Balanced phono cable

[Arkless Electronics]

And your point is? You seem knowledgeable about electronics and I'm sure I would be wasting my time in explaining the advantages of the 3 op amp topology over the single one.

 

[ChrisPa]

It's not to do with the advantages of the 3 op-amp configuration,

It's to do with the fact that most(?) single ended input stages are inherently balanced as there's a corresponding -ve input within the circuit. So balanced can often/usually be achieved with no additional active components

No you don't need more than one op amp or an extra transistor and I didn't say you did....but both are poor methods of doing it.

There is a clear advantage - avoiding ground loops. And why would you need to add 3 op-amps? If you already have one op amp in the input circuit, it will have an inverting input too, so no added components are needed at all. If the input stage is discrete, you need one extra transistor - and it is only "in the signal path" in the same way as the single-ended one is.

We'll have to agree to disagree

You seemed to previously disagree with julf when he said "no additional components, or only 1 transistor"

 

[Arkless Electronics]

It's not to do with the advantages of the 3 op-amp configuration,

It's to do with the fact that most(?) single ended input stages are inherently balanced as there's a corresponding -ve input within the circuit. So balanced can often/usually be achieved with no additional active components





You seemed to previously disagree with julf when he said "no additional components, or only 1 transistor"

I repeat that my argument has nothing to do with how few components is is possible to get away with and everything to do with doing the job properly. I would not use the circuits presented by Julf as there are much better ways of doing it.... and no I'm not going to go into detail as lifes too short...

 

[Julf]

I repeat that my argument has nothing to do with how few components is is possible to get away with and everything to do with doing the job properly. I would not use the circuits presented by Julf as there are much better ways of doing it.... and no I'm not going to go into detail as lifes too short...

"Balanced is bad because the circuitry is complicated. I know it doesn't have to be complicated, but I think it has to be complicated to be good. Thus balanced is bad because it is complicated."

 

[adamdea]

"Balanced is bad because the circuitry is complicated. I know it doesn't have to be complicated, but I think it has to be complicated to be good. Thus balanced is bad because it is complicated."

Leaving aside the combative element of the exchange, is there any obvious problem with your 1 transistor approach vs the 3 transistor eg can you achieve similar CM noise rejection?

 

[Arkless Electronics]

Leaving aside the combative element of the exchange, is there any obvious problem with your 1 transistor approach vs the 3 transistor eg can you achieve similar CM noise rejection?

Presuming you mean op amps and not transistors, the issue with the single op amp circuit is that it only performs well with an ideal low (and matched + to -) input impedance source. It also presents different impedances to the + &- inputs, hence the need for very low impedance source. When the source is less than optimum in this respect the CMRR goes out the window with the single op amp version...

Another useful feature of the 3 op amp instrumentation amp is that gain of the stage can be changed by changing just 1 resistor, or made variable with a single variable resistor, and all without affecting the CMRR. The other resistors do need to be closely matched as in the single op amp version.

There are several other possibilities with two and more op amps, such as the "Superbal" topology.

 

[Julf]

Presuming you mean op amps and not transistors, the issue with the single op amp circuit is that it only performs well with an ideal low (and matched + to -) input impedance source.

Thank you for providing some rationale for your view. Yes, the impedances of the + and - inputs to ground are different (but the source signal drives the differential input impedance, not the separate unbalanced input impedances). Yes, 3-op-amp circuits are used in instrumentation applications where extremely high (>100 dB) CMRRs are required, but considering an unbalanced input stage provides no common mode rejection at all, the CMRR provided by the standard "no extra opamp" circuit is more than enough.

There are several other possibilities with two and more op amps, such as the "Superbal" topology.

Yes, but none of them are required in a typical audio application. The argument that a differential stage adds complexity and component count in the signal path is still moot. Yes, you can keep adding stuff to make it even better, but the same applies to an unbalanced input circuit.

 

[Arkless Electronics]

Again we will have to disagree. The diff inputs do see the impedance to ground. I would NEVER under any circumstances use the single op amp version at any interface with the outside world...

I also prefer unbalanced to balanced as a generality. There is no need for balanced in domestic audio and in most cases it greatly complicates the circuitry, often needing literally double the circuitry. It's just a fad intended to sell more high end gear, like so much in hi fi....

 

[Julf]

Again we will have to disagree. The diff inputs do see the impedance to ground.

