LM2700 Compensation- no solutions

I think you are correct.  I tried a few different inputs, always the same R = 20k and C = 4.7nF.  It is very suspicious that there is no option to view the Bode plot either.  I'm going to move this post to the modeling forum to find out what is going on with webench.

Thank you very much John

Andy

Andy,

In WEBENCH, under the "Op Vals" button, you should be able to see the crossover and the phase margin. You can also edit the compensation components in the schematic (click on the component, then click edit) to see the effect on the crossover and phase margin with different components.

Thanks,

So why does it pick the same values no mater the input parameters?   WB does not produce a bode plot.  How can you expect customers to have any confidence in the WB results?  please feel free to contact me directly.

Hi John,

I tried different design cases and Ccomp doesn’t change much - it stays at 4.7nF - but the Rcomp value is different for different Vin/Vout/Iout combinations.  However, Rcomp is capped to a max value of 20Kohms, per the datasheet. Perhaps you were using similar design inputs that reached the max limit for Rcomp.

Here is the equation WEBENCH uses for the target value for Rcomp:

Rcomp_R_Tgt = (Rload * (1-D) * (1-D) * Cout * Rload * Rout * Ccomp * Vout)/ (L1 * 10 * Ccomp * Gm * Rout * Vref * (((1-D)* Rload) /(M_Rdson * (1+r))));

with the minimum and maximum limits set to 5Kohms and 20Kohms.

Here is the equation WEBENCH uses for the target value for Ccomp,

Ccomp_C_Tgt = (1 / (2 * PI * 40 * Rout));

where Rout is 875Kohms, the output impedance of the error amplifier, and the minimum and maximum limits set to 1.82nF and 4.7nF.

I suppose that is all fine.  I forwarded this, but I have no special knowledge of LM2700.  From my experience, many of these equations are iffy at best.  You will need to contact someone from that product line.  keep me informed though.

Hi John,

Here is some more information to try to understand the equations used:

The frequency for the pole used in the Ccomp target value is set to 40Hz, and Rout is the error amplifier output impedance. The datasheet recommends setting the value between 10Hz to 500Hz.

Ccomp = (1 / (2 * PI * 40 * Rout));

The Rcomp target value is simplified as this:

Rcomp = (1/10 * wRHPz) / (Ccomp * wP1 * wP2 * ADC)

where

wRHPz is the right half plane zero

wP1 is the pole from Cout and Rload

wP2 is the pole from Ccomp and Rout(EA output impedance)

ADC is the gain of the closed loop, the product ACM * AEA * AFB

So the target zero placed by the Rcomp and Ccomp is at (ADC * wP1 * wP2) / (2 * PI * 0.1*wRHPz). It's not an exact science, but the goal is to try to place this zero at a frequency lower than the RHPz (in this case, 1/10 the RHPz frequncy) and above the two poles.  Also you would like to try have this zero close to the overall system crossover frequency.

I also found an app note that describes the boost compensation design method here:

http://www.ti.com/lit/an/snva067c/snva067c.pdf

Are you in contact with anyone from that product line?  They would  be much more able to address this.

Thank you very much

I will study this and compare with my values

Andy

Hi

It is quite a challenge for a novice like me to try to calculate the best values for the compensation network

I have difficulties to calculate ACM (see the document snva067c.pdf)

ACM is calculated : D'*RLOAD/2*RSN

But how is the value of RSN for the LM2700?

Thank you very much

Andy

Hi

I think webbench indeed cap the Rcomp on 20kOhm.  I have tried to calculate with an excel-sheet. I don't know If I can upload this:

LM2700Compensation.xlsx