Active Speaker Project


Ever since I was a young boy, I have been interested in Hi-Fi and started at a very early age with an old record player that I was given. I stripped out the single valve amplifier and built a transistor amp to replace it. This was not entirely successful but produced some noise and a lot of distortion but not bad for an eleven year old.

The next big improvement came when I built my first speaker. Yes, single speaker as in those days, stereo was not common and indeed, only available on a few LP records. Yes, I know... and dinosaurs roamed the earth !

I knew a little about electronics and realised that if I wanted bass then a large speaker was needed. I duly saved up my pocket money and bought a 12 inch, dual cone speaker.

With the help of my Father (he did all the work!) we built a large box using ¾ inch chipboard with plywood baffle and back, and the monster was complete. The box was unstuffed and undamped as I simply didn't know about this.

I now had no suitable amplifier to drive my speaker so I found a design in Practical Wireless for a 5 W mono amp using a pair of OC35's. Built on Veroboard, it actually worked and was finished by installing it in a wooden sleeve with aluminium front and back panels.

My speaker finally came to life, with my Father's Medium Wave radio and my amp, it sounded truly wonderful. (To me at that time)

Thirty-five years, or so , have passed and I have now enjoyed the benefit of training as a BBC engineer. My standards are a bit higher now with many years spent monitoring BBC radio and TV output at work.

After slowly improving my modest set-up at home, I decided that a quantum leap was required to improve my present system of Quad 44/306 and Spendor BC1 speakers. These were the speakers that I had grown up with at work and I although I knew they were not perfect, at least I was accustomed to them.


I new nothing about designing speakers at this time so spent many hours doing research on the Internet looking both at commercial and DIY designs. A steep learning curve followed where I bought two unknown drivers that were on special offer from a surplus warehouse for £3 each and measured their Theil/Small parameters.

Armed with this new knowledge, I set about designing a small speaker for typical A/V use. The starting point for this was the size of enclosure and for this design, I wanted to limit this to a ported box of about 15 litres.

I then discovered something which made everything much easier. A piece of free software called "WinISD" ( www.linearteam.dk). Absolutely amazing, it has a database of many popular driver parameters, works out the optimum box volume and plots the response. Although written for the dreaded "Windows", it runs perfectly in Linux using Wine.

Using a mail order electronics catalogue, I put various drivers into Win ISD and viewed the results. I couldn't use my surplus drivers at this time, they would require too big an enclosure.
For this excercise, I limited myself to the Audax range of drivers. They were cheap, readily available from Maplin Electronics in the UK and "WinIsd" already had all their parameters in its database.

The drivers were bought, boxes built and performance measured the best I could in the garden. I had by this time bought myself a Behringer ECM8000 measurement microphone and built Rod's balanced mic preamp.

The design had been successful, the response was as predicted, 3dB down at 46Hz.

I had initially thought that this response would be adequate but I was not happy. They lacked the lower octave and it was very noticeable, otherwise they sounded OK, very neutral which I like.

My small A/V speaker project

Again using WinISD and two of my surplus drivers, I built an isobaric Bandpass Sub woofer with its own 100W amp and active low pass filter. This certainly gave me the bottom octave with much better bass than my Spendors. Measured response was now -3dB at 29Hz which I now feel is "adequate".

I had previously discovered Rods site ( www.sound.au.com ) during my research and was intreagued with his own active speaker system. Many hours of reading later, I decided that this was indeed the way to go.


I listen to a great variety of music and want to reproduce the lowest practical frequency found in electronic music, about 20 Hz. I also want this bass to be well defined and without masses of phase shift. To achieve this a sealed box design is necessary.

I want a very neutral sounding midrange with a crystal clear top-end. My Spendors are good in this area although they are very directional at high frequencies. I want to minimise this as far as possible.

Whatever I build has to fit into my lounge and meet the approval of the Wife who thinks that speakers should be as small as possible and hidden behind the sofa.

I thought I could get away with a 60 litre box size.

The tweeter should be at ear height when seated in the listening room.

