| title | location | attendees | date |
|---|---|---|---|
HPM5G Radio Tests |
Toronto |
8 |
2018-05-30 |
This document summarizes a series of tests on the Dragino HPM5G mPCIe radio module and the LibreRouter's 12 dBi antenna that are expected to ship with the soon-to-be-released LibreRouter bundle. Toronto Mesh acquired six radio boards and antennas from Dragino, and in this test, used a pair of the Marvell ESPRESSObin running prototype to test the radios.
We conducted many tests that are described in detail below, but our key findings are as follows:
- The radio and antenna combo creates stable links over hundreds of metres at 150 Mbps
- Wall penetration is excellent even when a client node has an omnidirectional antenna on a radio with average sensitivity
- Continuous transmission for 10 minutes at 26 dBm show no performance drop or heating-related issues
- Not all boards with a mPCIe interface can source enough current to this board, but external power input works great
- The 12 dBi antenna makes an excellent 5G antenna even for other radios
- Linux driver support is trivial, as expected from a
ath9kdevice
Toronto Mesh is overall very impressed with the performance of these devices.
Using our ESPRESSObin setup, the HPM5G board was automatically identified and brought up on boot with the ath9k driver. Throughout testing, we observed a couple issues that we were able to work around.
During the initial setup and testing we noticed that the ESPRESSObin would restart at what seemed to be random times. We tried different combination of radio boards and ESPRESSObin boards but the issue was persistent. We tried to reach out to the ESPRESSObin community without any luck; however the Armbian community mentioned that this seemed like an issue with insufficient power. Identifying this as a potential power issue we tried using larger amperage power supplies (1.5 and 2 A at 12 V) however this did not seem to solve the problem.
Work Around
After identifying a potential power problem we tried to limit the amount of power the radio would need by reducing the TX power of the radio. We found that keeping it around 14 dBm would let the ESPRESSObin operate without reboot. Our initial testing was then done using a 14 dBm TX power.
We did run into an issue where the TX power would not be persistent after being set, due to the Network Manager Service changing its value. The solution was to change the regulatory database to prevent TX power higher then 14 dBm. This held the TX power at 14 dBm.
Long-term Solution
This issue will occur on the ESPRESSObin and other boards that do not source enough current through its mPCIe interface. The LibreRouter mPCIe interfaces are designed to meet the power requirements of the HPM5G at 19 dBm. In later parts of this document, we will also describe a board modification that will allow us to power the HPM5G properly on the ESPRESSObin.
Both HPM5G radios were brought up with the same MAC address of 00:02:03:04:05:06, which prevented two devices from talking to each other.
Additionally, the ESPRESSObin interfaces themselves seemed to be the same as well. This is an unrelated bug in Armbian.
Work Around
Setting the MAC address manually allowed the two devices to mesh together. For example:
sudo ip link set dev wlan0 address 00:28:C7:0A:42:A2
sudo ifconfig wlan0 hw ether 00:28:C7:0A:42:A2
The ESPRESSObin interface issue was resolved by regenerating a unique machine ID and rebooting:
sudo mv /etc/machine-id /etc/machine-id.old
dbus-uuidgen | sudo tee /var/lib/dbus/machine-id
sudo cp /var/lib/dbus/machine-id /etc/machine-id
Long-term Solution
This issue is specific to the sample units and will not be an issue when the HPM5G is manufactured for production.
The first test we did to confirm the device was functioning was to connect an ESPRESSObin + HPM5G node to a Orange Pi Zero + TOP-4M02 node, using 802.11s mesh point. These two devices were about 25 feet apart separated by three drywalls, without line-of-sight, and with desks and file cabinets in between. The ESPRESSObin has the HPM5G transmitting at 15 dBm, its 12 dBi antenna pointed in the general direction towards the omnidirectional antennas on the Orange Pi Zero.
