Running RF connections Between Modules

[German_EE]

If I am connecting two RF modules together how do I connect them to ensure that there is minimum amount of radiated signal? The standard reply to this question is 'everything in screened boxes and connect using coax cable' but this ignores two things:

1) The RF equivalent of 'hum loops'

2) I was always taught that an RF screen is only effective if it is not passing any current, and RF will pass along the coax shield, the outside of the case (using skin effect) and along the chassis.

I have been considering using TWO connectors at each end and sending signals using balanced lines, maybe this would be better?

Thoughts please from those who know more about RF than me.

[tfr000]

Yes, I have used twisted pairs to transfer high-freq signals around inside a box. I didn't do any rigorous measurments or anything, but it seemed reasonably interference- and radiation-free. The particular application required a very quiet electrical environment.

This was basically taking two lengths of (22 awg) insulated wire and twisting them up tightly with a power drill. One signal, one ground. Characteristic impedance of this is in the 100-120 ohm range. Much easier to deal with than coax.

[rhb]

In HPAK RF gear all the connections at RF are commonly SMA & coax.  All the control signals use feedtrhu capactiors.  That's what I plan to do as I have a long list of things to either package (eBay modules) or build.

The ground loop issue is commonly  overstated.

[tkamiya]

As I understand it, shield in coax works in a way to trap the electrical field between center wire and inside of the coax, not radiating outwards.  Return current due to mismatch will flow on outside.

In order for twisted pair shield to work as you said, I'd think it'll have to be one connector at each end and shield to entrap both center wires at the same time.  In this case, I think, the electrical field takes place between the twisted pair.  Shield is merely a shield, and it is to be connected only on one side.

[TheUnnamedNewbie]

Ground loop is less of an issue at high frequencies since the inductance of the loop is very high. That is the power of transmission line models - you incorporate both feed and return currents.

[dmills]

Ground loops are pretty much a non issue because the admittance of the loop is very small compared to that of the cable screen (Because of the massively larger loop area giving rise to a massive inductance).

At low frequency, or if you are doing really crazy things you can also trivially add a ferrite tube on the coax line to raise the common mode impedance and further reduce the loop currents, the ability to do this after finishing the layout is one very good reason to like cable interconnects at least in prototypes.

Note however that screening cans and indeed coax is only really effective once the thickness of the screen exceeds a few skin depths, below that it is reasonably effective against E fields, but hopeless against the magnetic component. 

Regards, Dan.

[CopperCone]

Do they have multi layer shielded coax with dielectric between the shield layers?

I guess like 50ohm triax but meant for rf not high impedance work? Would it work better then twinax?

But you would need some kind of triaxial rf connector where the outside of the enclosure is connected to the outer shield.

But then you have the issue that the shielding can has inside and outside surface and that typically you cant design the shiekding can to slip over the pcb to form a 50ohm transmission line, so to prevent a loop you would need a quadax cable that has signal
, signal ground, shield interior and shield exterior?

Im really not sure whats going on at this point if you had triax rf connectors with floating signal layers.

And wont a ferrite cause some dm effects too?

I know its nuttery, given that 0.1db in rf is considered something like 5 digits at dc (not literarly but figuratively speaking with how difficult the measurements can be) but im curious if anyone wants to explain.

What kind of trickery would be required to impedance match the enclosures to the line? Like for stuff absorbed by the interior of the enclosure? I know you normally dampen them with ferrite sheets but what if you wanted more? Is it possible?

Whats the path of current from the inside of the enclosure on a regular coax connected box? Stays on the inside of the coax right? Kinda wierd to think about a radiation loss actually having a return but i guess this is only for very very high frequencies

I mean, you can only detune an antenna right? So even if capacitively detuned you have some kind of far field? But this would be only with like high mm wave stuff bordering on optics?

I think i have a misconception here, but how does this parameter look like on a telegraphers equation? Can you even have far field in an enclosure?

[David Hess]

If I am connecting two RF modules together how do I connect them to ensure that there is minimum amount of radiated signal? The standard reply to this question is 'everything in screened boxes and connect using coax cable' but this ignores two things:

1) The RF equivalent of 'hum loops'

Usually this does not matter as the RF will be AC coupled with the high pass cutoff greater than the potential ground induced noise.

2) I was always taught that an RF screen is only effective if it is not passing any current, and RF will pass along the coax shield, the outside of the case (using skin effect) and along the chassis.

That is true however alternative paths have a higher loop area (inductance) so they will not carry the RF current as well.

I have been considering using TWO connectors at each end and sending signals using balanced lines, maybe this would be better?

That is one good solution where the ground difference between the endpoints would create considerable common mode noise.  A common mode choke is sometimes included to enforce balanced operation.