Since a solid piece of steel or aluminum is essentially a very large gauge wire with near-zero resistance, why not come directly out from speaker ground terminals, rca shield terminals, amplifier grounds, the ground wire from the AC plug, and everything else with a less than 1-inch wire and bolt it straight into the chassis?
Or to look at it another way, why add a length of wire of any length at all from these connections to ground, introducing lcr, even if those values are small?
If the goal is to provide the most direct path to ground, I don't understand elaborate "star ground" schemes or other variations.
I'm especially interested in the answer today as I have just finished putting my latest amp together. Everything works, but the moment I plug in any RCA to the amp - instant noise. This happens whether I use CD, Preamp, or computer as the source. The only thing that doesn't give terrible noise is the headphone jack of a smartphone. I've been trying different ground schemes in this amp now for several hours. The only idea I haven't tried yet is "everything to chassis." I'm about ready to give it a shot as I just can't get rid of the 60Hz hum plus the switch-y noise from my PC.
If connecting everything straight to the nearest point on the chassis is the wrong idea, what does the right idea look like? Anyone have pictures of a good grounding scheme implementation?
Or to look at it another way, why add a length of wire of any length at all from these connections to ground, introducing lcr, even if those values are small?
If the goal is to provide the most direct path to ground, I don't understand elaborate "star ground" schemes or other variations.
I'm especially interested in the answer today as I have just finished putting my latest amp together. Everything works, but the moment I plug in any RCA to the amp - instant noise. This happens whether I use CD, Preamp, or computer as the source. The only thing that doesn't give terrible noise is the headphone jack of a smartphone. I've been trying different ground schemes in this amp now for several hours. The only idea I haven't tried yet is "everything to chassis." I'm about ready to give it a shot as I just can't get rid of the 60Hz hum plus the switch-y noise from my PC.
If connecting everything straight to the nearest point on the chassis is the wrong idea, what does the right idea look like? Anyone have pictures of a good grounding scheme implementation?
Last edited:
Yes, words and pictures http://hifisonix.com/wordpress/wp-c...re-up-a-Power-Amplifier_Updated-Autosaved.pdf
So I studied the document scottjoplin linked to.
Solutions it provided were:
1) Using a full wave bridge rectifier to "lift" the 0V connection to ground. While it gave this suggestion, it also said it's dangerous and illegal. Because of that, I don't consider this a solution.
2) Use a different transformer - one with a screen and/or flux band. Mine is already installed, I'm not going to use a different one. I'm not sure how to add a flux band to a toroidal transformer, or if it is even possible. I don't know if this comes with a screen.
3) Add hbr resistor to eliminate cross-channel ground loop. I tested the amp for this problem, it's silent. Noise only happens when I plug an RCA into it (Whether or not that rca is connected to anything on the other end makes no difference.)
4) Separate high current ground paths from signal ground paths. Aren't they both electrically connected and therefore impossible to separate? Physical distance separation is not possible within the same chassis anyway, and moving them to separate chassis introduces more problems than it solves.
5) Twist high-current wires tightly together. I cannot imagine how this would make any difference compared to running them side-by-side for the entire length, which I've already done. Twisted wires are twisted to reduce the size of the "loop" they encompass, so that they don't pick up huge amounts of magnetic flux inducing voltage within them. Wires run closely together side-by-side is just about the same. Somebody correct me if I'm wrong here.
6) "T" or "Star" ground. This seems to me to be an effort to add impedance to some ground references while not to others. For the life of me I can't understand why this would help.
The document BesPav linked had a lot to say about common mode noise reduction through subtraction in differential amplifiers. The amp boards I'm using do use differential amps, however I'm not using balanced inputs because my DSP is RCA-only. I'm using 2-conductor +shield cabling inside the amp to go from the RCA terminals to the inputs of the amp boards, with the negative signal wire and ground soldered together at the RCA terminal shield. There were some interesting diagrams near the end of the article, however and I will be referencing them as I continue.
At this point I'm both frustrated and convinced that the problem here isn't reference voltages, impedance, or length of wire, but rather what is referenced in mathematics as "topology." That is to say that I think the shape, orientation, and dimensions of all of the conducting paths add together to give me a loop which is picking up changes in magnetic flux and then amplifying it. Without the resources to analyze this meticulously, all I'm left with is guesswork and trial-and-error experimentation.
For the record I'm using an Antec AN-5220 transformer, the DIYAudio power supply board, and four Neurochrome Modulus 86 amplifier boards.
Solutions it provided were:
1) Using a full wave bridge rectifier to "lift" the 0V connection to ground. While it gave this suggestion, it also said it's dangerous and illegal. Because of that, I don't consider this a solution.
2) Use a different transformer - one with a screen and/or flux band. Mine is already installed, I'm not going to use a different one. I'm not sure how to add a flux band to a toroidal transformer, or if it is even possible. I don't know if this comes with a screen.
3) Add hbr resistor to eliminate cross-channel ground loop. I tested the amp for this problem, it's silent. Noise only happens when I plug an RCA into it (Whether or not that rca is connected to anything on the other end makes no difference.)
4) Separate high current ground paths from signal ground paths. Aren't they both electrically connected and therefore impossible to separate? Physical distance separation is not possible within the same chassis anyway, and moving them to separate chassis introduces more problems than it solves.
5) Twist high-current wires tightly together. I cannot imagine how this would make any difference compared to running them side-by-side for the entire length, which I've already done. Twisted wires are twisted to reduce the size of the "loop" they encompass, so that they don't pick up huge amounts of magnetic flux inducing voltage within them. Wires run closely together side-by-side is just about the same. Somebody correct me if I'm wrong here.