The results of this first test were very impressive. The HPM5G was the transmitter, and the TOP-4M02 was the receiver:
Connecting to host 192.168.100.4, port 5201
[ 4] local 192.168.100.1 port 52190 connected to 192.168.100.4 port 5201
[ ID] Interval Transfer Bandwidth Retr Cwnd
[ 4] 0.00-1.00 sec 5.02 MBytes 42.1 Mbits/sec 0 465 KBytes
[ 4] 1.00-2.00 sec 5.25 MBytes 44.0 Mbits/sec 0 465 KBytes
[ 4] 2.00-3.00 sec 8.85 MBytes 74.3 Mbits/sec 0 465 KBytes
[ 4] 3.00-4.00 sec 8.79 MBytes 73.8 Mbits/sec 0 465 KBytes
[ 4] 4.00-5.00 sec 6.63 MBytes 55.6 Mbits/sec 0 465 KBytes
[ 4] 5.00-6.00 sec 8.20 MBytes 68.8 Mbits/sec 1 465 KBytes
[ 4] 6.00-7.00 sec 9.44 MBytes 79.2 Mbits/sec 0 465 KBytes
[ 4] 7.00-8.00 sec 10.5 MBytes 88.4 Mbits/sec 0 465 KBytes
[ 4] 8.00-9.00 sec 10.0 MBytes 83.9 Mbits/sec 0 465 KBytes
[ 4] 9.00-10.00 sec 10.0 MBytes 84.2 Mbits/sec 0 465 KBytes
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval Transfer Bandwidth Retr
[ 4] 0.00-10.00 sec 82.8 MBytes 69.4 Mbits/sec 1 sender
[ 4] 0.00-10.00 sec 81.9 MBytes 68.7 Mbits/sec receiver
iperf Done.
These are the highest speeds we have observed from the TOP-4M02. The HPM5G is transmitting a very strong signal through all these physical barriers, and with low jitter:
Accepted connection from 192.168.100.2, port 43694
[ 5] local 192.168.100.4 port 5201 connected to 192.168.100.2 port 47881
[ ID] Interval Transfer Bandwidth Jitter Lost/Total Datagrams
[ 5] 0.00-1.00 sec 120 KBytes 983 Kbits/sec 0.299 ms 0/15 (0%)
[ 5] 1.00-2.00 sec 128 KBytes 1.05 Mbits/sec 0.545 ms 0/16 (0%)
[ 5] 2.00-3.00 sec 128 KBytes 1.05 Mbits/sec 0.560 ms 0/16 (0%)
[ 5] 3.00-4.00 sec 128 KBytes 1.05 Mbits/sec 0.494 ms 0/16 (0%)
[ 5] 4.00-5.00 sec 128 KBytes 1.05 Mbits/sec 0.432 ms 0/16 (0%)
[ 5] 5.00-6.00 sec 128 KBytes 1.05 Mbits/sec 0.466 ms 0/16 (0%)
[ 5] 6.00-7.00 sec 128 KBytes 1.05 Mbits/sec 0.388 ms 0/16 (0%)
[ 5] 7.00-8.00 sec 128 KBytes 1.05 Mbits/sec 2.240 ms 0/16 (0%)
[ 5] 8.00-9.00 sec 128 KBytes 1.05 Mbits/sec 1.051 ms 0/16 (0%)
[ 5] 9.00-10.00 sec 128 KBytes 1.05 Mbits/sec 0.800 ms 0/16 (0%)
[ 5] 10.00-10.04 sec 8.00 KBytes 1.79 Mbits/sec 0.802 ms 0/1 (0%)
HPM5G reported RX:
signal: -75 [-75, -77] dBm
signal avg: -72 [-73, -75] dBm
TOP-4M02 reported RX:
signal: -50 dBm
signal avg: -50 dBm
Placing two ESPRESSObins with HPM5Gs about 20 cm from each other, and lowering the TX power to 5 dBm, I was able to confirm speeds of over 160 Mbps.
We married an Orange Pi Zero with a TOP-4M02 USB radio to the 12 dBi LibreRouter antenna for a long distance test. In the past, we have not had good range or speeds with this board, but since the board supported 5 Ghz and was 2x2 MIMO it was a good test.
We mounted the device with antenna outside to test, with the ESPRESSObin + HPM5G at the other end of the connection to compare RX quality:
The devices were about 220 m apart:
The positioning of the antennas were not optimal, however with this quick test the TOP-4M02 was able to see the HPM5G with -70 dBm, while the HPM5G could see the TOP-4M02 with -54 dBm. Speed was only 10 Mbps, however the link was very stable.
