With my amplifier I would like to include a diagram that shows how to ground the amplifier and make it easy to read and unambiguous as possible. Sometimes I think some of the explanations can be confusing and I would like to try to have a diagram that minimizes the confusion. I have included a PDF of this diagram. If there are any comments on the grounding and how it is diagrammed, please let me know. Thanks.
Seen from the input your amp are really needs to mix reference point (input ground) with power output ground.
This anyway creates a relatively huge loop through interwinding capacitance of transformers.
Due to surrounding field in such a loop will flow derivative current.
This current will anyway produce voltage drop even on wire resistance (except of voltage drop by huge load current damped to your amp's ground and rectifier hammering currents)
All of this will make your signal reference point noisy in common mode.
Unfortunately noninverting amps have nonlinear common mode error vs common mode itself and this will provide you with a classical "unacceptable solid-state sounding" despite of the quality of the amplifier itself.
Yours first design goal is to break ground loop or make it's current to flow outside a signal path, then to change signal reference point to something another then input ground.
Perfect article from Bruno Putzeys are attached.
View attachment WP_The_G_word.pdf
This anyway creates a relatively huge loop through interwinding capacitance of transformers.
Due to surrounding field in such a loop will flow derivative current.
This current will anyway produce voltage drop even on wire resistance (except of voltage drop by huge load current damped to your amp's ground and rectifier hammering currents)
All of this will make your signal reference point noisy in common mode.
Unfortunately noninverting amps have nonlinear common mode error vs common mode itself and this will provide you with a classical "unacceptable solid-state sounding" despite of the quality of the amplifier itself.
Yours first design goal is to break ground loop or make it's current to flow outside a signal path, then to change signal reference point to something another then input ground.
Perfect article from Bruno Putzeys are attached.
View attachment WP_The_G_word.pdf
Big loop between power earth (ground) and the signal ground.
See the ground loop link in my signature below as there are some useful guidelines in there.
🙂
See the ground loop link in my signature below as there are some useful guidelines in there.
🙂
Could you explain where the ground loop is? Maybe my diagram is unclear. There is actually one point of contact between the earth ground and the input ground at the connector panel. The earth ground then connects to the earth ground pin on the power entrance socket.
The Safety Ground/Protective Earth should attach to the chassis near where the AC line enters the chassis.
The DC supply common and the audio circuit common should attach to the chassis at a single point near the audio input connectors.
The DC supply common and the audio circuit common should attach to the chassis at a single point near the audio input connectors.
This is what I intended in the diagram. If that is not clear, perhaps I could try a different symbol to indicate this. I use the chassis symbol to indicate this point. Some of what I read suggested that a wire should be connected between the rings of the two audio inputs and then to the ground point where the safety ground attaches to the chassis.
I read over the two other documents I saw. From what I understood, one said that the problem is unsolvable without balanced inputs. The other basically said that the problem is unsolvable. I was attempting to go by at first the advice given in Doug Self's book but try to make a diagram that was a little clearer than that diagram.
I read over the two other documents I saw. From what I understood, one said that the problem is unsolvable without balanced inputs. The other basically said that the problem is unsolvable. I was attempting to go by at first the advice given in Doug Self's book but try to make a diagram that was a little clearer than that diagram.
The audio GND is connected to protective earth (chassis) at the input panel. With two mono amps wired in this way, you'll get a large loop through the mains cables. If the sources (pre-amp, CD player, whatever) have a safety earth connection, they would also form loops. I'd suggest to separate the audio GND from the protective earth using a ground loop breaker.
Your second diagram is better and makes it clear there is only one connection point to the chassis.
Your scheme is similar to Self’s which I must say I have not tried, but he does make the point that it is very robust in dealing with ground loops.
Normally, I would bond the two input grounds together where they enter the chassis but not to the chassis at that point. I then run the signal wires and the shields separately back to the amplifier modules. The reason bond the input grounds like this is to trap any internal loop currents inside the amplifier. You don’t want those currents flowing out through the interconnect shields to the source equipment as this will cause volt drops across the shield resistance that appear in series with the source signal, degrading the SNR.
The humbreaking resistor (HBR) acts as an attenuator with the interconnect screen cable inside the amplifier to further reduce cross channel ground loops. If you look on page 37 of the ground loop presentation, it shows how to connect the HBR - it is very important that this is done correctly because if done as you have shown you will exacerbate any cross channel ground loops.
