YAAGLI - Yet Another Aikido Ground Loop Issue

[enigma_swe]

Good day Gents,

This is my first post here so bear with me. I'm a complete beginner when it comes to electronics but i have the motto that you have to try to learn something.

Either way my current learning project is an Aikido pre-amp based on an all-in-one octal kit. Everything is assembled and working fine when not using a volume potentiometer.

I have looked everywhere on how to actually connect those things to not create ground-loops, but i seem to have failed quite drastically there

I have used my non-existing image editing skills to create a sort of overview of how all inputs and outputs are connected which can be seen in the image below ->




The input connectors are to the top left, and all hot signals goes to the signal selector and then continue to the pot. The grounds of the inputs are connected channelwise and then connected to the input grounds on the PCB.

The signal output from the potentiometer are connected to the signal inputs on the PCB. The ground from the pot are connected together (the left&right and the pot housing ground) and then directly to the main G in the center of the PCB. I guess that this approach is quite wrong but i'm unsure how it actually works. The ground from the pot, isn't that just something that needs to be "dumped" out somewhere?

The safety ground (third prong) is connected via a set of diodes (image below) to J7 (to the upper left of the pcb in the picture above), and the pcb is then connected to the top metal chassis trough with a screw.



Does anyone have any tips here on what i'm doing wrong, bring out the teacher within and give me the lecture of a lifetime!

Thanks for taking the time.

Andreas

[SY]

First, connect safety ground directly to the chassis; this is very important for safety! Use the diodes to lift signal and PS ground from safety ground. Make sure that the "grounds" of the input and output connectors are isolated from the chassis.

[enigma_swe]

Hi Sy, thanks for your response. I apologize for being a bit slow but i'll try to explain how i understood your comment.

Instead of connecting the safety ground through the diodes to the pcb (which is connected to the chassis) i should isolate the pcb from the chassis and connect safety ground directly to the chassis and then from that connection point go through diodes to connect to the pcb "earth"?

The input and output connectors are isolated from the chassis.

So you believe that the issue with audible hum (which gets louder with increasing the volume knob) comes from the safety earth being connected in a bad way (with the diodes).

Maybe it could help if i used a photo to illustrate how things are connected ?

Many thanks!

/Andreas

[enigma_swe]

Good morning,

I was having a short 'meditation session' during breakfast about the grounding issue and decided i would do yet another sketch

Maybe if someone could just give an indication that the ground part of the volume pot is ok then i can focus on the safety ground as this might be the root cause for the hum that i have in my setup.

Based on Sy's comment yesterday i thought about how to connect the safety ground. Would it be ok to do as i've done in the image below ?



To go from the third prong directly to a bolt in the chassis top (aluminum, rest of chassis is wood) and then from that bolt go through the diodes and to the J7 jumper on the board.

With this approach i guess that i should use the rubber o-rings so that the only connection to the chassis is done at the bolt and not through the top-left mounting hole (which is what i am doing at the moment)

Any input would be dearly appreciated.

/Andreas

[SY]

That diagram looks better. The idea of the safety ground is to have EVERY exposed metal part be connected to a place that has zero potential with respect to the ground that we stand on. So if the third wire is not connected directly and firmly to the chassis, there's a danger that this connection will be broken, causing a shock hazard.

With that disposed of, now we need to look at the circuitry. The "ground" in that context is a different animal, but confusingly has the same name. In this context, it's a common reference point for all of the signals. So that's why one isolates the jacks from the chassis and returns their signal grounds to the single point signal ground reference (probably on the board, if it's well-designed, which I'm sure Broskie's is).

OK, so now we have two different "grounds." One is the signal ground, one is the safety ground. Many times, these can be connected directly together. Sometimes, though, this will cause a hum when interacting with the grounding in other components. One way around that is a "ground lift," where the connection between the two "grounds" is removed. This also can present some safety issues, so as an expedient, the back-to-back diode method is often used. Another common method is a small value resistor (10-50 ohms), but there are some safety concerns there that the resistor could go open in the event of a power fault- diodes tend to fail short-circuit, which will cause a breaker to trip or fuse to blow if there's a power fault. Clearly that's better than electrocuting the user (well, depending on the user; I have a little list...). Really, all you need is a single diode in each direction, not two in series, though.

Hope this clarifies things a bit. Dave Davenport wrote an excellent article on grounding which should be appearing shortly. And this gives me the second opportunity today to gripe about the ambiguity of the word "ground."

[janneman]

Originally Posted by SY

That diagram looks better. The idea of the safety ground is to have EVERY exposed metal part be connected to a place that has zero potential with respect to the ground that we stand on. So if the third wire is not connected directly and firmly to the chassis, there's a danger that this connection will be broken, causing a shock hazard.