The antenna case can even be modified to fit the Orange Pi Zero inside:
On a Tuesday afternoon we took the pair of ESPRESSObins to the bridges where we ran our Cantenna Field Test last year. The distance is approximately 300 m.
We powered one ESPRESSObin with a portable AC power source. The other we attempted to use a 3-cell lithium polymer 1600 mAh battery wired into the DC barrel connector.
When fully charged the battery put out 12.5 V (12.6 V being the absolute max) and the ESPRESSObin seemed to work well with this setup. The test drained the power of the battery to about 50% of its capacity.
We split up into two teams and connected a laptop to the USB port of the ESPRESSObin to record results, pointing the antenna in the general direction of the other team. One device had TX power of 15 dBm (but it occasionally reboots due to drawing too much power over the mPCIe) and iperf3 recorded speeds close to 100 Mbps. The other device with TX power of 14 dBm never rebooted, but it does not reach the same speeds when it's the one transmitting.
This is what one side looks like:
And looking out to the peer:
Dragino confirmed that what we were experiencing sounded like it could be a power issue, where the radio was draining more power from the ESPRESSObin than it could provide. After speaking with the engineer they provided a hardware modification to power the radio module externally, instead of via the mPCIe interface. It involves disabling of the 3.3 V to 5 V up-conversion circuit so that we could apply our own 5 V external power, by removing a surface mount resistor and grounding one of the pads.
The modified HPM5Gs look like:
Using the 5 V power from the ESPRESSObin's hard drive connecter, we were able to successfully power the boards externally using a modified cable:
After removing the regulatory lockout in Linux, we were able to bring the radio module up to 26 dBm and run an endurance test for 10 minutes using iperf3 in both directions without the ESPRESSObin rebooting. This shows the earlier issue was indeed power related and the modification successfully fixed it. The power levels observed here is inline with what was reported by LibreRouter in their earlier tests, which took the HPM5G to 27 dBm TX power.
Repeating the same 220 m antenna test with two fully powered ESPRESSObins and HPM5Gs using external power:
These pictures are taken by placing the camera flat against the front of the antenna (point-of-view from the antenna):
The ESPRESSObins were configured at 26 dBm TX power.
Remote side:
signal: -60 [-62, -65] dBm
signal avg: -60 [-62, -64] dBm
tx bitrate: 243.0 MBit/s MCS 14 40MHz
rx bitrate: 135.0 MBit/s MCS 6 40MHz short GI
expected throughput: 53.557Mbps
Near side:
signal: -62 [-70, -63] dBm
signal avg: -59 [-67, -60] dBm
tx bitrate: 121.5 MBit/s MCS 6 40MHz
rx bitrate: 243.0 MBit/s MCS 14 40MHz
expected throughput: 42.388Mbps
iperf3 forward test:
root@tomesh-fd90:~# iperf3 -c 192.168.100.2 -t 60
Connecting to host 192.168.100.2, port 5201
[ 4] local 192.168.100.99 port 35900 connected to 192.168.100.2 port 5201
[ ID] Interval Transfer Bandwidth Retr Cwnd
[ 4] 0.00-1.00 sec 8.04 MBytes 67.4 Mbits/sec 0 308 KBytes
[ 4] 1.00-2.00 sec 9.91 MBytes 83.2 Mbits/sec 0 376 KBytes
[ 4] 2.00-3.00 sec 11.6 MBytes 96.8 Mbits/sec 0 826 KBytes
[ 4] 3.00-4.00 sec 12.0 MBytes 101 Mbits/sec 0 826 KBytes
[ 4] 4.00-5.00 sec 11.9 MBytes 99.5 Mbits/sec 0 826 KBytes
[ 4] 5.00-6.00 sec 12.0 MBytes 100 Mbits/sec 0 826 KBytes
[ 4] 6.00-7.00 sec 10.1 MBytes 84.7 Mbits/sec 0 826 KBytes
[ 4] 7.00-8.00 sec 11.1 MBytes 92.9 Mbits/sec 0 826 KBytes
[ 4] 8.00-9.00 sec 12.9 MBytes 108 Mbits/sec 0 826 KBytes
[ 4] 9.00-10.00 sec 11.3 MBytes 95.0 Mbits/sec 0 826 KBytes
...