6) "T" or "Star" ground. This seems to me to be an effort to add impedance to some ground references while not to others. For the life of me I can't understand why this would help.
The document BesPav linked had a lot to say about common mode noise reduction through subtraction in differential amplifiers. The amp boards I'm using do use differential amps, however I'm not using balanced inputs because my DSP is RCA-only. I'm using 2-conductor +shield cabling inside the amp to go from the RCA terminals to the inputs of the amp boards, with the negative signal wire and ground soldered together at the RCA terminal shield. There were some interesting diagrams near the end of the article, however and I will be referencing them as I continue.
At this point I'm both frustrated and convinced that the problem here isn't reference voltages, impedance, or length of wire, but rather what is referenced in mathematics as "topology." That is to say that I think the shape, orientation, and dimensions of all of the conducting paths add together to give me a loop which is picking up changes in magnetic flux and then amplifying it. Without the resources to analyze this meticulously, all I'm left with is guesswork and trial-and-error experimentation.
For the record I'm using an Antec AN-5220 transformer, the DIYAudio power supply board, and four Neurochrome Modulus 86 amplifier boards.
Last edited:
Hi, just referring to the highlighted text alone, I think this is a mechanical problem not an electrical one.
Can you check the RCA socket? Is it mounted to the chassis? Are there plastic/nylon/insulating washer between them? Does plugging RCA cable to it move it out from the insulation and make contact with the chassis?
I came across this kind of problem before on phono amp and it was mechanical.
That is not the goal. I guess this explains your confusion?uzernaam said:
No ground loops. Every signal is referenced to what it should be referenced to (there is no such thing as 'a voltage') by one and only one path. All current loops (both PSU and signal) identified and, where appropriate, have their loop area minimised by techniques such as twisting. Wires establishing a reference point do not carry any significant current. At a suitable point there is a single connection from a signal reference point to the chassis (i.e. safety ground).
You are wrong. Twisting does two things: it ensures that the two wires are about as close as they can be; it frequently reverses the polarity of the magnetic induction so the net effect is near zero.
Star grounds are overrated, in my opinion, but they do provide a quick and easy way to achieve reasonable performance without having to do any thinking. I prefer a bus ground, combined with some thinking.
There are a lot of commercially made amps about with no connection to mains earth.
Two wire mains leads.
'Double Insulated' construction standards.
I use a Residual Current Circuit Breaker if the lid's off.
Audio grounding is a bit of an art.
I've worked with radio equipment, where ground planes are the norm.I don't think that way of think completely transfers to audio!
I think to make ground planes work, we'd have to make them massive.
Although the resistances and inductances are small, the mains-related currents can be large, and the audio input is very sensitive.
I've found Doug Self's book helpful.
One way to think about the problem is to consider that there will be 'common mode' hum and noise on the outer of the screened cable at the input.
Your amplier's job is to ignore that and amplify the difference between the input and the return.
You could try drawing a schematic where every ground wire (or other bit of metal!) is replaced by a significant resistor and try to see where noise is coupled in.
Don't think of the box as zero, only the phono outer is zero. Ground, earth, chassis, box and transformer centre tap are all different voltages.
Assume the rectifier is feeding the PSU caps with big fat current pulses.
There is a big handful of common mode on the mains, the transformer is coupling some of that into the box.
Hope that Helps?
Two wire mains leads.
'Double Insulated' construction standards.
I use a Residual Current Circuit Breaker if the lid's off.
Audio grounding is a bit of an art.
I've worked with radio equipment, where ground planes are the norm.I don't think that way of think completely transfers to audio!
I think to make ground planes work, we'd have to make them massive.
Although the resistances and inductances are small, the mains-related currents can be large, and the audio input is very sensitive.
I've found Doug Self's book helpful.
One way to think about the problem is to consider that there will be 'common mode' hum and noise on the outer of the screened cable at the input.
Your amplier's job is to ignore that and amplify the difference between the input and the return.
You could try drawing a schematic where every ground wire (or other bit of metal!) is replaced by a significant resistor and try to see where noise is coupled in.
Don't think of the box as zero, only the phono outer is zero. Ground, earth, chassis, box and transformer centre tap are all different voltages.
Assume the rectifier is feeding the PSU caps with big fat current pulses.
There is a big handful of common mode on the mains, the transformer is coupling some of that into the box.
Hope that Helps?
I was just wondering about common practice of wiring in the heyday of tube technology. There's not much difference between NF gear (e.g. a Wurlitzer Sideman) oder HF gear (e.g. Tektronix 511) as there's a bundle of wires bundled together with a lineman's knot, not much twisting of wires anywhere, no star grounds...
I know it is slightly off topic but I was wondering what I am missing here?
I know it is slightly off topic but I was wondering what I am missing here?
Does the noise change at all when you connect an input compared to no input? This is a power amp I presume, an unconnected interconnect will pick up noise. What happens if you short the inputs?
I used to own a commercial recording console, a Neotek Series II, that did basically that. The frame was made of thick aluminum extrusions, and everything was grounded to the frame as many times as possible. The entire circuit was unbalanced, except for the mike amp inputs, and the noise was extremely low. The console did use an external AC power supply, which IIRC obeyed normal grounding rules separating the chassis and audio ground, but the chassis of the console itself, all 6' by 4' of it, was grounded aluminum extrusions.
A ground loop via the interconnects then, could be a cross channel loop via the DSP, unlikely though. Is the DSP earthed?
Do not attempt to put anything like a flux band around a toroid. If it makes any connection that goes around the outside and through the hole (i.e. same as the windings) it will act like a winding with 1 turn and get very hot, very fast.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.