Many thanks for your informative post, nemophyle.
>anyway, we seriously lack information about your system, your room , the drivers you use. the modification you made is huge and there are many things that could have led to your preference, so please enlighten us !
My system was built by myself about 10 years back. Some of you might have read the old Philips publication “building amplifier systems”. There is a 40 W/channel transistor amplifier circuit (class AB) given in that, which uses a quasi-complimentary output stage and + 60 volts power supply. The output stages used the lowly 2N3055 transistors. In todays standards this circuit is a humble design, but it was quite popular at that time, and sounds good even now. For this stereo amp I made sealed speaker boxes, each housing a 40 W woofer, and matching midrange and tweeter. All “Philips India” make. The company doesn’t give any info on the drivers except that they are 8 ohm. Both the woofer and midrange are paper cone. It is very likely that the midrange is nothing great – even I had a feeling that it had a low sensitivity compared to the other drivers. Simple 3-way passive crossovers were used, which had two inductors and two or three capacitors to the best of my memory (these were purchased, and I still don’t know the crossover frequencies). Anyways I liked the sound, and things were going well. Off and on I tried chip amps based on STK, TDA etc series, but the Philips amp sounded somewhat better than these. All these amps distorted at high volumes, but I usually listen to music at half the maximum loudness the Philips amp can provide, so given the logarithmic response of the ear, this should be about 4 W electrical power per channel.
Then last year I chanced to read about the P3A amp of Rod Elliot, and also about biamping and active crossovers. First I made the P3A. Even with 2N3055 output transistors, it sounded definitely better than the Philips amp. Then I made two more P3As. All had the same good performance. Then I removed the passive crossovers and measured the frequency response of each driver. Based on that measurement I chose the crossover frequencies and made the three-way 4th order L-R active filter. After removing the passive crossovers I hooked up everything, to hear a much better performance. But the result was definitely below my passive system, as I have detailed in the original post.
My drivers are not time-aligned. All-pass filters are not used anywhere in the circuit. I was not even aware of these things when the tests were made!
The room is rectangular, about 20 feet long and 10 feet wide. Furniture is scarce. For the comparison tests I always tested with one channel (the same channel of about 10 songs from a CD), as I had only three P3A amps with me. The speaker box was placed on top of a 3 feet tall table, at least 4 feet away from the nearest wall. On week days I can do the testing only in the night. The volume is quite low, as my better half is not that interested in profound analyses of music and its various nuances in the middle of the night! Hence I would estimate the electrical power used only at about 400 milliwatts.
Nevertheless, during weekends I have made a number of more tests at much higher volume levels, at daytime. For these tests I had to move the system to another 8 feet by 8 feet room, and the speaker box was very close to one corner, maybe half a foot away from the walls.
Since the active configuration did not give the desired results in both of the above situations, I went back to the passive 3-way crossover using one P3A amp, which sounded better. That is when I heard the 2-way speakers of a friend, and thought of trying out the same in my speakers. Yes, for doing this I removed the midrange, and connected the woofer directly to the amp, and the tweeter was connected in parallel to the woofer, with a 10 microfarad bipolar electrolytic capacitor in series with the tweeter. The result was better than my earlier 3-way passive and 3-way active configurations.
Sorry for the long description, but I thought it is better to write down all possible details.
> The TL082 is not a particularly good sounding opamp. I'd expect to hear its effects, period.
I am also a little suspicious of these opamps.
> In my experience, many listeners tend to mistake added distortions for "loudness" and or "presence."
I agree. But to the best of my belief, I am hopefully out of this category! It is gratifying that the discussions here are giving valuable inputs which seem to explain experiences such as mine from a more technical point of view, without perhaps invoking this aspect of the subject matter.
reji
>anyway, we seriously lack information about your system, your room , the drivers you use. the modification you made is huge and there are many things that could have led to your preference, so please enlighten us !
