Updated: 27-Sep-03

Measurements with 2 alternative filter configurations


This is the most disputed and most discussed part of speaker design I guess..... Designing a Cross Over seems to be as a voodoo thing as a lot of other high end equipment modifications and accessories. Basically you have to make a choice of 6dB, 12dB, 18dB slopes, electrically as well acoustically. Parallel or serial filter ? 2 or 3 way ? or fully active ??? Pffffffffff.......

Do you know how many possible combinations you can make of all those alternatives ??? I know how to calculate it, but I already know, it would discourage me :-) I also had several emails and posts with totally different opinions !!! So HOW  to proceed here ????

BACK TO BASIC DESIGN PRINCIPLE !! define, by measurement, the options/problems areas, the most likely solutions and then just start trying..... The proof is in the eating, isn't it? So just start simple and from there on start tweaking and trying, with some measurement to make sure nothing strange is happening, so more a sanity check....... The first part is done already (see previous page) and now, as I try out and test filters I will test and describe them here: 

Quick Navigation:

Filter Alternative 1&2, Serial 6dB/12dB
Next Filter...

Open the Excel Sheet with Filter Calculations


Filter Alternative 1&2, Serial 6dB/12dB

gydinel_filter_sep6.gif (7141 bytes)


At the right, you will recognize a very basic 6dB Serial Filter in the 1st alternative version. This was basically the filter I used in the test versions. The 680uH and 11uF are calculated to give a cross over of aprox 2kHz and relatively normal slopes, similar as you would expect from  a parallel filter. I have developed an Excel File for this purpose, which also will give you insight of the effects when making the C smaller and at the same time the inductor larger. When you do so, the Filter  factor ( I just call this "factor" ) will become higher and the units will play longer together. Simply to understand if you would say that the C would be infinite small and the Inductor infinite large. In that case you only have the 2 units in series and therefore both will play at half over the whole frequency band.....

Any way, my experience is that you can tweak around with this and that with a high factor the sound is very smooth, but lacks warmth and body. In the other case you get a very pronounce sound and good focus. Just find a place on this scale where you feel comfortable.... I choose for a good compromise between these 2 ......

OK, having said all this, I already noticed at listening tests, that the focus was lacking a bit and the highs were a bit harsh sometimes, which I blamed on the cone hick ups from the C79..... So the first easy to decide alternative 2 was to add an inductor in series with the C79 and start measuring and listening. So said, so done ! Read on below .....



Frequency response Curves

freq_curve_C12_filter.jpg (23819 bytes)

Lets see, as this, after all, is a serial cross over, if the tweeter is affected much by this !! The red line is version 2, the black one version 1. As you can see the effect is little, just a small bonus on the roll off in the low frequency area. Also, the output in the mid range is gaining a few dB, as a result the curve is more flat and this is, very wanted, a compensation already for the somewhat too brilliant high response from the C12........ Some times you are just lucky, haha!

freq_curve_C79_filter.jpg (23680 bytes)

For the C79 of course we see a huge difference, as expected. The roll off is much more steep, which will leave the work to be done with the C12..... Specially the peak round 10kHz is well damped now !! Also very noticeable is the compensation of the baffle step The low / mid range bump has disappeared now and a flat response is the result. Looks pretty good now.

freq_curve_C79C12_filter.jpg (24255 bytes)

Now a comparison between the 2 versions as a total box. The black one is version 1 the red again Nr 2..... Clearly the extra inductor has been smoothing things up and the whole curve is now almost flat with small lift at the end of the spectrum. Which makes the speaker a bit more brilliant, but the few dB here are well below some commercial speakers !! Too flat sounds, in my taste, too English ;.) Matter of taste also...

freq_C79C12_ind_tot_1mH.jpg (24497 bytes)

Final Curve from the system and the 2 units individually..... What you can see very nicely is, that in the mid / high range the C12 is doing the work and in the lower mid range still both contribute. Never the less, this results are looking very promising already !!! Not expected things would go that quickly.... but I still need to measure with my ears, haha, read on !

Conclusions and next steps........

I have been listening for a week or so now with a lot of different material and the inductor is still there !  The version 1 compared to the #2 is relatively rough, I must admit, very dynamic, but #2 is also dynamic, so no showstopper.... but #1 is clearly suffering from the non well enough damped C79 and the fact that both units form a sound source for the high frequencies. At these short soundwaves, interference is unavoidable and is blurring the sound stage. #2 is clearly the winner when it comes to sound stage and focus / pin point. The sound is smooth, but not too soft. It is just a polishing from the sometimes almost brutal character from the #1. So I stay with this #2 first :-)

OK, I have already did some small listening experiments/tests with a full 12dB version, which means to add a C in series with the C12. Now both units get 12 dB slopes. Value was 22uF. This was leading to an even MORE smooth sound and some more pin point, what, reading the results above, was actually to be expected. BUT, I consider it to be too much. The Gydinel are then playing like a smooth snoring pussy-cat, OK, may be a wild tiger is too much, but I still want some dynamics and brilliance. The extra C was killing the sound a bit and the gain in pin point and focus was just minimally, so I you like KEF stuff, it might be nice, but then you don't use Thiel units, hahaha !!

REAL "next steps" are tests with full active approach, Guido is visiting me next week and bring more amplifiers with him!!!!

You will read about in a few weeks !!! For now, I am enjoying every single day the great sound of the Gydinel !!! It is truly a great speaker......


Back to Filter Design Page


Happy listening and building !!

Doede Douma

IMPORTANT: The information provided on this page is intended as guide for DIY activities and therefore free to copy and or publish. If any one wishes to use any of the information from my WEB site, please make sure to refer and footnote to my URL Link as source! Doede Douma