Excellent! That's exactly the kind of stuff I was looking for earlier, but you lot made it sound too easy and scientific. So it's something where 'optimal' is hard to define; the designer is aware that this may be a problem, but only listening could tell him how much of one. He has to make a decision whether to permanently alter the frequency response for 90% of all listening in order to improve the 10%, or have the speaker sound a bit crap occasionally at high volumes.
With that kind of skill and experience I'd be reluctant to give way to the digital upstarts, too.
Okay, so we have one reference. What passive setup is outperformed by that active analog crossover?
It is interesting to see what an OPA2134 can do. I have the opamp, and I have many other opamps much better than that. Do you have the schematic for the active crossover?
Bingo. I'm spoken in person to 'phools about active systems, and every single time, I've been asked 'but what if I want to try a different amp?' or similar. It's more about playing with this week's shiny new toy.
That might be your idea of fun, but if you can distinguish the sonic signature of an amp, it is faulty.
I disagree.
Distortion = magic? I'll pass on that thanks. Don't like the IMD that goes along with it.
a) To my ears every passive speaker I've ever heard. I listen at pretty much every level, from background to lifelike. Hence I want speakers capable of 115-120dBspl without running out of puff. I am also a firm believer in plenty of headroom.
b) Try the BSS website. I know there are better op amps but they are not drop in replacements for the ones which were in there. Also I got the 2134 very cheap indeed (about £0.70 ea.) as they were old non-RoHS compliant stock.
The better ones cost between £4 and £7 each and required extra bypass caps etc.
I build my own speakers but I do not do electronics.
Audio system should be seen and designed as a complete chain from source to driver. Because they affect one and other to achieve the final objective.
Don't forget that tubes are "slow". Anyway, like this (attached), or something much better?
Attachments
Problem is I don't know what you have ever heard. An active analogue crossover will have a hard time to compete with high speed transparent system. Non-digital system? Class-A amps?
But I have mentioned several times that part of the difficulty is taste and the ability to listen to fatigue (and how important this fatigue is to the listeners).
When we are talking about taste (and experience to some extents), it is really difficult. Why? If you sell amps to the low level crowd, they always want the boom-boom-criss-criss system no matter what. It's the time for the theoreticians to give them what they want, instead of debating what system is best.
Thanks, hopefully I can get something good from the site.
I build my own speakers and electronics. I wish I can eventually build a top class active analog system... tho I'm very pessimistic about it.
My OPA2134 are of course of good quality as I removed them from commercial stuffs. And so are my other opamps
Exactly and what better way to ensure system compatibility then to have the crossovers amplifiers and loudspeaker drive units all built into the same cabinet, by the same designer/manufacturing team.
Claiming that having it the other way around, with a passive loudspeaker and then allowing some random no body to pick the amplifier, is a better option, is just silly.
Yes. Probably
I'm not a theorist (many theorists don't know what they think they know). But I build my own amps and speakers. So I'm more a practitioner. Long time before I laid my hands with electronics I already know and planned to move to active someday. I have started, but I'm not there yet. You know what, you show me the best schematic, and I will build it. Is there anything else more practical than that?
You mentioned about tube crossover. I thought you want to proof that the crossover doesn't affect the amp because it doesn't sit between amp and speaker
So from your point of view (taste) tube is not a solution for a system with active crossover
How important do we think linear phase filters are? Passives and analogue actives both seem to mangle the overall phase in seemingly arbitrary ways depending on topology. I'm not claiming to be able to consciously hear the difference, but isn't it nice to remove that variable from the equation as well?
They don't mangle them in arbitrary ways, they mangle them in a way that is 100% predictable and for the sake of the filters in question, desirable too. If you shape a standard drivers response into say a 4th order acoustic slope with a Q of 0.5, then it's called a Linkwitz-Riley type slope and this has an associated phase shift where the phase rotates once through 360 degrees as you approach and pass through the crossover region. Now which ever way you go about arriving at this 4th order roll off with a Q of 0.5, it will have the associated phase shift unless you go with a crossover type that actively cancels out this phase shift so that the end result is linear phase but still with the same amplitude response.
Now if you take two drivers and make them co-incident, such as a co-axial driver, and apply filters so that both the high pass and the low pass on the respective drivers are perfect Linkwitz-Riley slopes and at the same frequency, then the two drivers, if wired in phase, will sum constructively with a flat amplitude response. If you reverse the phase of one of the drivers then they both cancel out completely at the xover frequency as one driver is now completely out of phase with the other. If the phase response of one driver happened to be changed using an all pass filter, or was made linear phase then the two drivers would no longer sum together properly around the xover frequency and they would no longer have a flat amplitude response as a result.
Now regardless of if you want to use linear phase filters you still need to make sure that the drivers are knitted together using suitable slopes. This means applying filters to the individual drivers that mould its frequency response into a standard slope type, so that after the filter is applied its measured frequency response = the desired slope type. Usually people will choose Linkwitz-Riley slopes as they tend to have the most desirable attributes. If you're not going linear phase the phase response of both the tweeter and the woofer will compliment each other, so that if they have their acoustic centres aligned they will sum properly. If they are not aligned then this is where you delay one driver to ensure that they are, bringing the two drivers into phase/summation with one another.
