Shop our Airplanes Products Drone Products Sales
Thread Tools
Dec 24, 2006, 01:57 AM
now that's a wattmeter...
simingx's Avatar
Build Log

DIY 250W (12V 20A) Switchmode PSU

First of all, I would like to state that any project of this type is absolutely NOT recommended to be built by anyone who is not familiar with the extreme danger of direct mains-connected equipment!!

"Advanced" test equipment such as an oscilloscope and (preferably) a signal generator are of absolute necessity because of the need to verify that the proper waveforms are being generated in the circuit! Failure to test the circuit throughly before connection to 230V mains will most assuredly result in (at least) blown semiconductors and (at worst) serious injury.

Practical it is not (buying one off the shelf costs less and is way safer) but it's a very good learning experience.

So here goes...

Inspired by the project here, I decided to get my hands dirty by building my very own 250W SMPS.
However, it is not so easy to find the required power supplies with the correct sized transformers (especially the main power transformer), so I decided to wind my own. It's not very difficult and saves you quite a bit of headaches.

First of all, the main power transformer is wound. I used an ETD39 core (pretty standard size and easily available). Primary is 2x20 turns, wound as 2 layers. Instructions follow:
Wind 20 turns of 0.8mm wire around the core. The ETD39 core is quite substantial, so you can probably use thicker wire too.
Bring out a centertap (it won't be used however). Wrap this first layer with some insulating tape (I used 3M Super 33+) and wind a further 20 turns [b]in the same direction[/i] on top of this.
This completes the primary winding. Wrap this with a further 2-3 layers of insulating tape.

For the secondary, it consists of a main winding of 14 turns centertapped and a further 4 turns at each end to drive the semi-synchronous rectifier. It's wound as follows:
First, use some thin wire (0.5mm is more than sufficient) and wind 4 turns around the (insulated) primary. Attach this to 2 pins on the bobbin. Wrap one layer of insulating or plumber's tape.
Next, take 2 strands of 1.0mm wire (in parallel) and wind 7 turns. Bring out a centertap in the form of a pigtail (make a large loop with the wire).
Wind a further 7 turns, all in the same direction. Anchor the 1.0mm winding. This is the main output winding.
Wrap this with some more plumber's tape.
Use the thin wire again and wind another 4 turns in the same direction.

Now, you should have (from inside out):
20 turns of 0.8mm wire
20 turns of 0.8mm wire
Lots of insulation
4 turns of 0.5mm wire
7 turns of 2x1.0mm wire
Insulation and a centertap
7 turns of 2x1.0mm wire
4 turns of 0.5mm wire

The 2 primary windings are connected in series.
The 4 secondary windings are also connected in series (in this order: 1st aux->first main->centertap->second main->second aux).
Make sure ALL windings (at least for the primary and secondary) are in the same direction!!!

To test this transformer, connect a audio signal generator to the primary side.
Adjust for ~50kHz, sine wave.
The output on the secondary side should also be a sine wave, related by the turns ratio. In this case, ratio is 14/40=0.35 for the main secondary wind, and 22/40=0.55 for the main+auxiliary secondary wind.
Using an oscilloscope, measure from centertap to one end of the main secondary winding, make sure the Vpp is correct.
Measure also from centertap to the end of the auxiliary winding and make sure the voltage is HIGHER than what you measure from the main winding. If it is lower, the phasing of the aux. winding is wrong and you will have to swap connections.
That concludes the main transformer.
Last edited by simingx; Dec 24, 2006 at 04:19 AM.
Sign up now
to remove ads between posts
Dec 24, 2006, 02:32 AM
now that's a wattmeter...
simingx's Avatar
Next, the driver and current sensing transformers are wound.
For these, you will have to raid a PSU for the cores.
They'll be soaked in varnish, so put them in acetone (nail polish remover). Use a water bath to warm up the solvent or it won't be effective.
After about 5 mins, the varnish should be soft enough for you to seperate the core halves. Be very careful, the ferrite is extremely brittle.
If they don't come apart easily, soak them some more.
I broke 4 transformers doing this. Get some extra

Unwind all the wire from the transformers.

