LM2735Y : How to compute a Bode diagram?

Hi CA,

I would propose that you try the following to calculate the compensation that you need. You can either use WEBENCH to work out your circuit needs or you can follow the datasheet using equations 1-6 on pages 12 and 13. Follow the sequence of the equations to get the compensation. As a fallback feel free to send me your schematic and test conditions and I can take a look at this for you. Do you have any Bode plots that you can also share?

Thanks,
Murray.

Hi Murray,

Thank you for your response. I have just tried to use Webench, it is a great tool. I did not know it before. To validate my model, I did a Webench simulation and put the same value in my Octave model. I have a problem in my code, the Bode is totally different. I used the equations in the datasheet as you suggested but I am not sure about my final equation ( H(p) = ... )

Here is my Octave code :

%% Data
R1 = 10e3 ;
R2 = 86.6e3 ;
Vin = 2.85 ; % Input Voltage
Cin = 10e-6 ;
Cout = 10e-6 ;
Ccomp = 470e-9 ; % External compensation
L = 33e-6 ; % Inductor
Rload = 120 ; % Load resistor to have 0.1 A

eta = 0.85 ; % Approximative value (depending of Iout) for global efficiency

Vref = 1.225 ; % Build-in reference voltage
N = 1000 ;
freq = logspace(1,6,N);
omega = 2*pi*freq;

%% Computed value
Vout = Vref * (1+R2/R1) ;
Dprime = eta * Vin/Vout ; % Duty cycle with efficiency correction

FreqZeroCF = 1 / (2*pi*R2*Ccomp) ;
FreqRHPZero = Dprime^2 * Rload/(2*pi*L) ;

FreqPoleGM = 0 ; % Pole at origin due to gm amplifier
FreqPoleRC = 1 / (2*pi*Rload*Cout) ;
FreqPoleCF = 1 / (2*pi* (R1*R2)/(R1+R2) *Ccomp) ;

%% Bode diagram
Hnum = (i*omega-FreqZeroCF) .* (i*omega-FreqRHPZero) ;
Hden = (i*omega-FreqPoleGM) .* (i*omega-FreqPoleRC) .* (i*omega-FreqPoleCF);

H = Hnum./Hden ;

Module = 20*log10(abs(H)) ;
Phase = 180/pi * arg(H) ;

%% Display
ax1=subplot(2,1,1)
semilogx(freq,Module);
title("Magnitude") ;
xlabel("Freq (Hz)") ;
grid on

ax2=subplot(2,1,2)
semilogx(freq,Phase);
title("Phase") ;
xlabel("Freq (Hz)") ;
grid on

linkaxes([ax1,ax2],'x')
-------------------------------------------------

Here are the Bode diagrams :

From Webench
webench.ti.com/.../SDP.cgi

My simulation
[URL=http://www.hostingpics.net/viewer.php?id=971427LM2735Y.jpg][IMG]http://img15.hostingpics.net/thumbs/mini_971427LM2735Y.jpg[/IMG][/URL]

As you can see, my simulation give me a very weird Bode diagram! I do not have any static gain in my equation... I guess I did not really understand how to use the equations in the datasheet. Do you have any idea? Thank you in advance.

About my stability problem, I guess my inductor has a too big value in my circuit (100uH) which affect the RHPZero might lead to stability issue according to the datasheet (end of page 13).

Hi CA,

I noticed in your code above that your Ccomp is a factor of 3 out. It should be 470pF. Your L as you stated will also impact the RHPzero which will affect the stability. Are you able to re-design your board with the suggested values based on the Webench simulation?

Thanks,

Murray.

Hi Murray,

Thank you for your response. Yes... I made a terrible mistake... nano is 10e-9 and pico is 10e-12.

My Octave simulation doesn't match either... Actually, I am not working on this project (and company) anymore. I help them sometimes, when they need me :) . I'll tell them the result of the simulation and ask them to change the inductor. The problem first appeared when they deceided to increase the output current.

I guess it will solve the problem. But do you know whether my equations are corrects?

Thank you again,

Best regards,

CA

Hi,

I have a great new. I still don't know why my Bode is not correct but my former colleague has just told me that everything works well since they changed the inductor for a smaller value!

Thank you!