Yes, I am familiar with distortion measurements using added resistor to increase resolution etc. That doesn't change fact that it would be helpful to see the peak value in your data. But we all know where this is going to go... you are not gonna show anything I ask for, simply because. So no matter. Just saying your data is hard to understand or appreciate its depth without a reference value. You are not doing yourself any favors by hiding the peak.
I see that this is going to go nowhere, so let's just stop this OT discussion. Frankly, I have no energy or desire to continue asking for blood from a stone. Just not going to get it no matter how hard I try.
It's presumptuous statements like this that make your arrogance obvious.
I see that this is going to go nowhere, so let's just stop this OT discussion. Frankly, I have no energy or desire to continue asking for blood from a stone. Just not going to get it no matter how hard I try.
It's presumptuous statements like this that make your arrogance obvious.
Last edited:
This is not so. There is a very long thread here on diyaudio regarding design of a low distortion audio oscillator and notching the fundamental is a specific technique used when after very low level harmonic measurement.
If I remember correctly, this is Dimitriy Danyuk, author of that "supply bootstrap" article - http://www.diyaudio.com/forums/members/dimitri.html
He is very nice person. Always glad to help and answer technical questions.
He is very nice person. Always glad to help and answer technical questions.
Thanks for the link Gaza. I have never had to measure ultralow distortion before. But notice that the measurements by TI for 1622 don’t appear to use a notch as they show fundamental to 0dB. If a notch is applied, it’s level of attenuation should be stated with the measurement if the fundamental is suppressed or not shown. A brief description of the measurement chain should also be stated.
Well, you should be automatically suspicious of distortion graphs that have harmonics below -130 dB and a full scale fundamental, because there are not many pieces of equipment capable of this without it being a test of the analyzer's own distortion.
I am not saying TI's measurements are not correct, just that plugging stuff into even an AP and clicking a button may give misleading results when you are down that far.
The SR1's spec says THD+N is -112dB. Maybe Shaq can describe his specific measurement chain that allows measurement of -165dB levels.
Hi xrk971,
Averaging is used to reduce the noise floor in combination with a notch filter. Some notch filters give amazing notch depths, but this isn't something you just walk up to your bench and do. Read through the threads on low distortion oscillators, the one by Frex is a good example and they go through what steps they had to take.
-Chris
Averaging is used to reduce the noise floor in combination with a notch filter. Some notch filters give amazing notch depths, but this isn't something you just walk up to your bench and do. Read through the threads on low distortion oscillators, the one by Frex is a good example and they go through what steps they had to take.
-Chris
A wide FFT also splits the 'N' component of 'THD+N' into a number of smaller bins, and therefore reduces the level of noise compared to the distortion products or signal levels. If you can use a wide enough FFT, then you can reduce the broadband noise levels by 30-40dB or more, allowing very small signals to be resolved within the noise floor.
Beside that, if system allows, there is also synchronous vector fft averaging available. It is mathematically equivalent to calculating only each N-th bin of N times longer fft. In other words it is reduces bin width without increasing number of bins, so it can play tricks on you if something is done wrong. But overall it lets to dig so much deeper in the noise.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.