GPS is highly disturbed by Lidar lite, how to fix it?

[Julien Dubois]

For those who have ever played with a Lidar lite, I wonder how they manage to make the GPS work with it.
As soon as I plug it, I loose all my sats!

Test made inside (I know that's not ideal) with a NEO-M6 GPS:
8 sats 2.16 HDOP

When Lidar is pluged, 0 sat. Sometimes I have 5-6 sats with high HDOP (around 10-14)... It seems to depend on the lidar orientation.
I've tryed to place aluminium foil around the lidar with no real improvement.

I think I'll get a M8N to compare perfs in this situation.

Every comment and idea welcome!

Julien

[Paul Riseborough]

I use a Lidar Lite and haven't had issues with it, but I do modify the I2C cable configuration as I have had conducted interference over I2C cause GPS problems. If you are using an external compass+GPS, any noise transmitted over I2C ends up in your GPS.

Make sure that you remove the pin for the un-used wire on the clickmate connector as that can radiate.

The I2c bus specification recommends the following for wiring patterns:

If the bus lines are twisted-pairs, each bus line must be twisted with a VSS return. Alternatively, the SCL line can be

twisted with a VSS return, and the SDA line twisted with a VDD return.

I was also recommended by Craig  to twist  one pair at about 1 turn /cm and the other at 1.5 turn / cm to reduce the enclosed loop area and minimize radiation. Doing this helped reduce the problem I was having with the PX4Flow sensor jamming the GPS.

The I2c cables as supplied by 3DR do not comply with the recommendations.

-Paul

[Julien Dubois]

Thanks a lot Paul for your answer, I'll try a custom cable then.
I've found Craig's original post here btw

Do you think we could lower the EMF interference by changing the I2C baudrate and clock frequency in order to be far from the gps working frequency or its harmonics?

[Andy Lee Robinson]

GPS at around 1575 MHz is way above I²C max bandwidth of 5 MHz - the signal would need to be powerful with very clean edges to generate any harmonics significant enough to disturb the GPS, imho but I wouldn't say it's impossible.

Twisting the I²C cable, running it as far away as possible and maybe shielding it is good practice anyway, but consider adding a SAW filter to the GPS unit to add additional RFI rejection.

Mod Installation --- DJI Naza GPS SAW Filter
http://youtu.be/yP2l96tqh1I

Not fancying the nanosoldering required I specifically ordered a NEO 7N which has a saw filter built in, and have no problems with it at all.

http://www.u-blox.com/en/gps-modules/pvt-modules/neo-7.html

[Ben Nizette]

Yeah I don't think it's the I2C frequency per-se, a few-hundred kHz
open-drain (which is kind of built-in slew-rate limiting) shouldn't
cause issues. Not that it's completely unheard of, 10/100 Base
Ethernet often causes issues with its 25MHz base clock, a harmonic of
which stomps right on 1.575GHz. Pretty different signal though :)

But yeah, I guess the weakly-driven I2C lines are just acting as an
antenna for other cruft around the place then helpfully conducting it
right up on to the GPS module.

Ben.

> --
> You received this message because you are subscribed to the Google Groups
> "drones-discuss" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to drones-discus...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.

[Chris Anderson]

I've had the same experience -- LIDAR-Lite killed my GPS reception on APM: Plane. 

I'm going to try twisting wires, shielding the LIDAR lite and a few other tricks to see if they'll help.  But I'm hoping this is something they can solve in the next rev of the LIDAR-Lite board.

-c

--

Chris Anderson

CEO, 3D Robotics

ch...@3drobotics.com

[Meier Lorenz]

Because today’s silicon is so effective at driving really sharp edges, every peripheral, even the slowest I2C, has the potential to create harmonics in the GHz range. You really can’t go just by the nominal operating frequency of the bus.

-Lorenz

[David Pawlak]

Not to mention that GPS regularly works in the -150dbW range, and often lower.

[Julien Dubois]

Lorenz, that's interesting, but I don't remember having any issue with the external I2C compass/APM. Here I'm using a PixHawk, I'll try tonight to plug an external compass instead of the lidar and compare.
So that would mean the Pixhawk and/or the lidar are driving sharper edges thats creates the disturbances.

And, as David say, we don't need so high signal strengths to kill the GPS signal.

Twisting and shielding the cable as suggested could lower the I2C EMF and fix this issue... everything seems logical.
Do you think we could round the I2C signal edges by using very small capacitor just to validate this theroy?

