Head-Fi.org › Forums › Equipment Forums › Sound Science › Measuring Amp frequency response with free software!
New Posts  All Forums:Forum Nav:

Measuring Amp frequency response with free software! - Page 2

post #16 of 25

Hi, I am a new member looking for information re measuring the freq response of T amps. These amps are internally bridged in each channel so that they do not have a connection to chassis ground on their outputs. Their outputs float independent of chassis ground. This can cause havoc to your sound card and computer. Any suggestions on how to get the job done safely? 

 

Thanks

post #17 of 25

Double post...Answered here:

http://www.head-fi.org/t/668485/measuring-a-t-amp-through-a-sound-card-new-member#post_9534110

post #18 of 25

What's the corner frequency for the anti-aliasing filter on a run-of-the-mill sound card? I wouldn't trust any measurement in the neighborhood of the Nyquist frequency....

 

Cheers!

post #19 of 25
Quote:
Originally Posted by ab initio View Post

What's the corner frequency for the anti-aliasing filter on a run-of-the-mill sound card? I wouldn't trust any measurement in the neighborhood of the Nyquist frequency....

 

Cheers!

Changes with sampling frequency.  But neither would I, though today's filters seem darn good.

post #20 of 25
Quote:
Originally Posted by jaddie View Post

Changes with sampling frequency.  But neither would I, though today's filters seem darn good.

 

Yes, but the filter (of a properly designed AtoD) has to be in full effect by the Nyquist frequency to avoid aliasing, so there will be nonunity frequency response  in some region before the Nyquist frequency. It's the same reason digital storage oscilloscopes might sample at something like 1 GHz, but the actual analog bandwidth will typically be some factor of 5--10 less than this.

post #21 of 25

Yeah, fine.  Point?

 

24/192K should be good enough for any audio measurements, assuming it's close to a real 24bits. 24/96 is probably good enough for most.

post #22 of 25
Quote:
Originally Posted by jaddie View Post

Yeah, fine.  Point?

 

24/192K should be good enough for any audio measurements, assuming it's close to a real 24bits. 24/96 is probably good enough for most.

 

My point is in regards to the measured frequency plots here:

 

Quote:
Originally Posted by chewy4 View Post

Asgard to Titanium HD line in (using monoprice Y cable):

HD Line out to line in:

 

These plots look like the soundcard has a chebyshev filter analog section prior to the ADC with a corner frequency somewhere before 20kHz. I think the ripples and frequency dive prior to 20kHz reflects the limitations of the soundcard hardware/implementation, and not the software, etc., as was being discussed earlier in the thread.

 

I completely agree with you that 24/192K should be good enough for any audio measurements, especially if you are benchmarking the performance of a DAC or amp that is being fed from a 16 or 24/44.1kHz data stream.

 

Cheers!


Edited by ab initio - 6/18/13 at 12:47am
post #23 of 25
Quote:

Originally Posted by ab initio View Post

 

These plots look like the soundcard has a chebyshev filter analog section prior to the ADC with a corner frequency somewhere before 20kHz. I think the ripples and frequency dive prior to 20kHz reflects the limitations of the soundcard hardware/implementation, and not the software, etc., as was being discussed earlier in the thread.

 

I think it is more likely to be a software problem. The frequency response shown on the graphs would be unacceptably poor performance for the Titanium HD; even my Realtek ALC887 onboard codec does not have ripples like that. On the other hand, Windows sound card drivers do tend to be buggy, and it is not always easy to avoid (low quality) sample rate conversion by the drivers or Windows. Another possibility is that the sample rate is not natively supported by the hardware (maybe the ADC is limited to 48 and 96 kHz), and resampling was needed for that reason.

 

A very steep analog lowpass filter is not needed with a decent delta-sigma ADC chip, such as the one used on the Titanium HD (which is a high quality TI model if I recall correctly), and the ADC chip itself is not likely to have such a poor digital filter either. Also, there is not much point measuring at 44.1 kHz when even onboard HDA codecs support 96 and 192 kHz.

 

In any case, for the purpose of amplifier frequency response measurements, the frequency response of the sound card can be measured separately, and the measured response of the amplifier can be compensated with an inverse filter.


Edited by stv014 - 6/18/13 at 1:35am
post #24 of 25

Yeah I wasn't really too concerned about attempting to get rid of the ripple/rolloff there since I was only trying to measure the Asgard. I'm not sure what sample rate I used but it was probably only 24/44.1kHz, or 24/48kHz at the most. The soundcard already has professionally done measurements by the manufacturer and I haven't heard any flaws so I wasn't concerned.

 

If you're interested I can re-do the measurements using different parameters.

post #25 of 25
Quote:
Originally Posted by ab initio View Post

 

My point is in regards to the measured frequency plots here:

 

These plots look like the soundcard has a chebyshev filter analog section prior to the ADC with a corner frequency somewhere before 20kHz. I think the ripples and frequency dive prior to 20kHz reflects the limitations of the soundcard hardware/implementation, and not the software, etc., as was being discussed earlier in the thread.

 

I completely agree with you that 24/192K should be good enough for any audio measurements, especially if you are benchmarking the performance of a DAC or amp that is being fed from a 16 or 24/44.1kHz data stream.

 

Cheers!

Yeah, those huge ripples in the response would be an issue and severely limit your measurement ability....but not if you're measuring speakers or any other transducer.   Some software has ways for you calibrate out system limitations  (TrueRTA for one, but not free). 

New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Sound Science
Head-Fi.org › Forums › Equipment Forums › Sound Science › Measuring Amp frequency response with free software!