Speakerbench...

Jump to Latest
#2
Underappreciated post.

Thanks for drawing our attention to this Wolf, and thanks to JCandy for the work building it. We need more content like this, a little sad to see it go unremarked.

I look forward to seeing comparative experiences between this new method and what say, DATS uses as standard.
 
#4
This is another awesome (and free!) addition to the DIY speaker building community.

They recommend against using DATS and REW however since they cannot do stepped-sine signal measurement under steady-state conditions - bummer....!
 
#5
InOtIn said:
This is another awesome (and free!) addition to the DIY speaker building community.

They recommend against using DATS and REW however since they cannot do stepped-sine signal measurement under steady-state conditions - bummer....!
Click to expand...
If you read a little closer you'll see they recommend not using fast (i.e. short) sweeps. I captured results for a range of sweep lengths and shared the outcome with Claus and Jeff, every sweep longer than 5 seconds gave essentially identical results. REW already indicates if a sweep is too short for an impedance measurement. More important for sweep measurements is the susceptibility to external noises, REW has a special noise filter mode to reduce that influence, there is more about it in the help, copied below for convenience.

Impedance Measurement Quality

The main source of measurement noise is acoustic noise and vibration during the measurement. Loudspeakers act as microphones, generating small voltages in response to sounds and vibrations that are picked up as part of the load voltage. To minimise this effect use long sweeps, low sense resistor values, avoid noisy environments and isolate the loudspeaker from vibration. Using a power amplifier to drive the speaker provides a much lower drive impedance, which reduces the effect of noise, and allows a low series resistor to be used.

REW provides a Noise filter option which applies a bandpass filter to the captured sweep, synchronised to the sweep frequency, to reduce the influence of noise. The High setting is suitable in most circumstances, but it may have a slight smoothing effect on very sharp resonances. If that is observed try the Medium or Low settings, or turn the filter off. When the filter is being used the sweep duration should be at least 10 seconds, a warning is shown if a shorter sweep is selected. Longer sweeps improve signal to noise ratios in the results with or without the noise filter.
 
#6
I would like to mention that Claus and I have added a low-frequency box-modeling capability to Speakerbench. We hope this will give the web app a more general appeal.

The original goal of Speakerbench was to enable the user to upload measurement data in order to calculate the advanced model parameters (that account for suspension viscoelasticity and motor semi-inductance not included in the traditional Thiele/Small theory). In addition to the model-parameter fit capability, we now include a box model tool that will be soon be significantly expanded to include legacy Thiele/Small modeling capability.

Because the web app and associated workflow are potentially confusing, Claus has produced a series of YouTube videos that explain and illustrate each of the steps:

Speakerbench step 1 - 4 - YouTube

We hope you'll give it a try and send feedback.
 
You must log in or register to reply here.
Top Bottom