A/P Servo Failure on TX

[Tasmag]

i all

I have a jabiru with 2 axis A/P that works a treat, until I broadcast on my second comm.

The antenna for this comm utilises crossed copper tape on the floor of the fuselage which continues around the Pitch Servo. The servo is effectively inside the ground plane, but not electrically connected to it.

When I transmit on this antenna for longer than a couple of seconds the pitch servo occasionally fails and requires a reboot of the A/P. (Seems to happen most when on 123.45)

Would the addition of an aluminium sheet or aluminium flyscreen over the copper tape shield the servo from the RF that is causing the failure?

 

And would that sheet need to be electrically isolated from the copper tape?

 

Thanks in advance

Mike

[Geoff_H]

Not quite sure of exactly what your system looks like but could be owing to am earth loop in the shield.  A cable shield MUST be only earthed at one end or the conductor inside can get all sorts of funny voltages appearing.  When RS 232/485 control systems were introduced I was a contractor and earned a tidy sum fault finding new installations.  Most time it was because the installer had earthed both ends.  A side cutter and disconnect one earth and bingo the control system fired up and worked.  

[RFguy]

mmmmmm. what is going on here is that the motor servo driver is getting RF into it and likely causing a cross conduction and thus a fail.

 

the pitch servo sounds like it is in the near field of the antenna . From memory most of those servos - the motor driver is in the main unit, so only motor currents are on cables.  The antenna is injecting current into the servo system because it is part of the antenna.

What you need to do here is run the cable to the AP servo in a shielded cable.  Attach the shield of the cable to the body of the servo motor. I'd suggest not connecting the other (AP controller) cable shield end at least to start. 

 

If you wrap the existing cable in copper or aluminium tape (again, connect shield to servo body) , you can acheive the same thing.

The shield needs to extend at least two x the antenna dimension.

 

What though may also happen is that this might affect the antenna .  Avoid shield lengths that are an odd multiple of 60cm +/- 15% .

IE make the shield 120cm or 240cm.  

What AP servo / AP system is it ?

Glen.

 

 

 

[RFguy]

"  A cable shield MUST be only earthed at one end or the conductor inside can get all sorts of funny voltages appearing. "
This is not correct.

 

[Tasmag]

Thanks Glen

 

The system is MGL and utilises CaBus for the control of the Servo.

 

Unfortunately, there are a number of other CanBus terminations within the ground plane area as well, so the shielding lengths you mention are problematic.

 

That was the main reason I was wondering if a metal shield would contain the lobing/Near Field of the signal and minimise the effect on the CanBus. (Probably not even remotely the correct terminology!)

 

Would a solid ground plane be an option to minimise the effect of the Near Field RF.

 

Mike

[RFguy]

Hi Mike OK so CANbus to the servo.  In this case.  and considering you have other CANBUS items,  then just brute force shield everything

IE all items in shielded cables.  foil is fine at this frequency although broaded is preferred (or partial briad and foil) .

What is probably happening (I do know a bit about CAN) is that the driver is self locking out  when it detects that condition- its used on CANBUS to avoid a sick controller jamming up the whole network.  If brute force shielding is too hard , 

Alternatives if you have access to rear of connector : 

Now another option for you , and which one somewhat depends  on the precise mechanism : 

1) two x 100pF disk ceramic capacitors from  CANBUS A and B  line to the ground / take it to the neg of the connector or servo body. 

That should kill the induced energy, take it to ground 

2) It's possible that you could end up with alot of energy in those caps. check if the VSWR changes after you have done this mod.

 

non intrusive ALTERNATIVES 

3) Additional benefit would be an RF choke ---- this would be feed the CANBUS feed through a few ferrite beads, or split beads.

you'll probably need at least 3 to 5 to make a dent, close to the servo. and put them on the other devices, also. and also put a coupl eon the MGL controller end. and put a couple on the COAX of the radio antenna. 

 

https://www.ebay.com.au/itm/354863733597?mkcid=16&mkevt=1&mkrid=705-154756-20017-0&ssspo=cjibypsiq0a&sssrc=2047675&ssuid=&widget_ver=artemis&media=COPY

 

https://www.ebay.com.au/itm/124994863518?mkcid=16&mkevt=1&mkrid=705-154756-20017-0&ssspo=SHweELfzT1y&sssrc=2047675&ssuid=&widget_ver=artemis&media=COPY

 

that 9mm ID and 5mm ID. use smallest that will clamp easily. 

 

also available in large IDs. You might need three in series to make the difference you need.  works betwen with the caps in (1) 

you can use the chokes every 60cm or so and should prevent the antenna getting any energy into the wiring...  (inserts a high impedance in series with the cable) .

