DIY 4 isolated 12V 20ma DC/DC convertor

[ali_asadzadeh]

Hi,
I have built this circuit, I'm generating the PWM with a spare timer on my STM32 MCU, also I have used an EE13 ferrite core, I have turned 10 turns of 0.2mm wire on each leg of the primary winding and I have not used any air-gap in the transformer,the primary inductance is around 120uH for each leg, I just wanted to wind 4 secondary windings with 20 turns each so I could make 4 separate  12V 20ma outputs,Now that I'm experimenting with this circuit the output is around 2V and it can only supply around 1mA if I sink more current the output would drop, Also I have changed the winding directions and the output seems very low, the input would take about 250ma  ad It would waste lost of energy with not enough output,

Do you have any Idea?

[SaabFAN]

I'm also building an isolated DC/DC-Converter.

In a completely different power-class, but the priciples are the same.

The biggest problem I'm having is the efficiency.
Through experimentation and by using this tool http://www.dicks-website.eu/coilcalculator/index.html, I realized that the output-winding is as cruicial as the input-winding.
Using too many output-windings causes efficiency to drop dramatically.

Looking at your schematic however, the problem most likely is in the diodes. If a transistor is off, there should be VIN * 2 across it - Transformer with center tap distributes voltages like this: GND - VIN - VIN * 2.
The voltage-spike can be removed by an RC-Snubber and/or by putting the diode in series with a small (1nF or less) capacitor.

Another question: How did you wind the transformer? I managed to screw that up myself.
Best result so far: Using two wires of equal length, solder them together and add a "tap"-wire to the place where they are soldered together, then wind the cable around the core as one large continuous coil.

[Psi]

Note that you have a high risk of blowing up your transistors if the switching ever stops or the frequency is too low.
Personally i would stick a large 5V light bulb in series with the power into the center tap. With no load on the secondary it should not light up much if at all. But if you have a short or you are not switching correctly the lamp will turn on and you wont blow up any of the transistors.  Later on when you get the frequency/efficiency right you can remove the lamp.

I recommend checking that your transistors are still ok and not already dead. It wouldnt take much current to kill a BC337

Check that top transistor, it should be going to ground, but i cant quite tell if you have it going to ground or to the 5V supply rail.

Check your two primary windings are wound in the same direction.
eg the center tap should be a tap, the wire should not stop and start rotating around the coil in the other direction.

The circuit will work with BC337 but you are better off with mosfets. Dont worry for now, you can change them later and get better efficiency once its working.

You either have to do the math and figure out the right frequency for the exact core and number of turns you have. Or just try different turn numbers and scan frequencies until you find the sweet spot.

I would start with no load on the secondary at all, just try to get the primary current as low as possible by changing turns and frequency.

[langwadt]

use the datasheet for something like this to get some design hints, https://www.maximintegrated.com/en/products/power/isolated-power/MAX253.html

[ali_asadzadeh]

SaabFAN thanks for your hints have you done building your own? if so, would youshare your circuit and results?
Psi thanks for your concern, I have used a 200ma current limit on the power supply that would supply the circuit, so if all the things goes wrong, the transistors would not blow.
langwadt thanks for the link