[craig sawyers]

A balanced Tortuga will be connected to the balanced input stage of typically a power amp.  Just about all of these use a simple balanced input looking like this



This actually has different loading for V+ and V-. Conventional wisdom is that the V+ sees 2R, but the V- sees only R (which is bad enough!), so as far as loading is concerned the simple balanced input is not balanced at all. But it is even worse than that.

The V+ impedance is 2R that is expected, but the V- is 2/3R, not R. This happens because of a sort of negative bootstrap. The + input of the opamp receives half the + input voltage. The difference between the + and - opamp inputs has to be zero, so the - opamp input has to be also 1/2V+, and hence the voltage across the V- input resistor is 1.5 what one would expect, reducing the effective input resistance to R/1.5 = 2/3R.

The typical balanced input will use 10k resistors, with the + input seeing 20k.  But the - input only sees 6.67k.  This difference in loading has a significant impact on any balanced passive pre, and the Tortuga is no different in that regard.  It will manifest itself as a balance shift as volume is adjusted, because the output resistance changes with volume setting.

Also just about all the "conventional" balanced input stages generate considerable noise as compared with a single ended input - as a result of the relatively high values of resistor used.  This is why just about all high end manufacturers quote power amplifier noise with the single ended input only and remain pretty much silent about the balanced input noise, which is usually worse by a good 10-15dB.

There are various balanced input stages which mitigate the impedance imbalance (but with noise problems too), but since you don't know what design philosophy has been used, and with amp manufacturers being pretty much silent on the topic,  you cannot assume anything.

I've just upgraded the balanced input stage of my active crossover (which was the vanilla type above) to a design which simultaneously gets around the loading problem (presents 47k to + and -, or any other value you like), and produces a major reduction in noise (15dB or so), to the extent you have to put your ear a few mm away from the loudspeaker cones to hear anything at all.  The effect on the sound quality of the Tortuga is not insignificant!

Craig

[tortugaranger]

  The effect on the sound quality of the Tortuga is not insignificant!

Very interesting info. Arguably the best way to handle incoming differential signals is to use transformers to convert the balanced signal to single ended. No impedance difference there. Jensen makes some very high quality transformers designed for this. Very pricey compared to op amps which is probably why you don't see them used commonly.

Regarding the sound quality comment, did the sound quality improve with improved impedance balance?

[craig sawyers]

Hi Morten

Absolutely.  I was always a bit puzzled why the image sounded to a degree off centre - and this was remarked on by others.  The main problem is that with a vanilla balanced input, the loading for V+ and V- is different by a factor of 3.  Which means that the loading of the lower arm LDR is very different for V+ and V-, and as the resistance of the lower arm LDR changes with volume setting, there is a volume dependent shift in balance.

This is not something that effects unbalanced systems - it is unique to balanced ones.

If it is active balanced with a very low output resistance this is not a problem.  But the unique characteristic of the LDR is musicality, with no active circuits - it simply needs to be loaded by the same impedance on V+ and V-.  And preferably around 10x the 50% volume setting if the volume change law is to be close to design.  So a 20k impedance setting has a mid range lower arm resistance of 10k, so the loading should be 100k.

The trick is to load V+ and V- with unity gain opamp buffers.  You can set the resistance to ground for each of these pretty much without restriction - >100k is not a problem with typical audio opamps like the LM4562 or OPA2134.  The outputs of these feed into a vanilla differential amp.  That the load impedances are different by a factor of three is not important since they are driven by the very low output impedance of the buffers.  In addition you can choose the resistances for the vanilla diff amp to be low - 1k typical - so the Johnson noise is very low.

[craig sawyers]

The technical explanation for perceived shift in balance is of course nonsense! 

Because the different attenuation on V+ and V- inputs is the same for both left and right channels.  But for whatever reason, the balanced LDR pre sounds significantly better when both plus and minus balanced outputs are loaded by the same resistances.

[Randy]

Huh? Can you translate into simple English for us non-techies?

[craig sawyers]

OK - I'll try.

The Tortuga LDR control works like a regular volume control, or any passive pre that uses switched resistors.  So you have something that looks like this



R1 and R2 are the LDR's, and are both adjusted to change the volume.  The sum of the two, R1 + R2 is equal to the resistance setting (20k by default).  The load presented by the power amp (for example) is shown by RL.  That is in parallel with R2.  If RL not sufficiently high, the volume control "law" that is in the manual (and spreadsheet) will be different.  If the amp has input resistance RL of around 100k the loading of the LDR at 20k setting the loading will be minimal.

The real problem that *could* arise is with a balanced passive pre where there are two of the above, one for positive and one for negative signals (V+ and V-), which are the same amplitude but one is inverted with respect to the other.  If the simple balanced input is used, as in the picture in the first post, the load resistance is very different for V+ and V- (by a factor of 3), which adds another layer of modification to the volume law.

If the power amp designer has done their job well, the simple balanced input will not have been used, and an alternative and more suitable circuit will have been used.  That is simple to achieve by using alternative input circuits such as an instrumentation amplifier http://www.allaboutcircuits.com/textbook/semiconductors/chpt-8/the-instrumentation-amplifier/ the buffered differential amplifier http://www.allaboutcircuits.com/textbook/semiconductors/chpt-8/building-a-differential-amplifier/, or choosing a specific chip that does the job like http://www.thatcorp.com/datashts/THAT_1200-Series_Datasheet.pdf .  They might even have used a design based on discrete semiconductors, or even tubes.  The problem is that you don't know what has been used, and it is impossible to tell from the amplifier specifications.

[poseidonsvoice]

If the power amp designer has done their job well, the simple balanced input will not have been used, and an alternative and more suitable circuit will have been used.  That is simple to achieve by using alternative input circuits such as an instrumentation amplifier http://www.allaboutcircuits.com/textbook/semiconductors/chpt-8/the-instrumentation-amplifier/ the buffered differential amplifier http://www.allaboutcircuits.com/textbook/semiconductors/chpt-8/building-a-differential-amplifier/, or choosing a specific chip that does the job like http://www.thatcorp.com/datashts/THAT_1200-Series_Datasheet.pdf .  They might even have used a design based on discrete semiconductors, or even tubes.  The problem is that you don't know what has been used, and it is impossible to tell from the amplifier specifications.

Exactly. Well said and explained. Sent you a pm.

Anand.

[tortugaranger]

Craig has done a great job highlighting the potential issue of some balanced amps presenting significantly uneven impedances (and thus uneven loads) to the 2 balanced audio signals ( each signal being 180 degrees out of have to the other) and that such uneven impedances can negatively impact audio quality when using a passive preamp.

It occurs to me that the adjustable impedance feature of the Tortuga preamp could be adapted such that each phase of the balanced signal could have different impedance levels that could effectively negate a balanced  amp's load differences.

To make this practical there would have to be a simple way to asses whether a given balanced amp has uneven phase impedances and what the magnitude of that difference is. Armed with that info, the preamp owner could enter that difference and a custom attenuation schedule could be generated by rerunning autocalibration. 

[craig sawyers]

Maybe a process that (a) set temporarily to 100k (to maximise any loading effects) and volume at 50% (b) inject a tone from the micro at a low level, say -40dB (c) measure resulting V+ and V- (d) work back to RL+ and RL-

Or use a higher level, but the user must unplug speakers first!

Would that be possible as a variant of autocal?  In any event even if RL+ and RL- are equal, or even with unbalanced, this (or a similar process) would enable autocal to produce the correct attenuation law even when loaded with a low-ish RL, such as the THAT bal to unbal 1200 series chips which are 24k, or single ended inputs which are frequently 33k or 47k.

[justubes]

Hi Craig,

I have also always been curious and didnt get around to ask Morten if this off balance issue was to do with autocal and balanced.I have a shifting of center image and resorted to adjustment of balance..

Thank for highlighting the techincals, but how do we do a temporary universal fix to this by adding some resistance to the v-?

Morten, is there a new board in the works for this?

Thanks

[jr4321]

I'm trying to understand this thread...  Is the summary that the balanced-style passive pre could have a balance issue (between left-right channels) depending on the amp it's used with?  And this can't be predicted as there are no specs showing the layout of amps?

[craig sawyers]

It is not an issue with L-R balance, but in significantly different effective gains between the V+ and V- inputs of each channel.  To put that into context, if the gain of V- (or V+)was zero, the balanced passive pre would  effectively be single ended.

