that is the only 'right way' to do it, and how its 'always' been done, when doing it properly
anything less is 'cheating'
though, simulations have proven to be impressively accurate, if you know what you are
doing
the pro guys doing passive designs 'the right way' will measure each driver raw
and then 'play' with simulations
mount the proto speaker with chosen components
and measure the completed speaker
and if not happy with result, back to try other simulation, and....so forth
The bass strategy is either 1) a nearfield measurement + simmed baffle step response, spliced onto the far-field measurement at a frequency before the gated far-field starts to lose accuracy. 2) A ground plane measurement done in the correct way and again most likely spliced onto a gated far-field response. The ground plane method actually measures the effect of baffle step rather it needing to be simulated. 3) An out door measurement with the loudspeaker and microphone lifted several meters above ground level to delay the arrival of the first reflection and to extend the far-field responses accuracy. If you can get things high enough, easily enough, this will mean you only need to take one measurement for all the relevant design work.
The first method tends to work very well, mainly because with typical loudspeakers in a typical vacated room, you will be able to measure the effects of diffraction without any problem. This is because they tend to occur at a high enough frequency as to be included within the standard gating required to remove the first major reflection. After the diffraction effects comes baffle step and this tends to be very predictable and smooth, so a simulation is fine, what you need however is the main diffraction effects + the start of baffle step to be shown in the far-field measurement so you can match the nearfield/simulated response accurately to it. This obviously doesn't work for open baffle loudspeakers as you need to remain in the farfield to measure the dipole cancellation.
Key ability is generation of inverse for measured impulse response(s). This is basis for Accourate and other room correction software.
Most of these packages stress room correction, and skip speaker correction. In efforts to make SW work well with measurements of speaker in room with microphone at listening position most adopt frequency dependent windowing methods to simultaneously gate out early reflections for high frequencies while keeping LF information.
Greatest difficulties occur at frequencies causing deep nulls in measured response because corresponding inverse filter generates big peak.
Nulls can be from both room and from speaker baffle.
I generate inverse filters with Nelson/Kirkeby based software: Aurora Plug-ins
Originally developed for Cool Edit/Adobe Audition, and recently compile for Audacity:
Index of /Public/Aurora-for-Audacity
Definitely a learning curve is involved.
Correction of room response in sparse mode region for LF is effectively speaker correction. Likewise windowing and gating techniques of listener position type measurements for high frequency correction is speaker correction.
So, IMO room correction starts with speaker correction.
With well behaved small speaker:
An externally hosted image should be here but it was not working when we last tested it.
Measurements are made from close up, and no gating/windowing of impulse responses is needed prior to generating inverse filters.
Above speaker is based on Linkwitzlab Pluto. System may be implemented with multichannel sound card and four channels of amplification.
This speaker effectively has no baffle step.
I have generated reverse null using swept sine, MLS, and band limited burst signals.
Regards,
Andrew
No, please don't leave us in the darkness!
You teach us that maximum of power transfer between the amp and the speaker is required. For maximum power transfer output resistance of the amp have to match input resistance of the speaker, so they have to be equal on all audio band. How do you match it in tube amplifiers? And you mentioned already Zobel to make load resistance of the amp flatter with frequency. That means on higher frequencies an energy that less transferred to the speaker is more transferred to Zobel. What is the purpose of such network, if it decreases maximum power transfer from amplifier to the speaker?
Here again, the graph that represents impedance of a typical so called "Tube - friendly speaker".
An externally hosted image should be here but it was not working when we last tested it.
I suspect part of the advantage of Zobel networks is that they can minimize the reactive component of the speaker load as seen by the amplifier, and that with the more constant load probably results in slightly better amplifier transient performance and less frequency dependent distortion products.
I found it interesting that I was able to get much of the same impedance leveling effect of a Zobel network in the Iron Lawbreakers by winding up an air coupled transformer which smoothed the input impedance to 18 ohms +/-20% between 40hz and 240hz by effectively varying the voltage/phase transfer function with frequency rather than dissipating power. This worked great with the low Qts woofer I used (JBL 2220A) BR loaded because it was able to convert the upper BR impedance peak to an amplitude boost of a couple db in the 60 hz region (with the damping approach I used the upper impedance peak was only about 32 ohms so not too much more boost was available when normalizing the impedance around 16 ohms) while cutting the amplitude around 200 hz and above by around 4db for an essentially flat impedance and amplitude transfer characteristic with almost no resistive power loss.
This air coupled transformer does double duty as a series pass inductance for the woofer LP filter. As far as 'theory' behind this, I'm not aware of anything specific to speaker xovers. This development process was very empirical, but it worked satisfactorily for the task at hand.
