Balanced Line Driver and Receiver Schematics

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

cpsmusic

Well-known member
Joined
Dec 3, 2013
Messages
292
Location
Melbourne, Australia
Hi All,

I'm planning on building an experimental distortion device for my home studio. The device needs to handle balanced input and output so I was thinking of using these:

Balanced Line Driver & Receiver

Before I launch into this I wanted to check that there aren't better designs out there.

Thoughts?

Cheers,

Chris
 
Hi All,

I'm planning on building an experimental distortion device for my home studio. The device needs to handle balanced input and output so I was thinking of using these:

Balanced Line Driver & Receiver

Before I launch into this I wanted to check that there aren't better designs out there.
There are. The line driver at Fig. 1 is plagued with what's usually referred to as the "Tascam problem".
If one of the legs is shorted to ground it results in loss of level and significant distortion. In a studio environment, one is never sure if an input is actually balanced.
You have two better choices:
Using a dedicated chip such as THAT1646 or SSM2142 or TI DRV134,
or
Using an impedance-balanced output stage. The cost of implementing that is minimal, just a couple of resistors and a capacitor, and it works remarkably well. There's a recent thread here
https://groupdiy.com/threads/increase-in-use-of-impedance-balanced-outputs.76180/
 
Thanks for the info about the Driver. What about the Receiver - are there better designs?
The receiver is ok; its performance depends very much on the resistors tolerance; should be 1% or less.
There are chips by THAT, SSM and TI that provide superb performance, but the measurable improvement in performance may not be really significant in a moderately clean studio environment.
 
Hi,
Please look in the attached document.
Cheers,
P.
Caveat: THAT1646 needs to be fed from a very low source impedance, for optimal CMRR. For example, many have been disappointed, while they fed the input from a potentiometer (which creates a variable source impedance). One must make sure the circuit is fed from an output that has no additional resistance in series.
 
Before I launch into this I wanted to check that there aren't better designs out there.

Before I launch into this I wanted to check that there aren't better designs out there.

What Abbey said. Plus if you will also want to feed the output to an unbalanced input then go with one of the IC drivers or look at a 'Ground Cancelling' configuration.
 
The THAT 1256 is a really good sounding amp, and being an op-amp designer, that says a lot. The +4 to -2 configuration can also be flipped around so it acts as a +6dB booster.

The 1646 is a good amp, but like everyone says, you have to drive it from an opamp output. I use a dual 1256, the 1286 or the 1283 as a balanced output stage, as it sounds better, and the console owners all know that you can't short one side. The trade off is worth the tone.
 
Hi All,

I'm planning on building an experimental distortion device for my home studio. The device needs to handle balanced input and output so I was thinking of using these:

Balanced Line Driver & Receiver

Before I launch into this I wanted to check that there aren't better designs out there.

Thoughts?

Cheers,

Chris
Back in the day the SSM2141/SSM2142 were the go to device for this. Even if you were doing a TL071/TL072 type circuit, if I recall correctly, the NE5532(AN?) was a cleaner quieter solution.
 
I use a dual 1256, the 1286 or the 1283 as a balanced output stage,
What are the aspects of performance that make them a better choice than let's say a 4562 in this application?
and the console owners all know that you can't short one side.
Not every engineer that works in a studio is the owner. Shorting one leg of a balanced connection is just a high-probability fact of life.
 
Last edited:
If you don't want to go SMD op amps, I'd stick to los cheapos NE5532 with meticulously matched resistors.

Better and easier are THAT 124X and THAT1646 respectively their TI INA and DRV equivalents.

Those TI and THAT dedicated line receivers and drivers cost a bit more but need very few peripheral components and are easy enough to socket and handwire on veroboard.
 
The most glaring error in the Rod Elliott's input circuit is that it fails to meet international standards (IEC) for input impedance - it's less than a third of the minimum required of 10 kΩ. All of the input circuits, except those using the THAT's InGenius ICs will have poorer noise rejection when driven by real-world signal sources instead of test generators.

For output circuits, you can't beat the THAT OutSmarts ICs. In the Elliot circuits, he warns to use 1% resistors - but in the output stage Fig 1, the critical resistors to CMRR are R4 and R5 (not R2 and R3 - signal symmetry has absolutely nothing to do with CMRR). You might consider the cheap, simple, and foolproof "impedance balanced output" - even though there's signal only at pin 2 (none at pin 3), the circuit has as much CMRR as any other ... all that's reduced is headroom - and no precautions are necessary driving either balanced or unbalanced inputs. Some consider it "illegitimate" but that just demonstrates how little they know about balanced interconnection systems!

Yes, I'm the inventor of THAT's InGenius circuit (which is now public domain BTW). For what it's worth, I'm the Audio Engineering Society's go-to expert on these matters and I convinced the IEC to change the way they test balanced input stages (because they, like most audio engineers believed that "equal and opposite signal voltages" are necessary to cancel noise ... they are NOT - and I can prove it).
 
