differential referencing ground

Hi,

"differential circuitry to forward and back reference audio signals"....

Are there any books/pdf.. that explain this subject... maybe this is part of the clues to proper grounding analog circuits...
I googled it, and found posts here... but not much details, because i suppose this is quite complex...

Thank you very much!

JAY X

No it is actually a simple concept. I ASSume you googled differential amp?

Just like a differential/balanced audio input subtracts the audio low (or audio -) from the audio hot (or +) to resolve a clean audio signal, the same differential handling can be accomplished between different small circuit blocks each with unique local grounds.

Use a brute force ground for a modest PCB area of single ended circuitry, then use active differential circuitry to pass that audio information as a local hot and a local low to the next local ground and circuit area.

It seems I have explained this about a hundred times, but many basic opamp stages can be converted to a differential stage, by adding a couple resistors, or selectively paying attention to values and how and where the existing resistors connect.  (Note: these different local grounds can also be connected to each through a common power supply, what matters is that the audio is referenced between these difference local ground voltages)

The classic example of the benefit from this is inside a 6' long console. The ground at the console input channel will never be the same voltage as ground in the master section several feet away. But passing the signal differentially from the strip to the master section can insure good signal integrity, as it cancels out the ground differences and delivers accurate signals. 
.
Simple in concept, not always trivial in execution but with planning and awareness we can generally leave brute force to the brutes and finesse away ground differences.

JR

PS: Referencing forward or back is two sides of the same coin... You only need a single differential between two internal local grounds, so that differential can either condition the signal (forward) being sent to reference to the ground the signal is being sent to, or on the input side ground is back referenced to the ground that is sending the signal in. For single ended interfaces (like an unbal insert point) we routinely differential both, so the output signal is referenced to be accurate wrt the insert jack ground. Likewise the unbalanced input from the insert jack is also read differentially wrt the jack ground. While 3 circuit interfaces benefit from both audio conductors carrying just audio not stray ground currents. 

http://electronicdesign.com/analog/whats-all-noise-rejection-stuff-anyhow

try this.

lookup the figures/pics, easy to miss them.

> try this.

OMG!! Bob Pease invented the differential input! Way back in 2009! 

yeah.. back to the future!

PRR said:

> try this.

OMG!! Bob Pease invented the differential input! Way back in 2009! 

Click to expand...

Well he couldn't invent it in 2013 (RIP).

OK here's a true story, apparently it was still novel to some at the USPTO in the '70s. One of my kit phono preamps (1980) used a differential input. Some guy from TX sends me a letter saying I am infringing on his patent... I ordered a copy of the file wrapper on his patent application to read his arguments with the examiner and the examiner had never heard of a differential input before his patent filing. 

I sent him back a zerox copy of an old transformer input tube preamp schematic that was balanced/differential input thanks to the transformer. The patent office routinely issues dubious patents, and some lawyers take advantage of that. I recall talking with one patent lawyer who didn't even advise doing a patent search before filing... He doesn't care as long as the patent gets issued, he gets paid. If theres a later problem he gets paid again.  If a patent search turns up a conflict, no patent gets filed and no money for the lawyer.

JR

Hi!

Attached is a simple schematic of a volume pot in its own pcb board, wired in between two stages on the main pcb: the signal comes from the output of the preceding stage, enters the pot and goes to the fader amplifier. To avoid gnd loops in the main pcb, i wired the gnd differentially.... i think it is correct... but  i'm not completely sure... ???

thanks for your advise!

Jay X

That´s a comparator

Hi!

Ok.... I see!  it needs the two resitors in the  non inverting input... to turn it into a differential amplifier.

So this arrangement (corrected) should work when there are stereo circuits, eliminating gnd loops..right?
Because right now to avoid gnd loops i pull out the gnd return from one channel...and route it in series with the other section of the dual pot.

Good reading on the topic: http://www.hypex.nl/docs/papers/The%20G%20Word.pdf

Samuel

Hi!

hmmm.... read this paper.... the reference translator on page 5.... is part of the solution...!!

Basically you can implement figure 10/11 in the text. Problems start when considering that for good CMRR, the receiver input resistance needs to be much higher than the pot resistance; for practical values this would impose a severe noise penalty. You could use an instrumentation amplifier (figure 19) but that seems a bit excessive for such a simple task.

Ensuring that the ground references of the driving and receiving amplifier are physically close to each other (without significant interfering currents passing nearby), and the three connecting wires to/from the pot tightly coupled, is probably the better approach.

Samuel

PS: Or making the driving amplifier the differential amplifier!

Samuel

JAY X said:

Hi!

Ok.... I see!  it needs the two resitors in the  non inverting input... to turn it into a differential amplifier.

Click to expand...

I think you missed the point that L'Andratté made.
The non-inverting input should not go to ground.

Quote from: PRR on March 15, 2013, 12:42:13 AM

    > try this.

    OMG!! Bob Pease invented the differential input! Way back in 2009! 

This is my understanding of the differential mode history. The Differential mode was first invented about 1870-1890 by Alexander Graham Bell. It was patented in 1897 as a Power Transformer by Nikola Tesla.
His patent did not talk about the benifets of differential mode. Later wide band transformers were developed at Bell Labs for telephone circuits. This was needed for long paths of audio.
Bob Pease gave us a real insight as how a real active solid state differential amplifier works.
Bob was a great mentor,  I only had a few casual meetings with him.  We all need to get to know  each component really works.
Bob Passed in June 2011 just after leaving Jim Williams memorial service.
Duke

> The Differential mode was first invented about 1870-1890 by Alexander Graham Bell.

