How to de-tune cables that are picking up RFI?

[e100]

I have a 50m long 'alarm type' non-twisted pair cable which I'm driving using max485 chips with appropriate bias and termination resistors.
Unfortunately it appears to be picking up interference from local AM radio transmitters. There's a 50kW 720kHz transmitter 6km away and there are also 585 kHz and 810 kHz transmitters in the same area.
To get the packet corruption below a tolerable 4% I had to drop the baud rate to 9600 and add a bunch of 10nf caps between the A and B wires to ground.

A Google search on RFI suppression didn't turn up much.
Is there a more scientific way of de-tuning cables in this kind of situation?

[Ian.M]

Common mode chokes and tuned notch filters would probably help.
Is there any way you can rewire with STP cable?

[e100]

Common mode chokes and tuned notch filters would probably help.
Is there any way you can rewire with STP cable?

I would prefer not to rewire as the installation is outdoors and the current cable has shown good weather and UV resistance.
Outdoor cat5 isn't cheap and installing conduit to house indoor rated cat5 would be expensive.

[johnkenyon]

I have a 50m long 'alarm type' non-twisted pair cable which I'm driving using max485 chips with appropriate bias and termination resistors.
Unfortunately it appears to be picking up interference from local AM radio transmitters. There's a 50kW 720kHz transmitter 6km away and there are also 585 kHz and 810 kHz transmitters in the same area.
To get the packet corruption below a tolerable 4% I had to drop the baud rate to 9600 and add a bunch of 10nf caps between the A and B wires to ground.

A Google search on RFI suppression didn't turn up much.
Is there a more scientific way of de-tuning cables in this kind of situation?

Two suggestions:

1) Clamp on ferrite bead clamped on cable
2) Clamp on ferrite bead, with the cable looped several times
2) Disconnect one end and wrap a few turns of the cable on a ferrite ring.

A Google for "ferrite ring rfi" then clicking on "images" will show some examples of clamp on beads, and ferrite rings, and also a handful of demonstrations of how they are used.

[vk6zgo]

I have a 50m long 'alarm type' non-twisted pair cable which I'm driving using max485 chips with appropriate bias and termination resistors.
Unfortunately it appears to be picking up interference from local AM radio transmitters. There's a 50kW 720kHz transmitter 6km away and there are also 585 kHz and 810 kHz transmitters in the same area.
To get the packet corruption below a tolerable 4% I had to drop the baud rate to 9600 and add a bunch of 10nf caps between the A and B wires to ground.

A Google search on RFI suppression didn't turn up much.
Is there a more scientific way of de-tuning cables in this kind of situation?

Let me guess,---You're in Perth,right?
It might have been a good idea to do some research before you or your customer installed this system---the station has been there since 1932!

You could try series resonant traps to ground instead of the 10uF caps.
Can you "retrofit" some sort of shielding to the cable?
How about using  two lengths of RG58CU (one for each conductor) & grounding the shields ?

[e100]

Let me guess,---You're in Perth,right?
It might have been a good idea to do some research before you or your customer installed this system---the station has been there since 1932!

You could try series resonant traps to ground instead of the 10uF caps.
Can you "retrofit" some sort of shielding to the cable?
How about using  two lengths of RG58CU (one for each conductor) & grounding the shields ?

Yes, I'm in Perth (Western Australia) and this is something that I've built for my own use. I'm using 10nF, not uF caps.
Are these radio transmitters notorious for causing RFI in the Perth area?
 

[CJay]

It's going to be a pain in the arse.

RS485 should be over shielded twisted pair, you're going to find it nigh impossible to get rid of the interference on what's essentially a nice antenna.

Basically, anything you fit to  the cable will reduce the data transmission speeds you can achieve but as others have said, ferrites may help, it may be possible to make it a balanced line though I doubt it'll work well.

Possibly some 'shotgun' satellite dish cable might make it workable, that should be dirt cheap and weather resistant enough, there are also outdoor rated CCTV cables that are designed for RS485 which would probably be my choice.

What speeds are you trying to achieve?

[e100]

It's going to be a pain in the arse.

RS485 should be over shielded twisted pair, you're going to find it nigh impossible to get rid of the interference on what's essentially a nice antenna.

Basically, anything you fit to  the cable will reduce the data transmission speeds you can achieve but as others have said, ferrites may help, it may be possible to make it a balanced line though I doubt it'll work well.

Possibly some 'shotgun' satellite dish cable might make it workable, that should be dirt cheap and weather resistant enough, there are also outdoor rated CCTV cables that are designed for RS485 which would probably be my choice.

What speeds are you trying to achieve?

The original plan was to run at 38400 baud (half duplex) so I could poll several devices within 500ms.
At 9600 it's just about doable but there is no time to deal with lost packets and retries.

[tggzzz]

"Detuning" is probably the wrong concept.

vk6zgo's resonant traps might help, provided there are no significant signal components at those frequencies.

Radio hams have to do this kind of thing, and they have developed many techniques. See, for example, https://www.amazon.co.uk/RSGB-Guide-EMC-Robin-Page-Jones/dp/1872309488

Since it is long and outdoors, consider galvanic isolation, spark gap arrestors, and similar protection mechanisms used on phones/modems.

[Ian.M]

Suppressing high level EMI <1MHz is *DIFFICULT*. I'd start with about 15 turns of the whole cable wound on the core of a microwave oven transformer or similar at each end to form a common mode choke.

[vk6zgo]

Let me guess,---You're in Perth,right?
It might have been a good idea to do some research before you or your customer installed this system---the station has been there since 1932!

You could try series resonant traps to ground instead of the 10uF caps.
Can you "retrofit" some sort of shielding to the cable?
How about using  two lengths of RG58CU (one for each conductor) & grounding the shields ?

Yes, I'm in Perth (Western Australia) and this is something that I've built for my own use. I'm using 10nF, not uF caps.
Are these radio transmitters notorious for causing RFI in the Perth area?

No,the radio transmitters are operating perfectly legally.
The onus is on you to design your system with sufficient immunity to their quite legitimate signals.
Do you have any problems with ADSL?
If not,that is an example of a system designed with sufficient immunity.

The "10 uF" was a typo.

[e100]

To follow up, the packet error rate is now less than 0.05% whereas previously it was about 4%.
I split the termination resistor at each end and added a 10nF cap with one side tied to signal ground as per the TI RS485 design guide http://www.ti.com/lit/an/slla272b/slla272b.pdf (Figure 5).
I also put a 10nF cap across the A and B lines which removes much of the high frequency differential noise. At 9600 baud the edges are slightly rounded, but not enough to cause problems for the Max485 receiver circuitry. I could probably run at 19200, but happy to continue with 9600 for the time being.
I also added 100uH chokes and 10nF bypass caps at either end of the Vcc line that powers the remote equipment.

From reading various documents I now realise that my early attempts at interference suppression simply turned some of the common mode noise into differential noise which made things worse.


Mike