dcdc converter transfer function caculation problem

[kkkhunter]

boost converter transfer function

when caculating the converter transfer function, the pole and zero are determined by the load resistance. when the load is a nonlinear device like LED, the large signal DC resistance is different from the small signal AC resistance. So which one should be selected to caculate the converter transfer function?
Thanks!

 

[nxtech]

Why are you including an LED in your calculation? Take it out. Your transfer function should be only for the DC-DC converter. Dont add stuff you dont need.

As for your LED you should only be using a current driving method to control it.

E

 

[kkkhunter]

Hi, nxtech, thanks. But my DC-DC converter is a LED driver, so the output load is LEDs. And the I-V curve is exponential curve. So how can I have the transfer function without the LED. The exponential resistance determine the output load impedance.
Waiting for your further advice!

 

[nxtech]

I take it you are using a integrated solution to drive your LEDs?

E

 

[kkkhunter]

yes.
So the loop transfer function must be calculated well and cannot be adjusted in application. But it depends on the load impendance.
And is there any difference between commom converter and LED driver in loop design?

 

[nxtech]

The fact that you are using an integrated solution usually eliminates the need to calculate a loop transfer function. What part are you using?

What do you mean by common converter? LDO, PWM, Switching? An LED specific driver (at least the ones I have used) should never require you to calculate a transfer function as the device is designed to compensate automatically to provide a constant current value while voltage is allowed to vary some usually due to declining battery voltage.

 

[kkkhunter]

My LED driver is a boost converter and now I have problem in the compensation design. What do you mean by compensate automatically? Do you mean that the converter employs a hysteretic controller? But the hysteretic method cannot be used in my design. So the compensation is determined by the loop pole depending on the load impendance. And which impendance should I use for the load LED, DC impendance or AC impendance?

 

[nxtech]

You say your LED driver is a boost converter? Then its not a purpose built LED driver. What supplier and supplier part are you using?

A purpose built LED driver IC is a PWM type of circuit that has a loop control circuit built in (they do not use hysteresis).
The IC has an internal synchronous rectifier to improve efficiency along with an internal inductive damping switch to reduce ringing. This combined with the feedback control circuit that is built in will drive your LEDs by out putting the correct voltage needed to turn the LEDs on and generating the required current to maintain brightness levels.

It sounds like you just grabbed a boost converter designed for non-LED applications. That does not mean you cant use it; it just means that you have your work cut out for you.

E

 

[kkkhunter]

You mean I cannot design a boost converter to be a LED driver? or a boost type LED driver donot need a exact loop compensation design? But there are many LED driver ICs which are boost converter. You can see the attached TI-tps61080. It can be LED drivers in its applications.
Although, lots of DC-DC converters are designed to be power supplies to drive nonlinear load. Linear load may be just ideal situation. So how to solve the nonlinear load problem?
Thanks!

 

[nxtech]

You're not paying attention. I never said you could not use a boost converter in an LED driver circuit and I never said that an LED driver circuit does not need a loop compensation design.

You have shown your inexperience which is ok so long as you dont get caught up in details or requirements which are not needed.

I have seen the TI TPS61080 device before and have even used it on two occasions. Are you using the TI part or are you trying to design a discrete circuit that does everything that the TI part does? Dont reinvent the wheel. By that I mean dont worry about the determining the IC's transfer function. In fact TI doesnt want you to know. That is why they tell you only what you need to know to calculate for certain items (i.e. ripple, startup times, transient response on the output,etc).

If you are using the TI part have you even read the datasheet? If you have, then you need to read it again and see that they have told you how to calculate values for only the items you need and you dont need the transfer function to use this part. In fact if you look at the section for the feed forward capacitor they not only tell you that the pole frequency is determined not only by R1 and R2 but also by C1 which if I remember correctly is a non-linear device. They also tell you what the crossover frequency is provided you use the inductor and capacitor they recommend. Doing so will allow you to better place your zero between 1khz and 10khz.

In the end you will never have a DC-DC power supply that does not have a non-linear device like a capacitor or inductor attached to its output if you expect it to provide a stable DC signal. These non linear parts are needed reduce output ripple currents.

To sum it up, you dont need to worry about calculating the entire loop transfer function. That is why we have integrated circuits. That is also why every IC manufacturer of DC-DC converters give you several reference designs to use as is or modify using the equations provided in the datasheet.

E

 

[kkkhunter]

Thanks,nxtech.
Now I want to have a discrete boost converter to be a LED driver. The function is just a little similar like TPS61080, so I take it as a example.
Because I must design not only the attached L, C but also the whole converter loop, I think I should concentrate on the loop transfer function and compensation design. Thanks for your advice. Nowadays, I know how to solve the problem.
Thanks!