[Build Log] High Quality PC Speakers Design and Build

I've decided I want to have a go at designing and building a pair (or more) of PC speakers.

As I'll go though all the trouble I might as well aim to build something good. I do not want them too bulky, so they will probably need to be supplemented by a sub in the lowest octaves.

I would like them to be active speakers, possibly with built in volume control and preferably something that will run directly off the PC's 12V supply.

I have a few ideas and some, hopefully, suitable drivers lying around. I am going for a 2 way design, although I do have some small full range drivers as well.

Small speakers are not very efficient, but still an 8-10W amplifier should be more than enough.

These are the drivers I will try to integrate, first a tiny 3/4" Vifa ring radiator tweeter (OT19NC00-4) I've did some waveguide tests with this tweeter a long time ago, it it looked pretty good, but I haven't decided on whether to use wave guide or not.




And a small 4" woofer from Dayton (RS100-8)





I got these drivers reasonably cheap during a sale at partsexpress.

I've designed and built a few speakers in the past, and in my experience the manufacturer data sheets for drivers are, put politely , a bit optimistic, with some exceptions. Besides I've found it's not possible to build a good speaker without doing measurements, listening is good, but much better to do measurements first, then listen and fine tune. Not that I want to get into the whole discussion of sound quality, but there are some things that can easily be measured and when done correctly provide a good starting point.

Anyway, plenty of measurements and simulations to come

For the amplifier, I'm thinking class D is a good choice as they are very efficient, can sound good and generate little heat. It will be possibly to get 10W out of a class D amp on a single 12V supply.

I have a few Tripath based amplifiers lying around, but I would like to build one from scratch. I do like the sound of the Tripath based amps, especially a well built TA2020, but TA2024 isn't bad either. Only problem is that the company Tripath went bankrupt a few years ago. While their chips are still plenty available, they might not continue to be, so not ideal to base a design on it, but so far I haven't really found any equal alternatives. Any inputs are most welcome on this.

I will have to choose between active or passive crossover, active means I will need one amplifier per driver, but it might not actually cost more as quality passive crossover components are very expensive.
Edited by jvjessen - 8/24/12 at 6:29pm

For reference a couple of manufacturer charts of frequency response and impedance, first the tweeter:



And the woofer:



Not that the frequency responses will look anything like that once they are in a box, but they are good to have for reference. Actually impedance is most interesting here as driver resonance and cone breakup will show on those plots, the plots are however not very high resolution.

I've just done a couple of impedance measurements for comparison;

Vifa tweeter:



It's hard to compare this with the manufacturer supplied chart as the impedance plot is so small, but the free air resonance is at the same place around 1KHz, which is good for this build. The plot is clean which usually indicates not much breakup is taking place, which is also good.

And the Dayton woofer:



Here the free air resonance (peak) is around 100Hz, that is somewhat higher than the 85-87Hz in the manufacturer chart. It usually goes lower with more break in of the driver, but this driver has already been broken in with a low frequency sine and some music for some time, so I doubt it will every go as low as 85-87Hz. Anyway 100Hz is OK and it's very good the plot doesn't show any major break up.

Edit (brain malfunction, higher not lower )
Edited by jvjessen - 8/24/12 at 6:43pm

Thanks .. I hope it will be interesting, might take some time before there are some actual speakers, but I would like to share the experience I've gained so far from many other speaker designs/builds. Any questions and other inputs are most welcome.

Cheers

For those interested in how I did the impedance measurements, it's very easy, I use a small device called WT3 from Dayton:



I used to have an elaborate home made setup with amplifier and feeding the signal back into the line in with some resistors and comparing the difference between the channels with a sweep. While it worked fine, it took forever to calibrate each time I needed it (20+ mins). So I bought this little USB thing, not expensive, and it's made this step so much easier. Having good impedance/phase measurements is very important for passive crossover design. It's also good to check a driver if e.g. the voice coil is stuck or damaged. And of course to know how easy/difficult a speaker is to drive. So for any DIY speaker builder doing their own designs it's a very nice tool.

Thanks man

More logarithmic plots incoming.. just to bring back more memories

I'm going to do some frequency response measurements for the drivers, first I'll do the tweeter, then I'll have a go at the woofer.

To measure a tweeter it's best to flush mount it on a reasonably sized square because of diffraction. I'm just using a piece of plastic something board with a hole in it, not ideal, but good enough to get a decent measurement and give more knowledge of how it really behaves. If I were to just measure it as is without mounting it in a plate the measurement would be uneven and very limited in lower frequencies.



To get a perfect measurement it should really be mounted in a large mdf board and made 100% flush with a perfect fit, but I only really need a guide, so this will be good enough.

I then use a calibrated microphone and some software to do the actual measurements, the setup I'm using looks like this:



These measurements are best done at a standard distance of 1 meter between driver and microphone and at 1W / 8 ohms (2.83V) to make them comparable to other measurements. It's important to do this as far as possible from any walls/ floor/ceiling to avoid early reflections.

In the picture the distance is not 1m, it's because I've set it up for a distortion measurement, I find it better to do at a shorter distance.

The microphone requires a phantom supply, so I'm using a small mixer as supply and microphone amplifier before I feed the signal back into my PC.

Did an on axis measurement of the tweeter, setup as described in prev post:




It actually looks very similar to the one provided by Vifa (see earlier post). I use an 80uF capacitor in series with the tweeter, just so it won't burn out when it's fed with a low frequency signal, it will not affect anything above ~500Hz. My setup will not go higher than to around 17-18 KHz, so ignore anything above that, it's not that important anyway.

Also did a distortion measurement, not sure how accurate it is, but it's good to have as a guide to see where it's best to cross over the tweeter:




From the look of it, a crossover around 3 kHz would probably be a good choice.

I need to do some off axis measurements as well to get an idea of the dispersion pattern.