Ferrite Beads on Inputs

[Bo Deadly]

Similar to the other topic about ferrites on output pins, does it not follow that there should be caps and ferrites on inputs as well like this?:

But because I'm just trying to shunt and block RF and not stabilize an output driver I can skip the 39R in parallel?

Is there any issue with running 48V through these?

Of course a transformer would block RF but in my immediate case, that is on a separate board (or mounted on the chassis) so this will block RF on those leads and generally prevent RF from getting past the connector pins.

UPDATE:

Here is the current schematic:

The double square symbols are solder jumps. This is one of about 20 boards that are supposed to be modular so many things are optional (like the 48V which was added after posting the layout).

Regarding the layout, the ground plane would ultimately be connected to chassis ground (again this is one of numerous separate boards that have separate ground planes) which is the one trace running off the graphic at the bottom. This is why the 48V ground is separate. The XLR connector is actually mirrored in the graphic but pins 1, 2, 3 are labeled as such. H33 is a header jump for configuring pin 1.

 

[abbey road d enfer]

Similar to the other topic about ferrites on output pins, does it not follow that there should be caps and ferrites on inputs as well like this?:

It depends on the topology of the input stage, but basically the impedance is much higher so the inductor must be of much higher value than a simple ferrite bead. In fact many electronically balanced input stages have resistors at the input, of sufficiently high value to make the inductors unnecessary.

But because I'm just trying to shunt and block RF and not stabilize an output driver I can skip the 39R in parallel?

  39R in parallels with an inductor that is inserted in a 10k input, you may as well get rid of the inductor.

Is there any issue with running 48V through these?

No.

Of course a transformer would block RF

That's not always true... depends on the inter-winding capacitance and how effective is the electrostatic shield (if any!)

but in my immediate case, that is on a separate board (or mounted on the chassis) so this will block RF on those leads and generally prevent RF from getting past the connector pins.

The 100pF are your friends.
BTW, I noticed you wired the 47R and 10n in series. They should really be in parallels.

 

[ricardo]

Of course a transformer would block RF but in my immediate case, that is on a separate board (or mounted on the chassis) so this will block RF on those leads and generally prevent RF from getting past the connector pins.

If your transformer is connected to the XLR by 2cm of wire (twisted or not), the RFI will be re-radiated from those wires into your enclosure.

How about posting a schematic of what you are showing too?

I'm too stupid to figure out what is happening in your pics.  It is not clear where the chassis connections are.

The clearer you show what you are doing, the more chance you will get constructive comments.

 

[Bo Deadly]

If your transformer is connected to the XLR by 2cm of wire (twisted or not), the RFI will be re-radiated from those wires into your enclosure.

That is the rationale for adding caps and ferrites.

How about posting a schematic of what you are showing too?

Updated.

 

[abbey road d enfer]

Soooo...it's a mic input!
What is the component on the lower left, inserted between G and the "ground" line?
Bottom of C70/71 should be connected directly to Pin 1 and chassis Gnd.
What value are the inductors?

 

[Bo Deadly]

Soooo...it's a mic input!

Not necessarily. The 48V is optional. This board can also be used for a line input.

What is the component on the lower left, inserted between G and the "ground" line?

It's the "shell" pin.

Bottom of C70/71 should be connected directly to Pin 1 and chassis Gnd.

The 100p are connected to chassis ground. The ground plane is actually chassis ground (for this particular board). Pin 1 is configurable as per Jensen line input transformer datasheets.

What value are the inductors?

Whatever fits in SMD 1206 package.

 

[abbey road d enfer]

Soooo...it's a mic input!

Not necessarily. The 48V is optional. This board can also be used for a line input. [/quote] OK, so C70/71 are ok for line, but for mics, you probably need 1n or more.

What is the component on the lower left, inserted between G and the "ground" line?

It's the "shell" pin.

I mean on the schemo; looks to me like a cap between chassis gnd and ground plaane...what for?

Bottom of C70/71 should be connected directly to Pin 1 and chassis Gnd.

The 100p are connected to chassis ground. The ground plane is actually chassis ground (for this particular board).

So you need to establish a separate connection?

Pin 1 is configurable as per Jensen line input transformer datasheets.

I can't access the Jensen site, cause they want me to sign in, but this seems dubious to me.

What value are the inductors?

Whatever fits in SMD 1206 package.

You will have no more than 100uH. It will produce attenuation only in the 1-7MHz range, and even negligible.

 

[Bo Deadly]

You will have no more than 100uH. It will produce attenuation only in the 1-7MHz range, and even negligible.

So then it's not worth it?

Something like Wurth 742792121 is ~50 ohms from 5 MHz to over 1 GHz. Yeah, I might still pick up a Mexican radio station but would it not block cellular?

And even if this is for an electronically balanced line input, the idea is to block RF directly at the input pins (thus the graphic of the pcb).

 

[Matador]

(Pretty much) every schematic program has separate symbols for 'chassis' and 'circuit 0V'...most even have digital and analog ground symbols as well.

