Balanced output w/ MC33079

[Script]

Is there any reason why an MC33079 would be a bad idea in a balanced output stage? Want to use them cos they have been lieing around.

Something like the attached schematic but with DC isolation caps on output and input.

Also, R3 and R6, why no pf caps for filtering ? Would MC33079 need some ?

 

[john12ax7]

This circuit has the tascam problem, can't drive unbalanced loads.

Never used MC33079, but looking at its specs it seems it can drive 600 ohm,  so that's an improvement over TL07x. Slew rate is a little low at 7V/uS.

 

[abbey road d enfer]

Is there any reason why an MC33079 would be a bad idea in a balanced output stage? Want to use them cos they have been lieing around.

They are a good GP opamp, and actually better than a TL072 in this application, better drive as John mentioned, although not as good as a 5532. Slightly better than a 5532 in terms of noise.

Also, R3 and R6, why no pf caps for filtering ? Would MC33079 need some ?

They are not absolutely necessary, in most cases, but, since they are due to be loaded with unknown lines, it is a good precaution, like also load isolation resistors.
And anyway, I would not implement this because of the Tascam problem.
To resume, it's not using the MC33079 that is a "bad idea", it's using this topology.

 

[Script]

I see. Thanks for pointing that out.

Would cutting the line from IN to R5 and instead connecting R5 to the output of U1A solve the Tascam problem? I.e. turn into floating ?
Could / should then probably drop R1 R4 and R3.

 

[JohnRoberts]

I see. Thanks for pointing that out.

Would cutting the line from IN to R5 and instead connecting R5 to the output of U1A solve the Tascam problem? I.e. turn into floating ?

no...  I was not aware this was called "the Tascam problem". The topology is a simple differential output (two opposite polarity driven outputs). The problem arises from interfacing with single ended gear that grounds the minus input.  The problem arises from driving full signal into that ground through the 100 ohm build out resistor. These tens of mA of signal current will find their way back to the sending unit's PS corrupting ground quality (crosstalk), and most op amps are incapable of driving 100 ohms (TL07x are rated for 2k loads). While distortion there should not influence the audio path, overloading 1/2 of a dual opamp could have some impact. .

Could / should then probably drop R1 R4 and R3.

This provides basic differential amplifier, that can be used to reference between local and distant grounds. R4 is grounded to sending ground, R2 to the receiving ground.

===

A simpler solution is "impedance balanced" where the - output would be a 100 ohm R from the sending units 0V.

JR

 

[Bo Deadly]

Would cutting the line from IN to R5 and instead connecting R5 to the output of U1A solve the Tascam problem? I.e. turn into floating ?
Could / should then probably drop R1 R4 and R3.

No. The "tascam problem" refers to when the - output is connected to ground, because you're driving an unbalanced input, the second op amp is going to be trying to drive signal into the ground (or through 100 ohms in this case). This will at least burn a lot of power and, more important, it may stress the op amp to the point where it produces distortion in the output of the other amp driving the + output.

These days the impedance balanced output is the prevailing solution. It's pretty much ideal for conventional connections. If you're driving really long lines and you want that extra bit if power, level and noise immunity by driving both outputs anti-phase, then use a THAT1646 IC.

 

[Script]

Thanks for explaining the Tascam problem in detail.

THAT1646 (+6dB) is actually my go-to solution. Just wanted to try something different here.

So I had a look at the ART Pro VLA output. Doesn't that one then too suffer from this same problem ? (Not shown on the schematic are the two 300R on the + and - lines to the XLR outs.)

 

[Bo Deadly]

So I had a look at the ART Pro VLA output. Doesn't that one then too suffer from this same problem ? (Not shown on the schematic are the two 300R on the + and - lines to the XLR outs.)

Technically yes but if there are 300R on each pin, that's enough to prevent the 5532 from freaking out. At some point the combination of op amp and buildout resistor effectively stops the "tascam problem" from really being an issue. An NE5532 with 300 ohms is almost definitely not a problem. There might be a measurable increase in distortion but nothing audible. Although 300 is a little high for a build out resistor IMO.

Another popular way to dodge the "tascam problem" is to have an XLR paralleled with a TRS but with the ring tied to ground through a resistor to make it impedance balanced. Just Googled "ART Pro VLA" and it looks like it does that. So I'd guess the 1/4" out is TRS impedance balanced.

 

[abbey road d enfer]

The "Tascam problem" is not only the issue with distortion when shorting one of the legs, but also the 6dB loss when using just one leg.
More importantly it does not provide ground loop noise reduction when connected to an unbalanced input.
A true floating output actually does reduce ground loop issues when connected to an unbalanced input, by sensing  the remote "ground" via its - leg.
An impedance-balanced output does not.
In the 90's Soundcraft developed very clever unbalanced remote-ground-sensing outputs.

 

[Bo Deadly]

A true floating output actually does reduce ground loop issues when connected to an unbalanced input, by sensing  the remote "ground" via its - leg.
An impedance-balanced output does not.

Yeah but impedance balanced output does have pretty good ground loop noise rejection.

 

[abbey road d enfer]

Yeah but impedance balanced output does have pretty good ground loop noise rejection.

It doesn't when connected to an unbalanced input.

 

[Script]

ART PRO VLA.

Another popular way to dodge the "tascam problem" is to have an XLR paralleled with a TRS but with the ring tied to ground through a resistor to make it impedance balanced.

Indeed. A TRS with 300R to GND.

 

[Script]

Another 'impedance balanced'-only output. Values picked at random.

 

[abbey road d enfer]

Much more clever, but difficult to implement on existing circuits:

 

[Script]

Link to an old thread discussing impedance balanced output (Jensen) and Tascam problem

https://groupdiy.com/index.php?topic=5264.0

(Now I know where I had copied the above schematic from.)

 

[Script]

Much more clever, but difficult to implement on existing circuits:

Apart from values being difficult to read, it's difficult because it needs a polarity-inverted input (as the first opamp is an inverter that flips to correct polarity) ?

R41/42 does the remote ground sensing ?

 

[abbey road d enfer]

Apart from values being difficult to read, it's difficult because it needs a polarity-inverted input (as the first opamp is an inverter that flips to correct polarity) ?

It can be non-inverting since it's a fully differential input between R37 and R40.

R41/42 does the remote ground sensing ?

R41 does. R44 balances the impedance of the lower leg.

 

[Bo Deadly]

It doesn't when connected to an unbalanced input.

True. I get the point about ground sensing so yes.

But having a resistor between unbalanced connectors can provide some noise reduction. It's not unheard of to get some ground bleed from something like a computer audio output or maybe something powered by a different wall socket and there's a slight difference in ground potentials. A 100 ohm resistor between the grounds is going to greatly limit the currents.

One time I plugged some audio interface into a tower computer and the ground was literally like 500 mV off from the wall powered device I was plugging it into and I got crazy results. I measured several mA of current between the grounds.

As long as we're dissecting the details of impedance balanced outputs, there is another scenario worth considering which is that it should not be used in portable devices that need the ground connection to establish the ground shield like in a guitar pedal that needs that ground to ground the guitar.

 

[abbey road d enfer]

Ground-sensing and true balanced have that in common that they need balanced cables.

 

[Script]

Much more clever, but difficult to implement on existing circuits:

Where does the problem hide that makes it difficult to implement (on existing circuits)?