DC-DC Converter/ Feedback/ Verilog-A

That's right. However, my doubt is where we put that source component.

The only sources that I have are the vpulse who is connected to the non overlap circuit and the vdc (power supply that supplies all the circuits).

If you look at the presentation that you posted, in page 16, they used a voltage source, connected between the output of the converter and at the feedback loop input.

Now, taking into account that I asked you if it's possible to do the bode plot of the converter without the feedback control loop:

"But, what if I want to get the bode plot only the converter itself (to see how the converter it behaves), not resorting to the transfer function (obtained for example through SSA) and then matlab, can I still use the PSS + PSTB analysis to do this?"

You said that if the loop is not closed, which is the case, because I'd like to make the bode plot without going to MATLAB, I could use the PSS + PAC analysis.

So if you look at page 16 of the presentation and imagine my circuit composed only by the power devices with respective non overlap circuit and drivers, LC filter and the load, I must add a voltage source like that presented on page 16 of the presentation somewhere on the circuit? In which part of the circuit? In series with the VPULSE connected to the non overlap circuit? Should I connect that voltage source in series with the power supply? I must connect that voltage source in another part of the circuit? Which one?

Perhaps I don't need to insert any voltage source to do that PSS + PAC analysis because I can take advantage of the VPULSE, but I don't know if that is possible.

Hope made myself clear.

I am looking forward for your reply.

Kind regards. 

 Dear Pyroblast,

 > However, my doubt is where we put that source component.

 The best place to include an injection voltage source for stability analysis is a the node where the impedance looking backward (toward the outputof the amplifier) is significantly less than the impedance looking forward (i.e., the feedback path). For example, in an opamp feedback circuit, a good location for the voltage source is between the output node of the opamp (low impedance) and the input node to the opamp (high impedance).

In your specific application, which I do not have access to, you might choose a location with similar properties.

I hope this is useful.

Shawn

Hi shawn,

Thanks for your reply. I appreciate. 

 I've understood that, howeve I am afraid that you didn't got my point.

 If you notice, at certain part of my posts, I've written:

 "But, what if I want to get the bode plot only the converter itself (to see how the converter it behaves), not resorting to the transfer function (obtained for example through SSA) and then matlab, can I still use the PSS + PSTB analysis to do this?"

 Putting into another words, I want to  plot the bode plot of the converter without any kind of feedback loop.

 What is the kind of procedure if you want to label it like this, that people do? They get the transfer funcion of the converter then they go to MATLAB and plot the bode response of the converter. Then, based on that they design the controller. Isn't it?

So, and if I don't want to go to the MATLAB environment can I plot the same converter bode plot (alone without the feedback loop) using cadence?

Mr. Mark answered yes by using a  "pss + pac analysis" but "obviously need a source component with the parameter "PAC Magnitude" set (usually to 1).".

Ok.

Since I don't have yet the feedback part of the control designed, I want to know where I must put the source component? NOTE: I odn't even know if it's possible to do that, get the bode plot of the converter only without the feedback loop. But as Mr. Mark said it is, I hope that someone can answer me to this.

I don't know if I am explaining myself clear. I think I am. But if not, please let me know that I will make a sketch on paper and put here a picture taken by my cellphone.

The design is pretty simple. Is a sync buck converter like that on the presentation the Mark has posted. There is nothing special. Imagina that circuit but without the feedback loop.

I am looking forward to hear from you.

Regards. 

 Dear Pyroblast,

>  Since I don't have yet the feedback part of the control designed, I want to know where I must
> put the source component? NOTE: I odn't even know if it's possible to do that, get the bode plot> of the converter

> only without the feedback loop.

You are correct. I did not understand you wanted to create the bode plot from an open loop simulation. I can relay to you what I have resorted to estimate stability from an open loop sense. I am not an expert, but I do not know of a way that spectre can perform a stability analysis without closing the feedback loop. Without the loading effect of the output to the input stage that occurs when you close the loop, I do not know how its estimate of stability would be accurate.