[ 4] 51.00-52.00 sec 12.4 MBytes 104 Mbits/sec 0 2.10 MBytes
[ 4] 52.00-53.00 sec 11.9 MBytes 99.9 Mbits/sec 0 2.10 MBytes
[ 4] 53.00-54.00 sec 13.2 MBytes 111 Mbits/sec 0 2.10 MBytes
[ 4] 54.00-55.00 sec 13.9 MBytes 117 Mbits/sec 0 2.10 MBytes
[ 4] 55.00-56.00 sec 13.5 MBytes 113 Mbits/sec 0 2.10 MBytes
[ 4] 56.00-57.00 sec 12.9 MBytes 108 Mbits/sec 0 2.10 MBytes
[ 4] 57.00-58.00 sec 13.8 MBytes 116 Mbits/sec 0 2.36 MBytes
[ 4] 58.00-59.00 sec 12.8 MBytes 107 Mbits/sec 0 2.36 MBytes
[ 4] 59.00-60.00 sec 13.9 MBytes 117 Mbits/sec 0 2.36 MBytes
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval Transfer Bandwidth Retr
[ 4] 0.00-60.00 sec 727 MBytes 102 Mbits/sec 0 sender
[ 4] 0.00-60.00 sec 725 MBytes 101 Mbits/sec receiver
iperf Done.
iperf3 reverse test:
Reverse mode, remote host 192.168.100.2 is sending
[ 4] local 192.168.100.99 port 35904 connected to 192.168.100.2 port 5201
[ ID] Interval Transfer Bandwidth
[ 4] 0.00-1.00 sec 15.6 MBytes 131 Mbits/sec
[ 4] 1.00-2.00 sec 18.3 MBytes 154 Mbits/sec
[ 4] 2.00-3.00 sec 17.6 MBytes 147 Mbits/sec
[ 4] 3.00-4.00 sec 17.2 MBytes 144 Mbits/sec
[ 4] 4.00-5.00 sec 18.2 MBytes 152 Mbits/sec
[ 4] 5.00-6.00 sec 17.9 MBytes 150 Mbits/sec
[ 4] 6.00-7.00 sec 18.6 MBytes 156 Mbits/sec
[ 4] 7.00-8.00 sec 18.8 MBytes 158 Mbits/sec
[ 4] 8.00-9.00 sec 19.2 MBytes 161 Mbits/sec
[ 4] 9.00-10.00 sec 18.4 MBytes 154 Mbits/sec
...
[ 4] 51.00-52.00 sec 18.9 MBytes 159 Mbits/sec
[ 4] 52.00-53.00 sec 18.1 MBytes 152 Mbits/sec
[ 4] 53.00-54.00 sec 18.7 MBytes 157 Mbits/sec
[ 4] 54.00-55.00 sec 18.3 MBytes 154 Mbits/sec
[ 4] 55.00-56.00 sec 18.4 MBytes 154 Mbits/sec
[ 4] 56.00-57.00 sec 17.2 MBytes 144 Mbits/sec
[ 4] 57.00-58.00 sec 17.8 MBytes 149 Mbits/sec
[ 4] 58.00-59.00 sec 18.9 MBytes 159 Mbits/sec
[ 4] 59.00-60.00 sec 18.7 MBytes 157 Mbits/sec
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval Transfer Bandwidth Retr
[ 4] 0.00-60.00 sec 1.07 GBytes 154 Mbits/sec 0 sender
[ 4] 0.00-60.00 sec 1.07 GBytes 154 Mbits/sec receiver
iperf Done.
