Another Cuk build

This is actually a rebuild of a module I did a couple years ago. It's based on my discrete flashlight PCB, which is a simple average current mode boost controller.

The original:

formatting link
formatting link
formatting link
The control board is mounted with 3M VHB tape, in an enclosure featuring two holes, one for the trimpot (0-20V range) and one for the switch (which handles control power only -- the power stage is always under bias).

formatting link
Didn't last forever, though -- the output side filter cap, the lead actually

*fused*! The resulting ripple fried the power stage, and I shelved it for a while. Considered replacing the transistors and gate driver, and the caps with polymers. Decided against it, as the build is just too damn awkward to service. Biggest offender being the inductor, which is neatly held to the board with its eight 16AWG copper leads.

So I rebuilt it in a slightly larger enclosure instead:

formatting link
formatting link

Of note, the filter caps are doubled up. They shouldn't be having any problems. There's also some ceramics in there for extra help.

Instead of TO-220 regulators, I used SOT-89s, one on the gate driver board (still a loose board in this build, hence the insulation) and one on the controller. These give 12V for the driver and 5V for the control.

The power board (not pictured, CBA to remove the top board for a pic) is on the bottom. First, a layer of copper tape provides a little heat spreading along the enclosure (which is only ABS plasic, so even this modest effort will make a difference thermally). Then a Gap Pad (20 thou) between the power board and foil. The power board is 0.8mm copper clad stock, with two D2PAK transistors and one D2PAK dual schottky. The coupling caps are 4 x

4.7uF 1206. The current sense resistor is in series with the transistor sources and the diode cathode, a pair of 5mohm 3637 chip resistors (actually skookum Vishay Foil parts, just because I had them leftover).

The switching loop is very tight, about a cm across. In fact, because the winding is doubled up ("star quad" arrangement), I decided to wire it as two stages in parallel.

So, think of the transformer as a four-lead cable with ferrite beads on it. Except it's wound around a powdered iron, for energy storage, but it works the same for AC purposes. That means no commutation currents flow up or down or within the cable. I have the cable starts (at AC ground) tied to the input and output nodes, which are bypassed to ground on the upper connector board. One pair to the input, one pair to the output. On the switching board, I have one pair to the diode anode, simply wired in parallel. The other pair is not wired in parallel, but actually routed to each transistor, with its own separate coupling caps. This is notable and weird, but perfectly equivalent to having them explicitly in parallel (there should be no currents flowing between the two input strands). The reason this was done, is because there was no layout room for an ordinary connection. (The transistors are sitting in the space between mounting holes -- the board is carved to just fit inside the enclosure.)

The top connector board has three pads on its bottom -- chunks of copper clad soldered on, just to get the right height over the power board to hold it in place.

So the top board is screwed down, and the screw bosses happen to be just tall enough that: a thermal pad, thinnish PCB, D2PAK, and some bits of average PCB, all stack up to just perfectly provide clamping force holding the whole stack down.

The control board is again mounted with double-stick tape, and I made the hookup wires longer this time, and with some strain relief, so they don't break off as frustratingly often as they did last time. (It's just solid

28AWG wire wrap wire.)

Power stage schematic: doesn't matter, boring and stock. As described above, which, I realize descriptions aren't always effective. You'll have to take my... word on it. *Cough*.

Schematic -- controller:

formatting link
This is for the CC output boost version. The changes necessary to adapt to a Cuk output stage are left as an exercise for the reader. :^)

The original power stage is here:

formatting link
Since it's boost, the current sense can be low side, at the battery. Which means it's negative going, hence the negative sign in the controller.

Only one trace cut was necessary, VREF2 (to insert a resistor divider; ILED is grounded, while the new node VREF2' has the correct feedback phase); everything else was done by DNPs, or partially placed components and bodge wires.

Well, I cut some ground plane to wrangle the regulator in there, but that's neither here nor there.

Oh, note this has thermal and under- and over-voltage protection -- well, the overvoltage isn't needed, it's voltage mode feedback (VREF2' comes from an output sense divider), but under- is implemented, and thermal is done by placing the thermistor on the power board, tucked in beside the diode.

Performance?

Efficiency is high. Transistors switch cleanly, very little overshoot, and with ~4mohm Rds(on), conduction losses are low. Can't expect too much of the powdered iron, but that's alright, it deserves to be beaten up. This isn't designed for continuous duty anyway.

So it's not bad at all, for a 200W converter that fits comfortably in the palm of your hand.

I really just wanted it to 1. spend a weekend building something, that 2. powers my cordless electric drill. :-)

Which... started out as a HF special, 18V, NiCd(!) drill. NiCd pack died, because of course it will. I put a cable on it so I can run it from the bench supply. It has become a corded cordless drill. Now I can run it from a small 12V battery. It's a battery powered cordless corded cordless drill.

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Design 
 Click to see the full signature
Reply to
Tim Williams
Loading thread data ...

Got pics of it running?

Reply to
bitrex

It doesn't have legs, sorry.

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Design 
 Click to see the full signature
Reply to
Tim Williams

seems like a design oversight

Reply to
bitrex

Does this help?

formatting link

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Design 
 Click to see the full signature
Reply to
Tim Williams

Perfect except for the lack of Yakety Sax!!

Reply to
bitrex

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.