My system was built by myself about 10 years back. Some of you might have read the old Philips publication “building amplifier systems”. There is a 40 W/channel transistor amplifier circuit (class AB) given in that, which uses a quasi-complimentary output stage and + 60 volts power supply. The output stages used the lowly 2N3055 transistors. In todays standards this circuit is a humble design, but it was quite popular at that time, and sounds good even now. For this stereo amp I made sealed speaker boxes, each housing a 40 W woofer, and matching midrange and tweeter. All “Philips India” make. The company doesn’t give any info on the drivers except that they are 8 ohm. Both the woofer and midrange are paper cone. It is very likely that the midrange is nothing great – even I had a feeling that it had a low sensitivity compared to the other drivers. Simple 3-way passive crossovers were used, which had two inductors and two or three capacitors to the best of my memory (these were purchased, and I still don’t know the crossover frequencies). Anyways I liked the sound, and things were going well. Off and on I tried chip amps based on STK, TDA etc series, but the Philips amp sounded somewhat better than these. All these amps distorted at high volumes, but I usually listen to music at half the maximum loudness the Philips amp can provide, so given the logarithmic response of the ear, this should be about 4 W electrical power per channel.
Then last year I chanced to read about the P3A amp of Rod Elliot, and also about biamping and active crossovers. First I made the P3A. Even with 2N3055 output transistors, it sounded definitely better than the Philips amp. Then I made two more P3As. All had the same good performance. Then I removed the passive crossovers and measured the frequency response of each driver. Based on that measurement I chose the crossover frequencies and made the three-way 4th order L-R active filter. After removing the passive crossovers I hooked up everything, to hear a much better performance. But the result was definitely below my passive system, as I have detailed in the original post.
My drivers are not time-aligned. All-pass filters are not used anywhere in the circuit. I was not even aware of these things when the tests were made!
The room is rectangular, about 20 feet long and 10 feet wide. Furniture is scarce. For the comparison tests I always tested with one channel (the same channel of about 10 songs from a CD), as I had only three P3A amps with me. The speaker box was placed on top of a 3 feet tall table, at least 4 feet away from the nearest wall. On week days I can do the testing only in the night. The volume is quite low, as my better half is not that interested in profound analyses of music and its various nuances in the middle of the night! Hence I would estimate the electrical power used only at about 400 milliwatts.
Nevertheless, during weekends I have made a number of more tests at much higher volume levels, at daytime. For these tests I had to move the system to another 8 feet by 8 feet room, and the speaker box was very close to one corner, maybe half a foot away from the walls.
Since the active configuration did not give the desired results in both of the above situations, I went back to the passive 3-way crossover using one P3A amp, which sounded better. That is when I heard the 2-way speakers of a friend, and thought of trying out the same in my speakers. Yes, for doing this I removed the midrange, and connected the woofer directly to the amp, and the tweeter was connected in parallel to the woofer, with a 10 microfarad bipolar electrolytic capacitor in series with the tweeter. The result was better than my earlier 3-way passive and 3-way active configurations.
Sorry for the long description, but I thought it is better to write down all possible details.
> The TL082 is not a particularly good sounding opamp. I'd expect to hear its effects, period.
I am also a little suspicious of these opamps.
> In my experience, many listeners tend to mistake added distortions for "loudness" and or "presence."
I agree. But to the best of my belief, I am hopefully out of this category! It is gratifying that the discussions here are giving valuable inputs which seem to explain experiences such as mine from a more technical point of view, without perhaps invoking this aspect of the subject matter.
reji
Quoted
---The TL082 is not a particularly good sounding opamp.---
A TL08x (well a TL07x's...) will never spoil the audio quality of a crossover as much as inadequately chosen frequencies and slopes. As far as components give the right measurable results, design stays much more important than parts. For example, I think it is much preferable to use TL07x's which, anyway, will give the intended frequency response in high slope crossovers than to use an expensive super-audio-op-amp as a buffer in a 6 dB/o crossover.
---The TL082 is not a particularly good sounding opamp.---
A TL08x (well a TL07x's...) will never spoil the audio quality of a crossover as much as inadequately chosen frequencies and slopes. As far as components give the right measurable results, design stays much more important than parts. For example, I think it is much preferable to use TL07x's which, anyway, will give the intended frequency response in high slope crossovers than to use an expensive super-audio-op-amp as a buffer in a 6 dB/o crossover.
Quoted
---The TL082 is not a particularly good sounding opamp.---
For anybody thinking so I recommend a blind A/B test comparing the input and output audio signal going through ten TL082 inverting stages with say +6dB and -6dB gains interleaved... Who is able to tell which is which by ear without any hints? (Hint: Listen for increased noise floor when there is no music, because there isn't going to be much more to look for...)