iperf UDP jitter test:
[ 5] 50.00-51.00 sec 128 KBytes 1.05 Mbits/sec 0.542 ms 0/16 (0%)
[ 5] 51.00-52.00 sec 128 KBytes 1.05 Mbits/sec 0.577 ms 0/16 (0%)
[ 5] 52.00-53.00 sec 128 KBytes 1.05 Mbits/sec 1.735 ms 0/16 (0%)
[ 5] 53.00-54.00 sec 128 KBytes 1.05 Mbits/sec 1.295 ms 0/16 (0%)
[ 5] 54.00-55.00 sec 128 KBytes 1.05 Mbits/sec 0.757 ms 0/16 (0%)
[ 5] 55.00-56.00 sec 128 KBytes 1.05 Mbits/sec 0.566 ms 0/16 (0%)
[ 5] 56.00-57.00 sec 128 KBytes 1.05 Mbits/sec 0.556 ms 0/16 (0%)
[ 5] 57.00-58.00 sec 128 KBytes 1.05 Mbits/sec 0.663 ms 0/16 (0%)
[ 5] 58.00-59.00 sec 128 KBytes 1.05 Mbits/sec 0.880 ms 0/16 (0%)
[ 5] 59.00-60.00 sec 128 KBytes 1.05 Mbits/sec 0.689 ms 0/16 (0%)
[ 5] 60.00-60.04 sec 8.00 KBytes 1.50 Mbits/sec 0.668 ms 0/1 (0%)
Noticing the discrepancy between the forward and reverse tests, the antenna was repositioned tilting it downwards a few degrees. The link quality and speed test improved:
signal: -54 [-57, -57] dBm
signal avg: -58 [-61, -62] dBm
Toffset: 273322777 us
tx bitrate: 240.0 MBit/s MCS 13 40MHz short GI
rx bitrate: 300.0 MBit/s MCS 15 40MHz short GI
expected throughput: 53.557Mbps
root@tomesh-fd90:~# iperf3 -c 192.168.100.2 -t 30
Connecting to host 192.168.100.2, port 5201
[ 4] local 192.168.100.99 port 35926 connected to 192.168.100.2 port 5201
[ ID] Interval Transfer Bandwidth Retr Cwnd
[ 4] 0.00-1.00 sec 15.9 MBytes 134 Mbits/sec 0 945 KBytes
[ 4] 1.00-2.00 sec 17.6 MBytes 147 Mbits/sec 0 1000 KBytes
[ 4] 2.00-3.00 sec 17.4 MBytes 146 Mbits/sec 0 1000 KBytes
[ 4] 3.00-4.00 sec 17.3 MBytes 146 Mbits/sec 0 1000 KBytes
[ 4] 4.00-5.00 sec 17.8 MBytes 150 Mbits/sec 0 1000 KBytes
[ 4] 5.00-6.00 sec 17.2 MBytes 145 Mbits/sec 0 1000 KBytes
[ 4] 6.00-7.00 sec 17.1 MBytes 144 Mbits/sec 0 1000 KBytes
[ 4] 7.00-8.00 sec 17.6 MBytes 148 Mbits/sec 0 1.09 MBytes
[ 4] 8.00-9.00 sec 17.4 MBytes 146 Mbits/sec 0 1.22 MBytes
[ 4] 9.00-10.00 sec 18.7 MBytes 158 Mbits/sec 0 1.22 MBytes
...
[ 4] 21.00-22.00 sec 19.0 MBytes 159 Mbits/sec 0 1.77 MBytes
[ 4] 22.00-23.00 sec 17.9 MBytes 150 Mbits/sec 0 1.77 MBytes
[ 4] 23.00-24.00 sec 18.7 MBytes 157 Mbits/sec 0 1.77 MBytes
[ 4] 24.00-25.00 sec 18.2 MBytes 152 Mbits/sec 0 1.77 MBytes
[ 4] 25.00-26.00 sec 19.0 MBytes 160 Mbits/sec 0 1.85 MBytes
[ 4] 26.00-27.00 sec 18.9 MBytes 158 Mbits/sec 0 1.85 MBytes
[ 4] 27.00-28.00 sec 18.9 MBytes 159 Mbits/sec 0 1.85 MBytes
[ 4] 28.00-29.00 sec 18.1 MBytes 151 Mbits/sec 0 2.20 MBytes
[ 4] 29.00-30.00 sec 18.9 MBytes 159 Mbits/sec 0 2.20 MBytes
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval Transfer Bandwidth Retr
[ 4] 0.00-30.00 sec 543 MBytes 152 Mbits/sec 0 sender
[ 4] 0.00-30.00 sec 542 MBytes 151 Mbits/sec receiver
At this point I was going to test lowering the TX power, but a van pulled up next to me. Although the line-of-sight was not obstructed, the fresnel zone must have been impacted as the speeds dropped to about 80 Mbps:
[ 4] 0.00-1.00 sec 8.43 MBytes 70.7 Mbits/sec 0 297 KBytes
[ 4] 1.00-2.00 sec 9.65 MBytes 80.8 Mbits/sec 0 522 KBytes
[ 4] 2.00-3.00 sec 9.44 MBytes 79.4 Mbits/sec 0 522 KBytes
[ 4] 3.00-4.00 sec 9.10 MBytes 76.4 Mbits/sec 0 522 KBytes
[ 4] 4.00-5.00 sec 9.20 MBytes 77.2 Mbits/sec 0 617 KBytes
[ 4] 5.00-6.00 sec 9.79 MBytes 82.1 Mbits/sec 0 617 KBytes
[ 4] 6.00-7.00 sec 10.6 MBytes 89.3 Mbits/sec 0 617 KBytes
[ 4] 7.00-8.00 sec 9.44 MBytes 79.2 Mbits/sec 0 1.37 MBytes
I found a new spot closer to the antenna that was not obstructed and performed the TX power test there. This position was only about 74 m from the antenna.
