Premium Content

New Signal Chain Resources from Texas Instruments:

Build Your Own Ultra-Low-Cost Isolated DC-DC Converter

Date Posted: April 07, 2010 12:00 AM
Author: Dwayne Reid

When low cost in small volumes using readily available parts is a primary concern, standard approaches to isolated dc-dc conversion may not be suitable. Integrated switchers from Linear Technology, Maxim Integrated Products, National Semiconductor, or Texas Instruments typically do not provide isolated outputs and so require expensive transformers that are hard to acquire in small volumes. Further, they typically require an opto-isolator and associated drive circuitry to provide feedback from the isolated output to the switch-mode controller.

An alternative using discrete components (see the figure) can provide 50 mA at 5 V from a dc input between 8 and 30 V dc. To avoid the need for feedback components, the power stage uses a constant current source design. This allows a shunt, such as the TL431 or even a simple zener clamp, to provide output voltage regulation.

Because the desired output voltage is less than the minimum input voltage, a simple 1:1 output transformer can provide the needed isolation. Low-cost, common-mode chokes can serve as the transformer. They have low capacitance between windings and excellent high-voltage isolation between the windings and cost only about $0.25 in quantities of hundreds.

A dc-dc converter requires a ramp generator and a comparator driving a power stage. Current feedback comes from a resistor in series with the transformer’s primary winding. It’s easy to make a ramp generator from an operational amplifier with some resistors and a capacitor. To save component count, however, the second half of the LM393 dual comparator replaces the op amp. A symmetrical ramp will work just fine, so there is no need for steering diodes. The ramp’s maximum duty cycle is set to about 50% to reduce the chances of saturating T1.

To drive the chosen transformer with current-mode feedback and achieve minimum loss due the current sampling, the ramp generator needs to run at about 400 kHz with an amplitude of about 0.5 V. The exact amplitude and frequency needed depend on the reference. An ordinary red LED has a forward voltage that is reasonably constant over indoor temperature ranges, so it serves as a low-cost reference.

The pulse-width modulation (PWM) comparator does not have the current capacity to drive the transformer directly. Its open-collector output, though, allows the use of a PNP transistor (Q1) and a single resistor (R12) to act as the power stage. The circuit also acts as a high-voltage clamp. When a transient spike exceeds the comparator’s breakdown voltage, the transistor conducts and clamps the spike.

Although the design is completely unorthodox, it works well and Spice simulations match the bread-boarded results. And, it’s extremely low cost. Total parts cost in hundreds of units, including the transformer, is about $0.60.

Part Inventory
Go
powered by:
 

 
You must log on before posting a comment.

Are you a new visitor? Register Here
  • Dwayne Reid
    2 years ago
    Apr 09, 2010

    I've been asked for parametric information on the common-mode choke that this design uses. I'm sorry, but I don't have that information available.

    However, its easy to tweak the oscillator to match most chokes. First, characterize the choke just by feeding it with the square-wave output of a function generator - just to get an idea of what frequency gives the best coupling.

    Then see how it behaves with a high-current pulse input. I use a home-brew test jig for that - basically a large FET in series with one winding. Its fed with a 10.0Vdc supply with several thousand uF of very low ESR capacitors as a reservoir (Sanyo OSCON). Feed the gate with a pulse generator and look at the current waveform with a current probe into my scope. Gradually increase the gate pulse width and watch the current on the scope.

    Its real easy to see where the inductor begins to saturate plus you can see where any non-linearity inflection points occur. Measure the slope before and after each inflection point and calculate the inductance for each region. That's why the supply voltage is set to 10.0Vdc - it makes calculating the inductance easy to do in your head.

    Use the saturation measurement from the above to adjust the duty-cycle of the oscillator so as to avoid saturating the core under heavy load or short-circuit conditions.

    I used Linear Technology's LT Spice (also known as SwitcherCAD) to find out the component values needed for the oscillator. That was strictly a trial-and-error process but LT Spice is so darned quick that it was easier to use the computer to get the oscillator doing what I wanted than it was to find the values using my breadboard. I found that the results that I got from LT Spice matched my breadboard results perfectly.

    dwayne

  • Dwayne Reid
    2 years ago
    Apr 09, 2010

    I've been asked for parametric information on the common-mode choke that this design uses. I'm sorry, but I don't have that information available.

