REW Beta Release Level sweep feature request

[Clayton Williamson]

REW has excellent functionality in terms of setting a level and sweeping through a wide range of frequencies to determine and improve the behavior of the system. A valuable addition would be a level sweep, where the frequency is fixed and the output is swept through a wide range of levels. This can reveal a lot about the linearity of the system, and how gracefully (or not!) it handles large signals. Thanks for your consideration!

 

[John Mulcahy]

Stepped sine allows stepping in frequency or level.

 

[Clayton Williamson]

Whoa, that's great! I never noticed that option. I'll test it out now. Thanks John!

 

[Clayton Williamson]

When using the stepped level sweep, the distortion reads unexpectedly low. For example, with a -10dBFS 1kHz output, I measure this clearly distorted output from the device I am testing..

But, the THD reading is roughly 0.2%. The distortion values are dominated by the odd harmonics as expected, but this looks like 20%+ THD, so I'm not sure what I'm doing wrong.

 

[John Mulcahy]

What are your distortion settings (particularly the distortion low pass setting) and how does that signal look on the RTA?

 

[Clayton Williamson]

Here are the distortion settings, which look reasonable to me.

Here is the spectrum, showing the artifacts of the heavy clipping, especially the odd order harmonics, yet a vanishingly small THD.

The input frequency to the device via REW generator is 1kHz, though the FFT peaks show 1.024k and multiples, which is odd and may be a factor in the odd readings.

 

[John Mulcahy]

That shift in frequency will be the reason for the low THD, since REW uses the generator frequency as the fundamental and looks for harmonics at its multiples. That's a very big frequency shift, far outside the tolerance of even the cheapest oscillators, so difficult to understand how it could occur. What is the whole signal path?

 

[Clayton Williamson]

OK, thanks. That makes sense. The signal is from a BTLE wireless device. The problem appears either in the LC3 codec or the ASRC -- I'm working to track that down now. Thanks for your insight!