RF Energy Harvesting Study


Ambient RF harvesting is a simple concept that involves capturing RF with an antenna, rectifying the AC waves to DC and then using that DC to power a load. In 1964 William Brown demonstrated the principal with RF energy used to power a helicopter1. In that case the RF energy was directed energy. For ambient energy harvesting the RF energy is simply whatever RF is available in the local environment. In the figure below; the matching unit is used to maximise the energy transferred from the antenna to the rectifier, the PMM is the power management module and can comprise a voltage converter, energy storage interface and voltage threshold monitoring.

There are a number of potential sources of RF energy including broadcast radio and television and mobile phone base stations. Energy harvesting has been previously demonstrated using these sources.

For ambient RF harvesting the RF signal strength is quite small and the rectifying diodes must have a relatively small forward voltage drop. For low frequencies below 10MHz, ancient technology in the form of glass germanium diodes can be used (OC47 diodes seem to work well). For harvesting digital TV and mobile base station signals microwave Schottky diodes are typically used. Some of these diodes have very low reverse breakdown voltages (1V to 2V) so the rectifier design needs to take this into account. RF energy harvesters designed for low power situations are likely to use single or multiple stage voltage multipliers. Fewer stages means less losses but more stages means a higher output voltage. The designer needs to choose a rectifier design to optimise the various trade-offs.

Rectifier Circuit

There are various Voltage Multiplier designs from which to choose. In this single stage voltage doubler the output is taken across C2. On the negative half cycle current flows via D1 and charges C1. On the positive half cycle the voltage on C1 is in series with the supply voltage and D2 conducts charging C2 to a voltage of (2 x Vp) - (2 x Vf). (Where Vp is the AC peak voltage and Vf is the diode forward voltage drop.) For circuits where Vp >> Vf then Vf is basically irrelevant but when Vp approaches Vf then Vf becomes significant. For example, if Vp=90V and Vf=0.7V then Vout = 178.6 or approximately 2 x Vp = 180V. If Vp=0.25V and Vf=0.15V then Vout=0.2V which is less than half of 2 x Vp. When current is drawn from the voltage doubler the output voltage will drop as the charge on the capacitors is drawn off.

To understand the importance of the forward voltage drop of a diode at low voltages consider this simple half wave rectifier (see images below).

The input waveform has a peak of 0.35V, the diode has a forward voltage drop of 0.15V and the output is a pulse of (0.35-0.15V)=0.2V

Where a voltage is applied to a diode the output voltage of the diode during conduction is equivalent to the input voltage less the diode forward voltage drop. This is the reason the voltage doubler output is only 2*Vp - 2*Vf.

As Vin get smaller and Vp approaches Vf the output gets disproportionally smaller until when Vp=Vf there is no output because the diode doesn't conduct.

RF Energy Harvesting: The Reality Check

When considering powering devices with ambient RF energy a reality check needs to be performed. This is just simple maths and needs to consider some of the factors involved in transferring power from the environment to a specific device. Starting with the load, consider the questions;

  • Does the load need a fixed voltage?
  • What is the average energy the load needs?
  • Does the load draw a steady current or a large current for a short period?
  • Will the system require energy storage?
  • How much energy does the voltage converter require?
  • How much energy does the storage element lose over time?
  • What RF energy is available?
  • What efficiency can I obtain from the rectifier circuit?

Example Scenario

A microprocessor based sensor wakes up at regular intervals, takes a reading and stores that reading. Once a week it uses a wireless connection to transmit the data to a base station. The microprocessor requires 2.8 to 3.2V and the system requires energy storage to handle the current draw during the wireless transmission. The wireless and microprocessor together draws 20mA during transmission and the transmission takes 3 seconds. Energy required per annum: 20mA*3.2V*3s * 52 weeks = 10 Joules (J) per year2. Assume for this example that another 10J p.a. are used making and storing the regular sensor readings. An energy storage element is required and it must be large enough to support the 3 second transmission without the voltage dropping below the required lower voltage limit.

Consider a 50 ohm antenna 'generating' -24dBm3 of RF energy and transferring this to a voltage converter via a matched microwave rectifier. At best only 50% of the energy generated by the antenna can be transferred to the rectifier and the efficiency of the rectifier gets lower as the input gets smaller. If the efficiency of the rectifier at this point was 20% of the input power then the output of the rectifier is 10% of the power generated by the antenna.

Power output of the antenna+ rectifier assembly = 10% of -24dBm = 0.4μW. Over one year this equates to 12.6J

Assume a voltage converter is required to charge a storage element to 3.2 V and that this voltage converter has a quiescent (inactive) current of 100nA. Over a year the energy required for the energy converter to do nothing is 3.2V*100nA*3600s* 24 hours * 365 days = 10J

Assume the energy storage element (a battery, capacitor or supercapacitor) has a long term leakage current of 200nA, over a year this contributes another 20J of losses.