---The TL082 is not a particularly good sounding opamp.---
For anybody thinking so I recommend a blind A/B test comparing the input and output audio signal going through ten TL082 inverting stages with say +6dB and -6dB gains interleaved... Who is able to tell which is which by ear without any hints? (Hint: Listen for increased noise floor when there is no music, because there isn't going to be much more to look for...)
Sorry I do not agree at all with the claim that one can not hear the diff between say TL082, 5532/43 & say OPA2604.
You might not hear the differences in your system.
Or you might not be able to hear the differences due to other causes.
I am here to assure you that it is completely trivial to hear differences both in the LF and the extreme HF audio spectrum between these opamps. Blind test or not.
Get the distortion down in ur system sufficiently, all will be revealed!
(fyi, "distortion" may or may not mean in this context merely a low measured distortion figure for any given component, as it has been shown that some spectra of distortion are orders of magnitude more audible than others... caveat emptor)
IF you hear no differences in your system or to your ears, then it makes no difference, so you can ignore the issue.
_-_-bear
PS. I completely agree that I'd prefer properly set up xover with "bad opamps" to badly set up xover with great opamps - but I'll take clean electronics plus properly set up xovers please!
You might not hear the differences in your system.
Or you might not be able to hear the differences due to other causes.
I am here to assure you that it is completely trivial to hear differences both in the LF and the extreme HF audio spectrum between these opamps. Blind test or not.
Get the distortion down in ur system sufficiently, all will be revealed!
(fyi, "distortion" may or may not mean in this context merely a low measured distortion figure for any given component, as it has been shown that some spectra of distortion are orders of magnitude more audible than others... caveat emptor)
IF you hear no differences in your system or to your ears, then it makes no difference, so you can ignore the issue.
_-_-bear
PS. I completely agree that I'd prefer properly set up xover with "bad opamps" to badly set up xover with great opamps - but I'll take clean electronics plus properly set up xovers please!
Hi,
As an aside, I thought this would be an interesting info:
See the thread titled “janneman” under the same category “solid state”. Go to post number 83 in page9, and download the .wav file from the link provided. It is a comparison between an LP and a CD of the same song.
It seems that for the first several seconds distortion is audible in the LP recording: however many people there seemed to prefer the LP recording, as it is more “lively” compared to the CD that was “flat”. There is no doubt that the CD would have measured better in terms of S/N ratio and other parameters.
I think the difference I heard between the 1st order passive and 4th order active filters was something of a similar kind (even though the source was the same CD for me in both cases) .
reji
As an aside, I thought this would be an interesting info:
See the thread titled “janneman” under the same category “solid state”. Go to post number 83 in page9, and download the .wav file from the link provided. It is a comparison between an LP and a CD of the same song.
It seems that for the first several seconds distortion is audible in the LP recording: however many people there seemed to prefer the LP recording, as it is more “lively” compared to the CD that was “flat”. There is no doubt that the CD would have measured better in terms of S/N ratio and other parameters.
I think the difference I heard between the 1st order passive and 4th order active filters was something of a similar kind (even though the source was the same CD for me in both cases) .
reji
Hi Bear,
>Let's skip this one?
I didn’t intend to make a CD-LP comparison in this thread..sorry to have made such an impression!
My only point was, the sonal difference I experienced was something of a similar kind, as heard from that .wav file. Don't know how good the analogy is, but it came pretty close.
reji
>Let's skip this one?
I didn’t intend to make a CD-LP comparison in this thread..sorry to have made such an impression!
My only point was, the sonal difference I experienced was something of a similar kind, as heard from that .wav file. Don't know how good the analogy is, but it came pretty close.
reji
The type of distortion from LP that is properly tracked is considered "euphonic". Whereas the type of distortion from tweeters that I was referring to is "anti-euphonic."
Perhaps you might be referring to an LP that is/was mistracking - in which case there is some similarity between "tweeter distortion" and mistracking. I think they sound somewhat different, but there is some similarity in terms of the nature of the distortion and the mechanisms at work.
_-_-bear
Perhaps you might be referring to an LP that is/was mistracking - in which case there is some similarity between "tweeter distortion" and mistracking. I think they sound somewhat different, but there is some similarity in terms of the nature of the distortion and the mechanisms at work.
_-_-bear
Hi Bear,
> Perhaps you might be referring to an LP that is/was mistracking - in which case there is some similarity between "tweeter distortion" and mistracking.
I really don't know whether the LP in question was mistracking..there is no mention of this in that thread. Perhaps it was.
> I think they sound somewhat different, but there is some similarity in terms of the nature of the distortion and the mechanisms at work.
That's it..it was similar, as well as different! Unfortunately now I am treading the subjectivist extremities, so let's come back to objective reasoning:
What I don't understand is the following - if the sonal signature of the first order crossover results from distortions, how could it be evident at such low volumes in which I did the comparison tests (I did these in the middle of the night, without disturbing others)? Perhaps the typical high sensitivity of tweeters is the culprit here?
reji
> Perhaps you might be referring to an LP that is/was mistracking - in which case there is some similarity between "tweeter distortion" and mistracking.
I really don't know whether the LP in question was mistracking..there is no mention of this in that thread. Perhaps it was.
> I think they sound somewhat different, but there is some similarity in terms of the nature of the distortion and the mechanisms at work.
That's it..it was similar, as well as different! Unfortunately now I am treading the subjectivist extremities, so let's come back to objective reasoning:
What I don't understand is the following - if the sonal signature of the first order crossover results from distortions, how could it be evident at such low volumes in which I did the comparison tests (I did these in the middle of the night, without disturbing others)? Perhaps the typical high sensitivity of tweeters is the culprit here?
reji
Note that with a different filter you are very likely to be getting quite a different frequency response and also a different polar response from your speakers (remember that off-axis sound radiation also ends up in our ears after being transformed by room acoustics). That's like comparing oranges to apples and is almost always going to sound different. Furthermore, a misaligned improvised first order filter is very likely to boost mid-high frequencies, which you may consider a pleasing effect at low volumes, specially if your midrange has a too low sensitivity (as usual) and you just skip it. I went through a similar experience with a pair of old 1975's 3-way enclosures back when I was a teenager, the midrange drivers lasted only a couple of weeks after lowering their crossover point, although they seemed to sound "better" that way...
On the other hand Matching filters of dissimilar orders to make a true comparison is not by any means an easy task.
On the other hand Matching filters of dissimilar orders to make a true comparison is not by any means an easy task.
What one thinks sounds good to one's ears need not be accurate sound. The brain can play all kinds of tricks on us.
Unless you hear the same sound in real life , you do not know what it should really sound like. Matter of fact is that most speakers are highly coloured and we tend to just get used to them ! Listen to an open baffle speaker and you will understand how boxy most boxed speakers sound .
I guess the only way most people can have a reference is to listen to well designed commercial speakers and then compare that against ones own system. This however leads one to the problem of room acoustics , music used , cables and electronic eqipment. Major problem ...huh ?
Unless you hear the same sound in real life , you do not know what it should really sound like. Matter of fact is that most speakers are highly coloured and we tend to just get used to them ! Listen to an open baffle speaker and you will understand how boxy most boxed speakers sound .
I guess the only way most people can have a reference is to listen to well designed commercial speakers and then compare that against ones own system. This however leads one to the problem of room acoustics , music used , cables and electronic eqipment. Major problem ...huh ?
The influence of the drivers resonance associated with crossovers on the overall response is not very well documented. There is an AES paper by Marshall Leach on passive filters, I lack other references.
So here are some simulations :
. Two ways crossover at 1000 Hz.
. First order filters (6 dB/o) ahead of the power amplifiers.
. The bass medium unit response is supposed to be ideal above 1000 Hz.
. The treble units have a frequency resonance at two or three octaves below the crossover frequency and various values of Q : 0.5, 0.7, 1.0 and 1.4.
Here are the deviations of the frequency response and the phase responses of the whole system,and the resonance rejection of the treble unit:
Q = 0.5
0250 Hz --> response -4.2 dB, phase -14° +12°, tweeter resonance -12 dB.
0125 Hz --> response -2.3 dB, phase -07° +06°, tweeter resonance -18 dB.
Q = 0.7
0250 Hz --> response -3.5 dB, phase -14° +10°, tweeter resonance -12 dB.
0125 Hz --> response -1.9 dB, phase -07° +05°, tweeter resonance -18 dB.
Q = 1.0
0250 Hz --> response -4.0 dB, phase -16° +09°, tweeter resonance -12 dB.
0125 Hz --> response -2.0 dB, phase -08° +04°, tweeter resonance -18 dB.
Q = 1.4
0250 Hz --> response -5.0 dB, phase -18° +10°, tweeter resonance -12 dB.
0125 Hz --> response -2.1 dB, phase -09° +04°, tweeter resonance -15 dB.
So, when using a first order crossover :.
