Why no fully differential circuits?

Status
Not open for further replies.
Jump to Latest
#23
No fully differential circuits

Good question

Phono cartridges and tape heads are (by there nature) "fully differential" - yet very few manufacturers ever tried to make a balanced input preamp - remember that this would involve rewiring the tone arm.

Speaker crossover networks tend to be "unbalanced" but could also be designed "balanced".

Charles

Charles
 
#24

Attachments

  • ulti.preamp.org_schematic_02.png
    ulti.preamp.org_schematic_02.png
    56.2 KB · Views: 394
  • ulti.preamp.org_built_01.jpg
    ulti.preamp.org_built_01.jpg
    833 KB · Views: 385
#28
dmills said:
Now try plotting the output stage bias current against common mode voltage, my instinct is that smoke is possible from R1,2.

Regards, Dan.
Click to expand...

Hi Dan, as the current flowing out of R1, R2 is proportional to the voltages on the nodes 3 and 6, I enclose transient analysis relating to the two nodes vs ground. The voltage on the two nodes is in phase with an amplitude of 5 microvolts peak.

On the other hand, this circuit I've made almost 15 years ago, and I use it still. In particular, it was never necessary to change those two resistors.

Someone want that I place photos of the two mono amps?

Best regards.
 

Attachments

  • comm_mode_sign_circlo_k1058_node number.png
    comm_mode_sign_circlo_k1058_node number.png
    22.3 KB · Views: 333
  • comm_mode_sign_circlo_k1058_transient_R1_R2.png
    comm_mode_sign_circlo_k1058_transient_R1_R2.png
    21 KB · Views: 313
#32
dmills said:
...
Ensuring near zero common mode gain is one of the many reasons to unbalance early in the design.

Regards, Dan.
Click to expand...
Yep, that is another problem with "balanced all the way through." The gain of the positive and negative paths must be matched to unusually high precision (using components matched to, say, 0.1 percent or better in EVERY position that affects gain - this costs real money for high precision parts or real time in matching lots of extra parts) to maintain good CMRR.

With the usual instrumentation-amp balanced input, you only need a very few well-matched resistors to get high CMRR. The ground currents on a well-laid-out PCB won't have a measurable effect on the "unbalanced" signal between the input and the output.

Unbalancing is the first thing done with (I presume "early in the design" actually means "early in the signal path") an input to a box, and for a balanced output, the "balanced output generator circuit" is the last thing in the signal path before a signal goes back to the outside world.
 
#35
vzaichenko said:
Hi Preamp, as a matter of interest - what is the idea behind using such an "array" of the small-signal transistors at the output follower?
Click to expand...

That is my approach to make some use of the remaining parts after matching the input pairs 😎.


mos57 said:
You are wrong, your circuit is complementary, not symmetrical, sorry. See post 30.
Also if there are 2 complementary differential stage the two main signals from input to output are in phase, so....... single-ended! 😀
Click to expand...

Okay, okay... 😱 😀
 
#38
benb said:
Unbalancing is the first thing done with (I presume "early in the design" actually means "early in the signal path") an input to a box, and for a balanced output, the "balanced output generator circuit" is the last thing in the signal path before a signal goes back to the outside world.
Click to expand...

By early I did indeed refer to the signal path, as in right after the RFI filters and common mode choke, but I hate the whole 'signal path' notion as it tends to ignore at least half and often more like two thirds of the bits that matter, not just the current in the 'ground'but also how this gets back to the appropriate power rail, these details are not IMHO meaningfully separable from the notion of a signal path (More like a series of "signal loops").

Mixing desks of the old school analogue sort sometimes did a rather odd differential scheme with the summing amplifiers but that was often the one place you saw differential (But usually not balanced!) signalling inside a box.

Regards, Dan.
 
#39
One of National's customeres was going to use the LM4702's in a new fully balanced high end power amp design. (The LM4702 was the first part in what became the LME series of analog audio parts built on a high voltage/low noise/low THD semi process.)

I heard the proto amplifier at Classe' Audio many years ago and it was incredible. It never made it to market though...as far as I know. So for a fully balanced power the LM4702 dual part is a good choice. (See AN-1490 on the TI website for notes on how to use that part...to bad TI removed the names of the National Semi Authors from all of the ap notes. I know the guy who wrote that ap-note ;-) )

audioman54 / Mark
 
#40
Completely differential circuits are quite common in analogue IC design. Advantages: cancellation of even-order distortion, low sensitivity to anything common-mode, like supply ripple, substrate interference, ground voltage drops et cetera, well-defined signal return paths.

I usually don't make domestic hobby audio circuits balanced because it tends to make them more complicated and they usually have to interface with unbalanced consumer equipment anyway. You can find an exception here:

http://home.kpn.nl/verwa255/esl/amp.zip

Another exception is this balanced to unbalanced and vice-versa converter that I made for a semi-professional local radio studio:
 

Attachments

Last edited:
Status
Not open for further replies.
Top Bottom