An amplifier or any other electronic device will work without a chassis and therefore will work with a chassis that is NOT connected to the amplifier's circuits.
The shielding effect of the enclosing chassis does NOT depend on connecting the audio circuits to the chassis.
Interference that is taken to the enclosing chassis is then capacitively coupled to the Earth (the big 12.8Gm diameter globe that we all live on).
I agree that it will work without chassis or without signal reference ("ground") connected to the chassis. I do not agree that floating the signal reference from the chassis has no effect on shielding, with or without safety earth (or Protective Earth, PE) connected to the chassis.
There's a reason why even the most sensitive measuring devices have signal reference connected to the chassis, even if that means connecting it to the "dirty" PE.
It's actually the first "rule" in the article I linked to in my previous post:
Just connecting PE to the chassis and then leave the signal reference floating from it doesn't work. That's why I wrote that signal reference (usually called "ground") must be connected to the chassis. Wether or not that chassis is connected to PE (class I or II) does not matter.
In my work, I mostly deal with the manufacture, testing and calibration of specific measuring devices. These are class I devices and have PE and signal reference interconnected in a carefully chosen location. There's also a "floating" variant which has the possibility to disconnect signal reference from the chassis for specific measurements. This has a negative effect on several specs (mostly noise related). So you can understand why I totally agree with "rule 1".
Last edited:
I am guided by PIN1 rules and this, although designed for balanced impedance connections, also applies to unbalanced connections.
Pin1 effectively connects the cable shield to the Chassis and these "enclose" the signal circuits. The shield is capacitively coupled to Earth.
The internal circuits do not need to connect to the shield.
The shield is independent of the internal circuits.
Compare the advice given by many Designers to us Builders:
Do not connect the input socket Signal Returns to the Chassis. Isolate these inputs from the Chassis.
To me this is all of the same design philosophy.
Later the output stage is required to take a voltage reference to the input stage. This "link" can also be independent of the shield.
The only rule that I am aware of that requires the signal circuits to reference the shield is :
All exposed conductive parts should be connected to the PE protected Chassis.
This SAFETY rule is not about effective signal handling.
Pin1 effectively connects the cable shield to the Chassis and these "enclose" the signal circuits. The shield is capacitively coupled to Earth.
The internal circuits do not need to connect to the shield.
The shield is independent of the internal circuits.
Compare the advice given by many Designers to us Builders:
Do not connect the input socket Signal Returns to the Chassis. Isolate these inputs from the Chassis.
To me this is all of the same design philosophy.
Later the output stage is required to take a voltage reference to the input stage. This "link" can also be independent of the shield.
The only rule that I am aware of that requires the signal circuits to reference the shield is :
All exposed conductive parts should be connected to the PE protected Chassis.
This SAFETY rule is not about effective signal handling.
All ClassI devices must comply with the "all exposed..." as in post24
So yes, it is
All ClassII devices MUST NOT connect to the PE of the mains socket.
Many do not have a shielding enclosure and they operate perfectly well because they incorporate interference attenuation at every cable entry.
look carefully at modern SMD PCBs for TV, Microwave, Signal decoders, etc and you see many (apparently extra) smd components at the cable/socket entries.
So yes, it is
All ClassII devices MUST NOT connect to the PE of the mains socket.
Many do not have a shielding enclosure and they operate perfectly well because they incorporate interference attenuation at every cable entry.
look carefully at modern SMD PCBs for TV, Microwave, Signal decoders, etc and you see many (apparently extra) smd components at the cable/socket entries.
Last edited:
I still have the feeling that I didn't make myself clear. Perhaps that's because all the different "grounds" are often regarded as the same because they're (almost) always connected together at at least one point.
- Safety Earth or Protective Earth (PE): the connection of the conductive parts that can be touched and the earth-prong of the outlet. This prong is wired to a long rod that is driven into the earth in or in close vicinity of the building.
- Ground: usually the power supply 0 V (return). When measurements are taken, these are with respect (referenced) to this point and hence the...
- ...Signal reference, signal 0 V (return) will need to be connected to it (as convention has it, but some British cars with battery + wired to chassis come to mind).
For argument's sake, I'm only interested in the degree of shielding of a system with ground connected to chassis and one with ground floating from the chassis. Whether or not the circuit is EMC compatible without case is not the issue here.
Let's at first also assume a stand-alone battery powered system with no wires connected. I still believe that not connecting signal reference to the chassis at at least one point in the system is detrimental to shielding.
Now, let's assume the same device connected to mains, without PE. The same goes.
Now, let's assume the same device connected to mains, with PE. The same still goes since earth potential and signal reference are still floating thanks to the galvanically isolated PSU's output if there's no connection.
The way I see is that as long as signal reference and chassis are floating from eachother, voltage differentials (noise sources) can occur, and the contribution of the chassis to EMI-shielding remains negligible.
- Safety Earth or Protective Earth (PE): the connection of the conductive parts that can be touched and the earth-prong of the outlet. This prong is wired to a long rod that is driven into the earth in or in close vicinity of the building.
- Ground: usually the power supply 0 V (return). When measurements are taken, these are with respect (referenced) to this point and hence the...
- ...Signal reference, signal 0 V (return) will need to be connected to it (as convention has it, but some British cars with battery + wired to chassis come to mind).
For argument's sake, I'm only interested in the degree of shielding of a system with ground connected to chassis and one with ground floating from the chassis. Whether or not the circuit is EMC compatible without case is not the issue here.
Let's at first also assume a stand-alone battery powered system with no wires connected. I still believe that not connecting signal reference to the chassis at at least one point in the system is detrimental to shielding.
Now, let's assume the same device connected to mains, without PE. The same goes.
Now, let's assume the same device connected to mains, with PE. The same still goes since earth potential and signal reference are still floating thanks to the galvanically isolated PSU's output if there's no connection.
The way I see is that as long as signal reference and chassis are floating from eachother, voltage differentials (noise sources) can occur, and the contribution of the chassis to EMI-shielding remains negligible.
Last edited:
Why would that be? Shielding is 'keeping out' or 'absorbing' radiated interference by whatever forms the shield, and I cannot see how this would be different whether it is connected to the circuitry or not?
Jan
we often see a recommendation that test measurement equipment should be battery powered to help avoid ground loop.
Sensitive battery powered test measurement equipment is usually shielded.
This type of "floating" and shielded equipment works !
Both the shield is floating and the internal circuits are floating.
Only when a connection is made to other equipment would either of these "floating" arrangements change.
If the DUT was also battery powered, then everything is floating and still they work !
Sensitive battery powered test measurement equipment is usually shielded.
This type of "floating" and shielded equipment works !
Both the shield is floating and the internal circuits are floating.
Only when a connection is made to other equipment would either of these "floating" arrangements change.
If the DUT was also battery powered, then everything is floating and still they work !
Hmm, so the dual rail LT1083 can't be built with a 12-0-12VAC transformer?
That is a shame since I was planning on building two of these and I only have one dual secondary transformer.
One for my BCL clone (has a 2x15VAC 1,5A toroid)
And one for a JHL headphone amp kit that is on it's way.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.