First, the driver transformer is wound.
Using 0.3mm wire, wind 26 turns on the core. This will be the primary.
Wrap it with some insulating tape.
Now this is important. Take 2 0.3mm wires and wind 16 turns together (bifiliar wind). Remember which end is start and finish. Seperate out the 2 windings. These are the secondary windings which will drive the FETs.
When its done, wrap some more insulation tape.

This step is critical!
Use a cable tie (steel lock preferred as it doesn't have "steps" in it. Use the cable tie to securely clamp the core back together! I cannot stress the importance of this. If the core is not properly assembled, the drive waveform will be seriously distorted = BANG.

Again, test with a signal generator to ensure correct ratio. Also test for polarity of the secondary windings. If you connect oscilloscope probes to the start and end of each winding in the correct fashion, the waveforms produced will be out of phase.

Next, the current sensing transformer is wound. I hope you have some patience!
Using the 0.3mm wire again, wind 100 turns on the core. Yes, 100 turns.
Bring out a centertap, wrap some plumber's tape and wind another 100 turns. That's 200 turns in total.
I didn't wind them too nicely as it's quite impossible to keep track of so many turns with such thin wire on a tiny core and wind it neatly as well.
Insulate the winding with some electrical tape. Make sure there's a gap left between the winding and the core for the single turn primary. Use a piece of 0.5 or 0.8mm wire for the primary.

The rest of the instructions can be found on the designer's homepage. I leave it to you to read it if you seriously decide to pursue this project.
Once again, make sure ALL testing at low voltage has unconditionally passed BEFORE connecting to 230V mains!
Dec 24, 2006, 02:35 AM
now that's a wattmeter...
simingx's Avatar
The stripboard for the control circuitry and completed circuit during load testing are shown here.
The heatsinks are slightly undersized because I ran out of PCB space, so I had to use a fan to keep the power semiconductors cool.

Don't build it like this though.. use a PCB.
Dec 24, 2006, 04:36 AM
Registered User
Sim, Good effort

Here's the version I made....

complete with new pcb
Last edited by Chippie; Dec 24, 2006 at 04:41 AM.
Dec 24, 2006, 04:39 AM
Registered User
Originally Posted by simingx

Once again, make sure ALL testing at low voltage has unconditionally passed BEFORE connecting to 230V mains!

Connect a 60watt lamp inseries with the mains supply will prevent anything nasty happening when you first fire it up....
Dec 29, 2006, 11:10 AM
now that's a wattmeter...
simingx's Avatar
13.2V @ 18.7A = 247W... everything is forced cooled using a 8cm fan. Burned it in for 5 hours, everything seems OK.
Dec 29, 2006, 02:35 PM
Registered User
VinceHaworth's Avatar
Nice project - but it seems like a lot of effort to go to when you can easily mod a pc power supply that only will only cost you an half hour of your time.
Dec 29, 2006, 06:49 PM
Micro Flyer, Big Poster
Pete P's Avatar
I was going to say the same thing- I have a spare Antec (Nice) 250W and it produces everything I need, 12v -12v 5V and 3.3V in usable amounts, plus I can add voltages when I need to.
Dec 29, 2006, 07:47 PM
now that's a wattmeter...
simingx's Avatar
Of course... Like I said earlier, it's absolutely impractical as a proper PSU is cheaper and safer (I myself have a very nice variable laboratory switchmode PSU which gives me anything from 0-20V at 36 amps), but as a learning experience, well as they say, priceless

An analogy in RC would be like buying an RTF vs. buying the balsa and cutting it yourself

Quick Reply

Thread Tools

Similar Threads
Category Thread Thread Starter Forum Replies Last Post
For Sale 250W 12v Power supply. foss1997 Aircraft - Electric - Batteries & Chargers (FS/W) 0 Apr 11, 2007 10:11 PM