I gather this : 

https://www.mglavionics.co.za/Docs/Servo user manual.pdf

so not so easy to put the capacitors (caps) in. Suggest some split  bead ferrite cores, in that case. 

 

when used with (1) , the beads will work more effectively. 

 

-glen

Edited August 9, 2023 by RFguy

[Tasmag]

Thanks Glen

 

Yes, that is the servo in use.

 

I'll get some Ferrites and give that a try.

 

Appreciate the thorough reply, especially considering the number of replies you give across the forum for such matters!

 

regards

 

Mike

[facthunter]

With all of this technical stuff not working well , does it have a proper breakaway  built into the design and does it power a tab Or the control surface itself.? Nev

[Geoff_H]

3 hours ago, RFguy said:

"  A cable shield MUST be only earthed at one end or the conductor inside can get all sorts of funny voltages appearing. "
This is not correct.

 

Well my experiences with instrument control systems says that it is. 45 years of instrument control system work I was wrong it appears.  When both ends of the shield are earthed a ground loop current can exist in the shield, this current can induce currents in the conductor it is shielding.  Earth only one end of the shield and no current from earth loops appears. But I submit to your superior knowledge.

[RFguy]

Hi Geoff

apologies, I should have written ' this is not always correct',  rather than ' this is not correct' . My point is that it  isnt a rule. 

 

In most control signal  cases, like RS232, 485, CAN,  earth or negative reference will be carried by its own conductor in a multicore, or on the shield. It MUST BE to ensure the common mode voltage is within limits- IE the voltage from either A or B side to ground or negative. or 'return'. Otherwise things blow up because the input circuitry is generally not galvanically isolated (layman : no DC path)  . An example of galvanically isolated is across  a transformer. There is no DC path .

Shields can be connected to various things at each end.  Shields are often connected to the same point the negative is.  But not always

There are some cases where the neg might be a separate ground in the device. Here are some examples :  - the power supply pairs might go through a common mode choke or filtering, and the shield is connected to the metal box   and not circuit ground , IE the negative is connected to the case via a choke, or via a high impedance- so that the AC impedance of the negative connection to case is >> zero.  

This is done in industrial settings to prevent UNINTENTIONAL  return currents flowing through sensitive signal shields. 

IE a shield is usually a large conductor, and due to its lower resistance, a portion of heavy currents to a DC motor (or imbalance in a 3 phase system)  might flow in the shield. In this case, only connect one end- as long as the multicore does have a negative return /reference assuming is DC referenced. 

For problematic systems, transformer isolation virtually eliminates ground. earth loop issues because a negative return does not need to be connected- the signals are purely referenced to each other  and never to ground .

 

All this is relative to the signal levels encountered. For dealing with  moving coil microphones with signals ~ 20mV, small parasitic currents in shields and grounds can generate a few mV  of interference.  

But so say, an RS232 signal -  +9, -9V, , that few mV of interference wont matter at all.

 

That's fine for DC and low frequency AC

 

When it comes to HF/ VHF  RF AC, its a difference ballgame.  In high RF environments, shields on multicore signals  at both ends are a must , and generally the cable shield must connect to a device shield of the device, whether a metal box, or a PCB ground.

When an shield is connected at one end, the RF will run on the outside, then back on the inside of the shield and couple to the internal cores. 

Long cables- the RF that gets on one end of the cable, say where the cable shield is device shield, the RF energy that gets on the outside of the cable , if long, may be substantially diminished by the time it gets to the end that does not have the shield connected, and not matter- this might be several wavelengths. 

 

There are some corner cases - If the shield is connected at one end and the shield is an odd multiple of a quarter wave at the frequency of interest, it will become part of the antenna system , but it may not if both ends are connected ! - multiples of quarter waves and half waves come into play here and the equivalent circuit becomes complex. So best method is to connect shields at both ends to stop the RF on the outside of the shield getting back on the inside. If the cable shield is connected at both device shielding ends, no RF can get into the devices

 

-glen

 

[RFguy]

Additionally-  while the antenna may have its "ground plane" elements , anything conductive in the near field - that the antenna can see - anything conductive between the antenna elements - essentially bounded by the dimensions of the antenna- will become part of the antenna. Your only option is to move the antenna or choke or shields the cables. 

You could stop the antenna seeing the cables by providing the continuous  conductive region (aluminium sarking 0.1mm is good) ABOVE the cables. The 'radiating element' of the antenna - in this case  a whip -  all the return currents will flow to the continuous ground plane as preference. 

gross example - a whip on top of a factory roof and feed through the roof- all the return current sflow between the whip and the top of the roof in the region of the antenna, no space currents are present  under the roof.