There are other losses to consider such as the energy used by the voltage converter when operating but this 'reality check' indicates that this application cannot use ambient RF energy harvesting.

Annual Energy Budgets
Description Value
Rectified RF energy available: +12.6J
Quiescent loses of voltage converter - 10J
Losses of storage element - 20J
Load - 20J
Total -37.4J

In this particular scenario there is a substantial gap between the requirement and the energy available. For comparison, a 1.2V AA NiMH 2400mAh rechargeable battery stores over 10,000 Joules and a 3.2V LiFePO4 AA sized 1200mAh battery stores 13800 Joules (about 690 years of energy)

To evaluate the potential for RF energy harvesting a survey of Glasgow City Central Business District was conducted.

RF Signal Strength

Radio frequency energy varies in strength over a geographic area. In the near proximity of a high powered, multi-channel TV broadcasting antenna there is more ambient RF available than there is in a remote rural area. There is a more even spread of ambient RF energy available in a large city's central business district where there is a reasonable density of cellular mobile base stations however individual base stations in a suburban area may transmit with a higher power.

The 'path loss' between a radio frequency transmitter and receiver is affected by both frequency and distance. The signal attenuation follows a logarithmic relationship with distance and the signal strength falls off very quickly near the antenna. The following figure represents the effect of the free space path loss on the signal strength of a 100kW transmitter. Two different frequencies are compared.

Using the graph below the approximate RF power expected near a base station can be estimated. For a 2600MHz 4G base station transmitting with maximum license power (65dBm), go up the left axis to 65dB, go horizontal until you intersect the 2600MHz line then go down to read the distance. By 20 metres from the base the RF power has dropped below 0dBm. If you have the distance, start on the distance axis, go up to the frequency of interest and find the path loss. To calculate the 950MHz RF power at 300 metres from a 25dBm microcell transmitter: Find 300 metres, go up until you intersect the 950MHz line and then left to the Path Loss axis. Path loss = 80 dB therefore expected RF power is 25dBm-80dB=-55dBm.

As mobile phone density increases the number of base stations increases and the power transmitted is reduced to prevent interference with nearby cells. When there is a very high density of mobile phones and base stations then ambient RF energy could actually reduce. Currently the 900MHz and 1800MHz GSM bands are being cannibalised with 3G transmission occurring within the 900MHz band and 4G transmissions in the 1800MHz band.

As broadcast technology, mobile phone technology and receiver sensitivity improves the available ambient RF energy may drop.

RF Spectrum Over Glasgow CBD

The link below displays a map of Glasgow and the signal strength for different bands can be displayed with markers. The markers indicate the availability of the signal for RF energy harvesting. The white-dots are mainly to give a context and don't indicate the availability of harvestable energy. The red pins indicate marginal harvestable energy that may, in some circumstances be useful. Orange pins are an improvement on the red. Green pins indicate those areas where RF energy harvesting should be considered viable - depending on the use-case. Clicking on a dot will give the frequency and signal strength.

The markers are indicative of the available energy but not indicative of mobile phone reception. Most phones will treat any signal strength better than -80dBm to -90dBm as '5 bars' or maximum reception.

The actual RF energy available at any one point may be more than that indicated by the marker. RF levels at any point are reported by the highest level at that point. RF energy harvesters will harvest any energy within the band covered by the antenna and rectifier design. Combining all the power available in a band.

To view the spectrum measured at a point, click on a marker pin and then click on the graph link in the pop-up information box.

It may take several seconds for data to transfer, especially for the larger data sets. Try the 2600MHz or WiFi buttons first to get a sampling. The GSM900 and several other data sets are much slower to load. The 900MHz markers give the best view of the full context or extent of this dataset.

[ Link To Maps ]


Dr Andrew Wixted,
Glasgow Caledonian University

This work funded by Stream Technologies and Innovate UK as a Knowledge Transfer Partnership with Glasgow Caledonian University

[1] W. C. Brown, 'The History of Power Transmission by Radio Waves,' in IEEE Transactions on Microwave Theory and Techniques, vol. 32, no. 9, pp. 1230-1242, Sep 1984. doi: 10.1109/TMTT.1984.1132833

[2] Joules per year is used because the numbers are easy to compare. 1W=1J per second

[3] dB is a unitless logarithmic measure. dBm is a measure of power relative to 1mW (milli Watt). 30dBm = 1W, 0dBm=1mW, -10dBm=0.1mW, -20dBm=0.01mW, -30dBm=0.001mW or 1µW etc.

LoRa is a registered trademark of Semtech Corporation.
LoRaWAN is a trademark of the LoRa Alliance.