It would make everything much clearer if you used the chassis symbol for stuff you intended to connect to the chassis (like your 100pF shunt caps, and perhaps the phantom power return).  You can then add your ground shunt resistor to bring the two together (if desired):  some layout programs even have a special symbol just for tying different ground planes together.

Like this (referenced to the Rane website):

Makes everything much clearer what was intended.

Doing this would make it clear that your 47R / 10n parallel combination would go in-between pin 1 on the connector and chassis ground.  As it stands, it looks like your 100pF caps dump right into the analog ground which I don't think was intended.

 

[ricardo]

The ground plane is actually chassis ground (for this particular board).

No.  Your ground plane is just some Cu on a PCB.  If this is connected to the chassis at the opposite end of your input XLR (with your 2x100p s and p1 thingy, the RFI will flow along your 'ground plane' and radiate to the sensitive parts of your circuit.

The EXACT PHYSICAL position of these bits and your EXACT connection to Chassis is important.

If you can't draw a diagram like Matador's Rane circuit, you don't know what you are doing .. and neither do we .. nor can we.

 

[Bo Deadly]

No.  Your ground plane is just some Cu on a PCB.  If this is connected to the chassis at the opposite end of your input XLR (with your 2x100p s and p1 thingy, the RFI will flow along your 'ground plane' and radiate to the sensitive parts of your circuit.

The EXACT PHYSICAL position of these bits and your EXACT connection to Chassis is important.

If you can't draw a diagram like Matador's Rane circuit, you don't know what you are doing .. and neither do we .. nor can we.

Ok, this is kinda getting out of hand for what is being asked but here is the entire schematic (click to enlarge):

But this is just one little separate board. Meaning this is the entire board:

So this particular board has 2 output and 2 input PCB mount XLR connectors. And they're are mirrored. Meaning they're on the bottom. The components like the caps and ferrites are on top.

The ground plane of this particular PCB is connected to chassis ground using a dedicated wire. That wire actually runs along the back of the enclosure about 20 cm to another board with a CPC connector on it and then over a 3m cable to an external power supply. Inside the external PS is another CPC connector PCB where the chassis ground is actually connected to the physical metal chassis and to the earth connection of the mains socket.

The ground planes of other boards may or may not be connected to chassis ground. Most are connected to PGND (power ground). Some ground planes are AGND (analog ground). There is also a DGND (dirty ground) but I have not found an occasion to use it for a ground plane so far. All of these ground are separate until they reach the ground plane of the external power supply CPC connector board (the "star" ground point).

This is the best answer I can give without redrawing all of these little boards into one big schematic. So if it doesn't meet your standards, I will understand if you choose not to answer.

 

[abbey road d enfer]

I believe I have the answer to my earlier question; "G" pin (which is actually the XLR shell) is connected (or not) to the ground plane via solder jumps, which implies it is a configurable option. I don't think of any good reason to do that. The "G" pin is actually your access to chassis ground; that's where Pin 1 should be connected, either directly or via a parallels-connected RC (not series). Minimizing length of PCB trace is paramount there.
You shouldn't really need a separate wire to connect the ground plane to chassis ground.
The connection between safety ground, or "earth" (that which is coming from the mains socket/lead) should be connected as shortly and sturdily as possible to chassis ground. Your use of multiple flimsy connectors does not appeal to me.

 

[Matador]

One more thing:  based on the layout, there is no phantom power return:  the net is isolated (the 'top' of the 100uF cap goes to the header, but nowhere else).  Phantom 0V needs to make it's way back to the shell of the XLR pin 1 to provide a return path for phantom power.

This has a good explanation of it:

https://www.dropbox.com/s/25va7g7kpn0jkxv/grounding101v2.pdf?dl=0

 

[ricardo]

This is the best answer I can give without redrawing all of these little boards into one big schematic.

Your schematic tells us what we need to know .. which is you don't really understand the priorities that Abbey, Matador and I have been trying to stress.

If you look at the Rane diagram, you'll see the cable shields are ideally connected to the OUTSIDE of the Chassis.  That's because the Chassis is a Faraday Cage and the aim is to stop RFI getting inside the cage.

Any wire that penetrates the cage must be decoupled DIRECTLY to the cage by the SHORTEST POSSIBLE LEADS.  Going via a PCB Earth Plane + 20cm of wire + more PCBs etc to the Star Point is completeley useless.

If those XLR sockets are Neutrik, there will be a pad/connection to connect p1 and the 'earthy' end of your 100p to chassis PROVIDED THE SOCKET IS MOUNTED TO MAKE DIRECT ELECTRICAL CONNECTION TO THE CHASSIS/FARADAY CAGE IN AN APPROVED MANNER.

The Faraday cage has holes in it for connectors etc and you have to think carefully about how to stop RFI coming in through them.

