My 1st order butterworth filter better than 4th order L-R: Why?

[philipreji]

Hello everyone,

I was reading the posts here for quite sometime, and learning many things. This forum is very informative, and thank you all for the great work..

I have a question that has confused me for sometime now, so I thought I would present it here. My 3-way home speaker system had a passive crossover with two capacitors and two inductors, and I had used it that way for about 10 years. The drivers are philips India make, of reasonably good quality. Then I chanced upon the tri-amping idea. So I removed the passive crossover, and made a 4th order L-R active crossover. The amps used were P3As (of Rod Elliott's design). With great expectation I connected the things together (preamp-crossover-power amps-drivers), but the sound was nothing exciting..it was in fact worse, and lost some detail. The bass lost all its quality In particular and the midrange was muddy (despite making frequency response measurments and gain adjustments using an excellent signal generator and digital storage oscilloscope).

While I was wondering what might have happened, I happened to hear a good-sounding pair of 2-way speakers, and the owner told me that his crossover is just a a 10 microfarad capacitor in series with the tweeter, and nothing else. I came back home, removed the active crossover and triamping from the circuit, connected a 10 microfarad electrolytic (it had polarity in fact) in series with the tweeter..removed the midrange too..and the entire dynamism of the music came back! It was better than my passive 3-way configuration and the active configuration.

Now I am puzzled. I had taken all precautions given in Rod Elliott's article on biamping, but still, the lowly single capacitor sounds better than the other two configurations.

Why is this so? Is it possible that the poor transient response of the 4th order L-R alignment is the culprit? But I thought group delays were not audible..Or should I blame it on the suspicious (possibly) quality of the opamps used in the active crossover? I used TLO 72 opamps, available for an equivalent of about 40 cents in india..

Has any of you ever faced such a thing before? Any thoughts from you guys will be a great help to me!

Thanks,

Reji

[audiobomber]

I made the same mistake. I read all kinds of talk about how I could just add active 4th order LR crossovers at the crossover points, and all would be great. It's BS.... It's not really any easier to design an active crossover than a passive. You still have to worry about BSC, and acoustic slopes vs passive slopes. Active operation is not an off-the-shelf solution.

[ashok]

A two way with just one capacitor on the tweeter is possible but probably hard to get correct in practice.
First the mid-woofer should roll off on its own at hf to match the tweeter response with the capacitor. The tweeter should be able to handle some lower frequencies as the roll off of a capacitor will be only 6db per octave. If it works well at low volume it might have problems at high volume depending on it's capability.

Some high end speakers do have single capacitors on the tweeter and an inductor on the woofer. The drive units will generally be very expensive !
On the other end of the scale you will also find cheap speakers with just one capacitor on the tweeter. While the sound could be attractive to some people it might not measure up to the performance of a good speaker system.

It's like the loudness control. Use loudness for several minutes and then turn it off and the sound appears to go 'flat' . That's just the way our hearing works .

Carefully implemented active systems sound very good . A passive implementation will not be able to match it for several reasons. Building an accurate active system however is not easy.

Cheers.

[CBS240]

Hi

I have a 4 way that I have played with, but it was designed to have a flat responce with different caps on low mid, high mid, and tweeter, and it sounds best the way it was designed. They are rated 120WRMS, but sound very much sweeter at not more than 50 or 60W. Not every house amp can drive them being 4 Ohms.

[x-pro]

Originally posted by ashok
A two way with just one capacitor on the tweeter is possible but probably hard to get correct in practice.

Epos Acoustics ES-11, ES-14, ES-12, M12 - all were made this way.

x-pro

[audiobomber]

Originally posted by audiobomber
You still have to worry about BSC, and acoustic slopes vs passive slopes.

I meant to say acoustic slopes vs electrical slopes.

[SY]

Originally posted by audiobomber


I meant to say acoustic slopes vs electrical slopes.

I'll bet that this is exactly the source of the problem. There is also the issue of EQ- since the drivers are not a constant resistance load, the frequency response with simple passive crossovers doesn't look anything like flat. Often, the BSC is taken care of this way. If I were going active, I'd first measure the electrical response of the crossover with the drivers attached to see what sort of compensation would be needed.

[phase_accurate]

It looks as if your speakers used first-order filters originally anyway in the first place.
If you want to copy this as a first-order active filter then put the passive x-over back into place and measure the frequency responses of the drive-voltages on each driver's terminals. Then derive the active circuits that will give the same transfer functions.

Regards

Charles

[AMV8]

Hi philipreji

I use a 5" mid range directly connected to the power amplifier. This is assisted in the treble frequencies by a treble unit with a capacitor/inductor filter on the treble unit and is assisted in the base frequencies by a base unit with an inductor/capacitor filter on a base unit. I think very similar to your prefered set up.

Using complex filters may provide the speaker with a smooth frequency responce curve and may help to smooth the speaker impedance curve but filters do add extra components. A direct connected mid range offers a cleaner more detailed and smoother sound that a complex filter system system.

You may conclude that I have used poorly designed filters and/or poor quality components. However I have tried 1, 2 and 3 way filters and used high quality components. I still find that I prefer the clean sound from a mid range unit with no filters and minimum filters on the treble and base units.

I think your ears are telling you the same.

Don

[philipreji]

Hi all,

Thanks a lot for your comments..they have been very helpful to me in forming the following conclusion: it should be possible to implement an excellent active crossover, but close results can be obtained from very simple passive crossovers, when you consider both electrical and acoustic aspects into the whole chain of sound reproduction.

I agree with Ashok that mathematically, a two way with just one capacitor on the tweeter is hard to get correct in practice – on needs great drivers indeed. But just hear such an alignment on decent drivers (that don’t cost a fortune by the way): the sound is forthright, pleasant, and carries the punch. At least that’s what I felt. In comparison my 4th order active filter lost most of the bass, and the midrange was veiled. I am not getting confused between sonic quality and loudness. The flatness of the music I felt with the 4th order filter was definitely not a lack of loudness – it was as if the whole life was drained from the music.

The crossover points for my active filter were decided from a careful measurement of the frequency response of the three drivers. After construction, the HF, LF and MF outputs of the crossover also were measured. The total frequency response was neatly summing to a flat line, and I didn’t observe any anomalies. The amplifier gains also were adjusted, considering the dispersion of the sound energy across the audio spectrum.

Everything else being identical for the comparison test, I am trying to explain my unexpected results on the basis of two factors: (1) poor transient response of the 4th order filter, and (2) lack of BSC in the active configuration.

I have read time and again that transient response shortcomings are not audible. This is something I haven’t digested yet, and as long as I cannot find another reason, I will be inclined to believe that the muddiness in the midrange was caused by this factor. What remains is the lack of punch at low frequencies. This can be related to the BSC issue, as Audiobomber says. It is true that since the drivers are impedance loads, the frequency response with simple passive crossovers doesn't look flat. Is the BSC taken care of this way, as suggested by SY, in at least some passive crossovers? Then that would be a real advantage.

For me, in the 1st order passive configuration the hearing experience was pleasant, which means that the anomalies that are expected to arise from intermodulation distortion, and limited bandwidth of the drivers, were inaudible, even when the amplifier was operating close to clipping. So one of my next experiments will be to connect a mid-range or full-range directly to the amp (as Don does), and connect the bass and treble drivers from there through minimalistic crossovers. Another direction of work would be as suggested by Charles: to calculate an active crossover with a similar transfer function as the passive, and implement the same.

Thank you all for your help, and more comments are welcome. And happy listening from your systems, CBS240 and x-pro!

Best regards,
Reji