NBitWonder

Class D Amplifier Assembled!

by on May.17, 2011, under Class-D Amplifier, Projects

The single-supply bridged version of the Class-D amplifier has been assembled, and I am most pleased to report that it works beautifully!

Input side of the amp

 

Output side of the amp

For those new to the project, a Class-D amplifier is a highly efficient amplifier that operates its output devices as switches instead of as variable resistors.  We posted a highly experimental version of this project last November, and it met with a surprising amount of positive feedback from the DIY community.  Therefore, we developed it as a formal project.

The amplifier has been converted into a single-supply version, which removes quite a bit of complexity from its power supply.  The ideal power supply used to be a custom-built bipolar supply capable of high currents, but now, an old laptop power brick that supplies 20V at a few amps is perfect.

A short demonstration video follows.  For more technical information, be sure to check out the github repository that contains a wealth of information on the development of this project.

Still remaining:  packaging the project.  I wanted to verify that everything worked before taking the time to stuff everything into a box.

:,

5 Comments for this entry

  • Vinicius Cordeiro

    Very nice project! The 60Hz hum is annoying and need to be solved but it’s a very promising project, anyway.

  • Alvie

    Hi,

    How fast can your drivers+mosfet switch ? Can they handle a sigmadelta output at about 5MHz ?

    Alvie

  • Ben

    Thanks for the comments!

    @Vinicius: You are right, the 60Hz hum needs to be solved. I think the cause is the poor power supply I used to test. If I replaced it with a regulated laptop power brick, the hum should be gone.

    @Alvie: I think 5MHz would be pushing it. 1MHz is probably more reasonable. The gate drivers have a 25ns rise/fall time. With their 2A source/sink capability and the MOSFETs 21nC gate charge, changing output states takes 21ns. One cycle is two of each of these time periods, or 92ns. So, if you didn’t give the output inductor time to charge at all, you could hit about 10MHz. To actually get an output, you’d want to clock it much lower than that.

  • van cong thuat

    link down

  • George

    @vang: All the links seem to be up, try again?

5 Trackbacks / Pingbacks for this entry

Leave a Reply

 

Google thinks you would like ...

-->