Also see pages 61-64 on how to route the input wiring from the rear panel to the amplifier modules (assuming you have two separate modules) again this done in the interests of keeping the loop areas as small as possible and avoiding the dreaded cross channel ground loop.
Take note of mbrennwa’s comments since this is still an issue and will arise because any stray magnetic fiield from the transformer will drive a loop current around the safety ground to the source equipment and back through the input cable shields causing noise. You don't get this problem with balanced interconnects and with single ended inputs, a ground lifter can solve the problem. Let me add, I am not using ground lifters anymore - seems all of the other practices to mitigate ground loops (covered in the presentation) basically solve the issue.
On the ground loop page on the link below I also gathered all of diyAudio member ilimnz’s posts on this subject - its the best summary around IMV concerning amplifier ground loops.
Your scheme is similar to Self’s which I must say I have not tried, but he does make the point that it is very robust in dealing with ground loops.
Normally, I would bond the two input grounds together where they enter the chassis but not to the chassis at that point. I then run the signal wires and the shields separately back to the amplifier modules. The reason bond the input grounds like this is to trap any internal loop currents inside the amplifier. You don’t want those currents flowing out through the interconnect shields to the source equipment as this will cause volt drops across the shield resistance that appear in series with the source signal, degrading the SNR.
The humbreaking resistor (HBR) acts as an attenuator with the interconnect screen cable inside the amplifier to further reduce cross channel ground loops. If you look on page 37 of the ground loop presentation, it shows how to connect the HBR - it is very important that this is done correctly because if done as you have shown you will exacerbate any cross channel ground loops.
Also see pages 61-64 on how to route the input wiring from the rear panel to the amplifier modules (assuming you have two separate modules) again this done in the interests of keeping the loop areas as small as possible and avoiding the dreaded cross channel ground loop.
Take note of mbrennwa’s comments since this is still an issue and will arise because any stray magnetic fiield from the transformer will drive a loop current around the safety ground to the source equipment and back through the input cable shields causing noise. You don't get this problem with balanced interconnects and with single ended inputs, a ground lifter can solve the problem. Let me add, I am not using ground lifters anymore - seems all of the other practices to mitigate ground loops (covered in the presentation) basically solve the issue.
On the ground loop page on the link below I also gathered all of diyAudio member ilimnz’s posts on this subject - its the best summary around IMV concerning amplifier ground loops.
Last edited:
You may find this thread interesting, the posts by ilimzn are thought provoking. The dozens schemes to wire an amp...
Ok, I think I might have it right now. Perhaps this is what is intended, looking at the figure on page 37?
Ok - if you are going for Selfs scheme that’s great. Bond the safety earth and the incoming signal ground together at the same point. This will prevent loop currents (which will tend to be at HF and couple via the pri-sec interwinding capacitance) flowing between the two points through the chassis. I would run the ground wire from the socket along the back panel and bond it with the input signal ground at one place. Try to keep it short and thick consistent with separating power and signal entry points. Check that all metal chassis parts are well connected together for safety.
A screen and a GOSS band are excellent ways of reducing noise if you can specify them. For reference, the interwinding capacitance on a 1.2 kVA toroid went from 1.3 nF to 100 pF with the addition of a screen. On a 500 VA transformer, the figure was 500 pF to 70 pF. The differential power BW on a mains toroid is 60kHz (measured) and the common mode BW > 3 MHz ( measured). My signal generator only goes to 3MHz - since the common mode coupling mechanism pri - sec is capacitive, there’s every reason to assume it’s actually 10’s of MHz.
The HBR looks good as well where you have located it.
A screen and a GOSS band are excellent ways of reducing noise if you can specify them. For reference, the interwinding capacitance on a 1.2 kVA toroid went from 1.3 nF to 100 pF with the addition of a screen. On a 500 VA transformer, the figure was 500 pF to 70 pF. The differential power BW on a mains toroid is 60kHz (measured) and the common mode BW > 3 MHz ( measured). My signal generator only goes to 3MHz - since the common mode coupling mechanism pri - sec is capacitive, there’s every reason to assume it’s actually 10’s of MHz.
The HBR looks good as well where you have located it.