# Phase correction on Opamp Output . It's possible to correct it?

#### Zeno53

Joined Sep 23, 2020
60
Hi guys, I would need a hand if possible, I have an opamp that at 100Mhz output gives me a phase shift ranging from +/- 3.5 to 5 ° degrees depending on the configuration. Both on the inverting and non-inverting outputs obviously. Since it is a few degrees it is possible to correct this small phase shift, for example with an input or output filter ?. Can you help me out please?

#### Attachments

• 16.4 KB Views: 21
• 11.4 KB Views: 20

#### crutschow

Joined Mar 14, 2008
27,210
Does the signal frequency change or is it constant at 100MHz?

#### Papabravo

Joined Feb 24, 2006
16,132
I think it will be a challenge to find components that you can use at that frequency for such a small correction. Since you do not have a fixed value of phase lead/lag it would have to be adjustable which adds another level of complexity. Do you have a datasheet for this part or is it "under development"?

Edit: since there is a frequency sweep and the phase change appears mostly linear as a function of log(ω), I guessing the 5° is about the max and 3.5° is like average. Of course I could be wrong which happens a lot when we have to "assume" things. This could all be avoided if the TS would include more details.

#### crutschow

Joined Mar 14, 2008
27,210
Why do you need to correct this phase shift?

#### Zeno53

Joined Sep 23, 2020
60
@crutschow
No the frequency is not fixed at 100Mhz but goes from 1 to 100Mhz. I'm not going to dwell on why but I need to have an exit with the same phase. The operational will provide a wave at 0 and 180 ° with respect to the input signal.

@Papabravo

The 5 ° phase shift occurs when the operational is single but if the circuit is multistage, the chain allows you to lower the phase shift to about 3/4 °. In fact on a chain of 4 OpAmp the output phase shift is on 17 °. I would have preferred if possible to make an intrestage correction to stabilize the phase as much as possible over the whole band.

#### Attachments

• 15.2 KB Views: 8

#### Papabravo

Joined Feb 24, 2006
16,132
@crutschow
No the frequency is not fixed at 100Mhz but goes from 1 to 100Mhz. I'm not going to dwell on why but I need to have an exit with the same phase. The operational will provide a wave at 0 and 180 ° with respect to the input signal.

@Papabravo

The 5 ° phase shift occurs when the operational is single but if the circuit is multistage, the chain allows you to lower the phase shift to about 3/4 °. In fact on a chain of 4 OpAmp the output phase shift is on 17 °. I would have preferred if possible to make an intrestage correction to stabilize the phase as much as possible over the whole band.

Take a look at the following article and tell me if you think the concept of a lead lag compensator will be workable. How you adjust over a 2 decade range is another question all together.

#### Zeno53

Joined Sep 23, 2020
60
I took a look and I think it can be done but is it possible to apply it to both outputs? In the meantime, I did research elsewhere and a user with a similar problem I think was suggested to switch the reference plane but what it is is not written. If the degrees of phase shift are too few to work on, i can increase the frequency which will certainly increase. I had given a limit to 100Mhz to contain the phase but surely the opamp can still go up.

#### Papabravo

Joined Feb 24, 2006
16,132
I took a look and I think it can be done but is it possible to apply it to both outputs? In the meantime, I did research elsewhere and a user with a similar problem I think was suggested to switch the reference plane but what it is is not written. If the degrees of phase shift are too few to work on, i can increase the frequency which will certainly increase. I had given a limit to 100Mhz to contain the phase but surely the opamp can still go up.
That's a pretty good question. It wasn't obvious from your original drawing with the square drawn around the amplifier that the outputs were differential and if you drew the square with wires that is why there are connection dots. If you want a non electrically conductive square around something you have to use the "Draw Line" or "Draw Rectangle" functions.

#### Attachments

• 56.6 KB Views: 3

#### Zeno53

Joined Sep 23, 2020
60
Sorry if I didn't say this before but I thought it was obvious it was differential. The outputs 3 and 4 are respectively 0 and 180 ° between them and both have this phase shift and more precisely they are 5.5 and 174.5 °. The rectangle is part of the opamp symbol (U1) which is originally light yellow. Moreover, in the opamp you can see that there is feedback and this makes me think that it is not further reducible but I asked anyway to be sure.