    However, its easy to tweak the oscillator to match most chokes. First, characterize the choke just by feeding it with the square-wave output of a function generator - just to get an idea of what frequency gives the best coupling.

    Then see how it behaves with a high-current pulse input. I use a home-brew test jig for that - basically a large FET in series with one winding. Its fed with a 10.0Vdc supply with several thousand uF of very low ESR capacitors as a reservoir (Sanyo OSCON). Feed the gate with a pulse generator and look at the current waveform with a current probe into my scope. Gradually increase the gate pulse width and watch the current on the scope.

    Its real easy to see where the inductor begins to saturate plus you can see where any non-linearity inflection points occur. Measure the slope before and after each inflection point and calculate the inductance for each region. That's why the supply voltage is set to 10.0Vdc - it makes calculating the inductance easy to do in your head.

    Use the saturation measurement from the above to adjust the duty-cycle of the oscillator so as to avoid saturating the core under heavy load or short-circuit conditions.

    I used Linear Technology's LT Spice (also known as SwitcherCAD) to find out the component values needed for the oscillator. That was strictly a trial-and-error process but LT Spice is so darned quick that it was easier to use the computer to get the oscillator doing what I wanted than it was to find the values using my breadboard. I found that the results that I got from LT Spice matched my breadboard results perfectly.

    dwayne

  • Dwayne Reid
    2 years ago
    Apr 09, 2010

    I've been asked for parametric information on the common-mode choke that this design uses. I'm sorry, but I don't have that information available.

    However, its easy to tweak the oscillator to match most chokes. First, characterize the choke just by feeding it with the square-wave output of a function generator - just to get an idea of what frequency gives the best coupling.

    Then see how it behaves with a high-current pulse input. I use a home-brew test jig for that - basically a large FET in series with one winding. Its fed with a 10.0Vdc supply with several thousand uF of very low ESR capacitors as a reservoir (Sanyo OSCON). Feed the gate with a pulse generator and look at the current waveform with a current probe into my scope. Gradually increase the gate pulse width and watch the current on the scope.

    Its real easy to see where the inductor begins to saturate plus you can see where any non-linearity inflection points occur. Measure the slope before and after each inflection point and calculate the inductance for each region. That's why the supply voltage is set to 10.0Vdc - it makes calculating the inductance easy to do in your head.

    Use the saturation measurement from the above to adjust the duty-cycle of the oscillator so as to avoid saturating the core under heavy load or short-circuit conditions.

    I used Linear Technology's LT Spice (also known as SwitcherCAD) to find out the component values needed for the oscillator. That was strictly a trial-and-error process but LT Spice is so darned quick that it was easier to use the computer to get the oscillator doing what I wanted than it was to find the values using my breadboard. I found that the results that I got from LT Spice matched my breadboard results perfectly.

    dwayne

  • Dwayne Reid
    2 years ago
    Apr 09, 2010

    I've been asked for parametric information on the common-mode choke that this design uses. I'm sorry, but I don't have that information available.

    However, its easy to tweak the oscillator to match most chokes. First, characterize the choke just by feeding it with the square-wave output of a function generator - just to get an idea of what frequency gives the best coupling.

    Then see how it behaves with a high-current pulse input. I use a home-brew test jig for that - basically a large FET in series with one winding. Its fed with a 10.0Vdc supply with several thousand uF of very low ESR capacitors as a reservoir (Sanyo OSCON). Feed the gate with a pulse generator and look at the current waveform with a current probe into my scope. Gradually increase the gate pulse width and watch the current on the scope.

    Its real easy to see where the inductor begins to saturate plus you can see where any non-linearity inflection points occur. Measure the slope before and after each inflection point and calculate the inductance for each region. That's why the supply voltage is set to 10.0Vdc - it makes calculating the inductance easy to do in your head.

    Use the saturation measurement from the above to adjust the duty-cycle of the oscillator so as to avoid saturating the core under heavy load or short-circuit conditions.

