DIY 3 way project active or passive?

Recommended Posts

I am looking for a bit of advice on a project that has taken far tooo long. Some of you may find the insight in to DIY speakers interesting if you are considering building your first pair. Someone here recently said that you will never stop tweaking them.....

I started this in 1996 :-) I also stopped moving forward on the project in 1996!

Anyway the basics:-

70 litre ported cabinet built from 25mm MDF with plenty of braces. All internally lined with 40mm thick sheeps wool carpet underlay.

Drive units:-

Dynaudio D260 Esotec tweeter http://www.gattiweb.com/images/dynaudio/esotec-d260_data.pdf

Dynaudio D76AF Mid http://www.gattiweb.com/images/dynaudio/d76af_data.pdf

Volt B250.8 Bass http://www.voltloudspeakers.co.uk/Loudspeakers/Studio_Range/B250_8/body_b250_8.html

The drive units were not cheap, starting with a nice flat frequency response surely makes the process of designing the cross-over & getting them to sound good much easier?

I modelled the drivers and started on cross-over using the old Visaton speaker designer software (unfortunately no longer available in english :-( )

The cabinets have been sat in the garage for 6 years and hence are looking tatty. The MDF has started to expand due to moisture. I decided to pull them out over the weekend and stick the drive units back in the cabinets to see how they sound and to check all drivers were still OK.


Ms Diceman is away working for a couple of weeks so I took the opportunity to fill the kitchen with a second system to test them with.


The shoes help to angle them up a bit toward the listening position ;-)

In order to orginally test they worked when I first built them I built a simple 6 dB/octave cross-over for the tweeter and mid which should cross-over around 4KHz and 650Hz. The bass driver never had a cross-over as I only ever used the passive cross-over to test them (for safety and to avoid blowing up the drivers!)


Once I tested they worked back in 1996 I quickly swapped over to running a car-audio based active cross-over driven by a wall-wart and 3 amplifiers in full active mode. I "borrowed" a calibrated sound level meter from work and purchased a signal generator to start trying to dial in the cross-over frequencies, relative sensitivities of the drivers and then play with port length. At the time Full computer based analysis software such MLSSA cost £600-£1200 (it was beyond my purchasing power back then - especially for a one off speaker project). Manually doing a freq sweep took about 2 hours per sweep in 3rd octave band so it was a very very timely procedure to tweak crossover settings and measure the effect.

I never really reached the point of being happy with the car audio cross-over settings - it was limited to 12 dB/Octave which required the phase to be reversed at the cross-over points and I suspected the odd suck-out in upper bass was a result of this. Whatever I tried I just could not get rid of it.

I stuck with these speakers for 10 years in their unfinished format until partner and house limitations meant they were just not viable. (They can't have been that bad then!)

Anyway 15 odd years on ...... The plan is.

I picked up a Behringer active cross-over that is user configurable between different orders, frequencies etc and I believe also allows you to time align/phase shift/stick bumps in freq and wipes your bottom. Also available to me is the XTZ room analyser - this allows real time analysis so it should be much much quicker to make some progress. If they sound any good I will have to build some new boxes now though!

First phase is to pick up another couple of matching amps to my existing units and then get busy trying to get a smooth freq response with the active cross-over. Once this is done I have the choice of either building a fixed setup active cross-over or trying to replicate the cross-over setup with a passive unit to allow 1 amplifier. Option 3 is to split them up and sell the drivers and bin teh cabinets if I can't live with them. If they compare to Martin Logans and Apogees than they may well stay :-)

Progress has been to date to retrieve the old passive cross-overs and plug them in to 1 behringer a500 amp - (nice and cheap in case I blew it!) and the bass units are running from a WAD Wd88VA Kt88 based valve amp - without a cross-over at all. I have to say they are sounding really quite promising in this configuration. Obvious failings in freq response but clarity, resolution and airiness are all there. Imaging is much better than I remember but they are terrible at a precise image, move your head and the image moves (I suspect the sharp edges to cabs are causing diffraction).

Tomorrow I hope to stick up a plot from XTZ of the nearfield freq response with my Tomy "my first passive cross-over".

If anyone has any bright ideas or suggestions for cross-over settings/modelling software/speaker analysis software I am all ears.