I have done something similar in large signal ring VCO based circuits by adding only DC feedback into the VCO to stabilize its DC operating point and then performing a series of AC tran simulations at various time points as the supply is raised. In this fashion, I maintain the DC operating point correctly, but can study the gain and phase from the AC analyses - which effectively is your bode analysis. If you are not familiar with AC tran analyses, this allows you to start a large signal transient analysis and, at times you specify, will perform an AC analysis using the operating points of the devices at those specific times.

 I am sure others may have other or better suggestions for you. However, this methodology has been useful for me.

Shawn

Hello guys, Hope everything is going well.

Frank, I appreciate your availability and kindness for the answers to my questions.

In attached, Frank and all can see the circuit schematic, that for the time being it is presented like that to keep it simple and transmit the idea of what I am trying to get. Nothing special and it is the well know DC-DC Buck Converter, composed by:

- VPULSE, which is simulating a fixed duty cycle (the duty-cycle that the converter will be working to get achieve the voltage conversion);

- Non overlap circuit, which devides the PWM signal (obtained by the VPULSE source) and connects to the Drivers;

- Drivers (mentioned before), fed by the VDC source;

- Power Devices, fed too by the VDC source;

- and the LC Filter, Load;

Now frank, when you say to connect a PAC source between the ground and the input of the circuit that I want to examine, you mean what by that? Because I want to get the frequency response of the converter in open loop (as you can see by the schematic), you say I need to add that PAC source. Well I assume that the PAC source that you're referring can be either a VSIN, VPULSE or even a VDC source provided that any of them has the field PAC MAGNITUDE to be filled.

(in this image you can see the PAC AMPLITUDE field. The VPULSE has that same field. 

To keep things simple, lets just consider the VSIN SOURCE.

Now, regarding the question on where I must connect the VSIN source working as a PAC SOURCE, it will depend where I want to examine (that is, where is the input of the circuit that I want to examine).

From what I have been reading, we can have 3 types of transfer functions on the converter: VLINE (I think it is the Vin) to VOUT, the CONTROL to OUTPUT and the INPUT IMPEDANCE. To be honest I am not sure what transfer function should I analyze, so that one can study which controller to use. I would say that the transfer function that really matters for the design of the controller is the Control-to-Output transfer function?

 If so, I need to insert the VSIN SOURCE in series with the VPULSE SOURCE? Which is the point where the control signal is?

Kind regards. 

EDIT: Mark I tried to run a simulation using VSIN source in series with the VPULSE as well as using the VPULSE it self with the AC Magnitude @ 1V but an error occured:

"Error found by spectre during pac analysis .. There is no AC source in the circuit." Here: (I didn't used the DC Analysis at the same time - don't know if it's needed):

Then I remember to use the PSIN source. I tried that in series with the VPULSE and in parallel (with a capacitor separating the VPULSE to PSIN) but the same error appeared. Here:

After that I have added a DC analysis and now a new error appears (but the other one desappeared) saying that the PAC analysis was skiped because PSS analysis must be performed first. But I have the PSS analysis configured, as you can see here:

On the PAC analysis, he detected the PSS beat frequency, I start the sweep @ 100MHz an stop @ 3GHz (step size 10MHz). Maximum Maximum sideband I put 2.

Any idea of what I am doing wrong? I appreciate all the help.

Regards. 

Hello Frank,

I did some progress, however with not any good results. In the circuit shown my previous post I filled the PAC MAGNITUDE with 1V and the simulation simpled converged (maybe because of what you said about the VDC SOURCE).

One doubt that I had was regarding the Maximum Sideband. Should I use 2 (I read somewhere that if the circuit is up to or 3rd order, we could use 2) or 0

Now regarding what you said,  just to confirm, you are suggesting me to include the PWM Block, that is, the modulator block, composed by a comparator + ramp generator + the control voltage (which is the error voltage that comes from the controller)?

I need to put a voltage source in the comparators pin where the error voltage comes and there fill the PAC Magnitude value to the analysis be performed there?

I cannot use the VPULSE SOURCE to simulate that? Why?

Thank you so much for your availability. I appreciate that.