The midrange and tweeter should be spaced in an attempt to get time alignment.

A three way, active, Linkwitz Riley crossover system is to be used with a Linkwitz Transform circuit to equalise the bass response.

Crossover frequencies to be of the order of 300Hz and 3KHz.

Crossovers to be ESP audio 3 way L/R active. Project 09

Linkwitz Transform to be ESP audio. Project 71

I have to be able to afford this !!! most important.


I started with the bass unit, looking at what was available and using WinISD to plot them in a sealed box. After a very short time, I found that most of the drivers that were available were very obviously designed for ported boxes and worked poorly in a sealed box. I did however find a driver that stood out from the rest and seemed to be designed just for my prospective box. The Peerless 831709 sealed box size worked out as 58 litres and gave me a -3dB point of 35 Hz. This I could equalise with the Linkwitz Transform to give my desired response.


Peerless 831709 bass driver

Next the Midrange driver. This was not as easy as the bass unit as almost any small speaker would cover the range required in a small sealed enclosure.

I very quickly found the best midrange, the ATC SM75.150 which I am sure must be very good but for the price of £255 each in the UK, I was never going to know just how good. After a fair bit of research, reading reviews of the DIY designs on the Internet, I decided on a midrange noted for its clarity and smooth response, the Vifa P13WH-00-08. Using WinISD and the Vifa gave me a sealed box of just 2.4 litres and a response flat to 200 Hz. A little bit tight at the low end but fits nicely with my crossover which will be -6dB at 300 Hz. As an added bonus, I could afford to buy this one.


Vifa P13WH-00-08 midrange driver.



The tweeter gave me similar problems. Almost anything would work and give me a response way past what I can actually hear these days. Again I tried to rely on what other reviewers had written although I am very cautious of anything written in a so-called Hi-Fi review. I had almost decided on the Vifa D27TG-45-06 when, whilst browsing another electronics catalogue, I came across something that grabbed my attention.



Visaton KE25SC ceramic dome tweeter.


The "Visaton KE25SC 1" ceramic dome tweeter" leapt out of the page at me. It has an incredibly hard ceramic dome which is set slightly back behind the front flange, protected by a slightly domed metal grill. A quick trip to the Visaton web site followed where its details were carefully studied and all in all seemed to be a well designed driver. As an additional nicety, built into the tweeter back is an LRC impedance correction circuit. Not of importance in my case but it just swung the balance in that this had been thought of in the design stage.

Of more importance was the fact that the dome is slightly behind the flange, the phase timing between it and the midrange would be closer than if its dome protruded beyond the flange like most do. As the midrange cone is fairly flat, I figured that I could obtain close timing by using this tweeter and even closer by offsetting the centre of the tweeter from the midrange to try to equalise the distances from each, to the listener.

Only one small drawback, the cost of this was double that of the alternative... oh what the heck, I bought them anyway.

During my practice run with the small A/V speakers, I bought the essential woodworking tools. A router and powerful orbital sander, both with dust extraction. It would have been nice to have a circular saw or table saw but I couldn't stretch to either at this time so all timber cutting was performed using my small electric jig-saw. This could produce quite accurate cuts in the material when run along a wooden batten, securely clamped to the work.

I used 18mm MDF for all the panels with a double thickness, laminated baffle.

I glued two thickness's of MDF, then used a router with “circle jig” to cut the holes.

Note that the left and right baffles are mirror image, the midrange and tweeter are both offset from centre.

Despite all the claims of stability of MDF, I found that I had errors errors creeping in. Even after careful measuring, some things were a poor fit. I eventually found the reason for this. Storage of the MDF panels in my unheated garage during a damp British winter had caused the MDF to swell. I made some measurements using my digital vernier callipers and was amazed to find that the thickness had swelled, in the worst case to 19.8mm. After this, I brought most cut panels indoors for as long as possible. I also did most of the glueing in a spare bedroom much to the consternation of my wife!



The laminated baffle and construction of the tweeter enclosure.



