Interference from 2.4 GHz telemetry

[Jon-Håkon Bøe Røli]

Hi!

I was wondering if anyone has had any experience with Piksi's susceptibility to interference from 2.4 GHz telemetry?

During testing this week we noticed that the SNR of satellites was severely reduced when our telemetry link was in close proximity to Piksi's antenna. Without the telemetry we were getting 8-10 satellites with SNRs between 12 and 8, but with the telemetry antenna powered on we dropped to 3-5 sattelites, with SNRs rarely above 5, and mostly at the noise floor.

We're using the Picostation M2, and Piksi had the v0.13 FW (we didn't upgrade to v0.14 as we hadn't yet done any testing with it, and reports from you guys spoke of frequent unit lock up) with an external helix antenna (M1227HCT-A-SMA from Maxtena).

The attached picture shows the setup on our hexacopters with the Picostation stripped of its original casing to save weight. The telemetry antenna is about 20 cm apart form Piksi's antenna, while the Picostation itself is about 10 cm apart.

When we discovered the problem we tried changing different settings on the Picostation: Setting output power to the minimum (-3 dBm), changing channel width between 20 and 5 MHz, changing the frequency from ch 13 to 1. However, we saw little or no change in the interference. 

I do find problem quite strange as we didn't notice any problems with interference in a previous test in november (Piksi then had FW v0.12, could this make any difference?), as we then had no problems acquiring 8-10 satellites. At this time we even had Picostation set to the maximum output power of 28 dBm.

I will be doing more tests this week to determine more precisely how the Picostation is interfering with the GPS signals.

Any help or useful insights would be greatly appreciated!

Regards,

JH

[Clive Turvey]

GPS signals at the surface of the planet are very weak, and very susceptible to jamming. I've jammed the things with CMOS switching current in an FPGA. The IF here is 4.092 MHz, and the chip rate is 1.023 MHz +/- doppler for the gold codes (LFSR noise).

It might be the radio side, but I wouldn't rule out noise in the power supply.

Perhaps some more shielding of the Piksi? The ceramic puck is still feeding directly into the MAX2769

[Blair Burtan]

Interesting observations.  It sounds like it would be easy enough to move the Picostation around while observing the Piksi chart recorder output to see the effects of proximity e.g. stick the Piksi on a tripod by itself hooked up via USB and move your copter around it.  Please keep us posted on this as I'm probably going to be doing something similar.  I'm also curious about that antenna.  It doesn't appear to be amplified.  Is it really worth the $285?

[Clive Turvey]

> Is it really worth the $285?

Quad Helix/Volute, probably not, the sarantel's we're small and expensive, and not particularly fantastic.

It's draws current so it's clearly got an amplifier in there. aka Active

http://www.farnell.com/datasheets/1681376.pdf

What's the mass of the thing? Didn't see it called out in the spec.

[Jon-Håkon Bøe Røli]

Price is a bit stiff, but it was recommended by experts in GNSS at another institute. Its size, weight (only 17 g) and good reception even when tilted fits our multicopter platform well. It is also ground plane independent, saving us even more weight.

[Clive Turvey]

I need to stick one of Leisten's original volutes up on the roof and characterize it, my recollections in talking to one of the Navstar guys is that there were phase centre issues, I don't ever recall seeing them (quad helix/volutes) used in survey applications, or in NGS lists. 

MAXTENA claim RTK applications, but I don't see any NGS data for them. My gut suggests they will overly favour 90 degree elevation satellites.

[Jon-Håkon Bøe Røli]

Here are some of the results from the interference tests:

The first two pictures shows a new thing we discovered. With only Piksi and the BeagleBone Black (BBB) and its cape turned on, distinctive spikes with ~4 second intervals sometimes appear. These spikes were especially big when Piksi was powered separately from the BBB and cape (first picture). The cape normally supplies the BBB, Piksi and Picostation with power (giving them common ground, seen in second picture). These spikes were only prominent with no other disturbance present.

The 3rd picture shows how severely the C/N0 is affected by the Picostation being turned on with its antenna right next to the external helix antenna connected to Piksi.

The 4th picture is with Picostation mounted on the hexa (as seen in the picture in the OP); first turned on and not interfering too much, then, after some while, reaching a stable state where the interference is much worse. We generally saw this behaviour whenever we turned the Picostation on, whether or not we were actually transmitting any significant amount of data.

The effect of moving an already powered Picostation first about 10 cm, then right next to the Piksi external antenna, is shown in the 5th picture with both max and min output power. As previously posted, we didn't see any distinctive difference in changing the output power.

Finally, in the last picture, the antenna of the Picostation was placed ~ 10 cm directly below and parallel with the Piksi external antenna, as the gain pattern is a flat donut with nulls directly above and below.

[Daniel Eckert]

I experienced similar interference issues with my Piksi recently. Interestingly the problem occured with almost the same antenna as you used (Maxtena M1227HCT-A2-SMA).

The problem occurs when I place the antenna close to our visual-inertial sensor (which is basically an IMU and two cameras). As soon as the VI sensor is powered and booting, there are small interferences, and as soon as the sensor is fully running the interferences cause the antenna to fail completely (see screenshot). For the antenna to fully work, I have to keep almost half a meter distance from the sensor. The interesting thing here is that the VI sensor doesn't use any wireless communication. It is only connected to the computer via USB for power and Ethernet for data exchange. I find it quite surprising that the interferences are that strong.

@Jon-Hakon, did you find any good solutions to shield the interferences, or what was your solution in the end?

[Clive Turvey]

>>The interesting thing here is that the VI sensor doesn't use any wireless communication.

I think there's a misconception here about what causes interference. It doesn't have to be in the L1 bands or an expected radio emission, CMOS switching currents and short lengths of wire are quite sufficient, a micro banging an un-terminated GPIO up and down is something an antenna can pick up. The MAX2769 down converts to 4.092 MHz, and the GPS receiver generates it's own local carrier and code signals, pulling out a pseudo-random signal from the noise at 1.023 MHz give/take the signal doppler. GPS signals are *very* easy to jam.

The antenna being L1/L2 GPS/GLONASS is going to have a very wide filter(s) on it, something doing L1 (C/A)+GPS only is going to have a significantly tighter filter at 1575.42 MHz, with a bandwidth of perhaps 2.5 MHz (which where the 2769 filter is configured). Here you have some powerful upstream amps with significant bandwidth, and antenna elements tuned to the signal.

Here the gain is pretty high, L1/L2 frequencies are transposed, and the L1 frequency has some 45 MHz bandwidth.
http://www.farnell.com/datasheets/1681376.pdf

[Kristian Klausen]

Hi,

As Clive also points out, or main source of interference was not the wireless transmission. In our case, it turned out we had two problems:
1. A DC/DC level converter in the radio we used caused interference in the L1 band. By hooking up a GPS-antenna to a spectrum analyzer, we detected a narrow but quite significant interference. We were not able to deminish this with shielding.
2. Simply using ethernet at 100mbit also caused problems in the mounting-configuration we had.

As a result, we switched to another radio (the slightly bigger Rocket M5), and forced 10mbit ethernet communication. That removed all our issues, and we've since had several successfull flights with the piksi and our maxtena antenna.

So for you Daniel, I would try to force 10mbit ethernet, and see if that helps!

Regards,
Kristian