[Paul Riseborough]

Julien,

I forgot to mention, when connecting to the I2C breakout board on the copter I used the following connection order:

1) Compass
2) un-used
3) to PixHawk
4) to Lidar
5) to PX4Flow

On the plane it was:

1) PX4Flow
2) un-used
3) to PixHawk
4) to Lidar
5) to airspeed

End connectors were used first to avoid having unconnected stubs on the breakout board that could radiate. The GPS and flow/Lidar sensor were separated as far as possible with the lidar and flow sensor positioned low and the GPS high. I still have some interference (mainly from the PX4Flow) but it is now manageable.

-Paul

[Gary McCray]

In addition to twisted pair, it might be worth considering shielded cable.

As Lorenz says it isn't necessarily the operating frequency that's killing you rather the square wave transitions.

Best,

Gary

[Julien Dubois]

I've made some additionnal tests to find out where that disturbance comes from... I'm not 100% sure of my conclusion but here are some new elements:
1- Lidar, Pixhawk and compass plugged on I2C splitter => GPS killed when lidar-GPS distance <20cm
2- Pixhawk, compass and "Lidar wires only" plugged on I2C splitter, GPS OK... even if I surround GPS antenna with I2C wires, no issue.
3- I've checked the I2C signal with a scope, It seems to be unchanged with and without Lidar plugged, I2C clock frequency seems around 166kHz.
4- Lidar powered from an external power supply => no change with 1.
5- If I let only the 2 power pins (VCC, VSS) on the lidar climate connector and power the lidar from external source or PixHawk I2C bus => GPS killed but at shorter distance (10cm).
6- If the lidar is wrapped into an aluminium foil, only its 2 power wires pluged on external source or pixhawk, I can touch the GPS antenna and keep a live GPS. But if there is an unprotected area of the lidar, that's enough to kill the GPS.

So, from these differents tests, I think the I2C is not only culprit. I guess there are components inside the lidar (maybe I2C, maybe others) that create these disturbances which are radiated in the air and transmited through wires. The air radiation only has an acceptable effect, contrarly to the ones transmited through wires.

So, I wonder if we could add some "pico capacitor" to make a low pass filter between wires... enough to stop that HF radiations but not too much to ensure a decent I2C signal shape at the current working frequency. The idea would be to stop radiations at the lidar instead of limiting its effect (by twisting/shielding).

Julien

[Randy Mackay]

 

     I just wanted to confirm that I’ve also seen the Lidar-Lite negatively affect the GPS.  Attached is a picture of my fancy set-up and a graph showing how the GPS HDOP climbs whenever the Lidar is attached and running.  I’m pretty sure the interference is radiated because I have performed similar tests on earlier prototypes in which I attached the LIDAR to a completely separate APM2 and still saw the negative effect on the GPS.  This is consistent with Julien’s aluminum foil test.  I’ve also seen a direct correlation between how close the Lidar-lite is to the GPS and the amount of interference.

 

Julien,

     could you try these two tests?

·         Confirm that the interference is radiated by attaching the LIDAR to a separate Pixhawk and see if it still affects the GPS on the 1^st Pixhawk.

·         The Lidar-Lite has 6 wires coming out of it but only 4 are used (VCC, GND, SCL, SDA) so try removing the extra two wires complete to see if it reduces the interference.

 

     By the way, interference on the GPS from external devices is fairly common.  The Odroid I used for the red-balloon-popper also negatively affected the GPS.  I resolved that by putting a layer of aluminum tape between the Odroid and the GPS.  So very similar to your aluminum foil solution.

         www.ebay.com/itm/5pc-Shield-Adhesive-Aluminum-Foil-Duct-Tape-10mm-X-20m-/130950907736

 

-Randy

--

[Paul Riseborough]

Randy,

I'd be interested to see what you get with some separation - say 15 cm and with the stray wires pulled and the remaining ones twisted. You have a lot of antennas in that setup.

-Paul

[David Pawlak]

Something like the following would help a lot!

http://rfexplorer.com/

Low cost and could zoom right in to the source of the interferance. Reduce elaborate and perhaps inconclusive testing schemes.

[Andrew Tridgell]

> Something like the following would help a lot!
>
> http://rfexplorer.com/

good idea! It just so happens I have one of these. I just hadn't though
to turn it on near my Lidar :-)

I'll have a play and report back

Cheers, Tridge

PS: I need to read the manual for the rfexplorer first .... I got it 2nd
hand recently off a HAM friend

[Dennis Corey]

As the manufacturer of LIDAR-Lite we are also looking into this issue and also talking with 3DR about possible solutions.  Until we have some definitive answers the suggestions from Paul and Lorenz are all very good and offer some simple solutions to the problem you're experiencing.

A little history on LIDAR-Lite might also be helpful in understanding the issue.  Very early versions included an reference clock oscillator, high voltage power supply (needed to power an optional avalanche photodiode detector) and an external micro that communicated with our Signal Processing Core (SPC).  As we tested this system we quickly discovered that there was with the GPS signal on the autopilot systems we had in our lab.