The tweeter was mounted inside its own circular enclosure made from 3 rings and one circle, cut from the MDF and glued together. This is within the midrange enclosure thus totally isolating it from any influence of the massive sound pressure in the bass enclosure. Several braces were used in the bass enclosure to make the whole thing very rigid. Inside, the midrange enclosure was lined with "Dedshete Panels" which are self adhesive 2mm bitumen sheet, laminated with 10mm foam. In the bass enclosure, all internal panels are lined with heavy, industrial carpet tiles which have a heavy felt/bituminous base and a rather attractive blue carpet finish, although this will never be seen. I got these second hand for nothing and they certainly damp the whole enclosure and add to the weight of course. Both midrange and bass enclosures were then stuffed with GAW (a replacement for long haired wool) that I had left over and filled up with glass-fibre wool.

All panels were clamped and glued using the "Instant Nails" type of adhesive which gives immediate grab, doesn't run and will fill any gaps due to poor woodworking. This was also used to form a “bead” along all internal joints to make the enclosure airtight. I only had one problem due to the use of this adhesive. This was during final finishing sanding. Make sure you remove excess adhesive before it sets and fill any gaps with standard decorating filler. Heat from the orbital sander can melt the adhesive and instead of removing it, deposit it elsewhere. It makes a real mess and is difficult to remove. Believe me.



Internal view showing midrange enclosure and bracing.


The drivers were mounted using either 5mm or 4mm 40mm long screws and T nuts, tapped and glued into place. I also added 4x 8mm T nuts to each base to allow for the fitting of carpet spikes. I made the spikes from a length of 8mm threaded stainless steel rod and turned the spikes using a lathe which I had access to. Ready made spikes are available together with threaded inserts for those unable to make them.  I never actually fitted the spikes, these things are too damm heavy !


I like speakers to have grills covering the drive units whatever the purists may say but also realise the effect that any discontinuities can have in causing diffraction at high frequencies. To this end, the front edges of the cabinets are slightly rounded although not enough to have much effect and all the inside edges of the 6mm plywood grill panels are lined with 15mm of felt, on its edge, built up in layers until it is the same thickness as the frame. Thus the grills should in theory reduce the diffraction at high frequencies and the effects thereof.

I chose a dark brown, open weave grille cloth which was glued around the edges of the frames. The frames having previously been painted matt black.



Rear view of one of the grills showing the layers of felt around the tweeter.

For finishing the MDF cabinets, I decided that a black paint finish, matching my wide-screen TV and stand was required. This was achieved by using two coats of matt black paint, applied using a fine finish roller followed by four coats of tough satin varnish, applied with a gloss roller, rubbed down lightly between coats. I am pleased with the results, the finish closely resembles the black plastic coated chipboard of my TV stand.


The left speaker in my lounge.


A block diagram of the active crossover, Linkwitz transform and amp system is here.

My original intention was to make the speakers totally self contained with just mains and phono and XLR, balanced line inputs. I soon realised that weight was going to be a problem and if I did this, in all probability, I would not be able to lift them.

After completing the cabinets and carrying them inside, I realised I was right, the amps would have to be external.

I had no problems assembling and testing the boards for the crossovers and Linkwitz Transform. Rod's excellent articles provided all the information needed.

Component values for the crossovers and bass equaliser were all calculated using the tools available on Rod's site. Using preferred values actually gave me crossover frequencies of 310Hz and 3.1KHz which was close enough as the absolute frequencies are uncritical. I then verified the frequency response of each crossover by doing a manual sweep with a signal generator and AC voltmeter to confirm that no mistakes had been made. To my relief, the graphs were perfect.



Linkwitz Riley crossover response


I had slightly more trouble with the Linkwitz Transform board, in that the closest that I could get to the calculated value of C1 needed 4x 0.47uF in parallel and the PCB only had space for three. I could have soldered one extra on the track side but I managed to find room on the board, drilled two extra holes and wired the extra capacitor under the board. The result was quite neat. Again the response was plotted to confirm that it was doing what it should, fortunately it was.

