DC-DC SPMS Boost Converter DCM Prevention

Question

Sorry for the long question, it's really opinions I seek which is why I felt I needed to explain the scenario and how much I know/don't know. I would really like some input on my design. I am designing a DC/DC SMPS(switch mode power supply) boost converter.

The general advice about DCM(discontinuous conduction mode) in SMPS boost converters is to try and avoid it. This advice is usually backed up by a barrage of equations. Unfortunately I am not very good at Engineering Maths, but if the DCM is anything like the equations, I get the point, I'll stay away :D ! Aside from adjusting switching frequency, increasing the inductance value looks the biggest way to avoid DCM. In my boost converter the output voltage can be varied(it's microcontroller based) and I expect the load to make the input current vary from 10mA(high inductance needed) to 4A(relatively high saturation current needed). I have had some problems in finding an inductor that would fit my specification, I used Digikeys' Parametric search.

Since inductance value is mostly inversely correlated to saturation current, I understand why it may be hard to find my golden inductor. I have found about 8 out of tens of thousands on Digikeys' site, but they are unshielded(which I don't want for this application), large :( and expensive :( .

I understand how DCM works well enough, that the converter enters DCM when the load current is so low that the discharge cycle of the inductor lets the inductor current slip past zero.

So here is my take, since this application is microcontroller based, couldn't one simply just stop the discharge of the inductor when the current nears zero? I know this means distorting the PWM. What I am suggesting is having two microcontroller pins connected to the base of the MOSFET(or gate driver to MOSFET) that controls the inductor charging. One is switched based on the PWM while another simply goes high when inductor current value drops to the microamperes range. I plan to use another MOSFET (Q2) in the inductor discharge path to measure the inductor current. I'll be using the Vds of the MOSFET (Q2) to calculate the inductor current and decide when step in to charge the inductor. So do I get the IEEE prize for SMPS design :D ? Well I doubt it, as while I was learning how to design a boost converter(which I did off the internet thanks to so many kind people), most of the designs I saw used a SMPS controller so I can understand that there wouldn't have been as much flexibility as with a microcontroller based approach because I never saw anyone discuss this.

So my question is, what are your thoughts? I haven't had the opportunity to test it on a breadboard yet, will do soon, but whether it works or not, I want to know if there is something I am missing, or the opinions of anyone on this idea. I am not an experienced circuit designer so I could do with some input.

Thanks for taking the time to read the question. :)

Answer 1

So here is my take, since this application is microcontroller based, couldn't one simply just stop the discharge of the inductor when the current nears zero?

Think about this - the only way to stop the "discharge current" is to prematurely ground the transistor in the boost device to start "charging" current thru the inductor - you have no option - to keep the inductor open circuit is to enter DCM and that is what you are trying to avoid.

The cycle for a boost converter (or flyback converter) can be: -

• Ground the inductor thus current ramps up and inductor stores energy (charge)
• Un-ground the inductor - energy gets released to the output cap and load (discharge)
• If inductor can't sustain current into load via diode you enter DCM and basically the inductor becomes open circuit except for the parasitic capacitance of the MOSFET switcher i.e. you get a damped oscillation until...
• The cycle begins again.

OK, so then what happens (in that cycle) is a tiny little too much energy becomes transferred to the output capacitor and load (during inductor discharge). This causes the output voltage to rise fractionally higher than what it would and, over a period of a few milli seconds, you might exceed the output voltage that is safe for the load. A few seconds later and you have a dead load and a few seconds later you have a dead boost regulator.

OK before you get to this, the control loop would probably have implemented cycle skipping but cycle skipping is noisier than ordinary DCM so why bother?

BTW, going into DCM isn't that bad - the line regulation of the output isn't as good but it's still quite controllable. There isn't much on the web about this but, due to the self-resonance of the inductor and MOSFET drain capacitance, once DCM is entered there is a small oscillating current in the inductor that is asynchronous to the PWM and when the inductor restarts charging it does so at sometimes a slightly positive or negative current. This can cause the noise on a simple controller such as a fixed-on-period controller.