Hi,
I have 3 components a DAC, Preamplifier and power amplifier (well actually also a couple of subwoofer amplifiers coming off the high level from the speaker output of the power amplifier, these are floating with no safety earth connection).
Every chassis earth is separate (independent) from signal ground. I found a really low impedance earth connection to the Pre-ampifier on the signal GND improved the sound quality when I did this a while back. I then added a low impedance signal ground to my DAC that was originally floating, this also improved the sound. So I thought great I would do the same to the power amplifier, BUT this created a hum....
So I put some thought into the mechanism and made all 3 signal GND connections to the same single point, and of as identical as I could impedance value.....and the power amplifier still hums when connected with this additional GND.
If I put a higher impedance value (longer and lower gauge) signal GND connection on the power amplifier the hum reduces, but you can still just hear it.
If the mechanism of hum pick up is only possible when there is a voltage difference driving the GND loop how can this make any sense?
Any big brains out there
I have 3 components a DAC, Preamplifier and power amplifier (well actually also a couple of subwoofer amplifiers coming off the high level from the speaker output of the power amplifier, these are floating with no safety earth connection).
Every chassis earth is separate (independent) from signal ground. I found a really low impedance earth connection to the Pre-ampifier on the signal GND improved the sound quality when I did this a while back. I then added a low impedance signal ground to my DAC that was originally floating, this also improved the sound. So I thought great I would do the same to the power amplifier, BUT this created a hum....
So I put some thought into the mechanism and made all 3 signal GND connections to the same single point, and of as identical as I could impedance value.....and the power amplifier still hums when connected with this additional GND.
If I put a higher impedance value (longer and lower gauge) signal GND connection on the power amplifier the hum reduces, but you can still just hear it.
If the mechanism of hum pick up is only possible when there is a voltage difference driving the GND loop how can this make any sense?
Any big brains out there
Long ago there was a cartridge tape player manufacturer who in their layout had the ground PCB trace run all the way around the entire case. As the power transformer was inside this loop, the trace acted as a single turn transformer to the flux that leaked out of the transformer core. Thus there was hum.
The cure was to simply cut the formed loop at what was effectively the start/finish. That dropped the hum away.
So one ground path is fine, two can actually create a loop and pick up noise!
The cure was to simply cut the formed loop at what was effectively the start/finish. That dropped the hum away.
So one ground path is fine, two can actually create a loop and pick up noise!
Hi all,
Get your hands on book Henry W Ott - Electromagnetic combatibility engineering and you can fix all your grounding issues for the rest of your life.
Never disconnect mains safety ground on any device, it protects your life and your property.
edit. I think another book of his is fine, cheaper and exactly to the point
Get your hands on book Henry W Ott - Electromagnetic combatibility engineering and you can fix all your grounding issues for the rest of your life.
Never disconnect mains safety ground on any device, it protects your life and your property.
edit. I think another book of his is fine, cheaper and exactly to the point
Noise Reduction Techniques in Electronic Systems
Last edited:
At a simple level, you need to take 2 things into account. The first is Kirchoff's law and the second is loop area.
The signal forward path goes down the central conductor but the return path for a single ended feed will take EVERY path back and this is why loop area can be important. If you dress your power cables at 90 degrees to signal cables (which has often been recommended) then you are maximising loop area.
beyond that, as Ed pointed out there are plenty of sub-optimal designs out there which are almost guaranteed to cause hum
The signal forward path goes down the central conductor but the return path for a single ended feed will take EVERY path back and this is why loop area can be important. If you dress your power cables at 90 degrees to signal cables (which has often been recommended) then you are maximising loop area.
beyond that, as Ed pointed out there are plenty of sub-optimal designs out there which are almost guaranteed to cause hum
Also try changing the orientations of the components' AC power plugs,
since you seem to be using only 2 prong AC power plugs.
With no signal cable connections, and just the AC line cords plugged in and the power turned on,
measure the 60Hz AC voltage leakage between each chassis and the AC line utility ground.
Orient each component's AC plug for minimum 60Hz AC voltage leakage.
since you seem to be using only 2 prong AC power plugs.
With no signal cable connections, and just the AC line cords plugged in and the power turned on,
measure the 60Hz AC voltage leakage between each chassis and the AC line utility ground.
Orient each component's AC plug for minimum 60Hz AC voltage leakage.
Thanks for all the replies, I understand the following:
I am also guessing that if the alternative path for the return is the interconnect cable AND the earth loop then the relative amounts travelling along each path are related to the impedance of each.
