Error Amplifier with both the measurement and the reference on the same input?

[Magnethicc]

Hi Everyone,
The following pictures shows 2 ways to produce an error signal.
The 1st circuit with the measurement and reference both on the inverting input with the non-inverting input at 0
The 2nd circuit is the more common way with the reference on the non-inverting input and the measurement on the inverting input.

I was wondering when it is best to use one or the other.
If you have any information on this I would kindly appreciate it.

 

[KlausST]

Hi,

The following pictures shows 2 ways to produce an error signal.

Where is the "error signal" you are talking about in your circuit?

***
Both circuits are of integrating character, thus they need a feedback. I miss this ... usually in the schematic
... textual descriptions often are too ambiguous for a detailed discussion.

****
Nomencalture:
You use "measurement" and "Ref" ... I don´t know what this means. The schematic also misses to show their sources.

I rather expect "setpoint" and "feedback".

****
The difference between both circuits:
Upper circuit: REF is connected via resistor to the inverting input. Thus it has NEGATIVE gain
Bottom circuit: REF is connected to the non vinverting input. Thus it has POSITIVE gain

So the circuits are not interchangable. Either use the one or the other. Which one: is determined by the application (wiring / schematic).

Klaus

 

[crutschow]

Both circuits act as voltage comparators, since there is no DC negative feedback.

In the top one, the output goes high when the input is lower than the reference voltage, with opposite polarity to the reference voltage and by a value determined by the relative values of R2 and R3.

In the bottom one, the output goes high when the input is higher than the reference voltage of the same polarity, and unaffected by the value of R5.

So which one is "best" is determined by your requirements.

 

[dick_freebird]

Do you want the sum or the difference (times gain)?

 

[danadakk]

Practical integrators -

http://www.tedpavlic.com/teaching/osu/ece327/lab2_fets/lab2_fets_integrators.pdf

Regards, Dana.

 

[D.A.(Tony)Stewart]

In both cases, Vin+ becomes the reference which defines the "virtual null input".
To have a gain other than zero, the output must in a linear range between Vcc&Vee or Vdd&Vss and the input must be at its expected input offset voltage due to input bias current Iin*Rin ( differential inputIio and external differential resistance to inputs) + Vio.

There are many Op Amps designed for working at or near the bottom supply rail and others that will never work at Vin= gnd using a single supply. So common mode Vcm are critical sheet specs in selection.

The lower circuit has the advantage on single supply designs with bipolar inputs to Vin- by raising the Vcm, often to V+/2 as a reference ground.


Other
Also, it should be apparent that using the non-inverting input does not make the circuit a non-inverting integrator due to the 1+ effect, which bypasses integration in Av+= 1+|Av—|. A non-inverting integrator requires an inverting amplifier in series with the output and a suitable reference for a single supply depending on the design requirements.

Any DC input offset Vio in either circuit must be considered in both integrator circuits. Not shown but there must be some is negative feedback on the output voltage to prevent reaching and trying to go past the output limit. (aka saturates or clips) to prevent malfunction. An analog switch is sometimes used to discharge the cap voltages.

 

[mtwieg]

I think I get the what the OP is asking: what's the difference between the "classic" way of connecting an error amplifier (with the reference at the noninverting input) vs doing a difference first and then passing that to the error amplifier (with the noninverting input at GND). Here I've redrawn the two circuits:

The two have the same exact transfer function with respect to MEASUREMENT, but are different with respect to REF. Here's example bode plots for each circuit (top is the "classic" circuit, bottom is the "difference first" circuit), with respect to each input (green is with MEASUREMENT as the AC source, blue is with REF as the AC source):

So for the second circuit, the transfer function is exactly the same with respect to both inputs. For the classic circuit, the transfer function with respect to REF is different. Not only is it different, its pole and zero frequencies are different, and it has an extra zero (preventing its gain from going below unity).

Basically, if REF is static (as is the case in most power supplies) then there's no real distinction between the two circuits. But in the case where REF is dynamic, then you may want to use the second circuit to maintain loop stability.

 

[KlausST]

I think I get the what the OP is asking:

to me it seems the OP is not interested in communication at all ...

Klaus

 

[Easy peasy]

Top one when Vout is negative. OR, ref is negative !

 

[Magnethicc]

Excuse my lack of presence, got caught up in this crazy week!
I first want to thank all of you who answered.

I think mtwieg hit the nail here, those bode plots shows the dynamics effects of the reference on the output so if there is any dynamics to Vref (especially high frequency stuff) than it's probably a good idea to go for the first circuit in the first post.

I always seem to be amazed that in electronics, some unseemingly insignificant change can make such a difference.

 

[KlausST]

I always seem to be amazed that in electronics, some unseemingly insignificant change can make such a difference.

Is it? Is it unseemingly?

Connectig VRef to an inverting vs non inverting input:
So translating this to math: (ignoring gain)

The one is: Result = -Meas - Ref
the other is: Result = -Meas + Ref

For sure it makes a huge difference if you get (+) $100 or you have to pay (-) $100

*****

And regarding gain: we know that (standard OPAMP circuits)
* for inverting OPAMP circuits: Gain = -R_fb / R_in
* but for non inverting OPAMP circuits: Gain = 1+ (R_fb / R_in)

--> thus even the gain changes .. depending where you connect Ref.

*****
And I´m not sure wheter mtwieg did this intentionally: (or maybe I misinterprete the simulation)
In his bottom circuit REF is inverted in E1 as well as by the OPAMP circuit, resulting in overall positive gain.
But in your circuit it is ONLY inverted in the OPAMP circuit, thus resulting in negative gain.

Klaus

 

[mtwieg]

Is it? Is it unseemingly?

It definitely caught me off guard before when designing a tracking power supply. Was certain the stability margins were good, but the step response had huge overshoot. Ultimately resolved the issue by adding a filter on the reference input to counteract the effect, rather than redo the board layout. Good learning experience though.

And I´m not sure wheter mtwieg did this intentionally: (or maybe I misinterprete the simulation)
In his bottom circuit REF is inverted in E1 as well as by the OPAMP circuit, resulting in overall positive gain.
But in your circuit it is ONLY inverted in the OPAMP circuit, thus resulting in negative gain.

I assume you're comparing my lower circuit to Magnethicc's upper circuit. Yeah in my case I preserve the overall polarity of REF and MEASUREMENT, while his circuits flip the polarity of REF (assuming this was unintentional, and not really relevant to the question they were trying to ask).