Oh BTW - they will never be finished. :nuts:

Share this post

Link to post
Share on other sites

Frequency response measured in room at a distance of 0.4m from speaker front face and half way between tweeter & Mid in the vertical plane.


Another one taken at a different height (IIRC this was closer to tweeter than mid).


I ran out of time last night and hence it was too late to test them properly. I need to pull them out from the corner and away form the kitchen cabinets. When carrying out other measurements I moved the mic up and down and across the front face, this shifted the peaks & troughs which to me is a sign that I am getting interference from the nearby boundary. (maybe from the off-axis drive unit response?)

I hope tonight to create a speaker tunnel with 50mm thick acoustic foam to restrict the dispersion and get a better idea of what is really going on. I will also measure port output and turn the volume up :-)

Considering there is no cross-over at all on the bass driver (and output extends up to 2kHz before dropping off!) a very basic cross-over for the upper freqs it is not too bad (it certainly isn't good either!). They are actually quite listenable in the current configuration (the bass is just bloody fantastic on these speakers - tight fast and they really go low with plenty of power) they have too much presence in the mid/lower treble which makes them sound amazingly detailed and I am hearing parts of the music I never have before very very clearly. Will this remain when they are actually flatter in response? I have no idea but it may be a good lesson in voicing speakers.

Now for my first request for advice...

What is the best way to actually measure speakers in room? I have only ever done it properly in an anechoic chamber with MLSSA. What distance? What angle? at tweeter height or half way between mid & tweeter?

My thoughts were to measure the reverberation time in 3rd octave bands of my room and then try and correct the freq response for the variations.

Any thoughts or advice?

Share this post

Link to post
Share on other sites

I am hoping for some advice from you guys.

I have now removed the passive crossover and used the Behringer digital crossover to form a fully active 3 way setup. This allows me to easily change crossover frequencies, adjust levels between drivers and change crossover slopes. I have so far spent about 25 hours adjusting, measuring and listening and not getting very far. I can get the response to be +- 3db across the board near field and on axis with some very very heavy band pass notches and very steep crossover slopes of 24-48dB/octave. They then sound absolutely terrible! I suspected something was not right as the drive units are pretty flat in response and have even measured my other commercial box speakers to ensure it is not a room effect (I am using shedloads of acoustic foam in the room to try and remove as many room reflectiosn as possible) All commercial speakers measure pretty flat.

I had a bit of a brain fart last night and started to measure response off axis - the reponse plots vary massively by 20-30 db within very short distance angles - say up to 15 degrees. Reverting to my passive crossover which is running a 6dB/octave filter between tweeter & mid I repeated the process. To cut a long story short I can get them to measure flat off axis when the distance between drive units is 1/2 wavelength at the big dip in reponse seen on axis. My thoughts were that something is that one of the drivers is wired out of phase! I have checked drive unit marking and all appear to be correct. In a moment of desperation I reversed polarity of a few bits and suddenly they now measure flat on axis and off. BINGO!

My thoughts are that one/matching pair of drive units are marked up incorrectly (probably mid-range but maybe tweeter).

I can check polarity for bass & mid using the 1.5V battery test. How do you check tweeter polarity? Can they be checked in the same way as bass drivers (with battery) and a movement of dome seen? Is this safe? They could both be marked wrong and hence sticking them both facing each other (my normal check for relative phase) will not provide any info.

The good news is that not only do they start to measure flat with the phase buggered with but they also sound pretty good both on and off axis.

So to summise

Question 1) how do you check tweeter polarity

Question 2) any suggestions for a decent quality internal speaker cable? (I built these on a budget when I was a student and used cable talk 4.1 for all drivers as I had some laying around)

Share this post

Link to post
Share on other sites

Hi Diceman.

Interesting project, I've built 3 way active speakers myself and know they can be brilliant performers.

Don't think the 'polarity' of the tweeters matters much. The half wavelength of sound at 10KHz is only 1.6 cm., and the human ear is not particularly sensitive to that, so just try one way and the other and see if it makes any difference (I never found any).

With 60 litres of cabinet volume I'd ditch the port and go sealed, you'll get much better bass transient response.

On another subject, will you be going to the Bracknell Audiofayre (just up the road for you) on the 17th April?