I am not sure you understand how balanced works. The source might not have any common earth/ground with the amp - what "ground" is your impedance measured against?

I would NEVER under any circumstances use the single op amp version at any interface with the outside world...

Maybe you should tell that to all the manufacturers of pro gear (who are the ones using balanced connections in the first place.

I also prefer unbalanced to balanced as a generality. There is no need for balanced in domestic audio and in most cases it greatly complicates the circuitry, often needing literally double the circuitry. It's just a fad intended to sell more high end gear, like so much in hi fi....

I agree that you can live without the benefits of a balanced connection in a domestic environment, but why settle with the danger of ground loops and no common mode noise rejection? As noted, a differential input circuit does not need double the circuitry. A fully balanced amp needs double the circuitry, but that is a completely different story.

 

[Arkless Electronics]

I find you offensive and overly combative so this really is the end of the debate. With you anyway. I'll discuss it further with others but any posts from you will be ignored.
What part of agree to disagree do you not understand?

You seem as confused over this as over balanced audio where you seem to think all balanced signals don't "see" a ground reference and not just floating transformer outputs and cartridges etc

I do not like balanced for domestic audio due to extra complication, longer signal path and the doubling of circuitry in many cases. It's an answer to a non existent problem in a domestic context where people don't usually have to quickly set up 50M runs of cable for a gig without having to worry about ground loops or if the cable is run next to mains cables etc.
What you or anyone else likes is up to you. Hence agree to disagree.

I would never use a single op amp diff input as they are shite. Yes they are the most commonly used commercially... because they are the cheapest.
If you like this compromised bit of circuitry then go ahead and use it. That's your business not mine.

 

[Julf]

What part of agree to disagree do you not understand?

I understand it all too well - the way you use it, it is a weak substitute for "I want to have the last word". If you agree to disagree, the please do so, instead of repeating the phrase and then responding anyway.

You seem as confused over this as over balanced audio where you seem to think all balanced signals don't "see" a ground reference and not just floating transformer outputs and cartridges etc

What part of "differential" is unclear? How about this? "The receiving circuit responds to the electrical difference between the two signals, rather than the difference between a single wire and ground."

I do not like balanced for domestic audio due to extra complication, longer signal path and the doubling of circuitry in many cases. It's an answer to a non existent problem in a domestic context where people don't usually have to quickly set up 50M runs of cable for a gig without having to worry about ground loops or if the cable is run next to mains cables etc.
What you or anyone else likes is up to you. Hence agree to disagree.

I would never use a single op amp diff input as they are shite. Yes they are the most commonly used commercially... because they are the cheapest.
If you like this compromised bit of circuitry then go ahead and use it. That's your business not mine.

You are of course entitled to your opinion. I am merely pointing out that your justifications don't hold water. You keep repeating the same arguments without any support for them, so let me just do the same - there is no "extra complication, longer signal path and the doubling of circuitry in many cases". If the input stage already has an opamp, no extra circuitry is needed, as opamps have differential inputs anyway. If the input stage is using discrete transistors, you need to simply mirror the input transistor. As to your "compromised circuitry" argument, it is like saying "but if you can make a car fly, you have to add a huge fuel tank, because otherwise it won't be able to fly thousands of miles", as the "compromise" is not having infinite common mode rejection ratio as opposed to not having any at all. I know I am repeating myself, but you don't seem to have properly read my previous comments.

 

[ChrisPa]

This document explains it rather well

Bruno Putzeys - The G Word
https://www.hypex.nl/img/upload/doc/an_wp/WP_The_G_word.pdf

Balanced doesn't care about ground. That's the point - there is no absolute ground, only the differential signal at the output and the input, or rather the point of reception by the next stage.

Balanced doesn't mean doubling of components, it means balanced impedences seen by the (differential) signal between the output of one stage and the input of the next. And it means understanding what your differential signal actually is at the next input stage.

Julf's earlier circuit examples are perfectly valid and are an improvement on their equivalent "single-ended" versions. There may well be (there certainly is) something "better still" but that does not negate the validity and benefits of taking a balanced approach to an existing circuit.

 

[Arkless Electronics]

I dropped a huge hint in an earlier post which seems to have gone over your head... remember when I mentioned floating balanced sources?

