The transmit power numbers can be confusing primarily due to confusing spectral density with total power. Here’s some clarification.
The regulatory limits typically are -41.3 dBm/MHz. This is a spectral power density which means no more than -41.3 dBm of RF energy can appear in a 1 MHz slice of bandwidth. Further, this is based on a 1 ms average detector.
For most channels, the nominal bandwidth is 499.2 MHz. If you could construct an RF signal which had the absolute maximum of -41.3 dBm/MHz spectral density across that entire channel, then it would be 499.2 times as much total energy as anyone 1 MHz slice. This adds up to -14.3 dBm total energy. This is where the -14 dBm number comes from.
But the spectral shape doesn’t perfectly fill every 1 MHz slice with the maximal energy, so you get something a bit less, and there is some energy outside the 499.2 MHz channel bandwidth as well. If you integrate all of that, you get something less than -14 dBm in total. This is where the -17 dBm number comes from, the DW1000 spectral shape can only fill the bandwidth about half the maximal power because it is simply not possible to do it perfectly using the 802.15.4a UWB signal.
Since achieving the most transmit power the regulations allow is critical to most UWB applications, Decawave provided the ability to finely adjust the transmit power so it can just be under the regulatory limit. Given that there is chip to chip, board to board, and design to design variations, Decawave provided some headroom in the adjustment to assure the regulatory limit can be reached. This is where the maximum TX power limit of -35 dBm/MHz number comes from, roughly a 6 dB headroom in the adjustment. You can make your device transmit at this higher power by changing the code to use the maximal number for TX_POWER. If you do so, then you will be most likely violating the regulations, but range will improve, of course.
For the DWM1001 module, Decawave provides a calibration of the TX_POWER for each unit when it was under factory test. Using this number provides the highest TX power for this serial number module, at least in theory. There are a number of variables which affect this, primarily temperature, which this setting does not account for. As the DW1000 chip gets hot, the transmit power drops and the receive sensitivity goes down, so range is lost. If you model the temperature effect, you can restore the lost TX power (which we do in our systems), but you can’t do anything about the lost receive sensitivity as this is just basic solid state physics.
Since the spectral density limit is done with a 1 ms average detector, any packet less than 1 ms long will not achieve the full spectral density allowed by the regulations. For example, a packet 500 us long will be 3 dB down, a packet 250 us long will be 6 dB down, even when transmitted at the same TX_POWER register setting as a 1 ms long packet. The averaging is occurring over dead air time.
This means you can increase the transmit power of the short packet so the average power over 1 ms is still under the regulatory limit. This is the “smart TX” power option. The DW1000 implements choosing higher TX_POWER numbers based on packet length in a few discrete steps. For our code, we don’t use that feature but explicit compute the best TX_POWER setting for each packet length. This helps by using the absolute maximum we can use as we find the smart TX feature a bit too “chunky” and can leave a few dB on the table, and we fight for every dB we can.
Note that you have to assure that you don’t transmit more than 1 packet in the 1 ms window of time or the smart TX feature can’t be used (or, at least, it gets more complicated). Usually this is not a major issue as most UWB systems naturally don’t transmit packets with 1 ms of each other.
Lastly, LNAs do help with receive sensitivity. For our current LNA design, we see a bit under double the range versus no LNA. This is effectively about 6-7 dB improvement, which we attribute to using a very low noise LNA design (NF about 0.5 dB). Most of our recent customer designs have included our LNA design for this reason, it is a significant increase in the service volume of a node.
Mike Ciholas, President, Ciholas, Inc
3700 Bell Road, Newburgh, IN 47630 USA
mikec@ciholas.com
+1 812 962 9408