. with a tweeter resonating at two octaves below the crossover frequency, additional correction is needed.
. it is only with a tweeter resonating at three octaves below the crossover frequency that the response is reasonably flat.
. in both the above cases, the phase deviations are low.
So here are some simulations :
. Two ways crossover at 1000 Hz.
. First order filters (6 dB/o) ahead of the power amplifiers.
. The bass medium unit response is supposed to be ideal above 1000 Hz.
. The treble units have a frequency resonance at two or three octaves below the crossover frequency and various values of Q : 0.5, 0.7, 1.0 and 1.4.
Here are the deviations of the frequency response and the phase responses of the whole system,and the resonance rejection of the treble unit:
Q = 0.5
0250 Hz --> response -4.2 dB, phase -14° +12°, tweeter resonance -12 dB.
0125 Hz --> response -2.3 dB, phase -07° +06°, tweeter resonance -18 dB.
Q = 0.7
0250 Hz --> response -3.5 dB, phase -14° +10°, tweeter resonance -12 dB.
0125 Hz --> response -1.9 dB, phase -07° +05°, tweeter resonance -18 dB.
Q = 1.0
0250 Hz --> response -4.0 dB, phase -16° +09°, tweeter resonance -12 dB.
0125 Hz --> response -2.0 dB, phase -08° +04°, tweeter resonance -18 dB.
Q = 1.4
0250 Hz --> response -5.0 dB, phase -18° +10°, tweeter resonance -12 dB.
0125 Hz --> response -2.1 dB, phase -09° +04°, tweeter resonance -15 dB.
So, when using a first order crossover :.
. with a tweeter resonating at two octaves below the crossover frequency, additional correction is needed.
. it is only with a tweeter resonating at three octaves below the crossover frequency that the response is reasonably flat.
. in both the above cases, the phase deviations are low.
My system is a two way bi-amped MTM system. I put together the Marchand active crossover, and liked it. Then I put together a line level passive crossover, using the BOSOZ pre amp as a buffer. The BOSOZ has gain controls that do not change the charactistic impedence of the circuit, making the BOSOZ very suitable as a buffer in a line level passive crossover, because you can change the gain of the treble or bass channel with out changing the response curve of the passive filter. The passive filter is an inductor and a capacitor.
This implementation sounds better than the Marchand Active filter.
I think it is because the Marchand uses op amps as buffers, and the signal is going through an awful lot of transistors. The BOSOZ is very simple, but flexible, and opens up the sound a lot.
This implementation sounds better than the Marchand Active filter.
I think it is because the Marchand uses op amps as buffers, and the signal is going through an awful lot of transistors. The BOSOZ is very simple, but flexible, and opens up the sound a lot.
It was a Nelson Pass project in Audio Xpress magazine. You can find it on Nelson's site or Audio Xpress. They have boards and directions for the project. This is a particularly good sounding pre-amp...naturally low distortion with not much feed back. The 60 volt rails have something to do with this.
Importantly, the BOSOZ (Bride of Son of Zen) has gain controls that change the intrinsic gain of the circuit without changing the output or input impedance of the circuit, making it perfect for a Line Level Passive Crossover. you can change the level on the high or low channel with out changing the response of the filter.
Importantly, the BOSOZ (Bride of Son of Zen) has gain controls that change the intrinsic gain of the circuit without changing the output or input impedance of the circuit, making it perfect for a Line Level Passive Crossover. you can change the level on the high or low channel with out changing the response of the filter.
Hi juergenk,
I am glad you liked my work. Note that the voice-coils are supposed to be verticaly aligned. A fact which many addicts of first order filters prefer to ignore is, that for constant voltage sinewaves at the system input, with these filters, the voice coil displacement of the treble unit under the crossover frequency increases until its resonance is met, although its level decreases. At one octave below the crossover frequency, the displacement has doubled, at two octaves, it is multiplied by 4.