To solve this problem, we eliminated the oscillator, high voltage power supply and moved the micro into the fabric of the SPC.  We also added ferrite inductors the the I2C and power line connections in order to reduce the the possibility of the radiation noise out those lines.  The I2C connections also have 4k ohm pull-up resistors which also helps in the reduction of radiated noise on those lines.  Finally, we modified the PCB to include a full internal chassis ground layer.  We tested with these modifications in place and the issue appeared to be resolved.

All of the LIDAR-Lite modules that are now in the field include these design features, so we were surprised to hear that the issue had popped up again.

We are installing our sensor on an IRIS system and will begin to troubleshoot the problem in a production drone rather than relying on lab testing.  Our first impressions are that the problem may be related to grounding and wire routing issues.  We think that it may be beneficial to have a common single point ground for all of the devices in the drone as well as simple shield for the GPS.  In our early tests with the "noisy" prototype sensor a simple grounded metal plate between the GPS receiver and the rest of the electronics did reduce interference significantly.

As we learn more we will let everyone know what we discover and offer our suggestions on how to mitigate this issue.

[Bret Chilcott]

Guys:

We are having the same issues you are describing above on our systems in Kansas.

I am ordering an RF Explorer to help diagnose the situation.

I have found when the LIDAR is near the Pixhawk, regardless of the distance that the GPS/Compass is from the LIDAR, the GPS becomes INOP.  We have the LIDAR positioned 18' away from the PixHawk and LIDAR, but still INOP.

Tomorrow we will experiment with shielded cables and ferrite cores.

Also we have found even when the LIDAR is separated from the PIXHAWK by a aluminum foil wrapped box, the results are the same - INOP GPS.

We will report back tomorrow.

Thanks

Bret C

[Dennis Corey]

Bret,

We're working on the problem and have some good ideas as to what might be causing the issue.  There are a couple of tests we're conducting to see if our suspicions are true.  Once we have more information we'll be posting it to the community.  

Dennis

[Chris Anderson]

One theory is that it's just hammering the power line so hard that it's killing the GPS. I think the GPS gets its power from the I2C connector. 

--

You received this message because you are subscribed to the Google Groups "drones-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to drones-discus...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[PL Dennis]

We’re narrowing it down to what we think is a grounding issue.  In simple terms any noise that gets into the ground lines for the electronics has no common ground to leak off to and is circulating around  in a ground loop creating RF interference.  We’re experimenting right now with a possible solution.  Let you know soon how it works.

Dennis

You received this message because you are subscribed to a topic in the Google Groups "drones-discuss" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/drones-discuss/VpQe2t4juk4/unsubscribe.
To unsubscribe from this group and all its topics, send an email to drones-discus...@googlegroups.com.

[Josh Welsh]

The serial connection provides power to the GPS and the GPS will function w/o the I2C connection connected, but are you stating that the GPS will prefer power from the I2C connection if both are connected?

[Craig Elder]

The GPS is only powered from the serial +5V.  The +5V in the I2C cable is not connected on the GPS side.

[Josh Welsh]

On Wednesday, December 10, 2014 1:18:51 PM UTC-8, Craig@3DR wrote:

The GPS is only powered from the serial +5V.  The +5V in the I2C cable is not connected on the GPS side.

Ok thanks, that was my understanding as well.. especially since I have three that are connected and working with no connections to the I2C port. 

[Robert Lefebvre]

Though it's still possible that the Lidar is dirtying up the Vcc and/or Ground.

[Craig Elder]

Bret

Would you mind taking a photo or making a schematic of your wiring so we (Dennis & I) can attempt to reproduce the issue you are seeing.

Thanks,

Craig

[Bret Chilcott]

Craig and Dennis:

We can share a photo with you but because of the tight compactness of the Pixhawk mounted on top of a mounting box, with the LIDAR underneath it, it  is very difficult to see the wiring properly.  However, we do want to share some updates from yesterday:

• A ferrite core with the cable wrapped three times around it was added to the cable that attaches to the LIDAR positioned near the LIDAR
• The plastic box on which the Pixhawk mounts is wrapped in aluminum
• The LIDAR and the ferrite core mounts underneath the mounting box
• LIDAR, airspeed and compass are all plugged into the I2C buss
• The above helped resolve the loss of GPS but by moving the LIDAR cable by hand, around the PixHawk we could occasionally cause the GPS to fail
• It is noteworthy to know that we also tried shielded wires between the GPS and COMPASS with no improvement in GPS reliability

Our chief engineer Javier, tried the following with good results except for possibly causing errors in the measured airspeed. 

