Yes it is also connected to Planet Earth. But that connect is not part of the problem nor the solution. That connection is for safety.
Noise, interference, leakage, lost Neutral and ground currents all want to get back to their voltage source. That voltage source is the power company Neutral. Little current will use Earth as a path back to it's source. None will use Earth as a sink or sump.
Re Occam's Razor: not if it does not work
You are creating earth loops like this. How do you minimize loop areas when you just connect everything to the chassis? You will end up with multiple loops
Do you have hard measurements to back up your assertion?
I would consider the likes of Self, Cordell, Ott, Jensen to be real experts in these matters.
I have never opened a good piece of audio gear and seen the grounding/wiring method you propose. Have you?
A good test on a power amp is to connect a pair of headphones directly to the amp output (no input signal of course). Because of the headphone sensitivity (typically 90 dB/mW) it's the equivalent of 30 to 40 dB gain. Any noise should be very faint.
Last edited:
Yes - classic earth loop where you get capacitive coupling from prim to secondary. I use a screen and a flux band on my transformers. Makes a big difference for about a 10% cost adder - worth it in my view.
Self makes a dc signal reference to chassis connection at the rca input. What's wrong with this?
It's not about the number of loops, it's the loop area that's the problem.
You can run 7.1 analog interconnects from place to place, if they all follow the same path then no problems.
Add Neil Muncy (RIP), Ralph Morrison, Jim Brown & Bill Whitlock to the list.
Everything is not "simply" connected to chassis, charge and discharge of reservoir capcitor don't flow in the chassis, neither earth current. I implement the "T" which bottom end is connected to chasssis in which only "audio" currents flow.
Multiple small area loops are not a problem if their impedance are low.
As you probably know, I published measurements on preamp. Their "success" does not motivate me to publish others.
My low frequency speakers sensitivity is 97dB/2.8V/1m and directly connected to LF amp, HF drivers are 109dB/2.8V/1m directly connected to HF amp : absolut silence the head in the horn or the ears nears the lf driver.
Nothing wrong - any single point along an amps 0V 'ground' circuit can be connected to chassis. But there may be subtle issues worth considering as to what currents pass along various 0V ground wiring and hence may cause noisy voltage drops along that 0V wiring.
There are parasitic capacitive currents looping to the chassis from the audio circuitry - this would predominantly be via the power transformer HT secondary, and the output transformer plate windings in a valve amp, but could also be to chassis grounded heatsinks from output stage ss devices. How much current occurs depends on capacitance to chassis, and voltage waveform. Any such loop current needs to pass back through the 0V circuit path to its origin.
There can be an external audio connection which introduces a mains power supply earth loop. Not a problem if your amp is only ever connected to a battery operated CD player, as compared to a preamp, or compared to the amp having multiple RCA connections to many and varied equipment.
Last edited:
Here is a good source of information, it describes the causes and gives you solutions beyond connecting every connector (PE, input and outputs) to the chassis and using transformers.
http://www.updatemydynaco.com/documents/GroundingProblemsRev1p4.pdf
http://www.updatemydynaco.com/documents/GroundingProblemsRev1p4.pdf
Also Tony Waldron.
And Keith Armstrong EMC UK.
Cant find Tony Waldrons EMC rants they seem to have disapeared.
http://www.audioemc.co.uk/wp-content/uploads/2014/04/Pages-from-HFC_issue33-4-Waldron-EMC-short.pdf
One interesting question I ask my self is how many low level analogue PCB layouts have I done in the last 30 years that have NOT had a ground plane...
None
How many designs have I done with spiders legs starred ground connections...
None
This is not to say that some designs have different GND planes, some may have separate analogue, digital, some outputs will have a starred ground for power etc. BUT ALL will have the signal and the return path in very close proximity, minimise the inductive loop is the first rule, when your ground track is some skinny liitle track miles away from the hot track you cant do this.
When I get back to the office I will do a survey of designs that are connected directly to chassis, connected by capacitors and totally unconnected from chassis (but with links just in case)...
