Grounding an aluminimum enclosure

[Jim]

Our device is inside a small aluminium enclosure. It gets 9V 500mA DC from a
wall-wart, no earth connection. The device comprises some digital TTL logic
running at up to 2MHz. There are also two RCA phono sockets (input from DVD
player, output to an amplifier).

Voltage levels on the RCA sockets are 1V pk-pk differential. The negative
pole of the input socket is connected via a cap to our board's ground plane.
The output RCA is isolated via a transformer.

For EMC and other considerations, should we connect the aluminium enclosure
to the board's ground plane or to any of the external connections?

Thanks,

Jim

 

[John Larkin]

Our device is inside a small aluminium enclosure. It gets 9V 500mA DC from a
wall-wart, no earth connection. The device comprises some digital TTL logic
running at up to 2MHz. There are also two RCA phono sockets (input from DVD
player, output to an amplifier).

Voltage levels on the RCA sockets are 1V pk-pk differential. The negative
pole of the input socket is connected via a cap to our board's ground plane.
The output RCA is isolated via a transformer.

For EMC and other considerations, should we connect the aluminium enclosure
to the board's ground plane
Definitely.

or to any of the external connections?

You could eliminate the input-lowside cap. It could cause several
problems.

"When in doubt, ground it."

John

 

[Jim]

You could eliminate the input-lowside cap. It could cause several
problems.

"When in doubt, ground it."

John

Thanks for your reply John. Connecting the enclosure to the board ground was
our first idea. Since then I've read in a newsgroup that the enclosure can
act like an antenna, picking up noise, and so it was better to leave it
floating.

Jim

 

[John Woodgate]

Thanks for your reply John. Connecting the enclosure to the board ground was
our first idea. Since then I've read in a newsgroup that the enclosure can
act like an antenna, picking up noise, and so it was better to leave it
floating.

That's bass ackwards. It can indeed act like an antenna, and if you
leave it floating it will re-radiate the hash it picks up into the
sensitive bits of your circuit. It won't do that if you earth it. But
the best thing to do is to get a copy of the Audio Engineering Society
Journal for June 1995, and learn about earthing and associated matters,
to which the whole issue is devoted, in depth.

The shells of the input and output connectors should be connected to the
enclosure, preferably its outside surface. The ground plane should be
connected at one place to the enclosure.

 

[James Meyer]

Our device is inside a small aluminium enclosure. It gets 9V 500mA DC from a
wall-wart, no earth connection. The device comprises some digital TTL logic
running at up to 2MHz. There are also two RCA phono sockets (input from DVD
player, output to an amplifier).

Voltage levels on the RCA sockets are 1V pk-pk differential. The negative
pole of the input socket is connected via a cap to our board's ground plane.
The output RCA is isolated via a transformer.

The cap at the input is redundant and can be replaced with a direct
connection.

For EMC and other considerations, should we connect the aluminium enclosure
to the board's ground plane or to any of the external connections?

The enclosure should be connected only to the output RCA jack common
(shell).

Jim "The other one." Meyer

 

[John Larkin]

The shells of the input and output connectors should be connected to the
enclosure, preferably its outside surface. The ground plane should be
connected at one place to the enclosure.

Many places, the more the better.

John

 

[Mac]

Our device is inside a small aluminium enclosure. It gets 9V 500mA DC
from a wall-wart, no earth connection. The device comprises some digital
TTL logic running at up to 2MHz. There are also two RCA phono sockets
(input from DVD player, output to an amplifier).

Voltage levels on the RCA sockets are 1V pk-pk differential. The
negative pole of the input socket is connected via a cap to our board's
ground plane. The output RCA is isolated via a transformer.

For EMC and other considerations, should we connect the aluminium
enclosure to the board's ground plane or to any of the external
connections?

