| Columns Retired Columns & Blogs |
May 12, 2007 - 12:57pm
#1
Class D HF Component
| Loudspeakers Amplification Digital Sources | |
Analog Sources Accessories Featured | |
Music Columns Retired Columns | |
Show Reports | |
Features Latest News Community | |
Resources Subscriptions |
© 2024 StereophileAVTech Media Americas Inc., USA
All rights reserved
I cannot answer you directly with regard to "clipping vs ultrasonic" so you will have to wait and see if JA or one of the more capable persons around here weighs in.
What I have just read is that a negative-feedback loop, which includes the output filter or similar functional circuit design is a requirement in all switching amps. So what amp designers do is place an output filter in the negative feedback loop to minimize the tendency for the ultrasonic frequency response to vary widely with load. In this way designers only have to compensate for it within a narrow range of values. What seems to be happening in one of the designs I have linked to is that by the time the noise comes to being output the frequency is being rolled off.
The following documents may be of interest to you. Search for the term ultrasonic:
http://www.stereophile.com/solidpoweramps/805cia/index4.html
http://www.hometheaterhifi.com/volume_9_2/ps-audio-hca-2-power-amplifier-6-2002.html
http://www.crownaudio.com/pdf/amps/136634.pdf
Thanks for the input. I was suprised to read the comment by John Levreault of PS Audio. I assumed that the low pass filter was just tacked on to the output.
Just as I typed that last sentence I had second thoughts about John's statements. There has to be an output filter because the output drivers are operating as switching devices and their switching frequency should be removed.
I had never thought about feedback in a switching amplifier. Being digital I didn't think that feedback was needed or would even work. But the input being analog could use output feedback. The output being PCM Digital would have to be filtered back to analog in order to be used at the analog input. How many ways could this stuff go badly?
By the way, how does a class D amplifier go into clipping since the output devices operate in saturation 50% of the time anyway? Must have something to do with max or min pulse width, but at the momment I can't visualize it.