[MOVED] PWM Boost Converter

[danmicro]

Hi. I am a beginner in analog design. I want to make a PWM boost converter (Vin=1.8 , Vout=5). My problem is, at the start, the feedback voltage is very low since the output of my boost is not yet reach. Now the output of my error amplifier is always high (near VDD=1.8). My sawtooth amplitude is only 1.5 and i cant go beyond VDD. Therefore there is no pulse generated. How do i fix this problem?

 

[scopeprobe]

You should provide some more information on your actual circuit so we know what IC your using and what value of components your using as this looks like a block diagram from some datasheet.

 

[danmicro]

You should provide some more information on your actual circuit so we know what IC your using and what value of components your using as this looks like a block diagram from some datasheet.

Hi, thank you for your response. I am designing a pwm boost converter in synopsis using tsmc.18um, so i dont use existing IC. I already design the opamp, comparator, sawtooth, and the reference voltage. My problem is as stated above that i cant generate a pulse.

 

[KlausST]

Hi,

so i dont use existing IC.

then please next time post in the "IC design" section.

Klaus

 

[danmicro]

Hi,


then please next time post in the "IC design" section.

Klaus

Sorry. Im new to this forum. I dont know there is a separate section for this topic.

 

[scopeprobe]

Hi, thank you for your response. I am designing a pwm boost converter in synopsis using tsmc.18um, so i dont use existing IC. I already design the opamp, comparator, sawtooth, and the reference voltage. My problem is as stated above that i cant generate a pulse.

To generate a pulse you need to verify you have the oscillating clock source in the form of a sawtooth as shown on your block diagram. The output of the error amplifier needs to operate in between the maximum and minimum of values of the saw tooth to generate pwm with duty cycle. For you to be not getting a Pulse width suggests either your sawtooth is too small pk-pk or the gain of your control loop isn't correct. If its not that then it could be that you've not provided limits to ensure you can always get some duty cycle, most commercial IC's are clamped to always operate and contain dedicated circuitry like pulse skipping etc at light loads to keep the device in operation to avoid operating rail to rail on their amplifiers so this may also be your issue.

Also of Note is most Mosfets usually require 3-5V volts of Gate to Source voltage to pass their operating threashold, I'm not sure of the case here as i'm not familiar with the Synopsis technology but 1.8V maybe problematic. Most commercial PWM controllers come with integrated regulator or fancy bootstrapping techniques to provide power adequate power rails with reasonable offsets of which you've perhaps not considered.

 

[mtwieg]

There's at least two options:
1. Clamp the output voltage of the error amplifier.
2. Create another logic signal which defines a maximum duty cycle and AND that with the PWM comparator. See below:

 

[Easy peasy]

implement a " soft start " that allows the demand ( or ref ) voltage to some up slowly from zero - this allows the EA to track the ref and allows a start and prevents overshoot when the ref reaches its target.

 

[danmicro]

H

There's at least two options:
1. Clamp the output voltage of the error amplifier.
2. Create another logic signal which defines a maximum duty cycle and AND that with the PWM comparator. See below:

View attachment 199176

Hi. I have tried your suggestion and got similar waveforms to yours. But it seems that the output is not stable. I also tried voltage divider in the output of error amplifier and for me it is a better waveforms. However i want to use the one you suggest. How can i improve it? I think it got something to do in the calculation of my type 3 compensation, maybe it is wrong. Here is my simulation results in voltage divider method and the one that you suggest.

 

[danmicro]

implement a " soft start " that allows the demand ( or ref ) voltage to some up slowly from zero - this allows the EA to track the ref and allows a start and prevents overshoot when the ref reaches its target

Hi. I already implement a soft start, but my error is still high until my output voltage reaches 5V which causes an overshoot. Maybe my design is not good. How do i correctly implement a soft start?

 

[Easy peasy]

Sorry - where is your soft start ? a proper soft start ensures the EA rises slowly.

 

[danmicro]

Sorry - where is your soft start ? a proper soft start ensures the EA rises slowly.

Sorry. I just realized your soft start is somehow different from mine because yours is for Vref. Mine is an increasing duty cycle but it does not affect the EA. Your soft start is also an increasing duty cycle but due to increasing EA, if im not wrong. Here is my soft start circuit. Can share the soft start circuit for vref?