What could be causing this waterfall to look so uneven?

[audio_tony]

I have a pair of vintage Linn Keilidh speakers. They have a tendency to sound a bit 'boxy' and there is an obvious issue with the midrange I think.

1kHz is sitting some 5 - 6db higher than the rest of the band. But looking at the waterfall plot, there's seems to be a lot of other resonances, and I wondered if cabinet resonances or lack of damping could cause this?

Can anyone offer any ideas?

I have also added the REW mdat file as I know some folk like to see the 'full picture'.

Measurements were taken in my lounge - the speaker was raised off the floor by 50cm. Microphone was 1m from the tweeter - and was centered on the tweeter.

 

[RickS]

Could speculate, but you really need to take some more measurements. Would try using the binding posts to just drive the bass or highs on their own. If you have some way to clamp the cabinet and take another measurement. Some more data will help eliminate and get focus on the likely cause or causes.

 

[Keith_W]

You have been playing with your window settings, haven't you?

First thing I saw when I opened up your graph was an overly smoothed appearance despite 1/12 smoothing, and a dashed line below 300Hz.

This is the reason why: you set your right window too short. It's only 3.29ms! What this means is that REW only analyses 125ms before the impulse, and 3.29ms after the impulse using the Tukey 0.25 window function.

Once I restore the window settings to something more sensible, i.e. 125ms before and 500ms after, we see what looks like a more normal looking frequency response. I applied 1/6 smoothing. NOW we can analyse it.

Without knowing your room dimensions, I don't know where the Schroder frequency is, so I am going to ignore all those giant room modes we see below 300Hz. It's impossible to tell whether that lack of bass is real, or if it's an artefact of microphone positioning. If you want to know, put your mic in the listening position.

There is a dip in the frequency response at about 3kHz. I googled an image of your speaker:

I am not sure if this is set up as a 2 way or a 2.5 way. But I would wager that improper summation between the woofer and the tweeter is the cause of the dip.

2.5 way speakers are also notorious for causing copious amounts of intermodulation distortion because woofer also has to handle the midrange frequencies. That might be why your mids sound a bit "boxy" (whatever that means). A sine wave sweep is not the proper measurement to take if you want to demonstrate this (since low frequencies and midrange frequencies are excited at different times), you need a two tone test or multitone distortion test. I am not sure if REW is able to do this, I have never tried.

As for those waves in your waterfall, you have set the window very short (3.75ms). Anything that shows up in that very short time window is likely to be a cabinet reflection, cabinet resonance, etc.

 

[RickS]

It is normal to change the window settings in REW for speaker measurements (vs room ones).

More testing is needed for other potential resonances but the higher frequency ones can also be mic reflections and so need to ensure your mic stand is smooth and sturdy. Surfaces around the mounting need to minimize diffractions.

 

[audio_tony]

You have been playing with your window settings, haven't you?

The settings I've used were recommended to me by a guy that restores speakers on Youtube.

Room dimensions are 10M x 3M x 2.2H - the speaker was measured about 2M away from the end wall and placed in the center of the room.

I also took measurements with Holm Impulse (As an occasional user, I find the multitude of adjustments / settings in REW to be a bit overwhelming).

By 'boxy' sound - I meant it's almost as if the sound is emanating from a cardboard box at circuit frequencies.

Anything below about 250Hz will be grossly inaccurate in the graphs below.

Coincidentally, the speakers pictured above are identical to mine (the wood finish). They are a two way speaker (I have attached a schematic of the crossover - I meant to do that in my OP - must have forgot!).

 

[IngMishael]

By 'boxy' sound - I meant it's almost as if the sound is emanating from a cardboard box at circuit frequencies.

it sounds boxy because your speaker's overall macro tonality leans boxy because of the overly recessed presence region and the overly present upper midrange. that leads to cuppy and boxy presentation. also the waterfall shows lingering in that region too

 

[Azrael]

According to this page, the Linn Keilidh is a 2.5-way speaker. It’s also already 34 years old. I’d take a look at the crossover and, if it contains electrolytic capacitors, check whether they still hold their rated capacitance.

In addition, the damping material inside the speaker may have crumbled after such a long time, which would increase cabinet resonances.

If the tweeter contains ferrofluid, it may have dried out, although the frequency response in the tweeter’s presumed frequency range does not necessarily indicate this.