• A new cable for the LIDAR was made where it pulled power for the LIDAR directly from the ESC and NOT the I2C
• The new cable cable fed the signal into the existing I2C buss but without the power into the I2C bus
• The ferrite core was left on the LIDAR lead positioned next to the LIDAR inside the aluminum wrapped box (The aluminum wrapped box is open on the bottom where the LIDAR sets underneath facing down.  Also aluminum was added to the bottom of the cavity where the LIDAR has visibility of the ground through a hole in the bottom of the aircraft)
• This appeared to completely resolve the loss of the GPS.  Wires were moved by hand without loss of GPS and the LIDAR signal as measured by the tuning page on Mission Planner indicated the LIDAR was operational
• The above setup was flight tested twice with excellent results except for dramatic changes in the autolanding which caused the aircraft to land 200 feet long consistently during the two test flights
• I asked Tridge to take a look at the logs.  He is not available at this time for in depth analysis but he saw dramatic changes in logs of the measured airspeed which may be caused by "dirty" signals in the I2C bus, but more testing needs to be done to confirm this.

Craig and Dennis, we are very eager to help you resolve this issue.

What do you think should be tested next?

Thank you.

Bret C

AgEagle.com

Br...@AgEagle.com

Bret Chilcott

117 S. 4th Street

Neodesha, KS

620-332-9264

www.AgEagle.com

www.facebook.com/AgEagleLLC

twitter.com/BretChilcott

The "Tractor Tough" AgEagle flying wing.  

Learn how the AgEagle can help yield higher results!

--
You received this message because you are subscribed to a topic in the Google Groups "drones-discuss" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/drones-discuss/VpQe2t4juk4/unsubscribe.
To unsubscribe from this group and all its topics, send an email to drones-discus...@googlegroups.com.

[Craig Elder]

Thanks Bret
Your observations are in agreement with the testing we did that. The results indicate the noise is conducted and not radiated.
We're going to have to work out the exact recommendations and details but using a separate power supply is quite viable.
If you post your log here or email it to me we can review any issues with I2C noise.

Craig

[PL Dennis]

Hi Bret,

Great progress and your findings do confirm that the problem is conducted not radiated.  We are going to be testing a cable modification today that we hope will resolve the issues.  It’s basically 100pf capacitors connected between the leads from the sensor to ground.  We’ll let you know soon how it goes.

Cheers,

Dennis
PulsedLight, Inc

[Bret Chilcott]

Dennis:

Thank you for the update.  Please advise if your test works as planned.   If it does work, we will duplicate here in our shop to verify.

Thank you.

Bret c

AgEagle.com

Br...@AgEagle.com

Bret Chilcott

117 S. 4th Street

Neodesha, KS

620-332-9264

www.AgEagle.com

www.facebook.com/AgEagleLLC

twitter.com/BretChilcott

The "Tractor Tough" AgEagle flying wing.  

Learn how the AgEagle can help yield higher results!

[PL Dennis]

If it does, we’ll let you know and then post a schematic for everyone.  Also if it does work we’ll add the caps to our sensor in addition to the inductors that are already in place.

Dennis
PulsedLight, Inc

[Julien Dubois]

Craig, from my tests (10 days ago), I had the feeling that the noise was MAINLY conducted, even if only power wires were plugged to the LIDAR (the 4 other wires completely removed) and even if the LIDAR was powered from an external source (other battety, other Ubec, with or without ground common).
But it is as well radiated at short distance, and the aluminium foil around the Lidar was the evidence.
Actually, I think that a part of the conducted noise is the "radiated one" collected by the Lidar wires.
A capacitor or ferrite help making a low pass filter on the wires and reduce the conducted noise

BTW Bret, every 4 wires are coiled around the ferrite core or only the 2 power ones?
Julien

[Bret Chilcott]

Julian:

In our test with the ferrite core, we wrapped all four wires.

Thanks

Bret C

[Julien Dubois]

Thanks Bret, I've done this test and that is a very good improvement.
2 turns with the 4 wires reduce greatly the conducted noise, I2C communication still works fine. I've placed the core at the lidar connector.
The radiated noise is still there, you can feel its effect if you place the Lidar close to the GPS (<10cm)
Lidar in a conductive bag, noise if well reduced.
Maybe it would be worth to change the plastic material of the Lidar case by a conductive plastic or a "carbon loaded" plastic.

BTW, I've compared the M6 with M8... and actually M8 helps a lot.
The Lidar noise kills 100% of the M6 GPS whereas M8 only looses a bit of sats intensity, HDOP remains unaffected (1.1 indoor) with lidar-GPS>5cm and the ferite core.

The M8 GPS I've used is very impressive, a bit big but so efficient!

To unsubscribe from this group and all its topics, send an email to drones-discuss+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[Bret Chilcott]

Julian:

The M8 GPS looks vey interesting.  What do you use for an external compass Egan you use this GPS?

TNX

Bret Chilcott

AgEagle

117 S. Fourth Street

Neodesha, KS 66757

620–332–9264

[PL Dennis]

We’re still working on testing cable with the caps on it.  Our current issue is getting the Pixhawk to work with our sensor…

Dennis
PulsedLight, Inc

[Julien Dubois]

For the moment I've not used any external compass with my PixHawk but I will probably use the one built in this GPS. If you're interested, I can give a try and check if it works well.

[Bret Chilcott]

Ok- it would be great to know how well the compass works.

Thanks!

Bret Chilcott

AgEagle

117 S. Fourth Street

Neodesha, KS 66757

620–332–9264

[Josh Welsh]

Interested in this too... Not to thread-jack, but, is there any reason to expect the support for the current popular compass (HMC5883L) to not work with the HMC5983 on this unit?

[gervais]

Don´t worry , the HMC5983 @ the CSG units with NEO8 works great , we are usings this already since a couple of months.No need to use the internal mag.

[Glenn Gregory]

Hi all,

What was the consensus about the best method to wire up the Lidar-Lite to prevent interference with GPS etc.?

Ideas presented so far:

• Ferrite core (2 turns with all 4 cores at Lidar connector end)
• Capacitors (LPF using 100pF, sensor to ground)
• Insulating via conductive bag / alluminium
• Separation of Lidar to GPS
• Shielded Lidar cable did NOT help.

My Lidar-Lite and PX4 have just arrived and I want to wire up correctly and try to avoid the lessons everyone here has learnt.

Very keen to get this setup running to help test and contribute in this area.

Thanks,

Glenn

[Randy Mackay]

Glenn,

 

      My understanding is that the most important thing is to power the LIDAR-Lite using a separate BEC.  Apparently the majority of the interference is travelling along the power wires from the range finder to the Pixhawk (or whatever flight controller).  So basically use a separate 5V BEC (which may pull its power from the same main battery) and attach the BEC’s 5V output to the Lidar-Lite’s 5V input.  Then connect the BEC’s ground wire to both the LIDAR’s ground and the Pixhawk’s rear servo rail’s gnd pin (the top one).

 

     I think Hamish is going to correct the LIDAR-Lite wiki page in the very near future to show this.

          http://copter.ardupilot.com/wiki/common-rangefinder-lidarlite/

 

-Randy

To unsubscribe from this group and all its topics, send an email to drones-discus...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--

AgEagle.com

Br...@AgEagle.com

Bret Chilcott

117 S. 4th Street

Neodesha, KS

620-332-9264

 

www.AgEagle.com

 

www.facebook.com/AgEagleLLC

 

twitter.com/BretChilcott

 

 

The "Tractor Tough" AgEagle flying wing.  

Learn how the AgEagle can help yield higher results!

--
You received this message because you are subscribed to a topic in the Google Groups "drones-discuss" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/drones-discuss/VpQe2t4juk4/unsubscribe.
To unsubscribe from this group and all its topics, send an email to drones-discus...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to a topic in the Google Groups "drones-discuss" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/drones-discuss/VpQe2t4juk4/unsubscribe.
To unsubscribe from this group and all its topics, send an email to drones-discus...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

[Glenn Gregory]

Thanks Randy I will implement this method after direct connection method to see the difference in radiation/interference.

Is there a reason for using the rear servo rail gnd over the I2C gnd? As I assume these are directly connected internally. Probably to move the interference further away from the GPS I2C.

[Randy Mackay]

Glenn,

I'm not sure about using the I2C GND or the rear servo rail but my guess is that it doesn't matter which you use.

-Randy

-----Original Message-----
From: drones-...@googlegroups.com [mailto:drones-...@googlegroups.com] On Behalf Of Glenn Gregory

[Robert Lefebvre]

I'm curious about why using a BEC solves this problem.  I guess it is acting as a really strong filter.  Shouldn't it be possible to build a really good filter that would function just as well, while allowing for simpler wiring?

[Dennis Corey]

Hi All,

While we haven't been able to reproduce the problem here (of course...isn't that the way it always is?) the feedback and testing we've done with Bret and others indicates that the issue is primarily conducted noise, not radiated from the sensor itself.  For now, we've come up with a cable modification and some other recommendations that we feel will resolve the issue should you be experiencing it.  I've attached a schematic for the cable modification.  Our next build of sensors in January, which will be released in February, will include 50pf capacitors (100pf is not necessary) on the board as part of the PCB design so the cable mods will not be needed.  

In our investigation of this problem we've come up with areas that could lead to GPS interference and recommendations;

• The GPS receiver should not be placed directly above the autopilot electronics and a ground plane shield (a grounded metal plate) between the GPS receiver and the electronics is highly recommended in all installations.
• Wiring lengths for all the sensors and electronics should be kept as short as possible and routed away from the GPS receiver.  Twisting of the leads on all of the electronics as has been mentioned in this post is also a good practice.
• Since the RF interference that is being seen in this issue is conducted and not radiated, an aluminum shield around the LIDAR-Lite sensor is not necessary or recommended as an ungrounded shield will act as a radiator for any RF noise that may be present and worsen the problem.  And shielded cables that aren't properly grounded will really give you headaches!.
• Power LIDAR-Lite from the BEC rather than the Pixhawk I2C bus and, optionally, run the power and power ground leads through a ferrite choke.  Depending on the number of devices connected to the Pixhawk, power up can be very slow due to the current load of all the devices.  Slow power ramp up can result in initialization problems with the sensor's Signal Processing Core. This both eliminates the initialization problem as well as reduces/eliminates the conducted noise on the power side by connecting the sensor ground to a common ground point.
  
• Connecting all electronics and shielding to a single point ground bus will eliminate ground loops and the resulting conducted RF interference.  This is a standard practice in all RF systems.

As commercial drones become more complex and additional sensors and electronic payloads are installed, the issue of RF interference with navigation and on-board systems will become increasingly important to address.  Implementing common practices in the reduction of RF interference, as is commonly done in modern avionics systems used commercially and in the military, will be an important next step in the evolution of commercial drones and drone safety.

Please let us at PulsedLight know of your success (or failure) with these modifications.  We are interested in hearing how effective these are in solving the issue for you.

Dennis

PulsedLight, Inc.

[Avidh bavkar]

if you are using an apm 2.6 maybe you could insulate the bottom of the gps module with an absorbent or reflective material so that any interference from the lidar system will not affect the gps system 

[Philip Rowse]

the interference is via conduction, ie, comes via the power cord.  powering it as recommended in the wiki fixes the problem.

That being said, we put copper foil under the GPS on IRIS+ to avoid unwanted interference, so it does not hurt to do this.  

Philip Rowse

Lead Systems Engineer

3DRobotics

424 Nicholson Street

Ballarat 3350

Australia

--

[Craig Elder]

We have a brand new sketch in the wiki showing the recommended connection  http://planner.ardupilot.com/wiki/common-rangefinder-lidarlite/#Connecting_to_the_Pixhawk

[Chris Anderson]

Craig, if I'm powering the LIDAR-Lite with an external BEC, as the wiki suggests, should I still build the filter board with caps that Dennis recommends for the signal wires?

-c

[Julien Dubois]

From my tests and results,  the external power supply was not the solution, only the distance or a low-pass filter on the wires reduced the noise.
Maybe you had a different conclusion with a different setup but obviously there's no general case with a single solution
Using external power supply will help to move the wires away from the GPS/FC depending on your setup... or not
The right solutions is probably a mix of all the good ideas suggested here:
- distance / external supply
- twisted cable
- capacitors (LP filter)
- feritic core

[Dennis Corey]

Thanks Craig for updating this information!  One other update would be to make sure that the maximum range parameter is set to 4000 and the minimum range parameter is set to 0.

[Dennis Corey]

Chris, I would try connecting directly to the BEC first, shielding the GPS receiver second and then the filtering caps on the signal and power leads.  

Dennis

[Dennis Corey]

Agreed!  We're finding that not every installation has the problem and if they do the resolutions varied based on a number of different factors.  In addition to your suggestions we would add shielding the GPS receiver, as has been suggested by Phillip, as an easy and effective option.

Dennis

[Chris Anderson]

As one of those who also had total GPS blackout with LIDAR-Lite, I'm keen to try every solution. Dennis, here's my cap board, as per your instructions. Do I also need the ferrite core I've shown here? I'm hoping I won't have to use a separate BEC (space reasons), but have one standing by in case it's needed.

[Jaime Machuca]

Chris,

Those two caps look too big to be 47pf are they the right ones?

Jaime 

Sent from my iPhone

On Jan 4, 2015, at 3:04 PM, Chris Anderson <ch...@3drobotics.com> wrote:

As one of those who also had total GPS blackout with LIDAR-Lite, I'm keen to try every solution. Dennis, here's my cap board, as per your instructions. Do I also need the ferrite core I've shown here? I'm hoping I won't have to use a separate BEC (space reasons), but have one standing by in case it's needed.