I found it interesting that I was able to get much of the same impedance leveling effect of a Zobel network in the Iron Lawbreakers by winding up an air coupled transformer which smoothed the input impedance to 18 ohms +/-20% between 40hz and 240hz by effectively varying the voltage/phase transfer function with frequency rather than dissipating power. This worked great with the low Qts woofer I used (JBL 2220A) BR loaded because it was able to convert the upper BR impedance peak to an amplitude boost of a couple db in the 60 hz region (with the damping approach I used the upper impedance peak was only about 32 ohms so not too much more boost was available when normalizing the impedance around 16 ohms) while cutting the amplitude around 200 hz and above by around 4db for an essentially flat impedance and amplitude transfer characteristic with almost no resistive power loss.
This air coupled transformer does double duty as a series pass inductance for the woofer LP filter. As far as 'theory' behind this, I'm not aware of anything specific to speaker xovers. This development process was very empirical, but it worked satisfactorily for the task at hand.
Last edited:
Zobel network is used in amps with global feedback that can be unstable without it. It adds losses on frequencies on which the amp rings or oscillates, they are usually above audible band. And they have nothing to do with amp- speaker frequency response. And absolutely irrelevant to "maximum power transfer"
"You can't do this with passive crossovers,"
Actually, you can, and I have. Perhaps not below 100hz, but I have gotten recognizable square waves from my Basement Blasters between 130 hz to above 2khz with an all passive xover. And JBL has done it better than I with some of their classic pro monitors, which they have written up on their website.
However, unless you have an ideal point source radiator in an essentially reflection free, near anechoic environment, getting such 'pretty' waves is definitely a 'head in vice' proposition for any fixed eq. whether digital or analog.
There is also the limitation with steep slope digital xover filters, separate from constant overall group delay alignment per se, that with any but an ideal inter-driver alignment, you are likely to get substantial Gibbs-type ringing on transients that manifests itself as an easily identifiable coloratioin.
Actually, you can, and I have. Perhaps not below 100hz, but I have gotten recognizable square waves from my Basement Blasters between 130 hz to above 2khz with an all passive xover. And JBL has done it better than I with some of their classic pro monitors, which they have written up on their website.
However, unless you have an ideal point source radiator in an essentially reflection free, near anechoic environment, getting such 'pretty' waves is definitely a 'head in vice' proposition for any fixed eq. whether digital or analog.
There is also the limitation with steep slope digital xover filters, separate from constant overall group delay alignment per se, that with any but an ideal inter-driver alignment, you are likely to get substantial Gibbs-type ringing on transients that manifests itself as an easily identifiable coloratioin.
Last edited:
From a loudspeaker/amplifier point of view, if you compare an 8 ohm load with no reactance, to an 8 ohm load with a severe phase angle, the current delivered to the load is exactly the same, it's that the phase angle changes the current delivery from inside the amp by placing greater stress on one half of the output stage.
The reason why a varying impedance affects the frequency response of a tube amplifier is because of the tube amps higher output impedance. If the output impedance is very low then regardless of what the loudspeakers impedance is like 99% of the voltage drop on the amplfier output will be delivered to the loudspeaker. However if the output impedance rises to a value that is significant in relation to the impedance of the loudspeaker then you're going to get a varying voltage applied to the loudspeakers terminals as its impedance changes.
If the loudspeakers impedance is flat then you wont get a frequency dependent voltage shift on the terminals that occurs as the relationship between the amplifiers output impedance and the loudspeakers own impedance varies (providing the amplifiers output impedance remains constant at audio frequencies too).
With a tube amp it isn't that the loudspeaker is easy to drive as per say, as the absolute impedance of the loudspeaker doesn't matter (within limits) providing it is matched to the amplifiers output stage via a suitable transformer. What matters is that the impedance is flat with respect to frequency.
As a result Wavebourn the impedance plot that you've posted isn't perfectly suited to a valve amplifier at all. For it to be better, a zobel needs to be applied to cancel out the inductive rise. Technically speaking this will most likely make the loudspeaker harder to drive as the impedance will drop, increasing the current demand placed on the amplifier. But from a voltage perspective as viewed from of the loudspeaker, in the presence of a higher amplifier output impedance, it will now appear far more constant with the zobel then without.
As has already been mentioned in this thread though, none of this matters providing the system has taken these effects into account when it was being designed.
The reason why a varying impedance affects the frequency response of a tube amplifier is because of the tube amps higher output impedance. If the output impedance is very low then regardless of what the loudspeakers impedance is like 99% of the voltage drop on the amplfier output will be delivered to the loudspeaker. However if the output impedance rises to a value that is significant in relation to the impedance of the loudspeaker then you're going to get a varying voltage applied to the loudspeakers terminals as its impedance changes.
If the loudspeakers impedance is flat then you wont get a frequency dependent voltage shift on the terminals that occurs as the relationship between the amplifiers output impedance and the loudspeakers own impedance varies (providing the amplifiers output impedance remains constant at audio frequencies too).
With a tube amp it isn't that the loudspeaker is easy to drive as per say, as the absolute impedance of the loudspeaker doesn't matter (within limits) providing it is matched to the amplifiers output stage via a suitable transformer. What matters is that the impedance is flat with respect to frequency.
