[3ll3d00d]

Given an RTA view of a CEA-2010 burst like this

when calculating distortion, do we measure the distance from peak to peak or from the peak of the 2nd harmonic to the flat line for the CEA-2010 overlay?

the former is ~30dB (~68 to ~98dB) = ~3.2%
the latter is ~27dB (~68 to 95dB) = ~4.5%

Is the calculated distortion of 1.74% (35dB) irrelevant because this isn't a sine wave?

Thanks
Matt

 

[EarlK]

That'll need to be an answer that comes from "John".

:sn:

 

[JohnM]

The CEA-2010 signal isn't very useful for measuring distortion, the envelope shaping of the signal broadens the spectral peak and the peak of the captured spectrum is often not at the centre frequency of the CEA-2010 signal. Better to use a normal measurement sweep or sine waves for distortion measurement.

 

[3ll3d00d]

The CEA-2010 signal isn't very useful for measuring distortion, the envelope shaping of the signal broadens the spectral peak and the peak of the captured spectrum is often not at the centre frequency of the CEA-2010 signal. Better to use a normal measurement sweep or sine waves for distortion measurement.

I was using it as it seems safer than cranking out 110dB sine waves. I'm only really looking for orders of magnitude here not precise values, i.e. the point at which distortion starts to rise rapidly. Is this a bad way to do this?

 

[JohnM]

No, it's OK, though sweeps are better if you need to know the actual levels of distortion.

 

[3ll3d00d]

No, it's OK, though sweeps are better if you need to know the actual levels of distortion.

Ok thanks. Does sweep length make a meaningful difference here? I understand this affects SNR so is a 128k sweep (ie as short as possible) ok?

 

[JohnM]

Depends how low the distortion is. If it is at or near the noise floor then a longer sweep would be needed to distinguish distortion from noise, but if you are just looking to see where distortion gets into the percent range then 128k sweeps are fine.

 

[prattwd]

Has anyone tried using the burst test at frequencies above say 300 Hz? I was looking to see if I could use the test to help get a feel for the headroom in my system across all frequencies and I ran into some problems.

Everything worked as it should while testing frequencies below around 200 Hz. When I started going higher I noticed that each pulse (I had the generator set to loop the bursts) started to sound slightly different and the spectrum reflected that. By the time I got to 1,000 Hz, some of the pulses weren't even audible. Some sounded horrible and some sounded right on.

I tried lowering the volume to make sure I wasn't distorting with a high volume setting and it made no difference. The speakers are brand new and sound fine during sine sweeps that cover the same frequencies. I'm wondering if it's an issue with the algorithm or with my sound card. I was using the HDMI connection to my AVR and a UMIK-1 microphone both connected to my laptop.

Any thoughts? Anyone else tried this?

 

[JohnM]

I've run the CEA-2010 test signal at frequencies up to 10 kHz and with sample rates from 44.1k to 192k, it stayed well behaved at all settings. Maybe worth checking REW is set to the same sample rate as the audio interface, or just try different sample rates in REW.

 

[prattwd]

Thanks. I'll look into it. Could it have something to do with the HDMI interface? Have you tested that interface with the burst tests?