Bluetooth 5 variations complicate PHY testing

Article By : Dorine Gurney

Bluetooth 5 is here and features a raft of improvements over the previous Bluetooth 4.2. Let's take a look at how the Bluetooth SIG is changing this ubiquitous and highly popular standard and how these changes will impact PHY-layer test and measurement.

Bluetooth 5 adds speed and flexibility to its low-energy (LE) variants. It offers twice the data throughput of 4.2, jumping from 1 Mb/s to maximum bursts of 2 Mb/s. To add versatility, bandwidth can now be decreased to achieve up to four times longer range while maintaining similar power requirements. With quadruple the range over which their devices can transmit and receive data, product designers creating home automation and security products can potentially provide coverage of an entire home, building, or locality. With these additions come new tests, particularly at the physical layer.

Bluetooth 5 more efficiently uses broadcasting channels on the increasingly crowded 2.4 GHz band with less broadcast time required for completion of tasks. These broadcasting channel improvements will enable developers to create experience-based apps that can bridge the physical and virtual worlds.

According to Bluetooth SIG, Bluetooth 5 will add significantly more capacity to advertising transmission. This means that it can convey much more information to other compatible devices without forming an actual connection. Advertising is extended to offload advertising data from the three traditional advertising channels to the full set of data channels for more frequency diversity as shown in Figure 1. The larger 255-octet data packet enables new threshold features such as asset tracking while maintaining backward compatibility with products developed for an earlier Bluetooth specification. These advertising data channels can convey information to other compatible devices without forming an actual connection, so accelerating the interaction.

Bluetooh 5 advertising chanels
Figure 1 In the 2.4-GHz band, Bluetooth 5’s advertisement channels fall between its Wi-Fi channels.

Not every application requires the same range, speed or broadcasting capability. Bluetooth 5 lets a product developer make the best choice for their implementation. With up to eight times the broadcasting message capacity over v4.2 and support for larger data packets (31-octet to 255-octet packages), the Bluetooth SIG estimates Bluetooth 5 will now be useable for Internet of Things (IoT) connections that extend beyond a room or even a house. They predict that more than 33 percent of all IoT devices will have Bluetooth in them by 2020.

Changes to Bluetooth PHY
Two variants have been added to the LE standard in Bluetooth 5. The first provides twice the symbol rate to the existing 1 Msps Low Energy standard is called LE 2M PHY (The earlier standard is now called LE 1M PHY). Both the LE 1M and LE 2M PHYs are part of what is called LE Uncoded PHY because they have no error correction coding stage in them.

The second variant is titled LE coded PHY. There are two coding schemes in the LE Coded PHY: S=8 and S=2 where S is the number of symbols per bit. In addition to Cyclic Redundancy Check (CRC) comes convolution coding followed by mapping, which increases redundancies and therefore reduces chances of error. As a result, the coded information can travel longer distances because it can be detected and corrected if needed. Table 1 summarizes the different modulation and coding schemes and resulting data rates.


Table 1 Bluetooth 5 PHY modulation and coding schemes and the resulting data rates

Figure 2 and Figure 3 show how the data payload is processed differently for Low Energy Coded and Uncoded. CRC generation and whitening are common steps. For LE Coded PHY, however, the payload goes through Forward Error Correction (FEC) and pattern mapping. The convolutional FEC encoder uses a non-systematic, non-recursive rate ½ code with constraint length K=4. Two output bits are generated out of the coder for each input bit and it goes through a convolution FEC encoder. Both output bits (out of the encoder) are mapped further. If S=2, they don’t go through any changes, but for S=8, 0 is mapped to 0011 and 1 is mapped to 1100. That is how 8 bits are created for every single input bit in LE Coded PHY S=8.


Figure 2
Payload bit processes for LE Uncoded PHYs.

Figure 3 Bit stream processing for LE Coded PHYs add steps not required in LE Uncoded PHYs.

The packet format defined for the LE Coded PHY is used for both advertising channel packets and data channel packets. The entire packet is transmitted with 1 Msymbol/s symbol rate. Each packet consists of the Preamble, FEC block 1, and FEC block 2 as shown in Figure 4.


Figure 4 Contents of Bluetooth 5 LE coded packets.

The Preamble is not coded. FEC block 1 consists of three fields: Access Address, Coding Indicator (CI), and TERM1. blocks use the S=8 coding scheme and always have the same number of final symbols.

The CI field determines which coding scheme is used for FEC block 2. The FEC block 2 consists of three fields: PDU, CRC, and TERM2. These use either the S=2 or S=8 coding scheme, depending on the value of the CI field. CI field is just a two-bit field that differentiates between S = 2 and S=8 schemes.

The Protocol Data Unit (PDU) can have a length varying from 2 octets to 256 octets. The smallest packet length therefore is 462 µs (if you add all values in the last row for S=2 with PDU being only 2 bytes of 16 bits) and the max packet length is 17,040 µs (obtained by S=8 with PDU being 257 octets).