The TX power was adjusted on both radios at the same time by running the iwconfig wlan0 txpower <txpower> command:
| TX power | Signal avg |
|---|---|
| 24 dBm | -52 [-55, -55] dBm |
| 20 dBm | -52 [-55, -56] dBm |
| 15 dBm | -59 [-60, -60] dBm |
At 20 dBm:
[ 4] 0.00-1.00 sec 18.4 MBytes 154 Mbits/sec 0 1.71 MBytes
[ 4] 1.00-2.00 sec 18.8 MBytes 158 Mbits/sec 0 1.71 MBytes
[ 4] 2.00-2.68 sec 12.9 MBytes 160 Mbits/sec 0 2.66 MBytes
At 15 dBm:
[ 4] 0.00-1.00 sec 15.0 MBytes 126 Mbits/sec 0 527 KBytes
[ 4] 1.00-2.00 sec 16.8 MBytes 141 Mbits/sec 0 792 KBytes
[ 4] 2.00-3.00 sec 16.8 MBytes 141 Mbits/sec 0 792 KBytes
[ 4] 3.00-4.00 sec 17.3 MBytes 145 Mbits/sec 0 905 KBytes
[ 4] 4.00-5.00 sec 17.1 MBytes 143 Mbits/sec 0 945 KBytes
[ 4] 5.00-6.00 sec 17.2 MBytes 144 Mbits/sec 0 945 KBytes
[ 4] 6.00-7.00 sec 16.8 MBytes 141 Mbits/sec 0 945 KBytes
[ 4] 7.00-8.00 sec 17.0 MBytes 142 Mbits/sec 0 945 KBytes
[ 4] 8.00-9.00 sec 16.6 MBytes 139 Mbits/sec 0 945 KBytes
[ 4] 9.00-10.00 sec 16.3 MBytes 136 Mbits/sec 0 1.63 MBytes
At 11 dBm:
[ 4] 0.00-1.00 sec 10.7 MBytes 89.5 Mbits/sec 0 385 KBytes
[ 4] 1.00-2.00 sec 12.4 MBytes 104 Mbits/sec 0 699 KBytes
[ 4] 2.00-3.00 sec 11.9 MBytes 99.7 Mbits/sec 0 996 KBytes
[ 4] 3.00-4.00 sec 9.17 MBytes 76.8 Mbits/sec 0 996 KBytes
[ 4] 4.00-5.00 sec 8.68 MBytes 72.7 Mbits/sec 0 996 KBytes
[ 4] 5.00-6.00 sec 12.5 MBytes 105 Mbits/sec 0 996 KBytes
[ 4] 6.00-7.00 sec 11.6 MBytes 96.9 Mbits/sec 0 996 KBytes
[ 4] 7.00-8.00 sec 9.78 MBytes 82.2 Mbits/sec 0 996 KBytes
[ 4] 8.00-9.00 sec 12.1 MBytes 101 Mbits/sec 0 996 KBytes
[ 4] 9.00-10.00 sec 10.7 MBytes 89.5 Mbits/sec 0 996 KBytes