With that disposed of, now we need to look at the circuitry. The "ground" in that context is a different animal, but confusingly has the same name. In this context, it's a common reference point for all of the signals. So that's why one isolates the jacks from the chassis and returns their signal grounds to the single point signal ground reference (probably on the board, if it's well-designed, which I'm sure Broskie's is).

OK, so now we have two different "grounds." One is the signal ground, one is the safety ground. Many times, these can be connected directly together. Sometimes, though, this will cause a hum when interacting with the grounding in other components. One way around that is a "ground lift," where the connection between the two "grounds" is removed. This also can present some safety issues, so as an expedient, the back-to-back diode method is often used. Another common method is a small value resistor (10-50 ohms), but there are some safety concerns there that the resistor could go open in the event of a power fault- diodes tend to fail short-circuit, which will cause a breaker to trip or fuse to blow if there's a power fault. Clearly that's better than electrocuting the user (well, depending on the user; I have a little list...). Really, all you need is a single diode in each direction, not two in series, though.

Hope this clarifies things a bit. Dave Davenport wrote an excellent article on grounding which should be appearing shortly. And this gives me the second opportunity today to gripe about the ambiguity of the word "ground."

Building on that, what you should do is to return each pot ground lug (which actually is the 'reference' for the signal) to the pcb pad where you also returned the input screens.
The best way, if you use a screened cable to get the signal from the selector to the pot and then onwards to the input on the pcb, is to picture the pot as a component halfway a screened cable from selector to pcb. The pot ground lug, in that picture, connects to the screen, the hot from selector to pot top, the hot from pot wiper to input on pcb, together with the screen.
I'll bet now your hum is gone.

jd

[chrish]

If the hum gets louder with increasing volume knob position, wouldn't that make the issue 'upstream' of the volume knob?

You say that your RCA inputs and outputs are isolated from the chassis, make sure that your volume pot and input selector are as well.

These things can be a pain to chase, but you usually learn a few lessons on the way!

Good luck,

Chris

[chrish]

Ah, yes, Janneman makes a good point...

Looks like you have put the audio signal ground reference (from the volume pot) to the ground reference of the power supply (the pad next to where you have connected the input signal ground is B+, the nomenclature for power supply positive voltage). Connect the ground reference from the pot to the ground reference pad next to the signal wiring input as mentioned by janneman. Make the power supply ground connection at that ground point in the middle of the PCB.

Hope that fixes it,

Chris

[enigma_swe]

Wow guys, you are overflooding me with lots of good information here that i am currently digesting (together with the pasta i had for lunch).

I DO see this project as a learning project and i'm not in any rush of getting to that "good enough" state but all these tips makes me want to go home from work early and put my hands on the preamp

I will do what SY mentions and change the way the safety ground is connected to the chassis just to cover the safety aspect (as i'm too young to die).

And just to quickly answer chrish's first post all the inputs/outputs and selector&pots are isolated from the chassis.

So just a clarification from my side here, the pot ground lugs should go to the signal input ground on the pcb together with the input connector grounds? Just thinking about this graphically wouldn't "that" created a loop? (this is the reason why i connected the pot ground lugs to the B+ ground as i thought the ground lugs from the pot was like a "trash bin" that should be disposed of somewhere)

Many many thanks, and i'll look forward taking these good pointers with me back home

/Andreas

[janneman]

Originally Posted by enigma_swe

Wow guys, you are overflooding me with lots of good information here that i am currently digesting (together with the pasta i had for lunch).

I DO see this project as a learning project and i'm not in any rush of getting to that "good enough" state but all these tips makes me want to go home from work early and put my hands on the preamp

I will do what SY mentions and change the way the safety ground is connected to the chassis just to cover the safety aspect (as i'm too young to die).

And just to quickly answer chrish's first post all the inputs/outputs and selector&pots are isolated from the chassis.

So just a clarification from my side here, the pot ground lugs should go to the signal input ground on the pcb together with the input connector grounds? Just thinking about this graphically wouldn't "that" created a loop? (this is the reason why i connected the pot ground lugs to the B+ ground as i thought the ground lugs from the pot was like a "trash bin" that should be disposed of somewhere)

Many many thanks, and i'll look forward taking these good pointers with me back home

/Andreas

Andreas, just to be sure: my term 'pot ground lug' referred to the bottom pin of the resistive element, NOT the housing of the pot. My term is admittedly not very accurate, sorry about that. The pot housing is connected to the chassis presumably.

In principle you are correct that there is a ground loop as the screens of the cables that bring in the signal eventually end up on the same potential, signal ground on the pcb. If they are also connected that way at the source, there is a loop.
The saving grace is 1) keep the screened signal cables close together to minimize the loop, and 2) by not connecting them at more than one place to safety ground, you also limit hum and noise injection.

It IS possible that you still get hum. In that case, just disconnect the connection of the screen of ONE channel only at the pcb input pad.

have fun!

jd