    I used Linear Technology's LT Spice (also known as SwitcherCAD) to find out the component values needed for the oscillator. That was strictly a trial-and-error process but LT Spice is so darned quick that it was easier to use the computer to get the oscillator doing what I wanted than it was to find the values using my breadboard. I found that the results that I got from LT Spice matched my breadboard results perfectly.

    dwayne

  • Dwayne Reid
    2 years ago
    Apr 09, 2010

    I've been asked for parametric information on the common-mode choke that this design uses. I'm sorry, but I don't have that information available.

    However, its easy to tweak the oscillator to match most chokes. First, characterize the choke just by feeding it with the square-wave output of a function generator - just to get an idea of what frequency gives the best coupling.

    Then see how it behaves with a high-current pulse input. I use a home-brew test jig for that - basically a large FET in series with one winding. Its fed with a 10.0Vdc supply with several thousand uF of very low ESR capacitors as a reservoir (Sanyo OSCON). Feed the gate with a pulse generator and look at the current waveform with a current probe into my scope. Gradually increase the gate pulse width and watch the current on the scope.

    Its real easy to see where the inductor begins to saturate plus you can see where any non-linearity inflection points occur. Measure the slope before and after each inflection point and calculate the inductance for each region. That's why the supply voltage is set to 10.0Vdc - it makes calculating the inductance easy to do in your head.

    Use the saturation measurement from the above to adjust the duty-cycle of the oscillator so as to avoid saturating the core under heavy load or short-circuit conditions.

    I used Linear Technology's LT Spice (also known as SwitcherCAD) to find out the component values needed for the oscillator. That was strictly a trial-and-error process but LT Spice is so darned quick that it was easier to use the computer to get the oscillator doing what I wanted than it was to find the values using my breadboard. I found that the results that I got from LT Spice matched my breadboard results perfectly.

    dwayne

  • Dwayne Reid
    2 years ago
    Apr 09, 2010

    I've been asked for parametric information on the common-mode choke that this design uses. I'm sorry, but I don't have that information available.

    However, its easy to tweak the oscillator to match most chokes. First, characterize the choke just by feeding it with the square-wave output of a function generator - just to get an idea of what frequency gives the best coupling.

    Then see how it behaves with a high-current pulse input. I use a home-brew test jig for that - basically a large FET in series with one winding. Its fed with a 10.0Vdc supply with several thousand uF of very low ESR capacitors as a reservoir (Sanyo OSCON). Feed the gate with a pulse generator and look at the current waveform with a current probe into my scope. Gradually increase the gate pulse width and watch the current on the scope.

    Its real easy to see where the inductor begins to saturate plus you can see where any non-linearity inflection points occur. Measure the slope before and after each inflection point and calculate the inductance for each region. That's why the supply voltage is set to 10.0Vdc - it makes calculating the inductance easy to do in your head.

    Use the saturation measurement from the above to adjust the duty-cycle of the oscillator so as to avoid saturating the core under heavy load or short-circuit conditions.

    I used Linear Technology's LT Spice (also known as SwitcherCAD) to find out the component values needed for the oscillator. That was strictly a trial-and-error process but LT Spice is so darned quick that it was easier to use the computer to get the oscillator doing what I wanted than it was to find the values using my breadboard. I found that the results that I got from LT Spice matched my breadboard results perfectly.

    dwayne

  • olddogg
    2 years ago
    Apr 08, 2010

    Hi Dwayne. Which common mode choke product did you use? I didn't find a product listing on Digikey or Mouser for "FL601". Thanks.

  • olddogg
    2 years ago
    Apr 08, 2010

    Hi Dwayne. Which common mode choke product did you use? I didn't find a product listing on Digikey or Mouser for "FL601". Thanks.

  • olddogg
    2 years ago
    Apr 08, 2010

    Hi Dwayne. Which common mode choke product did you use? I didn't find a product listing on Digikey or Mouser for "FL601". Thanks.

  • olddogg
    2 years ago
    Apr 08, 2010

    Hi Dwayne. Which common mode choke product did you use? I didn't find a product listing on Digikey or Mouser for "FL601". Thanks.