I have been working on a half bridge SMPS project to deliver 13.8V at 30A or so. During the course of my experiments I have found the ultrafast rectifiers to be rather fragile things compared with the traditional 50Hz rectifiers, which do seem to withstand the odd short circuit. I have burnt out 30A rectifiers with a load of only 8A on the DC side, I am wondering if this is due to voltage spikes on the waveform, would secondary side snubbers eliminate this problem?
I have come across several articles where Mosfets are used as rectifiers, but you need some extra turns on the power transformer to get the 10V or so gate-source voltage. My idea is to wind a small 1:1+1 transformer on a small toroid, say 13mm, and couple the primary of this small transformer to the secondary winding on the power transformer to provide gate drive to the Mosfets. This avoids putting extra turns on the power transformer which would take up valuable winding space.
Your thoughts please. Regards,
John
I have come across several articles where Mosfets are used as rectifiers, but you need some extra turns on the power transformer to get the 10V or so gate-source voltage. My idea is to wind a small 1:1+1 transformer on a small toroid, say 13mm, and couple the primary of this small transformer to the secondary winding on the power transformer to provide gate drive to the Mosfets. This avoids putting extra turns on the power transformer which would take up valuable winding space.
Your thoughts please. Regards,
John
If you didn't check the waveforms, then it must be your fault, you probably didn't use RC cross diods, I assume they are schottky's
In the end synchronous rectification should be more efficient, you also have IC's made fot this
In the end synchronous rectification should be more efficient, you also have IC's made fot this
Having checked the waveforms, there is a leading spike of about 50V on the rectifiers. I have since added an RC snubber across each rectifier (im using 2x STPS3045CW double diodes, with the 2 diodes per package parallelled) and the spike is of much lower amplitude. How do you calculate the snubber component values by the way? I used the 'cut and try' method.
I didn't know about these custom made ICs for synchronous rectification, but im going to stick with the gate drive transformer idea for now.
One other thing, I was amazed when, after a shorted rectifier and another incident where the power transformer output was shorted, the power supply didn't go bang. I wonder if there is enough leakage reactance in the power transformer to limit the current to a safe value; I have not as yet included any short circuit protection, Im manually controlling the pulse width with a variable resistor on to the DTC pin.
Thanks for the reply, regards,
John
I didn't know about these custom made ICs for synchronous rectification, but im going to stick with the gate drive transformer idea for now.
One other thing, I was amazed when, after a shorted rectifier and another incident where the power transformer output was shorted, the power supply didn't go bang. I wonder if there is enough leakage reactance in the power transformer to limit the current to a safe value; I have not as yet included any short circuit protection, Im manually controlling the pulse width with a variable resistor on to the DTC pin.
Thanks for the reply, regards,
John
well that spike is probably what killed diodes.
You have to zoom as much as possible, to see as much as you can into this spike.
what you will see is dampened ringing, that should be in order of MHz.
R is chosen (not sure how, I would pick low value) to suppress the amplitude, while use C that is big enough to pass that ringing with not much impedance, but have high impendance to switching waveform, several time lower in freq.
You have to zoom as much as possible, to see as much as you can into this spike.
what you will see is dampened ringing, that should be in order of MHz.
R is chosen (not sure how, I would pick low value) to suppress the amplitude, while use C that is big enough to pass that ringing with not much impedance, but have high impendance to switching waveform, several time lower in freq.
- Status
- Not open for further replies.