Hello everybody, i'll be building my first off-line SMPS soon. I'll be using either 2-transistor forward or halfbridge configuration. Since i haven't built any isolated SMPS supplies before, i'll have a few questions.
I was thinking of using optocouplers to drive the MOSFETs but i want to work at 100-150kHz and i understand an opto is too slow for that. I have some gate drive transformers from ATX power supplies, will those work? In one of the supplies there are MJE13009 bipolar transistors in the primary, and between the transformer secondaries and transistor bases there's 2.6kOhm resistors. Will the same transformer work for MOSFETs or do i have to use bipolars? For my ~400W output requirement i think the 13009s will do fine so i don't have a problem using them.
And now Q number two, feedback isolation. I have some 12v/200mA transformers. If i power my PWM controller board using one of those, i won't need any feedback isolation anymore, right? I can just connect the output of the comparator to the feedback pin of my controller, no opto, no nothing. Please correct me if i'm wrong.
I was thinking of using optocouplers to drive the MOSFETs but i want to work at 100-150kHz and i understand an opto is too slow for that. I have some gate drive transformers from ATX power supplies, will those work? In one of the supplies there are MJE13009 bipolar transistors in the primary, and between the transformer secondaries and transistor bases there's 2.6kOhm resistors. Will the same transformer work for MOSFETs or do i have to use bipolars? For my ~400W output requirement i think the 13009s will do fine so i don't have a problem using them.
And now Q number two, feedback isolation. I have some 12v/200mA transformers. If i power my PWM controller board using one of those, i won't need any feedback isolation anymore, right? I can just connect the output of the comparator to the feedback pin of my controller, no opto, no nothing. Please correct me if i'm wrong.
Hi, your questions sound basic but do adress a few holy grails in SMPS design, I think....
first, your power supply needs to adress the isolation topic somewhere, since the amplifier you want to power should not be dangerous to touch. Two alternatives - control circuit on the primary side, or on the secondary side.
The overwhelming majority of SMPS today are build with a PWM controller on the primary side, with an optocoupler + reference on the secondary side for regulating the output voltage. THis is a well-known and safe topology. Putting hte controller on the secondary will cause exactly the problem you mention - how to drive the gates of the power switches on the primary side? Plus, how do you power up the whole thing - you must have an auxiliary power supply in that case. In conclusion, a lot more signals need to cross the isolation boundary.
second, for the switching frequency you are looking at you'll probably want power MOSFETs, not bipolars, much better in swithcing losses.
I would recommend you look at some reference designs as available on the websites of semiconductor companies and study those first, that should answer many questions. If you need directions just PM me....
first, your power supply needs to adress the isolation topic somewhere, since the amplifier you want to power should not be dangerous to touch. Two alternatives - control circuit on the primary side, or on the secondary side.
The overwhelming majority of SMPS today are build with a PWM controller on the primary side, with an optocoupler + reference on the secondary side for regulating the output voltage. THis is a well-known and safe topology. Putting hte controller on the secondary will cause exactly the problem you mention - how to drive the gates of the power switches on the primary side? Plus, how do you power up the whole thing - you must have an auxiliary power supply in that case. In conclusion, a lot more signals need to cross the isolation boundary.
second, for the switching frequency you are looking at you'll probably want power MOSFETs, not bipolars, much better in swithcing losses.
I would recommend you look at some reference designs as available on the websites of semiconductor companies and study those first, that should answer many questions. If you need directions just PM me....
Well like i said, i will be using an auxiliary power supply for the PWM IC. For now it just seems like a logical thing to do.
I'll be looking at some designs on the web... My main reference is the book "Switching Power Supply Design, Third Edition", and it does a good job in explaining the topologies, giving transformer calculations and so on, but how you actually drive the power switches isn't that well explained.
I'll be looking at some designs on the web... My main reference is the book "Switching Power Supply Design, Third Edition", and it does a good job in explaining the topologies, giving transformer calculations and so on, but how you actually drive the power switches isn't that well explained.
I think i figured out which way to go. I already have a working 5v -> 12v boost converter (i dunno why i built it anyway but it works fine and with >85% efficiency). It's complete with controller, driver transistors, comparator and reference. Oh and you'll laugh if i'll tell you what i used as a controller.
I lifted the boost inductor and the mosfet off the board and i will operate the controller at 12v input to get better drive. I also have a ~45v AC transformer capable of 300W. I'll slap a diode bridge and filter capacitors on that, and use it as my "mains" supply for the switching regulator, so i can try various gate/base drive circuits, transformers, switching transistors and whatnot, without working on live mains (thus less chances of disaster
). Since the drive circuit will only depend on the voltage applied to the PWM controller, once i figure out a good one it shouldn't matter whether the transistors are switching 50 or 300 volts.
I lifted the boost inductor and the mosfet off the board and i will operate the controller at 12v input to get better drive. I also have a ~45v AC transformer capable of 300W. I'll slap a diode bridge and filter capacitors on that, and use it as my "mains" supply for the switching regulator, so i can try various gate/base drive circuits, transformers, switching transistors and whatnot, without working on live mains (thus less chances of disaster
). Since the drive circuit will only depend on the voltage applied to the PWM controller, once i figure out a good one it shouldn't matter whether the transistors are switching 50 or 300 volts.
An externally hosted image should be here but it was not working when we last tested it.
This works for constant PWM in the 10-90% range, let's say. Good for single ended applications, including buck/boost converters, class D amp, etc. Rapid changes in duty cycle may saturate the GDT and cause unexpected transient effects (like nonlinear duty cycle, slow gate edges, etc.)
A balanced waveform, such as the TL494 is intended to produce, can drive a GDT without worrying about coupling capacitors. This approach is used here:
http://myweb.msoe.edu/williamstm/Images/Tubescope_Supply2.png
and in most commercial power supplies. Note: a shorting-mode GDT is used to provide turn-off current to the bipolar transistors used in cheap power supplies. It wouldn't be a bad idea to do the same for MOSFETs, either.
Totem pole outputs (including TL598, UC3525, etc.) may be capable of driving a GDT directly; beware that these chips are bipolar, so the saturation voltage of 1-2V causes a step offset during dead time, which can push MOSFETs past threshold when they're supposed to be off. This costs efficiency, or worse, might make power output uncontrollable.
Unfortunately, I don't know of any CMOS-output voltage mode controllers. For that, you'll have to follow a TL494 by a gate driver like TC4432. Cheap complementary dual MOSFETs can also be used.
Tim
Thanks a lot for the schematic, it helped. I managed to get some decent drive from that ATX power supply transformer. At low voltage everything checked out okay, but unfortunately when i connected the thing to mains it worked for a couple seconds then both MJE13009s blew up.
I'll have to check for any unwanted spikes, perhaps some snubbers are in order. With their 400v Vceo rating and 230v mains, the 13009s are working close to the limit even if all is well, so i'm quite sure it's spikes that killed them.
I'll have to check for any unwanted spikes, perhaps some snubbers are in order. With their 400v Vceo rating and 230v mains, the 13009s are working close to the limit even if all is well, so i'm quite sure it's spikes that killed them.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.