Most balanced sources are not floating and do see earth. This is where your beloved single op amp balanced input fails so badly. The + & - inputs have unequal input impedances to earth. These form voltage dividers with the source impedances and, as the inputs are not the same impedance to earth, the CMRR goes out the window unless the single op amp version is driven from near 0R source impedances, which cannot be guaranteed and is indeed often not the case.... and I rest my case. Strangely enough I won a bet over EXACTLY this same argument with my then boss in the R&D department I was working for at the time...

 

[ChrisPa]

I dropped a huge hint in an earlier post which seems to have gone over your head... remember when I mentioned floating balanced sources?

Most balanced sources are not floating and do see earth. This is where your beloved single op amp balanced input fails so badly. The + & - inputs have unequal input impedances to earth. These form voltage dividers with the source impedances and, as the inputs are not the same impedance to earth, the CMRR goes out the window unless the single op amp version is driven from near 0R source impedances, which cannot be guaranteed and is indeed often not the case.... and I rest my case. Strangely enough I won a bet over EXACTLY this same argument with my then boss in the R&D department I was working for at the time...

Over whose head?
It would be nice if this could be approached as an exchange of information amongst equals, rather than what appears to be an all or nothing/can't be bothered to explain

We shouldn't confuse improved CMRR with "perfect"/optimum CMRR.
Of course balanced CMMR is at its optimum when the source output impedence is properly balanced. That's an inherent part of what makes balanced, balanced - which many don't realise.

Referring to the Balance section on page 3 and the accompanying equations on page 4 of Bruno Putzeys document. You can see the effect of an imbalance in the output impedance on the CMRR. And obviously, The higher the input impedance the greater the CMRR despite the imbalance in output impedance; and of course an instrumentation amplifier circuit has a nominally "infinite" input impedance and therefore provides a "perfect" CMRR when fed an imperfect output impedance.

However, balancing the input of a stage will improve the CMRR over that of an unbalanced stage. How well it improves it is dependent on the ratio of (unbalanced) output impedance to (balanced) input impedance. So the lower the output impedance and higher the input impedance, the better it will be.

If we knew the output impedance of the preceding stage/device we could add a balancing output impedance into the output ground connection of that device (going into the interconnect) and bring it even further into balance.

In Putzeys case he then goes on to apply balanced connections internally throughout an amplifier, but the principles and maths apply externally.

 

[Julf]

We shouldn't confuse improved CMRR with "perfect"/optimum CMRR.

Indeed. I really don't get the "if it isn't perfect it isn't any better" argument. A pretty good CMRR, even if not infinite, is still much better than no common mode rejection at all (and that is what unbalanced suffers from).

It seems Arkless is saying "there is no point in flying cars because they won't be able to do supersonic speeds".

 

[Arkless Electronics]

Over whose head?
It would be nice if this could be approached as an exchange of information amongst equals, rather than what appears to be an all or nothing/can't be bothered to explain

We shouldn't confuse improved CMRR with "perfect"/optimum CMRR.
Of course balanced CMMR is at its optimum when the source output impedence is properly balanced. That's an inherent part of what makes balanced, balanced - which many don't realise.

Referring to the Balance section on page 3 and the accompanying equations on page 4 of Bruno Putzeys document. You can see the effect of an imbalance in the output impedance on the CMRR. And obviously, The higher the input impedance the greater the CMRR despite the imbalance in output impedance; and of course an instrumentation amplifier circuit has a nominally "infinite" input impedance and therefore provides a "perfect" CMRR when fed an imperfect output impedance.

However, balancing the input of a stage will improve the CMRR over that of an unbalanced stage. How well it improves it is dependent on the ratio of (unbalanced) output impedance to (balanced) input impedance. So the lower the output impedance and higher the input impedance, the better it will be.

If we knew the output impedance of the preceding stage/device we could add a balancing output impedance into the output ground connection of that device (going into the interconnect) and bring it even further into balance.

In Putzeys case he then goes on to apply balanced connections internally throughout an amplifier, but the principles and maths apply externally.

Indeed. I really don't get the "if it isn't perfect it isn't any better" argument. A pretty good CMRR, even if not infinite, is still much better than no common mode rejection at all (and that is what unbalanced suffers from).

It seems Arkless is saying "there is no point in flying cars because they won't be able to do supersonic speeds".

Like I said, point proved. I rest my case.

Chris. The first comment was aimed at Julf.... and yes of course you can deliberately "unbalance" the balanced input to give improved CMRR with a specific known input impedance. Not exactly ideal for a universal balanced input! The single op amp version is best kept for internal use in a piece of equipment where it can be driven from known low source impedances from a previous op amp stage.

