Boost converter minimum load

[shadewind]

I'm designing a boost converter using the LM3478 PWM controller and looking through the equations, the minimum inductor value is determined by the minimum load current. You can also say that the inductor value determines the minimum load current. If I understand this correctly, dropping the load current lower than this moves the operation into DCM, discontinuous conduction mode. This is going to be used for bosting 12V to 40V for driving high power LEDs which will be dimmed by PWM so chances are high that I cannot avoid dropping below this minimum load current unless I want a very large inductor (i.e. 56 µH for 744 KHz and 50 mA min load current).

What will happen if it goes into DCM? Will it still work properly?

[dopplershift]

Hard to tell what it will do from the datasheet. It does say "[the CCM] mode of operation has higher efficiency and usually lower
EMI characteristics than the DCM"

So, I think it will still work, but will be relatively inefficient and produce a noisy output. Usually the manufacturers like to show off all the great features of their parts.... the fact that DCM is not shown in the datasheet may tell us that the output is not pretty!

[Zero999]

Going from what you've said, this seems to be the wrong approach. To drive LEDs you need a constant current source. You could of course boost the voltage to 40V and use a resistor to limit the current but this is inefficient and if you're going to be doing PWM the current will periodically drop to zero.

[shadewind]

Going from what you've said, this seems to be the wrong approach. To drive LEDs you need a constant current source. You could of course boost the voltage to 40V and use a resistor to limit the current but this is inefficient and if you're going to be doing PWM the current will periodically drop to zero.

I need the higher voltage to drive LEDs in series so the boost controller is only for getting the 40 volts. I'm of course going to be using a LED driver for the LEDs themselves. I thought it was obvious in my post that I was not asking about the whole solution, only about the boost converter?

[shadewind]

You can control the current through the LEDS by sensing the output current through feedback instead of the voltage. That way you don't need a separate driver. The boost converter in CC mode can be used to regulate the current independent of the Vf of the LED's,much less dissipative then a linear driver.

A lot of SMPS CC controllers can also be strobed (dimming). You could use your controller LM3478 in CC mode by taking feedback from the led strings current. The error amp doesn't care what your controlling all it wants to do is keep the inverting input equal to the positive input which is Vref. Usually though you use a controller with a reference of a couple hundred millivolts  to keep power dissipation low in the sense resistor. Read through the data sheets I linked too.

Oh and to answer your question your converter will still regulate when transitioning from CCM to DCM under light load. You could run unto problems when you design for DCM and go into CCM; CCM has RHPZ that needs to be taken into account,slope compensation is also needed for peak current mode control.An explanation of RHPZ and slope compensation as well as a boost LED  driver is in the tps40211 datasheet full design example for 35V 700mA LED driver.

If you dont want to  bother with RHPZ and slope compensation the NCP3065 doesn't require any.

Thanks for the great answer, this is what I was looking for. I think I'll take the plunge and learn about RHPZ since I've been reading a lot about complex frequency and Laplace tranforms lately. It's probably good knowledge to have anyway since I'm doing this for learning.

[shadewind]

I've been looking through the datasheet for the NCP3065 and I don't quite understand how the COMP pin is to be used for PWM dimming when it's already used for current regulation feedback? If get this correctly, one applies a voltage higher than the comparator voltage which then turns off the output?

[shadewind]

A question about the TPS40211, how is PWM dimming done. Judging from the application circuit, it uses the COMP pin but how does this work?

[thmjpr]

A question about the TPS40211, how is PWM dimming done. Judging from the application circuit, it uses the COMP pin but how does this work?

Which app circuit?
http://focus.ti.com/lit/an/slyt302/slyt302.pdf :
"The brightness can be programmed by altering R1, current injection into the FB pin, or by PWM dimming."

edit: I see it now page 34 of the datasheet. Good question, it seems overly complex. Not sure why you can't just use the EN/DIS input.

[shadewind]

Yeah, I found the demo board manual description. Since the COMP pin is simply the output of the error amplifier into the logic, what they're doing is pulling this down to ground to inhibit switching? The diodes seem to be for raising the voltage just a little bit above ground and not for any rectification. Thus, a logic level FET would be enough for my purposes. Can anyone confirm my thinking?

Also, in the same application circuit in the data sheet, what is the purpose of the resistor R13 which goes from the feedback pin to the current sense resistor. Is that necessary?