Following our previous post on dealing with power consumption of components, we will try to explain our solution to the issue with peak power consumptions.
As stated in that previous post, this is an important characteristic of the device you are using but most of the times you will never see it on the datasheet. You will have to test and measure yourself to make sure your design is effectively dealing with that sensor.
If you are working on a battery assisted design, the effect of these peak power consumption may be considered negligible. However, this over-current consumption issue is difficult to solve in energy harvester devices in which the energy is small and the time to attain stable voltage is long.
System oscillation issues
Since we work with wireless battery-free sensors, we have constantly had to battle against these power consumption peaks. This is what we see in Farsens when connecting our passive RFID tag directly to these ultra-low power devices:
- You connect the passive tag to the device (i.e.: the tag connected to a LIS331DLH) and then power up the system with the RFID reader.
- The tag starts harvesting energy from the RF field and the voltage starts increasing.
- The device reaches a point where it starts switching-on its internal blocks.
- A peak current consumption appears due to the turning on of the device.
- The voltage falls to the point where it was when the device is off.
- There is a decrease in current consumption due to the decrease in voltage.
- Once again there is enough power to try switching-on the device.
- Again, the tag tries to switch-on the device, the device starts consuming a lot of current again and once again the voltage falls.
This loop happens continuously. The system oscillates.
Start-up circuitry solution
The way we solve this problem is by maintaining the critical part of the system isolated and without power until we reach an appropriate voltage level.
For almost all of our RFID passive systems we use a very simple architecture based in a capacitor, a voltage monitor and an NMOS transistor. The next diagram is an example of how we use this architecture.
This block diagram shows the architecture of a Farsens Kineo tag, which is a battery-free orientation sensor tag. This tag is composed of:
- FARSENS ANDY100 tag IC
- LIS3DH 3-axis accelerometer from ST Microelectronics
- MAXIM’s MAX6427 voltage monitor
- Capacitor and an NMOS transistor
The ANDY100 is the RFID energy harvester with an SPI master port. The LIS3DH is the ‘ultra-low power’ device we want to read. The MAX6427 plus the transistor are the start-up and isolation block.
You want this architecture to work as follows. You power the system by harvesting the RF energy from the reader and charge the capacitor to maintain the peak current before the sensor is switched-on.
Capacitors with big capacitance allow a bigger peak current but slow down the charge, and thus this slows down the switch-on of the entire system. At Farsens we develop specific reader software to supply as much RF power as possible to charge the capacitor before we start reading the sensor data. You can follow the posts on reader SW optimization in this blog.
Voltage monitor selection
While the capacitor is charging, the voltage monitor checks the system voltage level, “VDD”. This monitor has a very low current consumption. It has an output connected to the gate of the NMOS transistor. During the charge, the “/LBO” signal is low, so the transistor effectively stops current circulation between the sensor ground and the system ground, so the sensor is isolated and not consuming. When a specific voltage in “VDD” is reached, the “/LBO” signal increases and then the NMOS allows the circulation of the current, thus permitting the activation of the sensor.
It is important that the voltage monitor has two voltage levels. Otherwise, a current peak can produce a drop in the main voltage level (in this case VDD) making the system oscillate, as we have already discussed.
In the next diagram we show how the MAX6427–MAX6438 family voltage monitors work.
“VBATT” is the voltage to control and “VHTH” and “VLTH” are the two voltage levels. “/LBO” is the output of the voltage monitor.
In the particular case of the Kineo tag, the “/LBO” signal rises in 2.4V and lowers in 1.8V. This rank was selected because the supply voltage of the LIS3DH goes from 1.71V to 3.6V. So you ensure that at 2.4V, when the sensor starts up, it will work at a correct voltage level. At 1.8V we will deactivate the sensor, thus avoiding malfunction due to operating below the minimum voltage specified.
Other important point regarding these two voltage levels is that you can afford a voltage drop of 0.6V due to a current peak without deactivating the sensor in the moment of the wake-up.
It is important to select the appropriate voltage monitor according to the device you want to feed and the current pull the capacitor can assume.
Once the current peak is avoided, you are ready to configure the accelerometer into a low power mode and then read the acceleration data.
Takeaway from posts on peak power consumption issues
As a summary of these two posts on power consumption of components, the main ideas for solving issues with peak current consumption are:
- Take a good look at all the information provided by the datasheets, not only the main charts.
- Choose the appropriate device working mode.
- Select a good transmission data rate to minimize current consumption with a good sampling rate.
- Characterize the current consumption of your devices during their wake-ups.
- Place a capacitor on the main power line and select the appropriate value to maintain current pulls.
- Check that you have enough power to ensure the charge of the capacitor.
- In case is it needed, design a good start-up and isolation circuit.
- If all these things do not work, your device may not be the best one to be fed with a full passive energy harvester.
I hope this information helps you design your own low-power solutions. Feel free to drop your questions in the comments below and share your own experiences with us.