I recently did an experiment where I compared the noise floor of 2 different ways of bringing the signal grounds into the preamp I was working on.
Way #1: Keeping RCA in/out isolated from the chassis using washers and bringing the ground signal in to the PCB to the common small signal attachment point and twisting the ground and hot wires together to that point.
Way #2: Not using the washers on the RCA's which connects the in/out grounds together at the chassis and then connecting a master chassis connection to the same common small signal attachment point as way #1.
I would have guessed that connection scheme #1 would have had less noise but it turned out that way #2 was a couple of dB quieter. Both connection schemes had acceptable noise floors, there was not a ground hum problem or anything like that. I would think that bringing the grounds closer to the op amp chips would keep loops to a minimum and translate to a lower noise floor. Is this experiment just a fluke?
Way #1: Keeping RCA in/out isolated from the chassis using washers and bringing the ground signal in to the PCB to the common small signal attachment point and twisting the ground and hot wires together to that point.
Way #2: Not using the washers on the RCA's which connects the in/out grounds together at the chassis and then connecting a master chassis connection to the same common small signal attachment point as way #1.
I would have guessed that connection scheme #1 would have had less noise but it turned out that way #2 was a couple of dB quieter. Both connection schemes had acceptable noise floors, there was not a ground hum problem or anything like that. I would think that bringing the grounds closer to the op amp chips would keep loops to a minimum and translate to a lower noise floor. Is this experiment just a fluke?
Did you try the correct way to wire an unbalanced connection: a shielded/screened wire, not a twisted pair?
#1 is correct for pure audio. Neither may be correct when RF is around (as it almost always is these days).
#1 is correct for pure audio. Neither may be correct when RF is around (as it almost always is these days).
No, I didn't use a shielded wire. It was a twisted pair that was about 1-1/2" in length to PCB.
In #1 the chassis was connected to the same point on the PCB that #2 went to.
Do you guys agree that #1 is generally the safest way to route ground or is tying the in/out's ground at connection point your preference?
In #1 the chassis was connected to the same point on the PCB that #2 went to.
Do you guys agree that #1 is generally the safest way to route ground or is tying the in/out's ground at connection point your preference?
The in ground should connect to the ground at the in end of the PCB. The out ground should connect to the ground at the out end of the PCB. This assumes that the PCB uses a ground bus or ground plane. The signal ground should follow the signal.
Where this connects to the chassis is another issue entirely. When working with audio grounds you should ignore the chassis as much as possible. The chassis is a safety ground, not an audio ground.
Where this connects to the chassis is another issue entirely. When working with audio grounds you should ignore the chassis as much as possible. The chassis is a safety ground, not an audio ground.
The "Earthed" input socket was a couple of dB quiester than the non "Earthed" input socket.
What were your measurements?
Try the coax in both arrangements and post the four measurements.
The Chassis is not part of the Audio. The equipment will work without a Chassis. Look at all the plastic cased audio and video equipment that you have.
The Chassis, if designed properly, is an interference attenuating system. It adds extra RF attenuation to what should already be fitted to all the cable inputs and all the circuits susceptable to RF.
But Safety is a separate issue
It's this connection that blows the Mains Fuse when there is a catastrophic mains wiring failure. It's this connection that helps prevent us being injured/killed, because the Chassis become LIVE for just a few µs/ms until the correctly rated fuse ruptures and the arc extinguishes. An incorrect, overly large, fuse may take tens of ms, or even seconds, to rupture. That increases the risk of someone touching the failed equipment while it is still LIVE, enormously.
What were your measurements?
Try the coax in both arrangements and post the four measurements.
The Chassis is not part of the Audio. The equipment will work without a Chassis. Look at all the plastic cased audio and video equipment that you have.
The Chassis, if designed properly, is an interference attenuating system. It adds extra RF attenuation to what should already be fitted to all the cable inputs and all the circuits susceptable to RF.
But Safety is a separate issue
All our ClassI equipment must be connected to Protective Earth (PE).
It's this connection that blows the Mains Fuse when there is a catastrophic mains wiring failure. It's this connection that helps prevent us being injured/killed, because the Chassis become LIVE for just a few µs/ms until the correctly rated fuse ruptures and the arc extinguishes. An incorrect, overly large, fuse may take tens of ms, or even seconds, to rupture. That increases the risk of someone touching the failed equipment while it is still LIVE, enormously.
Last edited:
Andrew, in both of these "ways" the chassis was connected to ground in the same location. The only difference was that in way #2, in/out RCA cords' ground shields were connected together immediately and shared the same path to circuit ground back through the chassis.
I only wanted to get some feedback on which of these connection methods everyone prefers, shielded coax from chassis to PCB aside.
I understand your safety concerns but I don't really want to get into that whole thing again. I'm using a double isolated commercial class II transformer in a wall wart which is perfectly legal here in the states.
As I said, the ground should follow the signal. Neither of your methods seems to achieve that. To that extent, they are both wrong.
I would have guess the way #2 would have less noise.
It's a way that has been recommended by a guy who works in scientific research of very high level with very small signals here in France.
The justification relies on the fact that the chassis offers the lowest impedance for the ground path.
The enclosing chassis is a shield. It has nothing to do with the audio signal.
The chassis is there to attenuate interference.
The chassis is NOT there to conduct audio currents.
The chassis is there to attenuate interference.
The chassis is NOT there to conduct audio currents.
I would imagine way 2 would give better RF protection, but not necessarily audio hum reduction. I always isolate the input connectors from the chassis.
As DF96 said, the idea is to keep the ground and signal wires on the PCB close together. This reduces the loop area and hence the area that can be cut by magnetic flux.
Could be that the loop area on your PCB is quite high, and when you go to method 2, the loop area is reduced - check it out and let us know.
Twisted pair is ok for the connections - remember at 50/60 Hz, a shield provides virtually no protection against mains frequency flux. Changing to screened cable in your case is unlikely to bring relief whether you use method 1 or method 2.
[NB: Wall warts are legal in the EU/UK as well as long as they carry the appropriate certifications (same as USA)].
As DF96 said, the idea is to keep the ground and signal wires on the PCB close together. This reduces the loop area and hence the area that can be cut by magnetic flux.
Could be that the loop area on your PCB is quite high, and when you go to method 2, the loop area is reduced - check it out and let us know.
Twisted pair is ok for the connections - remember at 50/60 Hz, a shield provides virtually no protection against mains frequency flux. Changing to screened cable in your case is unlikely to bring relief whether you use method 1 or method 2.
[NB: Wall warts are legal in the EU/UK as well as long as they carry the appropriate certifications (same as USA)].
Such a dogmatism will not prevent some guys to have obtained better results with Shredhead's way #2 than with his way #1.
A coaxial shield ensures that both signal and ground conductors receive the same voltage from magnetic induction, so it is as good as twisted pair from that point of view.Bonsai said:
A coaxial shield also ensures that the signal wire receives no voltage from electric fields, so it is superior to twisted pair (which provides very little protection) from that point of view. Twisted pair comes into its own for balanced circuits with differential receivers.
Most coupling inside an amp is magnetic.
If you read Jensen et al's seminar papers, they specifically discuss this point. In Shredhead's case, I don't think screen cable will help him.
If you read Jensen et al's seminar papers, they specifically discuss this point. In Shredhead's case, I don't think screen cable will help him.
That may be true, but you would need all coupling inside an amp to be magnetic in order for twisted pair to equal what coax can do. If there is any electric field then coax is better.Bonsai said:
I suspect, although cannot prove, that the common preference for twisted pair for unbalanced connections (both inside and outside equipment) comes from faulty logic:
1. balanced is better
2. balanced uses twisted pair
3. therefore twisted pair is better
i don't think twisted is better, I just don't think it will solve Shredheads problem in this instance.
When I used way #2 the chassis connection was very close to incoming RCA's so I was able to keep the in/out wires close to their shared ground wire until they reached the PCB. For ways 1 and 2 I kept loop area the same.
For way #2, because in/out was sharing a ground wire to PCB, I increased the gauge of that wire from 20 to 18 gauge. The difference in noise floor was about 2.5dBV RMS between the 2 ways. This surprised me, I thought way #1 would be superior. It makes sense that it was because way #2 was blocking more RF like Bonsai suggested.
I thought I'd ask about this because I vaguely remember a thread where a member suggested that in a multichannel amplifier, it was beneficial to connect all in/outgoing RCA grounds together at the chassis. I can't find that thread to link to for reference. It doesn't make sense to me to do that in any situation so I thought I'd bounce it off of you guys. Thanks for the feedback
For way #2, because in/out was sharing a ground wire to PCB, I increased the gauge of that wire from 20 to 18 gauge. The difference in noise floor was about 2.5dBV RMS between the 2 ways. This surprised me, I thought way #1 would be superior. It makes sense that it was because way #2 was blocking more RF like Bonsai suggested.
I thought I'd ask about this because I vaguely remember a thread where a member suggested that in a multichannel amplifier, it was beneficial to connect all in/outgoing RCA grounds together at the chassis. I can't find that thread to link to for reference. It doesn't make sense to me to do that in any situation so I thought I'd bounce it off of you guys. Thanks for the feedback
Just to be clear, I'm not having any kind of problem with hum in either of these 2 ways. The noise floor for both is acceptable (-92 to -95dB unweighted). I was just doing an experiment. Every little bit counts! 🙂
This one agrees with that :
http://www.diyaudio.com/forums/digital-source/275585-mains-earth-signal-grounding.html#post4353497
forr, that link seems to focus on grounding for safety in a worst case fault scenario. I am only concerned with lowering the noise floor and increasing dynamic range. Can you link something that claims that tying all in/outgoing grounds together at chassis has the effect of lowering noise?
- Status
- Not open for further replies.