Why didn't you ask these configurations when I published the tests results ?
Why don't you show your own test results ?
One and only one connection to the box is not a norm as you never showed it is better and suspecting is not a technically approved behavior. Suspect is not always guilty.
As I reply to Andrew, why didn't you say that when I published my tests results ?
You cannot proclaim your grounding scheme is the good one without any proof.
I made and showed measurements , where are yours ?
Some twisting to reduce the effective loop area (by periodic reversal) may be sufficient.AndrewT said:
Although I think I participated in that thread, I don't remember seeing your results before. Maybe I was busy at the time?herve00fr said:
A single connection from signal ground to chassis is certainly the norm in audio, to avoid hum loops. Whether it is good for EMC is another issue.
Last edited:
Once again, you cannot say that without any proof. Audio is a part of analog electronic signal processing, and obeys to the same laws.
I can say that. I just did. If you choose to dispute it then that is up to you. I could ask for "proof" from you to the contrary, but I really can't be bothered to argue. "Proof" of what is normal practice would require a near exhaustive evaluation of grounding details on thousands of commercial and DIY items going back over the past 60 years. Do you have the time to do that? I certainly do not.
I can say that over the years I have seen a plethora of grounding schemes implemented on numerous types of electronics, from audio to sensitive instrumentation, EE's for their own beliefs and implement them as they see fit. Eventually every system has worked, some have required more tweaking than others, especially during EMC testing or for signal integrity requirements. Whatever the scheme though the biggest issue is when unforeseen high impedance paths have to compete with low impedance return paths, then you get ground loops and the fun begins...😀
I am not asking you what is the "grounding & shielding" norm in commercial or diy audio.
I just suggest you publish some numbers, as I did, which can show that your suggested scheme is the optimal one, instead of telling me I'm lucky or my measurements results are poor.
If we consider as norm what is the most used in the world, then I can say mp3 is the norm in term of digital audio coding. Do you think mp3 is the best one ?
Sorry, you appeared to be asking for proof of two separate statements:
1. single ground is the norm (post 24)
2. single ground is best (post 22)
I got these two demands for proof mixed up.
For the issue of hum, I would expect single ground to be best because it does not in itself introduce a hum loop (unlike multiple ground), although it may form part of a loop with another item. It should be better than no ground because it helps ensure an equipotential around the audio circuit, with little or no difference in voltage between the equipotential and the signal ground.
1. single ground is the norm (post 24)
2. single ground is best (post 22)
I got these two demands for proof mixed up.
For the issue of hum, I would expect single ground to be best because it does not in itself introduce a hum loop (unlike multiple ground), although it may form part of a loop with another item. It should be better than no ground because it helps ensure an equipotential around the audio circuit, with little or no difference in voltage between the equipotential and the signal ground.
Getting a true single ground is actually quite hard though... hense so many problems with ground loops, often the PE wires of the mains leads form a ground loop as they have lower resistance than co-ax screens, proving a lower impedance path for lower frequency signals. Many views disagree with complex single point grounding schemes, Tony Waldrons' EMC rants are worth a read as are others. It is the hidden paths that cause the problems. As said I have seen every scheme imaginable to man, the most problematic have always been ones where a single point has been the aim, these may work in a single box system, but once you get multiple boxes, then problems can appear.
True, but a test of multiple grounds per item vs. no grounds per item omits a common audio option: one ground per item. This option is the one often recommended for audio (at least for hum) - as I said it might not be the best for RF.
By the way, by 'single ground' I certainly don't mean a star point. I mean a single connection from whatever audio ground is used within an item (star or bus or something else) to the chassis - possibly via some sort of loop breaker. I assume that any test of the grounds of an item must use floating source and test equipment; otherwise one is just testing the whole grounding arrangement, not just the item under test.
By the way, by 'single ground' I certainly don't mean a star point. I mean a single connection from whatever audio ground is used within an item (star or bus or something else) to the chassis - possibly via some sort of loop breaker. I assume that any test of the grounds of an item must use floating source and test equipment; otherwise one is just testing the whole grounding arrangement, not just the item under test.
I experienced the contrary : multiplying ground connections to a metal plane or box always reduce the noises.
When I make precise measurements on instrumentation devices, boards are at least connected to a copper plane with multiple spacers.
Concerning the measurements with no connection to the box, remember that some diyaudio expert told me that chassis is not a part of audio device which has to work properly without any box.
Once again, in an unbalanced RCA interconnect, how do you suggest connecting the screen/shield to the chassis if you don't use the RCA connector shell?
The ideal situation is that the coax screen/shield connect to the outside of the chassis for RF purposes, and to the signal ground arrangements for audio purposes. An approximation to this is possible.
The chassis hole already couples the inside to the outside. A cap from RCA outer to chassis on the inside is thus an approximation to 'shield to chassis'. You could put an RF choke in the audio return lead inside to further isolate the audio ground from RF.
If you are in a bad RF environment you may have to compromise on audio hum.
If you are in a bad hum environment you may have to compromise on RF ingress, or use other measures.
If you are in a bad RF environment you may have to compromise on audio hum.
If you are in a bad hum environment you may have to compromise on RF ingress, or use other measures.
Connecting RCA to chassis is simpler and works at least as well.
Measurements I made with RCA connected to box did'nt show any trace of hum.
The details of grounding (or not) affect what causes of hum are relevant.
No ground to chassis means the circuit is open to electric fields, but fairly immune to magnetic fields.
A single ground more or less eliminated the electric field issue, but opens it up to ground currents or magnetic fields coupled via adjacent items.
Multiple grounds add the possibility of problems from internal fields and currents too.
So what hum you measure depends not just on the arrangement, but also the local hum sources and how they couple to everything in the setup. Hence what you were measuring was not the hum of your grounding arrangement, but the hum of your grounding arrangement in your particular environment with your particular connections.
No ground to chassis means the circuit is open to electric fields, but fairly immune to magnetic fields.
A single ground more or less eliminated the electric field issue, but opens it up to ground currents or magnetic fields coupled via adjacent items.
Multiple grounds add the possibility of problems from internal fields and currents too.
So what hum you measure depends not just on the arrangement, but also the local hum sources and how they couple to everything in the setup. Hence what you were measuring was not the hum of your grounding arrangement, but the hum of your grounding arrangement in your particular environment with your particular connections.
No because you have loops - input signal + return, loops on pcb, output & return etc - sensible to mag fields.
I agree on current potential issue which need careful implementation, but the more you strongly connect to ground the more you are immune against mag fields.
That's true for all measurements, environment is a part of the setup.
I tend to think that my test environment is near the mean test setup with oscilloscope, laptop soundcard and solder iron on and the number of different possible connections between a soundcard and the dut is low.
Anyway, measurement is better than no measurement.
Those loops are unavoidable. I did say "fairly immune", not totally immune.herve00fr said:
No. It is loops which make you pick up magnetic fields.
Yes, provided that the measurements are measuring what the measurer thinks they are measuring and the results are correctly interpreted - all of which require a good theory about what is going on.
Open loops don't exist, it is always closed by some parasitic devices.
Low impedance closed loops are more immune against mag fields. See measurements.
This is the common verse of those whose theory doesn't agree with measurements. It is easier to question the competence of the measurer than questionning his own understanding of the theory.
- Status
- Not open for further replies.