Thanks,

Jim

Are you sure the input signal is differential? I was under the impression
that RCA is always single-ended ground referenced I would probably ground
the RCA shield directly to the case and directly to the circuit board
ground, as others have suggested. But if it really is differential,
connecting the signal directly to circuit ground could have consequences.
In particular, this will mean that negative differential signals will go
well below ground potential on the center conductor. Can you handle that?
If not, then I think you will have to keep the capacitor between shield
and circuit ground, which means you have to either not connect shield to
case or not connect case to circuit ground. If you do keep the blocking
cap on the shield, you should use one on the center conductor, too. Note
that you may need to explicitly set the DC bias on the circuit side of the
two blocking caps. Also note that the caps are making an RC high pass
filter with the differential input impedance. If this is an audio
application, you want the cutoff frequency down around 1 Hz or something.

Let me know if anything I said doesn't make sense.

Mac
--

 

[John Woodgate]

I read in sci.electronics.design that John Larkin <jjlarkin@highSNIPland
THIStechPLEASEnology.com> wrote (in <6afq3058ug42cml6i2tu0mu8pv2710op98@
4ax.com>) about 'Grounding an aluminimum enclosure', on Wed, 25 Feb
2004:

Many places, the more the better.

If you do that, you create unknown multiple current paths in the metal
of the enclosure. Murphy's Law shows that one or more of those paths
will divert currents to places where you do not want them to flow.

 

[Terry Pinnell]

Many places, the more the better.

One? Or many? Does this depend on the main signal frequency? If so,
what is the frequency above which 'one' should become 'many'?

 

[John Woodgate]

I read in sci.electronics.design that James Meyer <[email protected]>

The cap at the input is redundant and can be replaced with a
direct
connection.
Indeed.

The enclosure should be connected only to the output RCA jack
common

If you do that, the interference picked up on the screen of the input
cable goes straight on to the board, where it is likely to be most
unwelcome.

 

[Jim]

<Q_7%b.166$%[email protected]>) about 'Grounding an
aluminimum enclosure', on Wed, 25 Feb 2004:
That's bass ackwards. It can indeed act like an antenna, and if you
leave it floating it will re-radiate the hash it picks up into the
sensitive bits of your circuit. It won't do that if you earth it. But
the best thing to do is to get a copy of the Audio Engineering Society
Journal for June 1995, and learn about earthing and associated matters,
to which the whole issue is devoted, in depth.

The shells of the input and output connectors should be connected to the
enclosure, preferably its outside surface. The ground plane should be
connected at one place to the enclosure.
--
Regards, John Woodgate, OOO - Own Opinions Only.
The good news is that nothing is compulsory.
The bad news is that everything is prohibited.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk

Thank you John for your advice. I will defintely try to get hold of the
journal you mention asap. In the meantime, I was trying to work out how to
connect the ground plane to our enclosure in just one place.

The app notes for the bits of circuitry tell us to connect the low sides of
both input and output sockets directly to ground (yes, that low-side cap was
wrong. Thanks everyone for correcting me).

The RCA sockets are PCB-mount with a single through-hole pin connected to
the socket shell. Could we connect the shell of the input socket to both the
enclosure and ground, and use that as the only link between ground plane and
enclosure? The output socket shell would be connected to just the enclosure,
thus grounding it indirectly. Would that be acceptable?

Thanks,

Jim

 

[Jim]

Are you sure the input signal is differential? I was under the impression
that RCA is always single-ended ground referenced I would probably ground
the RCA shield directly to the case and directly to the circuit board
ground, as others have suggested. But if it really is differential,
connecting the signal directly to circuit ground could have consequences.
In particular, this will mean that negative differential signals will go
well below ground potential on the center conductor. Can you handle that?
If not, then I think you will have to keep the capacitor between shield
and circuit ground, which means you have to either not connect shield to
case or not connect case to circuit ground. If you do keep the blocking
cap on the shield, you should use one on the center conductor, too. Note
that you may need to explicitly set the DC bias on the circuit side of the
two blocking caps. Also note that the caps are making an RC high pass
filter with the differential input impedance. If this is an audio
application, you want the cutoff frequency down around 1 Hz or something.

Let me know if anything I said doesn't make sense.