All of the input circuits, except those using the THAT's InGenius ICs will have poorer noise rejection when driven by real-world signal sources instead of test generators.

It would be great if you could explain in a little more detail why this is so.

For output circuits, you can't beat the THAT OutSmarts ICs. In the Elliot circuits, he warns to use 1% resistors - but in the output stage Fig 1, the critical resistors to CMRR are R4 and R5 (not R2 and R3 - signal symmetry has absolutely nothing to do with CMRR). You might consider the cheap, simple, and foolproof "impedance balanced output" - even though there's signal only at pin 2 (none at pin 3), the circuit has as much CMRR as any other ... all that's reduced is headroom - and no precautions are necessary driving either balanced or unbalanced inputs. Some consider it "illegitimate" but that just demonstrates how little they know about balanced interconnection systems!
I would just like to mention here that Rod published a new article two years later (2002) where he dealt with this topic in much more detail and better.
https://sound-au.com/project87.htmHe concluded with the following:
"With all balanced interfaces, the impedance balance is the thing that counts. There is no (and never was) any requirement for the signal to be balanced. It doesn't matter if the signal levels are the same or radically different (including having one line with no signal at all). In contrast, an impedance mismatch of only a few ohms is enough to reduce the common mode rejection quite dramatically. "

This article was published almost 20 years ago.

Yes, I'm the inventor of THAT's InGenius circuit (which is now public domain BTW). For what it's worth, I'm the Audio Engineering Society's go-to expert on these matters and I convinced the IEC to change the way they test balanced input stages (because they, like most audio engineers believed that "equal and opposite signal voltages" are necessary to cancel noise ... they are NOT - and I can prove it).

At the end of the day, the most important is achieved the S/N ratio and not the signal or noise alone in its absolute values, IMO.
 
Regarding why extremely high common-mode input impedances (like those of an input transformer, 50 MΩ at 50 or 60 Hz) give much better noise rejection in the real world, see the attached AES paper - paying particular attention to Figure 3. Making those impedances 1,000 times those of ordinary diff-amps (which includes THAT's 1240) is a tremendous advantage. Note that the "imbalance" scale is log, so the left portion of the graph is in the realm of connector contact resistance. But, in real-world audio sources, output impedance imbalances in the 5 to 10 ohm range are very common. Diff-amps will produce their advertised CMRR only when driven by laboratory signal sources! Transformers and InGenius will produce very high CMRR with real-world sources that are less than perfect.

I had these same arguments with the IEC back in 1999, when I proposed a new test for CMRR. I finally convinced them, but suggested they put an "informative annex" in the standard to explain why the old test didn't correlate at all with real-world use of balanced interfaces. I've attached a copy of that as well.

Rod Elliott and I had lots of "educational" exchanges before he did his later stuff.

I hope everyone will understand that I've had this argument about what's responsible for noise rejection in a balanced interface (it's the impedances, not equal and opposite signals!) hundreds of time since writing my original 1994 AES Paper about it. Unfortunately, old ideas die slowly and this misinformation is still in lots of textbooks and websites. It's a bit of a sore spot and sometimes makes me want to just scream and rant!
 

Attachments

  • InGenius IC AES6261.pdf
    381.1 KB · Views: 40
  • IEC60268-3AnnexA.jpg
    IEC60268-3AnnexA.jpg
    86.9 KB · Views: 66
(y) I don't know whether it's a sound industry thing but many people who aren't confident enough in their own understanding to argue with you about technology, will still privately question and dismiss you (I mean me). I encountered that with my old 1980 console performance article, people didn't believe all my claims, but waited for somebody else to say I was wrong. The silence was deafening.:cool:
====
Ironically just a few months ago a friend on another forum tried to argue with me that "impedance balance" was somehow inferior. The Icing on the cake was that he cited Bill as his source of expert information for that claim. :rolleyes:

It was especially fun to provide him with a link of Bill posting on this forum that he supported impedance balanced interfaces when appropriate and properly executed.

Is there an emoticon for mic drop?

JR
 
Regarding why extremely high common-mode input impedances (like those of an input transformer, 50 MΩ at 50 or 60 Hz) give much better noise rejection in the real world, see the attached AES paper - paying particular attention to Figure 3. Making those impedances 1,000 times those of ordinary diff-amps (which includes THAT's 1240) is a tremendous advantage. Note that the "imbalance" scale is log, so the left portion of the graph is in the realm of connector contact resistance. But, in real-world audio sources, output impedance imbalances in the 5 to 10 ohm range are very common. Diff-amps will produce their advertised CMRR only when driven by laboratory signal sources! Transformers and InGenius will produce very high CMRR with real-world sources that are less than perfect.

Thanks a lot for this lengthy post. I am familiar with your work and published articles on AES.org where I am a member myself. I was just confused with what you meant by "laboratory signal sources".
I’ve always wondered why this InGenius topology wasn’t used in the mic preamp ICs that THAT produces, and where a large CMRR would be far more necessary because the cables can be long and the signal very small.
 
Back
Top