Share this post

Link to post
Share on other sites

Hi Henry,

Yes I hope to get along to the audiofayre, I went to the last one and whilst it was smallish it was still a nice day out.

as the crossover is only 6db/octave the phase makes a massive difference to both measurement and how they sound. Main peaks/troughs are around 2-4K near the crossover region.

Share this post

Link to post
Share on other sites

Have you read the Nelson Pass article referred to in here:


He mentions about crossovers and how you want them, when the drivers are close to the XO point, to add together rather than subtract - if the mid is subtracting from the tweeter, say, then that is a bad thing. Reversing the polarity of one of the drivers is one way of achieving this addition. It may be related to phase, not sure.

I certainly don't think that it is automatically bad to have the drivers out of phase with each other - there are valid reasons for doing so.

Another gratuitous link that might be interesting!


Share this post

Link to post
Share on other sites

Thanks Adam, some interesting reading. I decided to try and go for 6db/octave passive crossover between tweeter and mid to maintain the phase and so as not to require the phase to be reversed on one of the drivers. It just seemed to sound more homogeneous than other filter types.

The suckout was very large indeed around the crossover region - if it had been 3-6 dB then I could understand it being partial interference between the drivers. It was much much larger at around 20-30 db.

I will review my measurements of the drivers in isolation of each other (but with crossover) - in line with Nelson Pass article I suspect this will tell me more truths.

If i ensure I have bass & mid polarity correct (which is easy) then I think I will just have to wire the tweeters the way they sound and measure well (as opposed to sounding & measuring very badly).

Share this post

Link to post
Share on other sites

Hi JP,

If you've got a PC and can stretch to around £80 for a preamp and mic I can recommend Holmimpulse software. It's freeware and a very capable bit of software for measuring loudspeakers. It was discussed on this thread. http://www.hifiwigwam.com/showthread.php?50405-ARTA-and-REW-measurements-sound-card-and-use..

I'm by no means an expert with it but will be at the pie show in room 42 so if you are travelling up I'd be happy to have a chat with you and bring some printouts to show what I've done measuring raw drivers and the like. Those are some very tasty drivers so it would be good to get them performing at their best.

I'd have been willing to bring my kit down to help you but have just moved in to a new house that's in needs a lot of work and my wife is expecting our second child in July so I think my free time may be at a bit of a premium this year. My cousin isn't a million miles away from you in Alton so it might be the sort of project he's interested in.

Share this post

Link to post
Share on other sites

Thank you for the heads up Clive. I will have a look at its capabilities this evening. I do have XTZ room analyser which I am using at the moment.

A couple of questions..

1) does holmimpulse allow gating to remove room effect? (MLSSA style - look for first major reflection)

2) can it measure impedance?

3) do I need a Behringer ECM8000 mic? What preamp is required? I would prefer to run this with a laptop (with minimal soundcard inputs) to allow me to take measurements in my garden which is as close to an anechoic chamber as my funds will allow!

Maybe it will actually work with the XTZ usb soundcard and mic?

Thank you for your offer - if my holiday is cancelled due to no snow and I make it to Scalford I will definitely look you up ;-)

Share this post

Link to post
Share on other sites

To give you an idea of the difference in response by swapping the polarities about. (edit - just noticed I had frigged with the scales on teh first plot - 50 dB range as opposed to 80 dB! what a twat)

Before swapping polarity


After swapping polarity


Measurement of mid-range unit running full range and without any crossover


And just to indicate that the measurements are reasonable a response plot of a Heybrook Sextet


Share this post

Link to post
Share on other sites

Hi JP,

Sorry for not replying earlier but I'm still waiting for the broadband to be connected at home.

In answer to your questions

1) Yes - The software is very good for looking at crossover points when used in gating mode. It's also very useful in terms of room response and where to site your speakers and listening position where you look at the total response including reflections.

2) No but I have a Dayton Audio WT3 tester which will. I could easily post this down to my cousin and I'm sure he would be up to try some testing.

3) You don't have to use an ECM8000, I think it would be definitely worth trying with the XTZ. If you can get it running you could always run a one off comparison against an ECM8000 to see how it looks. The ECM8000 is by no means completely flat but general consensus is that it's pretty good across the likely crossover frequencies. I would like to get mine properly calibrated one day.