Story I heard (utterly unverifiable), is that a teleGRAPH operator "discovered" it. Telegraphs were usually worked earth-return, saved half the wire cost. Most lightning does not induce enough surge to clack a telegraph sounder. But out on the prairie, sometimes all telegraph traffic had to be held until the storm passed.

So sitting on his butt, with several idle lines, some guy managed to get a clue to the guy on the other end, to connect across two lines instead of line to ground. That worked really well despite lightning. While most telegraphy stayed earth-return, differential at half capacity was a fall-back for severe interference.

Ground-ref telephone has been done, but "most" analog telephony was floating-audio because super-small interference (especially parallel power lines) is very annoying.

I won't argue Tesla's contribution because the Martians stole all his papers and left bogus papers for the FBI to seize when his body died.

Pease's practical advice is greatly missed.

JAY X said:

Attached is a simple schematic of a volume pot in its own pcb board, wired in between two stages on the main pcb: the signal comes from the output of the preceding stage, enters the pot and goes to the fader amplifier. To avoid gnd loops in the main pcb, i wired the gnd differentially.... i think it is correct... but  i'm not completely sure... ???

Click to expand...

That is not differential. I think you want to learn more about grounding between multiple circuits which is a different topic.

You don't need differential connections between boards in the same piece of gear. One could argue you don't need differential connections between two pieces of gear on the same rack. The extra circuitry would add more noise than it fixes.

Differential connections use two signal conductors where one of the conductors carries the same exact signal as the other but inverted. The receiver then takes the difference between the two signals and in doing so any in-phase noise will cancel out.

Also, unlike a regular "single ended" connection where one signal conductor is used referenced to a ground, the differential connection (also called balanced) is not referenced to ground and therefore it cannot carry ground currents and thus it is immune to ground noise (ground loops).

But circuits in the same piece of gear are not (should not) have long runs of cables that would be picking up noise. And ground wires should run to each board from a single point on the power supply out to each circuit in a branching or star fashion such that there are no loops that could carry ground currents between boards (again no ground loops). Also, any signal wires that are high impedance, should be shielded but the shield wire should only be connected at one end.

So you probably want to just learn about grounding multiple circuits together. Incorrect grounding is almost always the most likely cause of excessive noise in a circuit. If something is buzzing, you have a grounding problem.

You might choose to make inputs and outputs differential to eliminate ground currents between other peices of gear or if the gear is going to be connected to long cables in an environment where there are other noise sources (like a radio station). But again, these are really two different topics.

squarewave said:

JAY X said:

Attached is a simple schematic of a volume pot in its own pcb board, wired in between two stages on the main pcb: the signal comes from the output of the preceding stage, enters the pot and goes to the fader amplifier. To avoid gnd loops in the main pcb, i wired the gnd differentially.... i think it is correct... but  i'm not completely sure... ???

Click to expand...

That is not differential. I think you want to learn more about grounding between multiple circuits which is a different topic.

You don't need differential connections between boards in the same piece of gear. One could argue you don't need differential connections between two pieces of gear on the same rack. The extra circuitry would add more noise than it fixes.

Click to expand...

Not in my judgement or experience.

Differential connections use two signal conductors where one of the conductors carries the same exact signal as the other but inverted. The receiver then takes the difference between the two signals and in doing so any in-phase noise will cancel out.

Also, unlike a regular "single ended" connection where one signal conductor is used referenced to a ground, the differential connection (also called balanced) is not referenced to ground and therefore it cannot carry ground currents and thus it is immune to ground noise (ground loops).

But circuits in the same piece of gear are not (should not) have long runs of cables that would be picking up noise. And ground wires should run to each board from a single point on the power supply out to each circuit in a branching or star fashion such that there are no loops that could carry ground currents between boards (again no ground loops). Also, any signal wires that are high impedance, should be shielded but the shield wire should only be connected at one end.

So you probably want to just learn about grounding multiple circuits together. Incorrect grounding is almost always the most likely cause of excessive noise in a circuit. If something is buzzing, you have a grounding problem.

You might choose to make inputs and outputs differential to eliminate ground currents between other peices of gear or if the gear is going to be connected to long cables in an environment where there are other noise sources (like a radio station). But again, these are really two different topics.

Click to expand...

Again the paper written by Bruno Putzey that Sam linked to is a good source of information.

JR

squarewave said:

Differential connections use two signal conductors where one of the conductors carries the same exact signal as the other but inverted.

Click to expand...

No. Only the first half of this phrase is true; indeed it takes two conductors, but you don't need balanced drive to operate a differential connection.

The receiver then takes the difference between the two signals and in doing so any in-phase noise will cancel out.

Click to expand...

That is true.

Also, unlike a regular "single ended" connection where one signal conductor is used referenced to a ground, the differential connection (also called balanced)

Click to expand...

Differential does not need to be balanced.

But circuits in the same piece of gear are not (should not) have long runs of cables that would be picking up noise. And ground wires should run to each board from a single point on the power supply out to each circuit in a branching or star fashion such that there are no loops that could carry ground currents between boards (again no ground loops).

Click to expand...

"Should" is a very important word here. Reality does not care about should; shit should not happen, but it does. Your proposed ideal solution is just not workable in a mixer of significant size. Very often "star ground" is presented as a panacea, but just try star ground in a power amp and you're up for big desillusion.