Mac
--

Hi Mac,

Thank you for your information. The SPDIF digital audio spec says it is 0.5V
pk-pk differential. All outputs are meant to be driven by an isolating pulse
transformer, but in practice this is sometimes not the case.

Pulse Engineering make one of the favoured pulse transformers for SPDIF.
They show an application circuit on page 2 (69KB):
http://www.pulseeng.com/pdf/T601.pdf

I'm confused by this! The diagram suggests grounding the shield of the
output socket - RCA is only two-pole, so does this mean grounding the
enclosure leaving the lowside output floating (not what people are advising
in this thread). Or does it mean ground the lowside of the output, in which
case there's a connection missing from their diagram isn't there?

Thanks,

Jim

 

[John Woodgate]

The RCA sockets are PCB-mount with a single through-hole pin connected
to the socket shell. Could we connect the shell of the input socket to
both the enclosure and ground, and use that as the only link between
ground plane and enclosure? The output socket shell would be connected
to just the enclosure, thus grounding it indirectly. Would that be
acceptable?

Well, there is an unresolved doubt. In another response to your post,
the word 'differential' that you used in connection with the signal
voltage suggested to the responder that you had *balanced* input and
output signals. But they require three-contact connectors, not the
2-contact RCA sockets. If you really do have balanced inputs and
outputs, then you need different advice about earthing.

However, assuming that you have unbalanced inputs and outputs, the
solution you propose above is suitable. If possible, use the type of RCA
connector that has a metal body, secured to the enclosure by a nut, not
the type that has an insulated body and requires a 'pigtail' connection
to the enclosure.

 

[Jim]

I'm confused by this! The diagram suggests grounding the shield of the
output socket - RCA is only two-pole, so does this mean grounding the
enclosure leaving the lowside output floating (not what people are advising
in this thread). Or does it mean ground the lowside of the output, in which
case there's a connection missing from their diagram isn't there?

Thanks,

Jim

I think I understand the diagram now - it is intended for "professional"
audio connections using XLR connectors that have a separate shield
connection.

Jim

 

[Jim]

Well, there is an unresolved doubt. In another response to your post,
the word 'differential' that you used in connection with the signal
voltage suggested to the responder that you had *balanced* input and
output signals. But they require three-contact connectors, not the
2-contact RCA sockets. If you really do have balanced inputs and
outputs, then you need different advice about earthing.

However, assuming that you have unbalanced inputs and outputs, the
solution you propose above is suitable. If possible, use the type of RCA
connector that has a metal body, secured to the enclosure by a nut, not
the type that has an insulated body and requires a 'pigtail' connection
to the enclosure.
--
Regards, John Woodgate, OOO - Own Opinions Only.
The good news is that nothing is compulsory.
The bad news is that everything is prohibited.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk

Thanks John for your patience! I realise I've been confusing the
professional AES/EBU standard for digital audio transmission with consumer
SPDIF (IEC-958). AES/EBU is differential and uses 3-contact (XLR) connectors
as you say, whereas SPDIF, what we are dealing with, uses a 2-contact (RCA)
connector and is grounded on the low-side. I think that then makes your
advice valid re the grounding. Thanks again.

I've been to'ing and fro'ing between chassis-mount and PCB-mount sockets.
PCB-mount are more convenient to us from a production point-of-view, and
look rather more attractive. I understand now though that chassis-mount
could provide better performance, so I'll look into it some more I think.

Jim

 

[Tim Shoppa]

You could eliminate the input-lowside cap. It could cause several
problems.

"When in doubt, ground it."

I've seen input-lowside caps added when there were ground loop problems,
but with a transformer-isolated output and a well-isolated DC supply
it seems odd that this would be necessary.

I don't know the history of this device, but I suspect that the input-lowside
cap was put in to make up for some deficiency in isolation elsewhere.

The sure way of finding what that deficiency is to remove the
input-lowside cap and ship a few ten thousand units :-(

Tim.

