Home Propeller Head Plaza

Technical and scientific discussion of amps, cables and other topics.

Objective Measurements to Compare Components Within a System

since new topics here in Geek Central seem to be few and far between.......

I was wondering what ideas some people may have relative to quantifying small changes in sound quality, or, less subjectively, system response. Since the topic is complex (and controversial) enough already, let's avoid speaker/room interactions and just discuss measuring electrical signals (the signal may be measured at different places in the signal chain depending on the goal of the measurement), and let's only measure unamplified signals for now to make it even easier.

I'm not interested in discussing DBTs or other testing methods using humans as the detection device. In this thread I'd like people to stay focused on using devices to measure differences in system response. The goal isn't to figure out if 'Component A' sounds better than 'Component B'. The goal is merely to determine that there is or is not a difference between A and B, and hopefully some indication of how large the difference is.

To give everyone a starting point, I'll propose a method that is likely very flawed (please forgive my ignorance if this has been proposed before). With any luck, the process of pointing out the flaws in my proposal will lead to some better ideas, and maybe even a method that eventually becomes reliable and accepted. Keep in mind that I'll be assuming any tests will be conducted scientifically, using the best methods possible to reduce potential errors or inconsistencies - in other words, I'll be omitting some of the details that are necessary to achieve precision and accuracy in the spirit of simplicity.

Start with a high quality digital recorder. Given that this exercise has no budget, we'll use either a DSD recorder or a 24/192 PCM recorder. Hook the recorder up as far upstream in the system as possible so that the recorder is connected to the output of the device in question. Now start recording samples of your reference material - you'll obviously want a digital source of the same format as your recorder, and it should be a complex music signal - no pure sine waves or square waves. Record lots of samples, because I'm sure no two samples of the same clip will be recorded exactly the same due to vibration and a slough of other factors.

Once you have enough samples, you can compare the samples to one another, and also to the source. If there's too much variation, you can stop testing and start thinking of a better way to do this. If you find a fair amount of consistency among your samples, then average the samples together. The difference between the average of your samples and the source is a rough gauge for your system's baseline distortion, including the recording device, of course.

Now, switch out the component and record the same number of samples you did in part 1. Now compare the averages of the two sets of samples. Are the averages different? If so, there's likely a difference in system response. How big is the difference compared to the baseline distortion?

Once again, this has probably been explored before and probably won't work, but why won't this work?

-Pete



This post is made possible by the generous support of people like you and our sponsors:
  Kimber Kable  


Topic - Objective Measurements to Compare Components Within a System - pburant 10:37:07 08/15/03 (138)


You can not post to an archived thread.