I use an Art tube pre (I guess I just like valves) which I picked up for around £30. Pretty much any mic pre will do as long as it has the phantom power so you could keep an eye on ebay to see what comes up.

Hope this is of help :)

Share this post

Link to post
Share on other sites

Looking at your XTZ measurements it would be interesting to run a back to back of the XTZ against holmimpulse to see what the differences are. It certainly looks as if it is getting you into the right ballpark.

Share this post

Link to post
Share on other sites

A small update:-

The aim was to run the tweeter and mid through a passive crossover with bass running actively (maybe through a plate amp at a later date) and allowing for some sweet valves to be used for the critical 500Hz to 20KHz region with grunt for the bass. My thoughts were that the benefit of doing this is that the mid/tweeter will not have to be padded down to match the senstivity of the bass driver (only 1 dB diff but with the baffle step correction occuring around 400 Hz it may need padding down a further 2-6 dB). I did not want to waste sensitivity if being driven by a small valve amp. Adding resistors in the signal path at high level just muddy the sound to me. (maybe uber expensive opnes don't!)

Holmimpulse downloaded, Behringer ECM8000 mic and M-Audio Mobile pre USB purchased. The gating that Holmimpulse allows is fantastic and allows you to dial out the room reflections from the measured response. For the £150 or so this setup costs it is fantastic value compared to alternatives and provides you with a good headphone amp and USB DAC/ADC for your PC all in one :-) Holmimpulse allows you to measure the frequency response, phase and THD within limits. Impedance is not covered.

I will upload some plots of Holmimpulse later when home when the phase plot has provided a lot of info on why the unusual frequency response resulted from wiring the drive units in phase with a 6dB/oct passive filter. The phase was swinging all over the place :-( I can only assume that it is due to either having a bandpass 6 dB/oct for the mid (2 lots of 90 deg phase shift?) or the driver alignment with the 3" dome mid.

I still have a bit of a peak around 10KHz but it is better than before. I may insert a notch filter later on.

As such the crossovers were mocked up using cheap caps (maplin electrolytics - http://www.maplin.co.uk/non-polarised-electrolytic-capacitors-12777?c=so&u=strat15 ) and Jantzen foil inductors (I decided to start with the end inductor products to save changing the DCR later on and having to reselect the caps). All were hooked up using speaker cable offcuts and spade connectors for ease of changing caps.

Frequency response could be made pretty flat but the sound was very grainy in the treble the 10KHz peakwas obvious and coarse sounding. I had a few spare caps of better quality from previous projects of the wrong values which I used for the treble section and a lot of the grainy sound was gone despite the odd frequency reponse! Caps were Jantzen crosscaps and cheap ICW audio grade caps.

A fellow wammer was selling some Claritycap DTAC and Jantzen paper in oil caps of a value close to my requirements which were tried. For my tastes the DTAC were a big improvment on all others tried. Treble was smooth and sweet without any grain - the 10KHz was no longer obvious.

The bullitt was swallowed and small value claritycaps (MR range for treble as DTAC no longer available) and SA range for bypass cap on the mid were purchased. The bulk of the capacitance is made up from motorstart caps for the mid. Wiring I am using multi-runs of 19/0.25 silver plated copper within a PTFE jacket - RS components stuff.

So far I only have 1 crossover built but it didn't stop me listening in Mono for 6 hours over the weekend ;-)

After a few hours the new crossovers and caps have become a little sweeter sounding, initially a little grain was present with treble a little spitty in nature.

I have some of the russian PIO Caps kindly supplied by Adam (AmDismal) which I hope to try out soon.

I have had this pair of speakers in a partial state of finish since the late 90s and I am happy to say the expense over the years and considerable time spent has been worth it - the end results suit me much better than any commercially available box speaker I have heard/owned to date. There is still plenty of work left to complete them but at presnet I am happy listening Mono ;-)

Now for a couple of requests:-

1) does anyone have a device to measure impedance so I can dial in the port length properly? I have so far just used freq response and added plenty of extra stuffing within teh cabinet to reduce port output substantially. It sounds close to me.

2) does anyone have a spare Waitrose port & stilton Gift box? I need another for the second crossover and can't wait to xmas when they will start and become available in the shops again ;-)



Share this post

Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

  • Recently Browsing   0 members

    No registered users viewing this page.