3
\$\begingroup\$

I am making a TL494-controlled boost converter from discrete components which has these specs:

  1. Max output power: 120 W
  2. Max output voltage: 100 V
  3. Max output current: 10 A
  4. Supported input voltage: 7-18 V
  5. Max input current: 12 A (Absolute max is 15 A, but not recommended)
  6. Switching frequency is 100 kHz

I will be using a high-current inductor so I don't have big inductance options unless I make the inductor myself. As far as I know inductance is important for current ripple; if I get a huge ripple it may overshoot the maximum current rating. Here is the circuit and simulation:

enter image description here

In the simulation the current ripple is 2 A which is acceptable. A closer look:

enter image description here

But when I calculate the ripple with the formula:

$$I_{ripple}=\frac{V_i\cdot D}{L\cdot f}=\\ \frac{12\cdot 0.87}{4.7\cdot10^{-6}\cdot 100000}=22\ \mathrm{A}$$

I don't know which one to trust.

Can you tell me if this is an acceptable design for a student project?

Is the way I do my calculations alright?

Does it matter when the converter operates at CCM or DCM?

Matlab:

    clear; clc;
%% Parameters
f = 100000;
Vi = 12;
P = 120;
Vo_max = 100;
Iomax = 10;

%% Requested Scenario
Vo = 100;
Io = 1.2;
deltaV = 0.1;

%% variables of the scenario
R = Vo/Io;
D = 1-(Vi/Vo)

%% Results for CCM
%L = (D*(1-D)^2*R)/(2*f);
L = 4.7 * 10^-6;
Ii = Vo/Vi * Io % = IL
%Iripple = Vi*D/(L*f)
C = D/(R*(deltaV/Vo)*f)

%% Results for boundary conditions
D1 = (Vo_max/Vi)*(2*L*f/(R*D));
IL_max = Vi*D/(L*f)
\$\endgroup\$
3
  • 1
    \$\begingroup\$ Your formula is just some numbers strung together. It isn't meaningful as written. \$\endgroup\$
    – Andy aka
    Commented Nov 10, 2022 at 9:07
  • \$\begingroup\$ ”Max output power: 120 W Max output voltage: 100 V Max output current: 10 A” How? Draw the voltage-current graph for the output. \$\endgroup\$
    – winny
    Commented Nov 11, 2022 at 8:36
  • \$\begingroup\$ Ofc, the max output current decreases with output voltage. Output is between 14V 10A and 100V 1.5A. Input can be at least 7V and output current can be 1.2A but this requires 15A input. So it is around 150W absolute max. There will probably losses so I said 120W. \$\endgroup\$ Commented Nov 11, 2022 at 12:33

2 Answers 2

5
\$\begingroup\$

Can you tell me if this is an acceptable design for a student project?

Not really, Depends on the student and the class, and would probably be up to the teacher.

Is the way I do my calculations alright?

Probably, but if you plan on building this thing you may want to simulate feedback. Boost converters usually use voltage feedback (with a divider) to regulate the voltage under a changing load. Another thing that is needed is something to change the switching if the voltage is changing (like a comparator). You may also want to simulate a variety of different loads.

Lastly, if you are going to physically realize this thing, make sure you also put in parasitics (each wire or trace has resistance and inductance, this can be a problem, so model it and put it in the spice model to see if it changes. Also check thermal performance (power dissipated by holding alt and clicking on the component). If the power dissipated is more than the component can handle, it might melt down if you turn on the circuit.

\$\endgroup\$
2
  • \$\begingroup\$ I will add feedback later, first I want to make sure open loop is fine. I can't model parasitic capacitances or inductances of the traces because I can't measure them, I am going print PCB at home so there is no way to predict what the PCB is going to look life beforehand :D. Power dissepation is part of the homework so I will sure do it. \$\endgroup\$ Commented Nov 10, 2022 at 12:36
  • \$\begingroup\$ You may want to put in some token parasitics (like 30mohms or 100mohms) make sure you also use a real cap you can right click on the cap and select a value. Same thing with the inductor, pick the same infinity that you will buy or make (you can right click and add paystubs to components) \$\endgroup\$
    – Voltage Spike
    Commented Nov 10, 2022 at 14:41
3
\$\begingroup\$

I don't know which one to trust.

Check your gate drive pulse definition. Your calculations are for 100 kHz (8.7 μs on-time) but in your simulation you apply 1 MHz (0.87 μs on-time).

Does it matter when the converter operates at CCM or DCM?

Yes, it does.

For your other questions refer to God of Mods' answer.

\$\endgroup\$
3
  • \$\begingroup\$ You were god damn right, thanks. When I fixed the simulation the MATLAB calculations and simulation confirmed each other. Now what I want to know is: As long as I don't go above the current limit of the inductor, does it matter how big the ripple is? I am planning to go for +-2.5A ripple. Also is it okay to have 0.1V ripple at 100V which is quite high? \$\endgroup\$ Commented Nov 10, 2022 at 12:42
  • \$\begingroup\$ @CaveScientist - 0.1v ripple in 100 volts is 0.1% 2.5A ripple in 10A is 25% - Power supplies are used for um, powering things. Different things have different requirements. A hot-wire foam cutter for instance, is going to care a lot less about the supply specs than say a motor-speed controller. The motor-speed controller in any optical drive or record-player or a cassette deck each tend to ruin the show when not working precisely. \$\endgroup\$
    – enhzflep
    Commented Nov 10, 2022 at 14:14
  • \$\begingroup\$ Alright, I will go for 0.8A current ripple this requires 136 uH inductor in the worst case (100V). I can even go over 120W with this configuration. The only drawback is that more inductance increases settling time as I checked from the simulation. Settling time is around 20ms, the overshoot goes to 60 Amps but only for around 1 milliseconds. Would the overshoot or the settling time be a problem? Also is 100uF output capacitor is too much? \$\endgroup\$ Commented Nov 10, 2022 at 15:16

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.