Some more simulations :
. Two ways crossover Linkwitz-Riley (12 dB/o)
. Second order filters at 1000 Hz ahead of the power amplifiers.
. The bass medium unit response is supposed to be ideal above 1000 Hz.
. The drivers are supposed to be verticaly aligned.
. The treble units have a frequency resonance at 1, 1.4 and 2 octaves below the crossover frequency and various values of Q : 0.5, 0.7, 1.0 and 1.4.
Here are the deviations of the frequency response of the whole system and the resonance rejection of the treble unit:
Q = 0.5
0500 Hz --> response -2.3 dB, tweeter resonance -20 dB.
0353 Hz --> response -1.6 dB, tweeter resonance -25 dB
0250 Hz --> response -1.0 dB, tweeter resonance -31 dB.
Q = 0.7
0500 Hz --> response -2.0 dB, tweeter resonance -17 dB.
0353 Hz --> response -1.0 dB, tweeter resonance -22 dB
0250 Hz --> response -0.7 dB, tweeter resonance -28 dB.
Q = 1.0
0500 Hz --> response -1.7 dB, tweeter resonance -14 dB
0353 Hz --> response -1.0 dB, tweeter resonance -19 dB.
0250 Hz --> response -0.5 dB, tweeter resonance -25 dB.
Q = 1.4
0500 Hz --> response -2.2 dB +0.5 dB, tweeter resonance -11 dB
0353 Hz --> response -1.3 dB +0.2 dB, tweeter resonance -16 dB.
0250 Hz --> response -0.7 dB +0.1 dB, tweeter resonance -22 dB.
So, when using a second order Linkwitz-Riley crossover, the crossover should preferably be set at 1.4 octave or more above the resonance frequency of the high-passed unit.
Note that now, for constant voltage sinewawes at the system input, the voice coil displacement of the treble unit under the crossover frequency remains constant until its resonance is met, although its level decreases.
I am glad you liked my work. Note that the voice-coils are supposed to be verticaly aligned. A fact which many addicts of first order filters prefer to ignore is, that for constant voltage sinewaves at the system input, with these filters, the voice coil displacement of the treble unit under the crossover frequency increases until its resonance is met, although its level decreases. At one octave below the crossover frequency, the displacement has doubled, at two octaves, it is multiplied by 4.
Some more simulations :
. Two ways crossover Linkwitz-Riley (12 dB/o)
. Second order filters at 1000 Hz ahead of the power amplifiers.
. The bass medium unit response is supposed to be ideal above 1000 Hz.
. The drivers are supposed to be verticaly aligned.
. The treble units have a frequency resonance at 1, 1.4 and 2 octaves below the crossover frequency and various values of Q : 0.5, 0.7, 1.0 and 1.4.
Here are the deviations of the frequency response of the whole system and the resonance rejection of the treble unit:
Q = 0.5
0500 Hz --> response -2.3 dB, tweeter resonance -20 dB.
0353 Hz --> response -1.6 dB, tweeter resonance -25 dB
0250 Hz --> response -1.0 dB, tweeter resonance -31 dB.
Q = 0.7
0500 Hz --> response -2.0 dB, tweeter resonance -17 dB.
0353 Hz --> response -1.0 dB, tweeter resonance -22 dB
0250 Hz --> response -0.7 dB, tweeter resonance -28 dB.
Q = 1.0
0500 Hz --> response -1.7 dB, tweeter resonance -14 dB
0353 Hz --> response -1.0 dB, tweeter resonance -19 dB.
0250 Hz --> response -0.5 dB, tweeter resonance -25 dB.
Q = 1.4
0500 Hz --> response -2.2 dB +0.5 dB, tweeter resonance -11 dB
0353 Hz --> response -1.3 dB +0.2 dB, tweeter resonance -16 dB.
0250 Hz --> response -0.7 dB +0.1 dB, tweeter resonance -22 dB.
So, when using a second order Linkwitz-Riley crossover, the crossover should preferably be set at 1.4 octave or more above the resonance frequency of the high-passed unit.
Note that now, for constant voltage sinewawes at the system input, the voice coil displacement of the treble unit under the crossover frequency remains constant until its resonance is met, although its level decreases.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.