Superbal differential amp

[Moses]

Hey all,

Regarding the superbal diff amp configuration, whereby an opamp is used in +ve feedback, is it possible to make this feedback opamp a DC servo?

Thanks

Mo

 

[tv]

In theory, you could try to break the dc-path of the NFB of the +Ve FB opamp with a fat capacitor. This way, a "servo" functionality would be realized, but there is a chance that cmrr would get bad at lower frequencies.

From the top of my head. Spice guys - any deeper insight?

 

[Moses]

Hey TV,

Thanks! Im not sure if I made myself clear... Im not entirely sure that we're singing from the same page!! I have uploaded a picture at http://bonningtonlanestudios.com/prodigy/ to make things more clear. Hopefully!!

What I think is happening is that I have taken the Superbal circuit (essentially this, without the capacitors) and turned the fb opamp into a 2 pole filter as well as an inverting buffer, which I'm hoping (with values chosen for a 0.1Hz or so cutoff) would negate the need for coupling caps...

Am I being clever or re=hashing stuff that's been flamed a million times?

Thanks!

Mo

 

[JohnRoberts]

Yes, you can make it a DC servo, but you lose the balanced input. It becomes a simple differential input with DC servo.

JR

 

[Crusty2]

Also, "ALL R's EQUAL" would give you .5 gain output...

 

[abbey road d enfer]

Also, "ALL R's EQUAL" would give you .5 gain output...

No. Unity gain

 

[abbey road d enfer]

Yes, you can make it a DC servo, but you lose the balanced input. It becomes a simple differential input with DC servo.

JR

To be more accurate, you loose the balanced input impedance

 

[JohnRoberts]

Yes, you can make it a DC servo, but you lose the balanced input. It becomes a simple differential input with DC servo.

JR

To be more accurate, you loose the balanced input impedance

I though I was adequately accurate, but you don't "loose" anything...  ;D

Yes, the input impedance is no longer identical for both inputs, so it no longer meets the criteria for being called a "balanced input".

i guess I could have said you lose the balanced input "capability" or "functionality". I'm sorry if my omission was confusing to anybody. 

I already posted to a thread in the LAB yesterday how differential circuit input impedances differ from each other, degrading balance. 

JR

 

[Samuel Groner]

The common-mode input impedance is still equal for both inputs so this actually still serves as differential amplifier with CMRR. The differential-mode input impedance becomes again unequal for the two ports but that does not cause many troubles.

On the other hand it is advisable to decouple the servo somewhat for noise reasons so this leads to at least slightly different solutions.

Samuel

 

[JohnRoberts]

Yes, for equal value R's. I just discussed that in another thread.

Pretty questionable merit in defeating the active balance to make a servo (IMO), but sure it could be done.


JR

 

[Crusty2]

Also, "ALL R's EQUAL" would give you .5 gain output...

No. Unity gain

Hi Abby
Unless I'm not understanding something - If all R's are equal, the output would stabilize at half the input voltage, G = -2

 

[abbey road d enfer]

Also, "ALL R's EQUAL" would give you .5 gain output...

No. Unity gain

Hi Abby
Unless I'm not understanding something - If all R's are equal, the output would stabilize at half the input voltage, G = -2

When the second opamp is turned into a VLF low-pass, as it is in the schemo, the stage becomes a standard one-opamp diff amp with unity gain.
The superbal with equal R's has 0.5 or -6dB gain.

 

[tv]

One embodiment of the present invention:
(as predicted by Hopis, strange things are likely to happen below the corner frequency)

Instructions: Only use the most quality Flux Capacitor. It should be of such proportion that even the most seasoned size-queens would faint by the sheer experience of exposure to said Flux capacitor.

 

[JohnRoberts]

One embodiment of the present invention:
(as predicted by Hopis, strange things are likely to happen below the corner frequency)

Instructions: Only use the most quality Flux Capacitor. It should be of such proportion that even the most seasoned size-queens would faint by the sheer experience of exposure to said Flux capacitor.

That will work, but it doesn't buy you any improvement over just putting that same cap in series with the output or inputs. The whole concept of DC Servos is to use an opamp buffer to allow you to shift the impedance higher and use higher quality film capacitors for their superior audio behavior. Making your LF pole with 10k means you would need a much larger than practical with film capacitor.

A servo operating into 10k means you will see the same capacitor effects as a passive blocking cap working at that impedance.

JR 

 

[tv]

This could be debated. I only posted it because there was a misunderstanding (at least I think so) regarding what I described in my previous post in this thread.

To my knowledge, such config is used when there is need to "cram" the functionality of a HPF and a "DC servo" in one FB-loop. Of course, the two 10K resistors in the second opamp's leg could be scaled to f.e. 100K but I don't think that this would be wise to do at the debalancing circuit, due lagging and snr - not to mention what would happen to cmrr due to (phase) imbalances introduced with a cap reactance. (Spice guys, show up.)

It is indded possible in theory, but it would be cheaper to side-line a dedicated DC-servo, because for the config in "my" schem, you would need a pretty mondo Flux Capacitor. "Exactly As Advertised".

 

[JohnRoberts]

You can debate if you wish... 

A servo with 10k Rs is technically a servo, but not a very good or practical one...

JR

 

[boji]

One embodiment of the present invention:
(as predicted by Hopis, strange things are likely to happen below the corner frequency)

Instructions: Only use the most quality Flux Capacitor. It should be of such proportion that even the most seasoned size-queens would faint by the sheer experience of exposure to said Flux capacitor.

I think I have found what moses needs to see your idea come to light TV:

 

[Moses]

Hey guys!

Sorry I've been away... Im only back so soon because my car died a rather nasty death. Have been towed the length of the country by a breakdown service today... Not much fun!!

So... (to skip out on the Back To The Future references!!) Are you guys saying that a 2 separate stages to do what I'm suggesting, or just 2 feedback paths? One for DC servoing and one for impedance balancing?

Mo