Doubting measurements

The other day I was reviewing some very high quality bookshelf-sized loudspeakers, and decided I’d check their bass extension. What I measured was quite at odds with both my listening impressions and the manufacturer’s specifications. The makers say 50 hertz more or less flat, 42 hertz at -6dB. That was also my sense. Very solid with what it did, but lacking a firm underpinning of deep bass.

But my measurements said flat to about 36 hertz, -6dB at 32 hertz. Which is such an impressive result I immediately doubted my measurements. I’ve recently acquired a new high-end microphone pre-amplifier/ADC (Focusrite Forte) and was quick to blame that.

To check, I tested the bass end of some mid-priced floorstanding loudspeakers and the result was remarkably similar. Could the Forte be doing something weird?

I still had my old FastTrack Pro interface around — Windows 8.1 broke its drivers so it now only works at 44.1kHz sampling, whereas the Forte goes up to 192kHz. It gave the same results. I filled most of the morning using different test tracks, different amplifiers, different digital decoders, different recording and analysis software, and even a different microphone (not a measurement one, but one with reliable bass so I could see if anything was grossly wrong).

Eventually the issue was more or less resolved. Instead of measuring on tweeter axis at one metre, I measured from my listening position which is sort of on-axis, but about 2.7 metres away. The bottom end now went down to 44 hertz and fell away quite quickly. Do’h! That’s what I was hearing!

Regardless, it does seem that the loudspeaker maker is being modest in its bass claims.

Despite all this, my morning’s efforts had left my confidence in my measurement rig a little shaken. This morning, however, I was measuring a different speaker (with a claimed low end extension of 25 hertz!) which gave a weaker than promised result. Indeed, a weaker response than the bookshelf. So I left the microphone in place and changed back to the bookshelf speaker for final confirmation. Yes, the rig is capable of measuring substantially different bass ends, from which I can conclude that it’s probably accurate!

This entry was posted in Audio, Testing. Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *