Correct way to bring external connections and power into a shielded circuit?

Thread Starter

Pinkamena

Joined Apr 20, 2012
22
Hello, AAC forum!

I have been thinking about how to do this for a while. How can I bring power and connections to a circuit that lies within a shield enclosure? Naturally, without rendering the beneficial effects of the shield useless in the process.

The optimal way would of course be to use shielded wires, but this is not always possible due to cost. As a real example, let's use the circuit I am currently attempting to build, and which forced me to look harder at this problem. The circuit is in short a relatively high gain (~40V/V) band-pass amplifier (with some other active components, not so important), connected to a silicon-based photon detector. The circuit is very susceptible to noise. Unfortunately for me, the band pass is smack middle in the FM broadcast radio frequency spectrum so I worry I may pick this up from my local radio tower. I have decided to keep the entire circuit within an enclosure so as to minimize the amount of cables that need to pass through the enclosure. The bare minimum is however:
  1. SPI communication wires
  2. Positive and negative power supply (20-30V) + ground return
  3. -56V bias for the detector + ground return
These wires are by necessity several meters long, nothing to do about that. And so, they will very likely pick up radio frequency noise. Is there some "magic trick" (probably not!) or guidelines for how to prevent bringing noise picked up by these cables into my enclosure?
Regards, Pinks

EDIT: I see now I posted this in a different subforum than I intended. This thread was meant for "general electronics chat".
 

GopherT

Joined Nov 23, 2012
8,009
Hello, AAC forum!

I have been thinking about how to do this for a while. How can I bring power and connections to a circuit that lies within a shield enclosure? Naturally, without rendering the beneficial effects of the shield useless in the process.

The optimal way would of course be to use shielded wires, but this is not always possible due to cost. As a real example, let's use the circuit I am currently attempting to build, and which forced me to look harder at this problem. The circuit is in short a relatively high gain (~40V/V) band-pass amplifier (with some other active components, not so important), connected to a silicon-based photon detector. The circuit is very susceptible to noise. Unfortunately for me, the band pass is smack middle in the FM broadcast radio frequency spectrum so I worry I may pick this up from my local radio tower. I have decided to keep the entire circuit within an enclosure so as to minimize the amount of cables that need to pass through the enclosure. The bare minimum is however:
  1. SPI communication wires
  2. Positive and negative power supply (20-30V) + ground return
  3. -56V bias for the detector + ground return
These wires are by necessity several meters long, nothing to do about that. And so, they will very likely pick up radio frequency noise. Is there some "magic trick" (probably not!) or guidelines for how to prevent bringing noise picked up by these cables into my enclosure?
Regards, Pinks

EDIT: I see now I posted this in a different subforum than I intended. This thread was meant for "general electronics chat".
Use a two conductor cable plus shielding.

Then use common mode rejection to kick out the noise picked up. The two conductors will be your +/- signal. The shielded cable is connected to your shielded enclosure and both connected to ground (at one point). Connect +/- signals to a differential amplifier. Use a 10-turn potentiometer in series with a resistor that is about 80% of your feedback resistor in series with a 10-turn potentiometer that is about 50% of your feedback resistor value. Adjust pot as needed to allow exact matching of common mode and exactly cancel out noise.
 

crutschow

Joined Mar 14, 2008
34,285
For additional reduction of any noise from wires carrying low frequency or DC going into the enclosure you can use feedthrough filters through the enclosure shield.
Also use shielded wires, which shouldn't be that expensive.
If you can't afford that, then you will have a real problem minimizing noise pickup.
 

WBahn

Joined Mar 31, 2012
29,979
Here's a small laundry list of things you might consider:

Optocouple those signals that you can, such as your SPI signals.

Minimize the longest dimension of the holes in the shielding -- it's generally better to have several small holes than one large hole.

Use differential signaling wherever possible, such as your SPI signals.

Use ferrite beads around all the conductors that penetrate the box.

Use conducting foil to shroud the holes into the casing and better connect the enclosure to the cable shields.

Use bulkhead connectors that are specifically intended for shielded enclosures.

Don't forget to deal with the long, thin gaps associated with the mechanical edge of the enclosure opening itself.
 

AnalogKid

Joined Aug 1, 2013
10,987
For standard size connectors like D-subs, many shielding products companies make shielded gaskets stamped to conform to the connector shape. These are essential for GHz frequencies, but still can make a difference at MHz freqs.

If you have a large hole for a plastic connector and there's no way to change that, put a 5-sided metal box (small casting or deep drawn can) around the connector on the inside of the enclosure, with a much smaller hole for the wires to pass through.

ak
 

WBahn

Joined Mar 31, 2012
29,979
For standard size connectors like D-subs, many shielding products companies make shielded gaskets stamped to conform to the connector shape. These are essential for GHz frequencies, but still can make a difference at MHz freqs.

If you have a large hole for a plastic connector and there's no way to change that, put a 5-sided metal box (small casting or deep drawn can) around the connector on the inside of the enclosure, with a much smaller hole for the wires to pass through.

ak
Out of curiosity -- why "5-sided"? I haven't done much beyond purely ad hoc enclosure stuff, so I don't recognize why this might be significant or relevant.
 

WBahn

Joined Mar 31, 2012
29,979
Out of curiosity -- why "5-sided"? I haven't done much beyond purely ad hoc enclosure stuff, so I don't recognize why this might be significant or relevant.
Oh, I just did a Google search -- are you referring to the kind of outlet box used for wall switches and outlets? I see that several sites refer to those as "five-sided boxes".
 

GopherT

Joined Nov 23, 2012
8,009
So they roll electrical boxes whenever the game calls for a five-sided die? :confused: ;)
I don't play, I just remember some weird kids with odd shaped dice about a hundred years ago. 5 cannot be done with all equal sides - but it was interesting to think about.
 

WBahn

Joined Mar 31, 2012
29,979
I don't play, I just remember some weird kids with odd shaped dice about a hundred years ago. 5 cannot be done with all equal sides - but it was interesting to think about.
And while it can't be done with equal sides, it CAN be done with equal probability (though it opens the door to thrower's bias).
 

AnalogKid

Joined Aug 1, 2013
10,987
Oh, I just did a Google search -- are you referring to the kind of outlet box used for wall switches and outlets? I see that several sites refer to those as "five-sided boxes".
Nope. Five sided because the panel the connector is mounted on is the 6th side. We do this often, but we've got CNC punches and brakes so we make them whatever size we need. The box completely surrounds the rear of the connector, creating a Faraday cage and completely sealing the connector hole (in terms of EMI). Then a much smaller hole is added for the connector's wires to escape the box.

ak
 

GopherT

Joined Nov 23, 2012
8,009
And while it can't be done with equal sides, it CAN be done with equal probability (though it opens the door to thrower's bias).
Well, we were wrong. There are some people with a much higher level of problem-solving creativity, motivation and tenacity than I have.

Click button to view

image.jpg
 

WBahn

Joined Mar 31, 2012
29,979
Nope. Five sided because the panel the connector is mounted on is the 6th side. We do this often, but we've got CNC punches and brakes so we make them whatever size we need. The box completely surrounds the rear of the connector, creating a Faraday cage and completely sealing the connector hole (in terms of EMI). Then a much smaller hole is added for the connector's wires to escape the box.

ak
Ah, I see. But I take it the shape is rectangular solid, just with one side missing?

I didn't think that those electrical boxes would be super good shields, but since they would already be inside the enclosure, the might help out quite a bit. I don't know.
 

WBahn

Joined Mar 31, 2012
29,979
Well, we were wrong. There are some people with a much higher level of problem-solving creativity, motivation and tenacity than I have.
Ah, yeah. I keep forgetting that we tend to make assumptions about the constraints, usually that the sides are planar, that CAN be relaxed. I think I now recall a friend saying that fair (equal-shaped faces) die for all values up to some fairly high number -- but I don't know if I'm recalling that correctly.
 
Last edited:

Thread Starter

Pinkamena

Joined Apr 20, 2012
22
Thanks for all the replies and good ideas! For the power, I will use feedthrough filters as recommended by Crutschow, and shielded wires. Here's question that may sound stupid, but is there anything preventing me from using the cable shield as a current return path?

GopherT's suggestions is good for the SPI signals, thanks. I like your laundry list, Wbahn, lots of good tips.
 

AnalogKid

Joined Aug 1, 2013
10,987
I didn't think that those electrical boxes would be super good shields, but since they would already be inside the enclosure, the might help out quite a bit.
They are excellent shields, but for our uses electrical boxes are too large, too heavy (we use .050" aluminum) , and have too many little holes all over the place. Our normal bending tolerance is good enough to make a box water-tight, but for TEMPEST we seam weld.

ak
 
Top