#### Papabravo

Joined Feb 24, 2006
16,132
Sorry if I didn't say this before but I thought it was obvious it was differential. The outputs 3 and 4 are respectively 0 and 180 ° between them and both have this phase shift and more precisely they are 5.5 and 174.5 °. The rectangle is part of the opamp symbol (U1) which is originally light yellow. Moreover, in the opamp you can see that there is feedback and this makes me think that it is not further reducible but I asked anyway to be sure.
Q: If it was originally yellow, why did it change back to uncolored?
1. It could be part of a hierarchial schematic
2. It is a symbol defined as a subcircuit schematic
The yellow color denotes a symbol which references a subcircuit text file.

All three methods are valid for design and simulation.

#### Zeno53

Joined Sep 23, 2020
60
Q: If it was originally yellow, why did it change back to uncolored?
1. It could be part of a hierarchial schematic
2. It is a symbol defined as a subcircuit schematic
The yellow color denotes a symbol which references a subcircuit text file.

All three methods are valid for design and simulation.
Because that image was a screenshot that I converted to black and white for other reasons and removing the color before the conversion was better. Another reason is the need to provide as little information as possible on the type of opamp used as it is not necessary for the initial question for the tread. I still don't understand what are the solutions you are talking about. Can you give me an example?

#### ericgibbs

Joined Jan 29, 2010
12,908
hi Z,
E

#### Papabravo

Joined Feb 24, 2006
16,132
With a hierarchical schematic you have a local symbol (.asy file) in your working directory along with a local (,asc file) which defines the implementation. On your schematic it will have a reference designator with the "X" prefix which is used for subcircuits.

The second alternative is the essentially the same thing with respect to files in the local directory, but you edit the definition to have a "U" for the reference designator, and can override any default parameters

The third alternative is to have the symbol (.asy) file in the ..\lib\sym path on your machine. There are some minor differences in the symbol definition. The implementation is not a schematic (,asc file), but a subcircuit text file like (.mod, .cir, .sub)
This is the one where the body of the symbol is filled with yellow or whatever other choice you make in the control panel.

I've attached some files that I use for an analog limiter. The first two, limit00 (1).asy and limit00.asc are used with the hierarchial or subcircuit schematic methods
The next two, limit00 (2).asy and limit00.sub are used with the symbol and subcircuit file. All three methods produce identical results.

#### Attachments

• 305 bytes Views: 1
• 1.6 KB Views: 1
• 484 bytes Views: 1
• 308 bytes Views: 2
Last edited:

#### Zeno53

Joined Sep 23, 2020
60
@Papabravo

Thank you so much for the explanation. I need the simulation to understand if it's possible to reduce the phase shift to mount everything on PCB. Is this solution (beyond my capabilities) applicable in a real circuit? That is, if I mount it on PCB, can I reduce the phase shift?

@ericgibbs

hi, i think not

#### Papabravo

Joined Feb 24, 2006
16,132
@Papabravo

Thank you so much for the explanation. I need the simulation to understand if it's possible to reduce the phase shift to mount everything on PCB. Is this solution (beyond my capabilities) applicable in a real circuit? That is, if I mount it on PCB, can I reduce the phase shift?

@ericgibbs

hi, i think not
I don't know the answer to your question, but I do know the process.
1. First you design the solution
2. Then make is small
3. Then make it elegant

#### Zeno53

Joined Sep 23, 2020
60
With a modest loss on gain moreover obtainable with a modification to the circuit I think I have solved it. Now we need to understand what the real drift is on FR4. Thanks everyone for the help.

#### LvW

Joined Jun 13, 2013
1,247
I am not sure if I could understand your problen in full detail. However, there exist severas methods for passive or active phase compensation for opamp-based amplifier stages. These methods can reduce the opamp-induced phase shift for increasing frequencies. In principle, the methods do nothing else than to extend the usable bandwidth of the main opamp.