 

[Julf]

Like I said, point proved.

Point indeed proven.

I rest my case.

I wish you had, But of course you didn't:

The first comment was aimed at Julf.... and yes of course you can deliberately "unbalance" the balanced input to give improved CMRR with a specific known input impedance. Not exactly ideal for a universal balanced input! The single op amp version is best kept for internal use in a piece of equipment where it can be driven from known low source impedances from a previous op amp stage.

You still keep offering totally unsupported subjective opinions, and don't address our main argument. To repeat: I really don't get the "if it isn't perfect it isn't any better" argument. A pretty good CMRR, even if not infinite, is still much better than no common mode rejection at all (and that is what unbalanced suffers from).

 

[Arkless Electronics]

The main argument was whether or not a single op amp balanced input was a good solution. I have proven that it is not.

I prefer unbalanced to balanced in general for the reasons we started with... ie unless the circuitry itself is completely balanced it means adding op amps to the signal path. Most so called balanced units are internally unbalanced. They then add op amps to the output to make it balanced just before it goes to the XLR socket. Usually the signal then goes through more op amps at the other end to convert it back to unbalanced where it then continues through unbalanced circuitry.

Outside of a pro audio scenario, where cables of 50M long etc must be connected up very quickly, and without a thought for where they are routed and what they are connected to, balanced is a solution looking for a non existent problem. It's just one of the many fads in hi fi designed to sell more units...

 

[Julf]

The main argument was whether or not a single op amp balanced input was a good solution. I have proven that it is not.

No, you haven't. You have pointed out, and I agree, that it is not perfect. It is still much better than unbalanced. This is the fact you keep trying to avoid.

I prefer unbalanced to balanced in general for the reasons we started with... ie unless the circuitry itself is completely balanced it means adding op amps to the signal path.

And as I have " proven" (to use your terminology), nothing needs to be added to a circuit that uses an opamp, and only the input transistor needs to be added to a discrete circuit. I even posted the circuit diagrams. Do I have to do it again?

Most so called balanced units are internally unbalanced. They then add op amps to the output to make it balanced just before it goes to the XLR socket. Usually the signal then goes through more op amps at the other end to convert it back to unbalanced where it then continues through unbalanced circuitry.

You are mixing up two separate things - fully balanced circuitry (that you seem to be talking about) and balanced/differential connections (that the OP is about). The latter only needs a differential input, not a fully balanced circuit all the way through the amp.

Outside of a pro audio scenario, where cables of 50M long etc must be connected up very quickly, and without a thought for where they are routed and what they are connected to, balanced is a solution looking for a non existent problem. It's just one of the many fads in hi fi designed to sell more units...

Glad to hear you don't belong to the "everything matters" school. In this case, as I have noted multiple times, a differential/balanced connection provides common mode noise cancellation and (often more importantly) avoidance of ground loops. I agree that in most domestic settings you get by just fine even with unbalanced.

 

[Arkless Electronics]

"No, you haven't. You have pointed out, and I agree, that it is not perfect. It is still much better than unbalanced. This is the fact you keep trying to avoid."

No I'm not avoiding it al all I'm saying the opposite is true. Unbalanced is better than balanced.


"And as I have " proven" (to use your terminology), nothing needs to be added to a circuit that uses an opamp, and only the input transistor needs to be added to a discrete circuit. I even posted the circuit diagrams. Do I have to do it again?"

You can post your circuits as many times as you like and will remain wrong as, apart from anything else, we are arguing different things. By your "logic" a unicycle is better transport than a car as it requires only one wheel...


"You are mixing up two separate things - fully balanced circuitry (that you seem to be talking about) and balanced/differential connections (that the OP is about). The latter only needs a differential input, not a fully balanced circuit all the way through the amp."

Differential and balanced are the same thing. It's just semantics. If I was going to use balanced I would use completely balanced discrete circuitry from input to output and not bastardise it by repeatedly converting the signal from balanced to unbalanced and vice versa with op amps!

"Glad to hear you don't belong to the "everything matters" school. In this case, as I have noted multiple times, a differential/balanced connection provides common mode noise cancellation and (often more importantly) avoidance of ground loops. I agree that in most domestic settings you get by just fine even with unbalanced."

So now you agree that in a domestic setting unbalanced is fine. Good.