Testing Bluetooth 5
There are a number of measurements required to test devices to Bluetooth compliance on the transmit side as detailed below. These tests can be performed using a mid-level performance spectrum analyzer equipped with Bluetooth 5 analysis software.

In band emissions: This test verifies that the in-band spectral emission of the Bluetooth transmission is within limits. The limits have been modified to accommodate LE 2M PHY. The limit lines for LE Coded PHY that operates at 1 Ms/s are the same limit lines used by LE 1M PHY. The entire Bluetooth band of 80 MHz is split into 80 channels of 1 MHz each and the integrated power in each band is calculated. The device transmits on an RF channel with the center frequency M and the center frequency of adjacent channels of 1 MHz bandwidth is denoted by N. For LE 1M, the integrated power in the band that is offset by 2 MHz should be less than -20 dBm and power in bands offset by 3 MHz or more should be less than -30 dBm. For LE 2M the limit comparison starts from frequency offset of 4 MHz on either side (instead of 2 MHz). For bands offset by 4 MHz and 5 MHz, the integrated power is expected to be less than -20 dBm and only for offsets greater than 6 MHz, the more stringent requirement of <-30 dBm is set.

In the screen capture in Figure 5, you can see that LE 2M power is calculated every 1 MHz and indicated by the blue lines. You will also notice there are three limits suggested by the standard. ±4 MHz, ±5 MHz, and ±≥6MHz.


Figure 5 LE 2M power is calculated every 1 MHz, indicated by the blue lines.

Modulation characteristics: The modulation used by Bluetooth is Gaussian Frequency Shift Keying (GFSK) with a bandwidth bit period product BT=0.5. The modulation index must be between 0.45 and 0.55. This test verifies that the frequency deviation for known test patterns is within specified limits. The measurement uses specific test patterns. In previous versions of Bluetooth, patterns such as 0x0F (00001111) and 0x55 (01010101) were used and the frequency deviation in each bit interval was calculated in a manner specified in the standard. In Bluetooth 5, the pass/fail limit has been changed for LE 2M PHY as the frequency deviation for a 2 Msamples/s modulation scheme is different. These limits double for LE 2M PHY. For LE Coded (S=8), the measurement patterns are different. The first pattern is generated by giving all 1s. After coding and mapping the pattern becomes 00111100. The second pattern, 00110011, is generated if the input pattern to the coder and mapper is all 0s. The standard also defines that this measurement starts from 33rd symbol.

Stable modulation characteristics: This is a new measurement and was not a part of the test specification of earlier Bluetooth releases. An LE device with a transmitter that has a stable modulation index may inform the receiving LE device of this fact through the feature support mechanism. The modulation index for these transmitters is between 0.495 and 0.505. A device should only state that it has a stable modulation index if that applies to all LE transmitter PHY it supports. A transmitter that does not have a stable modulation index, while still within one percent margin of the ideal 0.5 modulation index, is said to have a standard modulation index.

Frequency offset and drift: The frequency offset is calculated by averaging the frequency deviation in a specified interval of alternated 1s and 0s patterns. The interval duration is 10 bits or 10 µs in the earlier low energy standard. This frequency offset is calculated in preamble and in payload. Then the drift in these frequency offset over a 50 µs interval (5 intervals away) is calculated. For LE 2M PHY, the intervals are still 10 µs but consist of 20 bits instead of 10 (as it is 2Msps). The drift measurement is still done in five groups or five interval durations away. For LE Coded PHY, a 16-bit interval is chosen instead of 10 and the drift is calculated 3 interval durations away (48 µs) because the patterns are 00110011.

20 dB bandwidth: Measures the bandwidth up to the point where the spectrum drops to 20 dB below the peak power.

Output power: Calculates the power in the complete packet.

In-depth Bluetooth analysis: In addition to the measurements described above, some Bluetooth analysis software offers additional information about the signal tested. Such analysis help you debug and optimize performance for the intended application and include the following:

  • Decoded packet information as a representation of all the header and packet information that has been decoded.
  • A summary or snapshot of all measurements and decoded packet information.
  • Displays that show frequency deviation against time, which can be useful while trying to debug or interpret Modulation Char and Drift measurements.
  • Drift table showing frequency offsets calculated in 10-bit intervals and the drift over 50 µs (5 interval durations away).
  • Constellation, Eye and Symbol table displays.

Get real
It’s also useful for Bluetooth applications to use real-time spectrum analyzers that can show failures hidden under broadband noise that otherwise would not be visible. The screen shot in Figure 6 (on the right side) shows what a swept spectrum analyzer sees in a 40 MHz sweep versus a real-time spectrum analyzer (left).


Figure 6 A real-time spectrum analyzer can show hidden failures missed by traditional swept spectrum analyzers.

Bluetooth 5 brings across-the-board improvements compared to Bluetooth 4.2 LE. By paying close attention to your test and measurement strategies, you can design devices that take advantage of everything the new standard offers.

Dorine Gurney has over 15 years of experience as a product planner at Tektronix. She has recently retired.

Related articles:

Leave a comment