<unnamed.jpg>

[Chris Anderson]

That's what they say. (I just happened to have tantalium caps in my parts box)

​

[Philip Rowse]

They are uF not pF :)

On 5 Jan 2015, at 08:21, Chris Anderson <ch...@3drobotics.com> wrote:

That's what they say. (I just happened to have tantalium caps in my parts box)

<unnamed (1).jpg>
​

[Dennis Corey]

Nice job!  We have an opening for an assembler... ;-)

We should have probably specified something along the lines of ceramic disk capacitors that would be a little smaller.  What you have should work, but I need to check with the real engineer in our company to get a confirmation.  As far as the ferrite core, I'd try the set up without it.  The one you're using is a little on the large side and probably only needs to be used on the power leads if you do install one.  As far as the BEC, you can use your existing one and wire the sensor in parallel with anything else that's connected.

Dennis

[Chris Anderson]

What's a factor of a million among friends??

Thanks for the catch, all!

[Julien Dubois]

Jaime is right, your capacitors are 1000x too big (pF, not µF)

About feritic core, I was using much smaller one with good results, like that http://www.directindustry.com/prod/tianjin-huigao-magnetics-co-ltd/inductor-ferrite-cores-79831-1036723.html

[Dennis Corey]

Wow!  Good catch!  X1000!

[Lorenz Meier]

Hi,

I’m a notorious dyslexic when it comes to metric prefixes, so I keep this table handy:

http://en.wikipedia.org/wiki/Metric_prefix

Don’t forget about nano between micro and pico, it would make the prefix very sad 8).

-Lorenz

[Philip Rowse]

Lol, this is an easy thing to do, I have done so myself too many times! :)

Philip Rowse 

Lead Systems Engineer

3DRobotics

Ballarat

Australia

[Chris Anderson]

Dennis, after checking through my parts box, I don't have any of the right size. Looks like I'm going to have to design a board to do it all with surface-mount components :-(

In the mean time, I'll try this one anyway (even with the wrong values) and see what happens. 

-c

[Lorenz Meier]

Hi Chris,

You can probably use the board you started - if you have your CC ready, I’m sure this one will be there in no time:

http://www.digikey.com/product-detail/en/K470J15C0GF5TL2/BC1009CT-ND/286631

Independent of that, a surface mounted board with connectors would be great - anyone interested to hack one in Eagle? I can help with the connector footprints.

-Lorenz

[Craig Elder]

Chris,
My bet is you would be better off running a BEC and forgoing the cap board. The caps probably need to be located within a few mm of the lidar unit to be very effective anyway. If you are using your Iris, you can pull a 5V supply off the quad ESC board and not add any weight or wires to the vehicle.
Craig

[Dennis Corey]

Worth a try, although it might give you problems with the I2C communications.  It might be better to try powering the sensor from the BEC before trying the cable mod.  Our next production run will include the caps on the PCB.

Dennis

[Dennis Corey]

Craig, You're correct.  I did the mod with the IRIS we have here and it seems to work well.   I also added a GPS receiver shield.  While we started with decent GPS reception, I did see a small improvement in signal afterward.

Dennis

[Andy Lee Robinson]

Guys, we should really know  this stuff... a µF is a million times bigger than a pF. nF sit in between...

Having read this thread, there seems to be a bit of stamping around in the dark going on!

Although twisting wires, adding shielding and LC circuits are good practice, these are still like bandages on a patient with a broken artery.

The components generating the noise, the circuit and physical layout need fixing, though I appreciate that could be difficult with short high energy pulses to manage.

Has anyone done any quantitative analysis of the radiated noise with a spectrum analyzer and the conducted noise with an oscilloscope to determine frequency distribution and characterization?

Perhaps opto isolation of the energetic stuff would help?

[Robert Lefebvre]

Wouldn't a PI filter give much better attenuation than a simple capacitor bank?

http://en.wikipedia.org/wiki/Capacitor-input_filter

[Dennis Corey]

It might be, although we don't know that it would have been significantly better than the solution Bob came up with, which, independently or along with the other proposed remedies, appears to address the issue in those instances where there has been GPS interference.  

Dennis

[Chris Anderson]

Success!  I was able to find some 33pf caps (not 47pf as spec'd, but close enough) and redid my cap board. Problem solved -- I get both solid GPS and Lidar reading!

Can't wait for this filtering to be built into the board.

​

[Dennis Corey]

Yippee!  I can start breathing again...  The build of systems scheduled for late January will have the filtering on the board.

By the way did you also connect the sensor power to the BEC?

Dennis

[Julien Dubois]

Nice done Chris,
could you please compare with u-center Sats intensity with and without the fix?

eg, with the feritic core, the sats intensity is decreasing by 20% if I place the Lidar close to the GPS, but not enough to lose the sats, that's why it's still working well, HDOP remaining acceptable but the filter is obviously not perfect. I guess it will be much better with this board but that needs to be compare-tested.