One thought I had was that the interconnect is a twisted pair, and not shielded so if the impedance of the earth loop was low this would be preferred and so the interconnect was no longer benefiting from a balanced signal and return to nullify it's capacity to pick up hum itself (ignoring that contributed by the loop)
I will keep experimenting
- loop area increases the chance of hum pick up
- removing the loop cuts the pathway for hum to travel
- if there is no potential difference throughout the ground path there is no loop current and no hum
I am also guessing that if the alternative path for the return is the interconnect cable AND the earth loop then the relative amounts travelling along each path are related to the impedance of each.
One thought I had was that the interconnect is a twisted pair, and not shielded so if the impedance of the earth loop was low this would be preferred and so the interconnect was no longer benefiting from a balanced signal and return to nullify it's capacity to pick up hum itself (ignoring that contributed by the loop)
I will keep experimenting
I have 3 pin mains plugs, but have no connection from signal GND to the chassis safety earth connection. The 3rd pin on the mains power supply plug deals only with chassis safety earth. I use an entirely separate plug to provide a connection to the earth pin for signal GND.
In the UK on 3 pin connections we keep Live and Neutral in this only phase configuration (or i do!)
I have 2-wire power cords on my turntable, phono preamp, AM FM tuner and CD player. Those commercial designs were designed to work that way.
I use RCA phono plugs and jacks for all low-level and mid-level signal connections (of course I use different connections for the speaker wires).
I have properly wired 3 wire power outlets; Hot, Neutral, Safety Ground.
My mono-block power amplifiers use an IEC power socket, and 3 wire power cord. The Safety Ground from the IEC socket is connected to the amplifier chassis, which also becomes the central ground point. I use Choke input B+ filters. There are some (carefully controlled) local ground loops in the amplifier, such as the power transformer secondary center tap that goes to the first filter cap (-). A second wire from the first cap (-) connects to the 2nd filter cap (-), and only then, with a 3rd wire, from there to the central ground point.
And, the signal input RCA connector signal ground is connected to the bottom of the input stage's volume control, tube Rg, and the bottom of the input stages cathode circuit (a local signal return loop); after that another wire connects from there to the central ground point.
Hum? What hum at the speakers? (less than 100uV, yes micro-volt).
Wires are resistive (no matter small of a resistance); and wires are inductive (no matter how small of an inductive reactance).
Any current along a ground wire will cause a voltage drop from one end of the wire to the other end of the wire. Consider what happens when that voltage differential gets applied to the input stage of an amplifier . . . it gets amplified, and becomes noticeable.
"Grounds are Commonly Misunderstood"
I use RCA phono plugs and jacks for all low-level and mid-level signal connections (of course I use different connections for the speaker wires).
I have properly wired 3 wire power outlets; Hot, Neutral, Safety Ground.
My mono-block power amplifiers use an IEC power socket, and 3 wire power cord. The Safety Ground from the IEC socket is connected to the amplifier chassis, which also becomes the central ground point. I use Choke input B+ filters. There are some (carefully controlled) local ground loops in the amplifier, such as the power transformer secondary center tap that goes to the first filter cap (-). A second wire from the first cap (-) connects to the 2nd filter cap (-), and only then, with a 3rd wire, from there to the central ground point.
And, the signal input RCA connector signal ground is connected to the bottom of the input stage's volume control, tube Rg, and the bottom of the input stages cathode circuit (a local signal return loop); after that another wire connects from there to the central ground point.
Hum? What hum at the speakers? (less than 100uV, yes micro-volt).
Wires are resistive (no matter small of a resistance); and wires are inductive (no matter how small of an inductive reactance).
Any current along a ground wire will cause a voltage drop from one end of the wire to the other end of the wire. Consider what happens when that voltage differential gets applied to the input stage of an amplifier . . . it gets amplified, and becomes noticeable.
"Grounds are Commonly Misunderstood"
Thanks, but the question is not related to good design and topology of ground within an amplifier, I want to more clearly understand the mechanism that drives an earth loop outside of a component in a system. Unless you are suggesting the reason for the ground loop hum on the amplifier IS the ground topology in the amplifier?
The reason for my question is that whilst I can remove the ground loop, and remove the hum; I have a better sound quality (marginal but noticeable) with a low impedance ground connected to a component from my personal experience, but in the case of my amplifier I create a hum doing this.
I will keep experimenting / reading / exploring and share anything useful.
I have just come across this, but not read it yet!
