Frequency response in relation to load in Class D

[orpheus]

I had a couple of T amps (both went belly up) that I briefly used with my AKG K1000's. They are a unique headphone that is powered by speaker outputs, and present a 120ohm load to the amplifier.

The TA2024 chip in the T amp varied frequency response with load, at least according to the chart in the data sheet.

I've been looking into getting a class D amplifier, but I am concerned that I won't be able to use them with my headphones. (I also use speakers, but I'd like my new amp to work with both).

Do all Class D amplifiers vary frequency response with load? I looked on the data sheets for some of the newer Tripath chips, but they left off the load/frequency response graph that was on the data sheets for earlier chips. This graph showed a vast increase in high frequencies (above 10k) as the load increased.

Nuforce's website has a chart claiming that their amplifier is load invariant but that none of the other manufacturers listed on the chart are load invariant.

I am seriously looking at getting an ICEpower amp, or perhaps building a tripath. Does anyone happen to know if frequency response varies according to load in all of the class D implementations outside of Nuforce?

Thank you for your time-
Aaron.

[fokker]

UcD from Hypex is also load invariant. and this generally holds true for amps with post-filter feedback.

[orpheus]

I'm thinking about the getting an ICEpower based amplifier, do you happen to know if they are load invariant as well?

Thank you for taking the time to respond.

-Aaron.

[phase_accurate]

I'm thinking about the getting an ICEpower based amplifier, do you happen to know if they are load invariant as well?

Since they also use post-filter feedback they are also "load invariant" to a certain degree (look at B&O's datashet). But UcD is better in this respect.


regards

Charles

[koolkid731]

Originally posted by fokker
UcD from Hypex is also load invariant. and this generally holds true for amps with post-filter feedback.

As phae_accurate pointed out, ICEpower has post filter feedback but is still load variant. "Load variance" depends on forward block gain vs. frequency. Typically gain goes lower with frequency in opamp-based designs, whereas discrete designs have higher gain-bandwidth product, therefore more load independent.

In electronics, everything depends on many factors.