 

[John Woodgate]

I've been to'ing and fro'ing between chassis-mount and PCB-mount
sockets. PCB-mount are more convenient to us from a production point-of-
view, and look rather more attractive. I understand now though that
chassis-mount could provide better performance, so I'll look into it
some more I think.

Yes, this is a dilemma. We need to explain the EMC issue to connector
manufacturers, but some of them are deaf.

The point is that the shell of the connector must make a low-inductance
(and resistance, but that is easy) connection to the enclosure,
preferably to its outside surface, to keep high-frequency currents to
the outer skin. It isn't impossible to design a PCB-mount connector that
does that, but it may well need an outer metal collar as a loose part,
which is not commercially attractive. It also means that the positioning
of the PCB within the enclosure is somewhat constrained, as may be the
positioning of the connectors on the PCB; you can't have connectors
projecting from each end, because you then can't assemble the PCB to the
enclosure unless you have loose panels on the enclosure, which in turn
is bad news from the EMC point of view.

Still, if it was easy, anyone could do it and it wouldn't be so
intensely fascinating. (;-)

 

[Jim]

Yes, this is a dilemma. We need to explain the EMC issue to connector
manufacturers, but some of them are deaf.

The point is that the shell of the connector must make a low-inductance
(and resistance, but that is easy) connection to the enclosure,
preferably to its outside surface, to keep high-frequency currents to
the outer skin. It isn't impossible to design a PCB-mount connector that
does that, but it may well need an outer metal collar as a loose part,
which is not commercially attractive. It also means that the positioning
of the PCB within the enclosure is somewhat constrained, as may be the
positioning of the connectors on the PCB; you can't have connectors
projecting from each end, because you then can't assemble the PCB to the
enclosure unless you have loose panels on the enclosure, which in turn
is bad news from the EMC point of view.

Still, if it was easy, anyone could do it and it wouldn't be so
intensely fascinating. (;-)
--
Regards, John Woodgate, OOO - Own Opinions Only.
The good news is that nothing is compulsory.
The bad news is that everything is prohibited.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk

I'm just wondering - protruding through the same panel of the enclosure we
have mounted a right-angle D-socket. It is PCB-mount but also has bolts
connecting its shell to the outside of the enclosure. If we run a short
track between the grounding pin of the D-socket to the pigtails of the two
PCB-mount RCA sockets, would this be likely to provide a decent solution? As
you say, getting a low resistance connection is quite easy, but low
inductance could be another matter.

Thanks,

Jim

 

[John Larkin]

I read in sci.electronics.design that John Larkin <jjlarkin@highSNIPland
THIStechPLEASEnology.com> wrote (in <6afq3058ug42cml6i2tu0mu8pv2710op98@
4ax.com>) about 'Grounding an aluminimum enclosure', on Wed, 25 Feb
2004:
If you do that, you create unknown multiple current paths in the metal
of the enclosure. Murphy's Law shows that one or more of those paths
will divert currents to places where you do not want them to flow.

If there were multiple current paths in the ground plane (or traces,
if no plane) of the PCB, paralleling the pcb with the metal of the box
reduces voltage drops between the grounds, which is generally good.

But whet's more important is that, if the pcb is tied to the can at
one point, the result is an RF resonance mode of the pcb circuitry
within the box, sort of the equivalent of the mechanical cantelever
resonance of a beam or plate fixed at one end. The more grounds, the
stiffer the box-to-pcb connection and the higher frequency the
resonant modes and the harder it is to excite them.

RF resonances allow external EMI to get into rectifying junctions and
become low-frequency noise and offsets. And RF sources are everywhere
these days.

John

 

[Keith R. Williams]

One? Or many? Does this depend on the main signal frequency? If so,
what is the frequency above which 'one' should become 'many'?

IMO, 1/20 wavelength is the beginning of "antenna". If your structures
are larger it's an antenna. I agree with the "ground early and often"
group. Single point/star grounds are a nice idea, in theory. In
practice they aren't (single point). Avoiding ground loops is futile,
it's best to make them as small and as low impedance as possible.

Of course, if it's audio use a tube and charge 1000x more.