.4 | 0 |
| 1.33 | 0.4 | 1 | |||
| ACT | 1G dimmer | ZDW230 |
TBD | TBD | 0 |
| ACT | 2G wall switch | ZRM230 |
TBD | TBD | 0 |
| TBD | TBD | 1 | |||
| Astral* | 300W 2G dimmer | LDM32S |
0.45 | 0.5 | 0 |
| Astral* | 2G wall switch | LSM12S |
TBD | TBD | 0 |
| TBD | TBD | 1 | |||
| TBD | TBD | 2 | |||
| TKBHome | On/off plug-in unit | TZ68E |
0.62 | 0.3 | 0 |
| 0.83 | 0.3 | 1 | |||
| TKBHome | 2G wall switch | TZ57 |
0.69 | 0.4 | 0 |
| 0.96 | 0.4 | 1 | |||
| 1.22 | 0.4 | 2 | |||
| Z-VIDAR | Dual nano switch | Z-PRL2-V01 |
0.19 | 1.0 | 0 |
| 0.53 | 0.4 | 1 | |||
| 0.85 | 0.4 | 2 |
* MK Electric
Note:
- All measured at 228-229 VAC, 50Hz.
- No significant difference was measured when in/out of the network.
- All Z-Wave devices are UK frequency (868.42 MHz).
The Aqara E1 series performs surprisingly well
At 130mW when off, and 280mW with one relay on, the Aqara E1 switch has the lowest consumption of all the devices tested.
They use the NXP JN5189 Zigbee IC, which draws just under 7mA in RX mode (10/22mA in TX @ 0/+10 dBm) at 3V.
Power is supplied via a 120/230VAC -> ~4.3VDC buck regulator, followed by a 3.3V linear regulator to the MCU. The 4.3VDC drives the relays. Assuming a draw of 7mA, that puts the SMPS efficiency at ~23% with the relays off and ~64% with one on:
IC power = 7mA * 4.3V = 30mW
Relay power = (4.3 / 125Ω) * 4.3 = 148mW
30mW / 130mW = ~23% efficiency when off
(30mW + 148mW) / 280mW = ~64% efficiency when on
The no relay consumption could be reduced a bit with a better power supply, but the only way for a material improvement with a relay active is to use a latching relay.
It’s a good design.
Unfortunately its firmware is lacking, failing to implement basic Zigbee endpoint/binding functionality. With good hardware and poor software, I ended up writing my own customer firmware for the E1(post coming soon) to address its software deficiencies.
Many devices waste power for no good reason
It’s disheartening to see designs that draw so much idle power (700mW from the TKB Home TZ57 unit for example). I doubt there are any cost savings. More likely just laziness during design due to a lack of incentive: 99% of customers have no idea and no ability to measure it. Regulations (in the EU at least) demand a maximum of 0.5W, but it’s not clear how much enforcement goes on.
Perhaps the Zigbee alliance could consider adding power consumption to their certification requirements.