The reason for my question is that whilst I can remove the ground loop, and remove the hum; I have a better sound quality (marginal but noticeable) with a low impedance ground connected to a component from my personal experience, but in the case of my amplifier I create a hum doing this.
I will keep experimenting / reading / exploring and share anything useful.
I have just come across this, but not read it yet!
Both Ott and Whitlock are very good. The real cause of the hum is that the return path for domestic audio is earth/ground. If you go the whole hog to a transformer balanced system, this is no longer the case and you can have earth loops with impunity because the signal no longer flows in them. That's what broadcast and professional audio uses. For domestic audio, we don't need to be quite as extreme because the cable lengths are much shorter.
Think currents. Hum currents are induced in all cables, but the shield of a coaxial cable is lower resistance than the inner. A current is induced in each wire, but because of the different resistances, the voltages developed along the cable are different; equivalent to a differential signal. Which means hum. If the resistances had been equal and each wire had experienced the same hum field, the voltages would be the same and would cancel. This would be a balanced system (balanced impedances, not voltages). Cable balance usually implies twisted pair. Interestingly, you can get a reduction in hum using twisted pair even in an unbalanced system provided the screen is earthed only at one end.
Hum loops are a problem when two pieces of equipment are at a different earth potential. So, put them at the same earth potential. That means plugging them into the same distribution board. And, as already mentioned, minimise loop area because mains hum is a very long wavelength, which means that greater distance intercepts a larger differential in the field. That means short cables to the (nearby) distribution board.
Think currents. Hum currents are induced in all cables, but the shield of a coaxial cable is lower resistance than the inner. A current is induced in each wire, but because of the different resistances, the voltages developed along the cable are different; equivalent to a differential signal. Which means hum. If the resistances had been equal and each wire had experienced the same hum field, the voltages would be the same and would cancel. This would be a balanced system (balanced impedances, not voltages). Cable balance usually implies twisted pair. Interestingly, you can get a reduction in hum using twisted pair even in an unbalanced system provided the screen is earthed only at one end.
Hum loops are a problem when two pieces of equipment are at a different earth potential. So, put them at the same earth potential. That means plugging them into the same distribution board. And, as already mentioned, minimise loop area because mains hum is a very long wavelength, which means that greater distance intercepts a larger differential in the field. That means short cables to the (nearby) distribution board.
I actually tried a 1KVA balanced transformer on each component and even on the DAC it slugged dynamics and lost sound quality, so whilst a lot of people rate them as a preferred solution my experience/preference to date is that they are not for me :-(
"Hum loops are a problem when two pieces of equipment are at a different earth potential. So, put them at the same earth potential. That means plugging them into the same distribution board. And, as already mentioned, minimise loop area because mains hum is a very long wavelength, which means that greater distance intercepts a larger differential in the field. That means short cables to the (nearby) distribution board."
My test was to take a low impedance equal length cable from each component signal GND to a single shared plug, used only for accessing an earth connection. I had hoped this would prevent any voltage creating the possibility of earth loop hum, but weirdly not in the case of the power amp.
My mains set up is a single spur for each plug socket taken back to a separate consumer unit
Thanks for the thinking
"Hum loops are a problem when two pieces of equipment are at a different earth potential. So, put them at the same earth potential. That means plugging them into the same distribution board. And, as already mentioned, minimise loop area because mains hum is a very long wavelength, which means that greater distance intercepts a larger differential in the field. That means short cables to the (nearby) distribution board."
My test was to take a low impedance equal length cable from each component signal GND to a single shared plug, used only for accessing an earth connection. I had hoped this would prevent any voltage creating the possibility of earth loop hum, but weirdly not in the case of the power amp.
My mains set up is a single spur for each plug socket taken back to a separate consumer unit
Thanks for the thinking
I actually meant balanced for the audio signal rather than the mains. Audio on XLRs, not horrible phono plugs.
As it happens, I power my amplifier (which is balanced input on XLR) from a 750VA Topaz isolating transformer. The virtue of the Topaz is that with a stated interwinding capacitance of 1fF, it blocks common-mode interference from the mains. It cleaned up the sound considerably. I wound a low capacitance power transformer and added linear regulator for the DAC in place of its noisy supplied switcher. That further cleaned up sound. I suspect there are a lot of non-CE compliant switchers and data-over-mains excrescences near me.
Seeing your comment about separate spurs back to consumer unit, you might want to try using one spur and short leads to all your kit. It might help.
