1 kW Push-Pull Converter

[H713]

This is another part of an ongoing larger project. The final product is a dual 0-80V 0-10A programmable power source for driving electromagnets. The buck converter preregulator has been discussed quite extensively in this thread:
https://www.eevblog.com/forum/projects/buck-converter-pre-regulator/

Originally, this was to be powered from an unregulated half-bridge converter. I'm pivoting a little to make this board more useful and more interesting to design. It is now being configured as a regulated current-mode push-pull converter.

Ultimately I decided that a push-pull converter made the most sense from a cost, complexity and size perspective. While it does require an additional winding on the transformer, it eliminates the need for bootstrap gate drivers. A half bridge would require a high-side current sensor, while a full-bridge would require an extra set of switching devices. The EMI filter and bridge rectifier will be located on a separate board.

I'm using more or less the equivalent of a FT-290-77, with a core area of right around 2 cm^2. With an input voltage of 175 V with a 75 kHz switching frequency and B_max = 1400 gauss, I should be okay with 20 turns. I could almost certainly get away with a smaller core (wind more turns, etc) and have acceptable losses, but I happen to have these cores on hand, and it should give me flexibility to use these boards for other purposes.

L2 was calculated to be about 60 uH for a 66% ripple fraction at 75 kHz and 80V output with a 1:1 transformer. Some of these values may change, but at this time I'm considering a T157-2 core wound with about 60 turns. I need to do some more calculations before I feel comfortable with this solution, however.

On the secondary side, I'm using a TL431 and PC817 for the error amplifier and isolator. I've got a few different feedback divider options for increased flexibility. Since it only added three parts, I included an option to use this board alone as a preregulator. Not sure if I'll use it or not, but it could be a useful option to have.

Because this should be somewhat adjustable, and because I already have control power supplies in this system, the primary and secondary side controls are powered externally.

[T3sl4co1l]

You may want an RCD rate or clamp snubber on the drains.  When one side turns on, the other side shoots up by transformer action, and the leakage between halves resonates with the far side capacitance, significantly increasing peak voltage.

Keep leakage low by winding pairs, or quads, or even multiple pairs/quads, in parallel.  That is, wind the two primary halves as a twisted pair, and connect them in series outside.  Or if quad, use opposite pairs in parallel.  Likewise, keep the secondary closely coupled, preferably in the same way, making series-parallel connections of pairs with primary windings.

The peak voltage is a concern, as it's already pretty high in the push-pull configuration.  What's +VBUS anyway -- is this offline?  Half bridge is probably better, and current sense can be had by current transformer into rectifier.

Push-pull is typically chosen at low voltages, where the trade for less current per switch is more favorable.  It's okay I guess at modest power levels and 120VAC input, but is kind of awful for universal or 240V input, especially at low power where winding capacitance dominates instead of leakage inductance.

Tim

[Benta]

Nice design, Thanks for showing it.
At 1 kW, I'd probably have gone towards a half-bridge design to keep the magnetics (transformer) smaller and simpler. The high-side drivers are no big issue today.
But it's a balanced decision, and if this is right for you, great.

[H713]

Interesting point on the drain snubber, that makes perfect sense.

VBUS will be about 175 V, probably sagging a bit under load. This thing isn't a universal device and thus won't be run at 240. Also, I just noticed that I still have the switches labeled as IRFP250s. Those were supposed to be a IRFP460s, but I neglected to change the part number while drawing the schematic.

[T3sl4co1l]

Keep in mind the high capacitance of those bad old devices.  I guess you're running at a fairly low frequency, so that's fine, but significant performance gains are available with newer parts.

Tim

[H713]

I haven't totally settled on the switching devices. Probably going to depend somewhat on what is in stock when I order parts for this one.

I'm sure I'll quickly discover why push-pull is so uncommon for offline converters (I can't remember the last time I saw one), but from my perspective (knowing this is a 120 V only device) the idea of 20 or 25 turns on a core doesn't sound so bad, and it saves the cost and board space of high-side gate drivers and potentially high-side current sensing.

[profdc9]

I like the simple design.  I think it could be a good design for educational purposes as well.  It's very easy to overengineer such things.  May it serve you well.

[Zero999]

Another disadvantage of this arrangement is the switching transistors need to handle double the voltage, since the autotransformer action of the primary doubles the voltage, when the device is off. A half bridge could use transistors rated to 250V, rather than 500V.

[Benta]

Not only the transistors, but also the transform, which means careful winding planning is needed.