I actually built six discreet power amplifiers using Rod's P3a boards with the intention of making two, 3-way power amplifiers containing the active filters etc.

After assembling and successfully testing all six modules, I had a failure of one of the output transistors and didn't have a spare.  I then read, on Rods site that a lot of fake semiconductors had been identified, falsely marketed as Toshiba originals. This worried me because these were the same type that had just failed on me. They were not cheap but I ordered a spare to repair the amp.
I emailed Rod and told him about the failure and he advised me to open the transistor package and look at the size of the die. Aparently, the fakes have a small die size and are poorly rated. I did this and it appears that I actuially had a genuine Toshiba device.

I was getting impatient to listen to my creation so I decided to put the power amps on hold and build a unit with just the active filters with its own power supply.

This was completed the next day and wired up to three stereo power amplifiers. I used a Quad 405 for the bass amp and a pair of Quad 306's for the midrange and tweeters. Using the quoted sensitivity of the three drivers, I used a signal generator at 100Hz, 1KHz and 5KHz to set the AC voltage across each driver to the correct level.

Now finally, I could listen to them.

I have a favourite track for testing things, Dire Straits "Private Investigations". I know every note of this track intimately. First impressions were that the bass was lacking but the midrange and treble were fantastic, but after listening for a few more minutes and to different types of music, I realised something. Bass was definitely NOT lacking but actually sounded different than what I was used to. Especially percussion. I am not a hi-fi reviewer and hate the language they use to describe esoteric things that can't be quantified scientifically. However, they sound good to me and all I can say is that I strongly suspect that they sound the way that all loudspeakers should sound.

The midrange sounds very smooth and treble was as I'd hoped it would be, clearly defined.

Further listening tests revealed things I hadn't heard before in tracks I was very familiar with.  instruments that were now discernible, phasing effects that were inaudible previously, stereo panning effects hitherto un-noticed.  Stereo imaging has proved to be excellent, I have run the speakers with and without grills and cant hear any difference.

Recently, I invited a colleague, who's opinion I respect, around to listen to them. We listened to several tracks, ending with Peter Gabriel and Kate Bush "Don't give up".

He said that the bass guitar solo was the best he'd ever heard it. He thought that the treble was a bit lacking and I probably agree. The beauty of this design is that I actually have control over this and can try a slight increase in the HF output level.

I have now turned up the treble by just 1dB and that, I think is about right.



Was the whole thing worth it. Absolutely. After living with the speakers for several weeks now and listening to a great variety of music, TV and DVD's, I am still pleased with the performance of them. They are very easy to listen to at any volume and sometimes I don't realise just how loud I have them. As far as I can tell, the speakers do not seem to have any characteristics of their own,  there is just music.  The bass response is absolutely fantastic, I never realised before, just  how loud the heartbeat on Pink Floyd's Dark side of the moon was. In fact, its not a heartbeat at all, but bursts of about 20Hz, however,  it makes the room shake !

The only limiting factor on a system such as this is due to the bass equalisation performed by the Linkwitz transform circuit. At high levels, the bass amp or driver is definitely going to overload. I have not noticed this happening, even at volume levels that are excessive for my lounge.

Well worth the effort and one day, I may even finish the power amps and sell the Quads... maybe one day ?

I still intend to attempt to measure the frequency response of my design but that will have to wait until the weather warms up a bit. The technique I have used with reasonable success is to set the test speaker up in the garden. This is as near anechoic as I can achieve. I have a Behringer ECM8000 measurement microphone which is stand-mounted on the speaker axis and about 1m distant. This microphone feeds a balanced line preamp with phantom power supply, then to my AC voltmeter. I manually sweep using an audio signal generator.

The box I have built containing the microphone preamp also contains a pink noise generator. (again from Rod's site). I hope to use this, one day with a laptop running a FFT spectrum analyser which should eliminate the manual sweeping and plotting.
A better way maybe to use an impulse generator and use the computer to do an FFT analysis.  Again, maybe one day ?

<-- Back