Oh Keith Armstrong (EMC UK) calls PE the noise distribution network, it is there for a safety PE connection only, use it as a signal return at your peril.
Last edited:
I don't think a GP is mandated for linear audio although the it does help. It's also not that easy to do with a 2 layer board and it is expensive if you go to 4 layers. At audio frequencies, good layout and minimizing loop areas and common impedance will get you 99% of the way there - I.e excellent results. Keeping all the input and output connection close together, twisting and a compact PCB layout work surprisingly well for 2 layer boards.
Digital audio is different. You have relatively high frequencies (clock, data etc) and lots of fast edges. GP definitely required
I'll stick to the tried and tested engineering approaches, continue to refine my understanding and hopefully break the -105 dBV barrier in the near future.
I won't be using the chassis as a signal ground anytime soon though.
Digital audio is different. You have relatively high frequencies (clock, data etc) and lots of fast edges. GP definitely required
I'll stick to the tried and tested engineering approaches, continue to refine my understanding and hopefully break the -105 dBV barrier in the near future.
I won't be using the chassis as a signal ground anytime soon though.
Last edited:
you are confirming what I said and also answering my last question.
The enclosure does not need a direct wire connection back to the mains distribution board to act as an RF screen.
the enclosure is capacitively coupled to "EARTH" via the air.
And the enclosure does not need a connection to "EARTH" that is effective down to DC.
What's the argument?
Is there some specific phrase that I use that is misleading, or confusing, or simply wrong?
Last edited:
I have measured the AC voltage along these enclosing "ground" traces using a crude scope set to 2mV/div or 5mV/div
Any input that taps into this corrupted "ground" will have interference injected into the amplifier and it will be seen at the output as an interference signal.
If there are charging pulses across this trace (because the +ve and -ve power inputs are at opposite ends) then those pulses also contaminate the signal input.
One of the worst was a DX amplifier PCB that had a "ground" trace that was about 7" long passing from +ve via signal Return to -ve.
But it's not unique. We see this style of PCB layout on ~90% of the PCBs submitted to the Forum.
Probably not. Most just connect Mains Earth to mains Neutral at the Mains distribution board. The earh connection to planet EARTH is outside the building and may be some distance away from your building/house.
enjoy the music, not the interference.
The ability of the connection to pass Fault Current back to PE.
I referred to D.Self's advice just a couple of days ago.
But he says it gives maximal rejection of interference (I paraphrase). He doesn't refer to fault conditions (it's quite possible I missed that)
Also, I'm glad you added coupled to Earth via the AIR. I suspect I'm not the only one that didn't understand exactly what you meant
Last edited:
The trouble today though is the ever increasing RF pollution. I was always taught EMC immunity begins with the PCB design, get that right and your on a winner (saving on expensive cases and shielding). I would say that going for a GP design (I can often achieve this on a 2 sided PCB with careful placement) on a 4 layer board gives you so much more freedom and is not that expensive these days as it use to be... For digital it is a must. Outside of DIY it is very rare to see a double sided board these days, just the spectre of EMC testing means that everyone plays safe and goes for at least a basic sig,gnd,power,sig layout. Though most I work on are in the higher layer count with numerous ground planes for signal integrity.
D.Self does not mention Fault Current.
That's my point.
If you make the MAG to protected Chassis Safety connection via the signal return then that ROUTE must survive longer than it takes the Mains Fuse to rupture and the arc to extinguish during the incident that is passing Fault Current, that may have exceeded 1kA.
http://www.diyaudio.com/forums/power-supplies/305956-proper-grounding-scheme-2.html#post5036575
That's my point.
If you make the MAG to protected Chassis Safety connection via the signal return then that ROUTE must survive longer than it takes the Mains Fuse to rupture and the arc to extinguish during the incident that is passing Fault Current, that may have exceeded 1kA.
http://www.diyaudio.com/forums/power-supplies/305956-proper-grounding-scheme-2.html#post5036575
Last edited:
Sorry, I thought the words "capacitively coupled to EARTH" meant via whatever is in between.
- Status
- Not open for further replies.