It isn't that stark, but depending on the input impedance of the thing that the passive pre feeds, and the impedance of the passive pre, the V+ and V- gains could be different by a factor of 3.

[jr4321]

OK, thanks.  So it doesn't show up as a L-R imbalance but a change from balanced to single ended.  And this would result in a lower volume, usually, since the source should have double the V-output as balanced vs single ended.

[TheMonkey]

Hello and sorry, I know it's been a long time since the last activity in this thread.

I'm also trying to understand all of this... Are there any news on how the upcoming V3 will handle this potential issue?

My setup is all truly balanced and I've been thinking about getting a Tortuga preamp. I'm having a hard time deciding whether or not to wait for the V3 and now (since reading this thread) I'm also a bit confused whether to use balanced or not...

Thanks
/ Simon 

[tortugaranger]

Hello and sorry, I know it's been a long time since the last activity in this thread.

I'm also trying to understand all of this... Are there any news on how the upcoming V3 will handle this potential issue?

My setup is all truly balanced and I've been thinking about getting a Tortuga preamp. I'm having a hard time deciding whether or not to wait for the V3 and now (since reading this thread) I'm also a bit confused whether to use balanced or not...

Thanks
/ Simon

You have to put this into perspective. It's really not issue or even a potential issue except perhaps in some extreme case with really really bad balanced audio gear.

A balanced audio source sends supposedly identical signals down 2 wires except those 2 signals are 180 degrees out of phase with respect to the other. Our balanced attenuators attenuate (i.e. control the volume of) each signal separately and reasonably close to identically. When those balanced signals arrive at the amplifier's balanced differential input stage there's a presumption that each phase will "see" the same impedance. That may not always true as it depends greatly on the detail design of the amplifiers differential input.

To the extent an amp's differential input has significantly different impedance levels between the 2 phases it could somewhat defeat the effectiveness of balanced signal transmission (noise immunity) and the stereo imaging could be skewed slightly left or right depending on the degree of the impedance imbalance.

As a practical matter this has not proven itself to be a big issue, a small issue or an issue at all with our balanced LDR attenuator. To the extent it were to occur, a simple tweak in the left/right channel balance using the remote would correct for it.

In short, we're not going to work on providing a solution to a problem that hasn't presented itself as a problem. There's nothing different in the V3 vs. the V2 in this regard so you can check this concern off your list.   

[TheMonkey]

Thanks for your clarification! True, no need for solving nonproblems... Then I will stick to my balanced setup
Is the V3 gonna come as a preamp-kit as well on it's initial release?

[Armaegis]

Just about all of these use a simple balanced input looking like this

Is it a fair assessment to say that all amps will use an opamp input like that? I would think the use of a differential receiver IC like the INA134 would make more sense/be more common, and that's been around for years now. The THAT receiver chips have been around quite some time as well, as well as others.

[tortugaranger]

Is it a fair assessment to say that all amps will use an opamp input like that? I would think the use of a differential receiver IC like the INA134 would make more sense/be more common, and that's been around for years now. The THAT receiver chips have been around quite some time as well, as well as others.

My interpretation of that comment you referenced is that he was talking about the same thing you are. Call it an op amp or call it a differential receiver - potayto/potahto. 

[Armaegis]

Hmm, it was my understanding that the specialized differential receiver chips (as opposed to regular opamps) would provide equal impedances on each leg as they are designed for that sort of thing... but I guess it comes down to scrounging through the data sheets and fiddling with the circuits yourself.

[tortugaranger]

Hmm, it was my understanding that the specialized differential receiver chips (as opposed to regular opamps) would provide equal impedances on each leg as they are designed for that sort of thing... but I guess it comes down to scrounging through the data sheets and fiddling with the circuits yourself.

You're correct and and I don't think the OP would disagree. My take is the OP was simply suggesting that to the extent a balanced amp design used less than specialized receiver chips or otherwise did not ensure that the input impedance of each balanced leg within an amp were equal, that using a resistive balanced passive preamp under such circumstances would be less than ideal.

Again, as a practical matter, our customers with balanced amps using our balanced passive preamps are not finding this to be an issue. And I do to hear about problems if there are any.