As it happens, I power my amplifier (which is balanced input on XLR) from a 750VA Topaz isolating transformer. The virtue of the Topaz is that with a stated interwinding capacitance of 1fF, it blocks common-mode interference from the mains. It cleaned up the sound considerably. I wound a low capacitance power transformer and added linear regulator for the DAC in place of its noisy supplied switcher. That further cleaned up sound. I suspect there are a lot of non-CE compliant switchers and data-over-mains excrescences near me.
Seeing your comment about separate spurs back to consumer unit, you might want to try using one spur and short leads to all your kit. It might help.
The audio GND usually has some connection to earth (for safety reasons). Are you sure there is no connection at all? What do you get if you measure the resistance from audio GND to safety earth with a DVM?
Also, how does your DAC get the data? Is it a wireless connection, or is there a cable (USB, ethernet, SPDIF, etc.)?
Uhhhhh,
I think:
The return path for power is Neutral;
And the return path for signal is the RCA plug cable's shield.
The safety ground does two things:
Keeps the chassis from floating at partway, or all the way of the power mains Hot lead (insulation leakage and distributed capacitance leakages of the power transformer needs a safety ground return);
And ground keeps the chassis from being the return path for the signal shield.
Why do you think I did not return the RCA 'ground' connection directly to the chassis (earth / power mains ground). I connect it to all the returns of the input stages volume control, Rg, and the return end of the input tube's self bias network. That forms a low-distance signal return ground loop within the amplifier, instead of a long wire with resistance and inductive reactance that causes lots of ground loop voltage interference.
Only after that local loop, do I bring that point off to the central ground point.
Generally, the two wire power cords of devices, like many phono preamps, tuners, and CD players are not meant to have a ground connected from them to power mains ground, nor off to a power amplifier.
The main exception to that is the ground wire from the turntable to the preamp, which are meant to be tied together. The 2-wire power cord does not have ground in either lead.
Often times, the transition from real world power mains transformers in all devices, to switchers has caused more problems than it solved.
But the power transformers were heavy, and expensive.
So, Switchers were used to make the Bean Counters happy (and the consumer happy, because he likes the lower price, even if it might sound worse).
Just my opinions.
I think:
The return path for power is Neutral;
And the return path for signal is the RCA plug cable's shield.
The safety ground does two things:
Keeps the chassis from floating at partway, or all the way of the power mains Hot lead (insulation leakage and distributed capacitance leakages of the power transformer needs a safety ground return);
And ground keeps the chassis from being the return path for the signal shield.
Why do you think I did not return the RCA 'ground' connection directly to the chassis (earth / power mains ground). I connect it to all the returns of the input stages volume control, Rg, and the return end of the input tube's self bias network. That forms a low-distance signal return ground loop within the amplifier, instead of a long wire with resistance and inductive reactance that causes lots of ground loop voltage interference.
Only after that local loop, do I bring that point off to the central ground point.
Generally, the two wire power cords of devices, like many phono preamps, tuners, and CD players are not meant to have a ground connected from them to power mains ground, nor off to a power amplifier.
The main exception to that is the ground wire from the turntable to the preamp, which are meant to be tied together. The 2-wire power cord does not have ground in either lead.
Often times, the transition from real world power mains transformers in all devices, to switchers has caused more problems than it solved.
But the power transformers were heavy, and expensive.
So, Switchers were used to make the Bean Counters happy (and the consumer happy, because he likes the lower price, even if it might sound worse).
Just my opinions.
How true! If one wants peace of mind then only connect PE directly to Audio GND in the power amplifier and lift Audio GND from PE in all the other source devices with the few famous tricks. Of the tricks I like the 10 or sometimes 100 Ohm 2W resistor the most as it is most effective with only 1 part.
BTW if there is no connection/reference at all between PE and Audio GND then don't be surprised of unexpected high voltage levels at various PSU secondaries. That counts both for + and -. Stuff leaks as stuff is not ideal. If one permits that high voltage levels to discharge to other devices when plugging cables in then please consider that devices do break down because of this (HDMI electronics is known for this).
This complies with my experience that a higher impedance additional connection to earth on a component that could create an earth loop does not because there is a preferred low impedance path along the interconnect that takes most of the return current, and if the 'loop' of an unbalanced interconnected that is a twisted pair is tiny then the pick up is minimal = no hum.
What is the advantage of a lifted 10R connection to Chassis Earth for the signal circuit in comparison to being fully floated?
I currently have the Preamp as the location of the primary GND, I suppose the choice is somewhat arbitrary (and it does sit in the middle).