DIY Bookshelf Speaker


Sharing my current build in progress.

It’s a 2 way bookshelf speakers using Scan Speak drivers from the Discovery series. Not the most premium range, but I understand are very competent in their category range.

I bought over these drivers from a fellow diy kaki’s collection. To have made it past his selection criteria means it is quite special.



The cabinet is from Taobao. I was browsing and came across it. Seems perfect - I have been wanting to do a build with nicer drivers, with chamfered baffle for diffraction mitigation.

The cabinet came with a single shelf bracing between the T and W. However, it is a large bookshelf, 20L and there is definitely work to be done if this speaker is to reach its potential.

First I added bracing with some damping tape in the woofer area. A bit messy with the acrylic sealant as adhesive, due to tight access via woofer hole only.

Next, I wanted to try just CLD (constrained layer damping) without bracing, on the cab around the TW area.

Prepped some ply with the damping tape.

Stick on internally on the top, sides and back walls around the TW.

Most speaker cabinets can be modified this way, access via the driver holes.

1 Like

Drivers installed. Nothing special here. Apply gasket tape. Drill and screw. Connect wires.

Looks pretty sharp!

Next, taking impedance measurements with Dayton DATS, for crossover modeling.

Next, we need to measure driver response in the cabinet, for crossover modeling. This is where ‘anechoic’ measurements would be ideal. However, we don’t all have access to a chamber…
Given that this is a more serious build, I tried my best to get cleaner measurements.

Pardon the mess… :sweat_smile:
I used the ottoman to manage the floor reflection. It will be the closest source of reflection. The next closest is either the sidewalls, or the wall behind the mic, where I have placed a couple of panels to try and mitigate slightly.

You’ve been busy :smiley: :+1: :+1:
Simulation using Scanspeak provided data is close enough for me.

This is also the first time I’m taking so many angles for the crossover simulation. Will try to sim with a different too that can calculate the spinorama curves.

Taking measurement off axis.

As I was saying, I’m pleasantly surprised how much ‘reflection free’ (edit : technically this is 2nd reflection, the tough one would be to get the first reflection out as far as possible, not 1.7ms) time I had on the measurement.
There is one blip at 1.7ms after first arrival, 23dB down. The next significant reflection is actually all the way out to 11ms.

A bit of computer aided voodoo, and we have a prototype crossover ready for listening test.