But if your gear is meant for a bedroom studio in a house without mobile phones, fridges, nearby radio transmitters etc .. it may be OK if your Ferrite beads & 100p s are purely decorative.

So yes.  All this is pertinent to your original question.

 

[Bo Deadly]

Your schematic tells us what we need to know .. which is you don't really understand the priorities that Abbey, Matador and I have been trying to stress.

If you look at the Rane diagram, you'll see the cable shields are ideally connected to the OUTSIDE of the Chassis.  That's because the Chassis is a Faraday Cage and the aim is to stop RFI getting inside the cage.

Any wire that penetrates the cage must be decoupled DIRECTLY to the cage by the SHORTEST POSSIBLE LEADS.  Going via a PCB Earth Plane + 20cm of wire + more PCBs etc to the Star Point is completeley useless.

If those XLR sockets are Neutrik, there will be a pad/connection to connect p1 and the 'earthy' end of your 100p to chassis PROVIDED THE SOCKET IS MOUNTED TO MAKE DIRECT ELECTRICAL CONNECTION TO THE CHASSIS/FARADAY CAGE IN AN APPROVED MANNER.

I am using plastic, PCB mount XLR connectors. So there is no way for the shell and pin 1 to make contact with the chassis without getting inside. There could be a screw and a short wire but in practice it's not worth it. One benefit of using modular boards is that I can always change it without rebuilding everything. Meaning if the gear really needs to work in a high RFI environment, I could remove that board and use metal connectors.

And I'm not sure it's necessary anyway because if the board is configured as a mic input with 48V power, pin 1 must be connected to a non-chassis ground for the 48V return current (as Matador pointed out my circuit was wrong, I needed a solder jump between pin 1 and the power ground instead of the chassis, thanks for spotting that error). If it's a line input, ideally pin 1 should just be left unconnected entirely in which case there can be no shield currents. I know some people insist it should be connected to chassis ground so it can be jumpered that way or with the Jensen resistor cap shunt.

Also, the last 3-4 posts have all been about the shield. Meaning pin 1 and the "pin 1 problem". I understand what you're saying and I appreciate the comments. But my original question was about RF filtering the input pins 2 and 3. It's not clear that that question was every answered although your points about the shield make me think that a) the shield is probably much more important wrt filtering than pins 2 and 3 and b) if there is RF on 2 and 3 it's already too late. The question ultimately becomes, how much attenuation is provided by blocking RF on the leads even though it can still be capacitively coupled to other parts of the circuit.

There are plenty of schematics with LC filters in inputs that shunt to ground on a PCB. Can they all be wrong?

 

[Andy Peters]

There are plenty of schematics with LC filters in inputs that shunt to ground on a PCB. Can they all be wrong?

Yes, they can.

 

[JohnRoberts]

Yes, they can.

8) 8) 8)

JR

 

[john12ax7]

A faraday cage doesn't need to be solid, a mesh can also be used. So holes in the chassis may or may not be a problem, it would depend on the wavelength of the RF.

A simple LC filter won't do much for signal that is radiated and coupled downstream in the circuit. Which is why proper shielding is important. However the LC filter certainly helps attenuate things in the conductor path. It's a two sided approach.

 

[kambo]

A faraday cage doesn't need to be solid, a mesh can also be used.

radio stations use mesh....

 

[ricardo]

I am using plastic, PCB mount XLR connectors. So there is no way for the shell and pin 1 to make contact with the chassis without getting inside.

The plastic Neutriks I've seen have a tin shield with connection to the PCB which also make contact with one or more of the mounting screws.  If the screws make good electrical contact with the chassis, this is probably as good as a metal XLR.

And I'm not sure it's necessary anyway because if the board is configured as a mic input with 48V power ...

A mike preamp is the most common case which needs good RFI treatment.  As I said earlier, you may not need any of this if you are using it in a bedroom studio without mobile phones etc ..

It's not clear that that question was every answered although your points about the shield make me think that a) the shield is probably much more important wrt filtering than pins 2 and 3 and b) if there is RF on 2 and 3 it's already too late.

Non est tantum facile ... but basically, "yes".  The important consideration is whether the RFI has an 'antenna' (which could be your 'ground plane') inside the Faraday cage to radiate to other parts.

There are plenty of schematics with LC filters in inputs that shunt to ground on a PCB. Can they all be wrong?

Yes.  If they are not connected in the correct physical place.  8) 8) 8)

There's some REALLY EXPENSIVE gear with ferrite beads, ceramics, CM inductors bla bla on inputs ... on a little PCB connected to the XLR with 2" of ribbon cable.  They basically do NOTHING for RFI .. except re-radiate it 

To answer your original question .. if you stick to your original layout & earthing system, without addressing the shield, pin 1, chassis etc. issues .. your beads & caps will be purely decorative.

However, they probably won't degrade performance compared to not having them so OK to leave them in.  But I recommend you use only beads & capacitors that are hand carved from solid Unobtainium by Virgins for the greatest clarity & midrange definition.  8)