[Chris Anderson]

I got 9 sats without LIDAR-lite,  zero with it and no filter,  and 8 sats with it and the filter. I didn't track the HDOP rigorously,  but it broadly tracked that.

[Julien Dubois]

Thanks Chris for your feedback.
I've found u-center was much more efficient to check it.
eg: we can have 8sats with 60%signal intensity for each and 8 sats with 80% intensity... MP will show always the same status (8 sats), maybe hdop will be more or less good.
with u-center, you can see directly the effect on each sats with their signal intensities.

Anyways, between 0 and 8... no need such details to figure out the effect of the fix! that's such a thought for fine tuning.

[Jonathan Challinger]

eg: we can have 8sats with 60%signal intensity for each and 8 sats with 80% intensity... MP will show always the same status (8 sats), maybe hdop will be more or less good.

My understanding is that HDOP doesn't change with signal intensity, only with the number of satellites being used and their positions.

[Wayne Holder]

I did a quick analysis using a spectrum analyzer and found a possible reason for the interference.  You can see the process I used in this video:

  https://www.youtube.com/watch?v=32gf5NJb3JY&feature=youtu.be

The short version is that the LIDAR lite unit I tested seemed to put out a conducted signal centered at about 528.5333 MHz, which puts it at about 1/3 the frequency of the GPS L1 signal of 1575.42 MHz (the primary signal used by consumer GPS modules.)  So, I suspect that a 3rd harmonic of this signal may be an issue.  I don't have the EMF probes to due a proper analysis, but LIDAR lite manual says the unit uses a 500 MHz "sampling clock" so I suspect this is what I'm seeing except shifted up a bit in frequency.  If this 500 MHz clock is not well controlled for frequency, it may vary from unit to unit, which could explain why some people seem to have this issue and others don't.

Wayne

[Craig Elder]

Thank you Wayne for performing this analysis for us.
I look forward to comments from Dennis and Bob about the clock frequency.

--

[Lyle Larson]

Hi Julien,
We're you able to observe any energy at the third harmonic of the 528mhz on your spectrum analyzer?
Lyle

[Julien Dubois]

Hi Lyle,

I think your question is more addressed to Wayne Holder. I have not the equipment to perform such tests.

Julien

[Lyle Larson]

Yep...;-}

[Andy Lee Robinson]

Thanks for sharing, I think you've nailed it.

500 MHz sounds a lot for a machine that goes ping! :-)

A square wave isn't square without a large 3rd harmonic so it must have a fairly high amplitude, and if it falls close to GPS frequency then it would be the obvious culprit even if the GPS was equipped with a SAW filter.

Perhaps there's a way of adjusting the clock frequency, or another method of sampling?

I don't know the lidar's power consumption, but a separate battery with opto-isolated signals in a shielded case should almost eliminate the interference.

[Dennis Corey]

Hey Wayne, thanks for taking the time to run this test.  It does confirm that the RF noise is conducted not radiated, so the fix we’ve posted, adding capacitors to the I/O and power leads, will solve the problem.  Also attention to grounding and wire routing will contribute to keeping RFI to a minimum from other sources.  In our investigation of this issue many of the sources of GPS interference can be resolve by these simple measures.

As a point of information, LIDAR-Lite has two clocks contained within the signal processing chip.    One operates at 125mHz and the other at 250mHz.  The signal processing gain achieved by our  algorithms allows us to achieve performance comparable to more expensive systems utilizing much less costly and lower power consuming emitter/detector combinations.  We are doing a serious amount number crunching within the chip - every 8ns we do 64 multiply and accumulates in our correlation processing algorithms.  We are using processing power to achieve our performance, not complex and costly hardware.

In addition to the measures that are already a part of the LIDAR-Lite design to eliminate RFI, a full PCB ground plane and inductors on the power and I/O lines, we are also implementing the capacitive filtering on future builds of the LIDAR-Lite PCB.  This should resolve any issues related to conducted RF interference.

I know I’ve said this before, but it bears repeating.  As more sophisticated devices requiring more powerful processors or FPGA’s, like LIDAR-Lite, find their way onto drones, more attention is going to need to be paid to issues like grounding, power distribution and the routing of wiring in order to avoid issues similar to what we’ve encountered here.  These are standard practices in the aviation industry in general and do need to be addressed by this community in order to make our drones safer and more reliable.

[Philip Rowse]

Thanks Dennis

    100% agree.  These are aircraft.... Please set up your aircraft well :) 

Philip Rowse 

Lead Systems Engineer

3DRobotics

Ballarat

Australia

--