If you are going linear phase though you will still go about things in exactly the same way, except that there will be no associated phase shift through the xover point. The two drivers still need to have amplitude responses that compliment each other and suit the drivers in question and they still need to be aligned correctly so that they sum properly. There might not be any phase shift, but the two drivers still need to be aligned correctly for things to work correctly.
Now you can either go about things by using linear phase filters from the start, or you can use typical filters and then apply a correction curve afterwards that linearises the phase of the system. For either way to work correctly though you need to measure the drivers so that you can apply the suitable correction so as to make the system properly linear phase.
Speaking of the phase shift that occurs though when you use standard filters, this is normally a good thing. Usually when you put two drivers on a flat baffle their acoustic centres wont match and they also wont be co-incident, this is fairly standard as most drivers aren't coaxial and most drivers don't have their acoustic centres in the same place. This might sound like a problem at first but it usually isn't, this is because when you're designing a passive crossover to integrate the drivers, you can massage the filter types gently (so that they aren't the same), but this allows you to bring the two drivers into phase with one another, while, if done correctly, having hardly any effect on the axial flatness. Usually using asymmetrical slopes will affect the lobing in the vertical off axis, but it's a small price to pay for bringing the two drivers into correct alignment with each other.
Now if you take two drivers and make them co-incident, such as a co-axial driver, and apply filters so that both the high pass and the low pass on the respective drivers are perfect Linkwitz-Riley slopes and at the same frequency, then the two drivers, if wired in phase, will sum constructively with a flat amplitude response. If you reverse the phase of one of the drivers then they both cancel out completely at the xover frequency as one driver is now completely out of phase with the other. If the phase response of one driver happened to be changed using an all pass filter, or was made linear phase then the two drivers would no longer sum together properly around the xover frequency and they would no longer have a flat amplitude response as a result.
Now regardless of if you want to use linear phase filters you still need to make sure that the drivers are knitted together using suitable slopes. This means applying filters to the individual drivers that mould its frequency response into a standard slope type, so that after the filter is applied its measured frequency response = the desired slope type. Usually people will choose Linkwitz-Riley slopes as they tend to have the most desirable attributes. If you're not going linear phase the phase response of both the tweeter and the woofer will compliment each other, so that if they have their acoustic centres aligned they will sum properly. If they are not aligned then this is where you delay one driver to ensure that they are, bringing the two drivers into phase/summation with one another.
If you are going linear phase though you will still go about things in exactly the same way, except that there will be no associated phase shift through the xover point. The two drivers still need to have amplitude responses that compliment each other and suit the drivers in question and they still need to be aligned correctly so that they sum properly. There might not be any phase shift, but the two drivers still need to be aligned correctly for things to work correctly.
Now you can either go about things by using linear phase filters from the start, or you can use typical filters and then apply a correction curve afterwards that linearises the phase of the system. For either way to work correctly though you need to measure the drivers so that you can apply the suitable correction so as to make the system properly linear phase.
Speaking of the phase shift that occurs though when you use standard filters, this is normally a good thing. Usually when you put two drivers on a flat baffle their acoustic centres wont match and they also wont be co-incident, this is fairly standard as most drivers aren't coaxial and most drivers don't have their acoustic centres in the same place. This might sound like a problem at first but it usually isn't, this is because when you're designing a passive crossover to integrate the drivers, you can massage the filter types gently (so that they aren't the same), but this allows you to bring the two drivers into phase with one another, while, if done correctly, having hardly any effect on the axial flatness. Usually using asymmetrical slopes will affect the lobing in the vertical off axis, but it's a small price to pay for bringing the two drivers into correct alignment with each other.
Last edited:
Thats a pretty unreasonable and self-centered approach. These are considerations for most people.
Of course your own system sounds best to you! Daa!
This is not quite true. There are things that can be done to make thermal changes in the design (dominately the drivers DC resistance) more or less of a factor. I am not going to give away my secrets, but just think about it for a moment - how could I make the drivers impedance less of a factor in the design?
I've check other components and only the drivers are prone to significant changes. I mean this too only makes sense - where is all the power disipated? Not in the crossover components I hope.
Further the amount of these changes is also a factor. Small coils in inefficient drivers fare far far worse than larger more efficient ones (sometimes as much as ten times worse). Basically, if done correctly thermal changes can be made to be insignificant - but they must be paid attention to!
You mentioned about tube crossover. I thought you want to proof that the crossover doesn't affect the amp because it doesn't sit between amp and speaker
So from your point of view (taste) tube is not a solution for a system with active crossover
If accuracy of reproduction is your aim than tube amps anywhere in the system are not the solution.
Years ago Bob Carver demonstrated that it is quite easy to use a cheap ss amp to mimic a very expensive tube amp well enough to fool practically any listener. It seems not possible to do the reverse, this should tell you something about the accuracy of tube amps.
And since according to him 90% of the job was to stick a resistor on the ss amps output it should also tell one something about the audibility of an amps damping factor.
In the end it all depends if you want to satisfy your taste or if you want to be accurate. I know what I want, your aims may differ.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.