How to 100% eliminate ground loop?

Status
Not open for further replies.
Jump to Latest
F

Fusion916

Member
Joined 2012
#1
I have gotten my ground loop down to ~20mV single ended but I want to get it down to an unmeasurable value.

This is my grounding setup (green lines are ground wires):

An externally hosted image should be here but it was not working when we last tested it.


Details: RCAs are shorted with close 14wg wire, then TWO 14awg wires off a single point are brought to the star ground. I did this because with one 14awg wire i got ~50mVpp of 120hz noise, adding the second brought it down to ~20mV.

Here is the oscilloscope output and pic of amp wiring:

An externally hosted image should be here but it was not working when we last tested it.


An externally hosted image should be here but it was not working when we last tested it.
 
#2
Fusion916 said:
I have gotten my ground loop down to ~20mV single ended but I want to get it down to an unmeasurable value.

This is my grounding setup (green lines are ground wires):

An externally hosted image should be here but it was not working when we last tested it.


Details: RCAs are shorted with close 14wg wire, then TWO 14awg wires off a single point are brought to the star ground. I did this because with one 14awg wire i got ~50mVpp of 120hz noise, adding the second brought it down to ~20mV.
<snip>
Click to expand...

It's possible you're introducing a ground loop with the scope. To verify, you could use an isolation transformer under the scope. Or TEMPORARILY use a ground lifter under the scope. BEWARE THIS DISABLES THE SAFETY FEATURE OF THE PROTECTIVE GROUND !! If you're not comfortable with this the don't do it but if you ask other members, many have done this in the past with no problems BUT IT'S IMPORTANT YOU KNOW WHAT YOU'RE DOING.

G²
 
#3
"Ground loops" can contain a mutually inductive component which is coupled from wire to wire. Looking at the wiring of the amplifier, I see no attention to lead dress to minimize this coupling ?
The "dirty" wire routing can easily couple undesired signals between "Gnd" as well as other signal and power lines.
 
#4
Good question, I don't have an answer for you but I have started to notice that the chassis makes the ultimate loop around everything and it is usually tied to ground. I've seen a couple of people who are smarter than me post that adding resistance to either the chassis to the source ground or adding resistance from the star ground to the source ground helps with noise. I have yet to experiment with this so I don't really know.
 
#6
A Power Amplifier with no hum can measure <0.1mVac on the speaker output.
That includes Hum and noise.
With speaker of <93dB/W you cannot hear this.

0.05mVac is equivalent to -115dBW

A 93dB/W speaker fed with a -115dBW sign wil have an acoustic output that is ~20dB below audibility @ 1m
if you reduce that 1m to 0.1m then the SPL becomes ~ 0dB, or just about audible in completely silent surrounds. i.e you can't hear 0.05mVac of hum and noise with 93dB/W speakers.
That is the target you should be aiming for, with normal speakers.
If you get to 0.1mVac of hum and noise from the completed "in Chassis" amplifier, then well done.
 
#8
a DMM that has a 199.9mVac scale measures very approximately the hum and noise on the amplifier output.

Don't buy DMM that stop at 2Vac, or worse 200Vac.

A scope that has a 2mV/div cannot show a 0.1mVac hum signal.
some scopes only go down to 5mV/div or 10mV/div
 
#9
Hi !
If anything doesn't help then is possibility to give into power amp's inverted input through big resistor&capacitor and potentiometer connected to dc-power rails so-called compensational signal...It can eliminate the hum. Anyway , how big caps You have on power rails ? And methodically , disconnect preamp and then check the hume at first , then You'll know the source of hum..
 
#12
For my last build, I fixed a piece of copper clad board (un-etched PCB) in the bottom of the case (not connected to the case by the mountings), and grounded everything to that by the shortest path. I mounted PCBs (e.g. power amp, pre amp) to it with spacers, connected the ground for each PCB directly to the copper clad board using a short wire link and added 0.1uF ceramic disk capacitors for all supplies to the PCBs to the copper clad board. It's totally silent even with 98dB @ 1W speakers with my ear right next to the cone. I imagine it's like the ground plane layer on a multi-layer PCB, but not quite as good.
HTH,
Brian
 
#13
brig001 said:
For my last build, I fixed a piece of copper clad board (un-etched PCB) in the bottom of the case (not connected to the case by the mountings), and grounded everything to that by the shortest path. I mounted PCBs (e.g. power amp, pre amp) to it with spacers, connected the ground for each PCB directly to the copper clad board using a short wire link and added 0.1uF ceramic disk capacitors for all supplies to the PCBs to the copper clad board. It's totally silent even with 98dB @ 1W speakers with my ear right next to the cone. I imagine it's like the ground plane layer on a multi-layer PCB, but not quite as good.
HTH,
Brian
Click to expand...

That is exactly what I did and showned in the OP, still have ~20mv 120hz noise.
 
#14
Mains breakthrough doesn't have to be an earth loop, it can be capacitive coupling from AC wiring to input circuitry. An earthed screen would be effective at finding/reducing this type of pickup, perhaps a sheet of tape insulated copper foil wired to chassis - move it around and see if the hum decreases.
 
#15
Fusion916 said:
That is exactly what I did and showned in the OP, still have ~20mv 120hz noise.
Click to expand...

Does your PCB cover the whole bottom of the chassis? Sorry, I can't tell from the photo.

TBH, your noise looks like it is caused by high current peaks when the PSU capacitors are being charged, or even a rectifier diode commutation issue. Does it go away if you put some resistance (maybe 1 ohm) in series with the transformer secondaries? If not, what about between the rectifier and the smoothing capacitors?

Some good reading here: http://www.diyaudio.com/forums/power-supplies/190638-rc-snubbers-diode-recovery-noise.html

Brian
 
#17
brig001 said:
For my last build, I fixed a piece of copper clad board (un-etched PCB) in the bottom of the case (not connected to the case by the mountings), and grounded everything to that by the shortest path. I mounted PCBs (e.g. power amp, pre amp) to it with spacers, connected the ground for each PCB directly to the copper clad board using a short wire link and added 0.1uF ceramic disk capacitors for all supplies to the PCBs to the copper clad board. It's totally silent even with 98dB @ 1W speakers with my ear right next to the cone. I imagine it's like the ground plane layer on a multi-layer PCB, but not quite as good.
HTH,
Brian
Click to expand...
Not try to explain with words,post a schematic!
 
#18
Here is a simple guide.

Short connection between the two filter caps
Tee off this connection
1st connection is speaker return
2nd is power ground return
3rd is signal ground

Twist tightly the wires to the bridge rctifiers from the transformer
Twist tightly the wires from the bridge to the filter caps
Twist tightly the wires from the filter caps to the amp PSU, incl the power ground return

Look at the link I posted earlier - that gives the details in the .pdf I posted.

For a practical example, you can take a look at the nx-Amplifier write up. There is a wiring diagram in there and quite a few photos.

On my power amps (15 W through to 250 W) I am getting 500 uV and below noise on the output. I really can't see accurately below this level because my scope resolution isn't there. My amps are silent on 89 dB speakers.
 
Status
Not open for further replies.
Top Bottom