As a result Wavebourn the impedance plot that you've posted isn't perfectly suited to a valve amplifier at all. For it to be better, a zobel needs to be applied to cancel out the inductive rise. Technically speaking this will most likely make the loudspeaker harder to drive as the impedance will drop, increasing the current demand placed on the amplifier. But from a voltage perspective as viewed from of the loudspeaker, in the presence of a higher amplifier output impedance, it will now appear far more constant with the zobel then without.
As has already been mentioned in this thread though, none of this matters providing the system has taken these effects into account when it was being designed.
I think the 'this' that he's referring to is really the combination of 16th+ order filters with the ability to reproduce square waves faithfully.
Thank you for the explanation. It reminds me,
"This transceivers are installed on tanks".
"Sir, are that transceivers on valves or on semiconductors?"
"Let me repeat for you and for the rest of dumbest soldiers: on tanks!"
I feel I need to study carpentry instead of physics and electronics, to understand your logic.
People are still discussing the supposed advantages of the holy passive series crossovers, but It amazes me that seemingly nobody has persisted to point out the big white elephant in the room that give active mulitway system a big advantage over a conventional mono-amplified passive-crossovered multiway speaker:
#1 - The reduction of intermodulation distortion (IM) at the amplifier(s) because each amplifier's working bandwidth is significantly narrowed. An amp no longer has to simultaneously cope with big 40Hz peaks and delicate 10-20KHz content riding on top of the same signal. The reduction of IM afforded by this is (in my opinion) worth the price of admission by itself, and in my opinon THIS is the main reason to go for fully active.
This was posted in page 1 or page 2, but it seems it got lost in later pages.
Other important advantages are:
#2 - Which allows us to use a low powered Class A (or similarly high quality) amplifier for the mids and tweeters. In other words, something closer to the ideal amplifier, with big gains in sound transparency.
#3 - A steeper filter slope (i.e. 24dB/oct for a typical active XO versus 12dB/oct) makes for better integration between drivers and better prevents them from receiving signals of frequencies where they are not linear, diminishing distortion further. The less need for overlap between driver frequency responses, the better the integration is.
#4 - Higher efficiency (due to no more power lost in the crossover), which enables lower power requirements, which means, again, we can go for a Class A solution and/or gain more headroom.
#5 - (actually part of #1) the big impedance peak brought by the woofer (and it's reactance too) will have no influence over the mids and highs, since they're handled by a different amplifier. Real amplifiers are far from ideal, and ideally they want to drive a flat impedance device.
#1 - The reduction of intermodulation distortion (IM) at the amplifier(s) because each amplifier's working bandwidth is significantly narrowed. An amp no longer has to simultaneously cope with big 40Hz peaks and delicate 10-20KHz content riding on top of the same signal. The reduction of IM afforded by this is (in my opinion) worth the price of admission by itself, and in my opinon THIS is the main reason to go for fully active.
This was posted in page 1 or page 2, but it seems it got lost in later pages.
Other important advantages are:
#2 - Which allows us to use a low powered Class A (or similarly high quality) amplifier for the mids and tweeters. In other words, something closer to the ideal amplifier, with big gains in sound transparency.
#3 - A steeper filter slope (i.e. 24dB/oct for a typical active XO versus 12dB/oct) makes for better integration between drivers and better prevents them from receiving signals of frequencies where they are not linear, diminishing distortion further. The less need for overlap between driver frequency responses, the better the integration is.
#4 - Higher efficiency (due to no more power lost in the crossover), which enables lower power requirements, which means, again, we can go for a Class A solution and/or gain more headroom.
#5 - (actually part of #1) the big impedance peak brought by the woofer (and it's reactance too) will have no influence over the mids and highs, since they're handled by a different amplifier. Real amplifiers are far from ideal, and ideally they want to drive a flat impedance device.
Last edited:
As long as the amplifiers are close enuff in character that the they don't give away the XO point.
dave
Dave, pick any two woofer and sqwawker (or a sqwawker and a tweeter) and they will be more mismatched than the most dissimilar amplifiers!!
Unless, of course, they're the same driver but with different cone diameter.
I like such threads, they are all miraculously created by freshly registered users, and questions in style, "What is better, red or sour?" can't leave intact neither those who like sweet taste, nor red color, as well as those who assume that red means apple, or those who assumes red means pepper, and so on.
I like such threads, they are all miraculously created by freshly registered users, and questions in style, "What is better, red or sour?" can't leave intact neither those who like sweet taste, nor red color, as well as those who assume that red means apple, or those who assumes red means pepper, and so on.
I don't understand your point. The goal here is the accurate reproduction of the original source signal. This shouldn't be about what suits your taste better, but what works better, and why should it work better.
Yes Wavebourne, I agree. And then we get strange postings like this:
What I see is basically those who understand the advantages of crossovers both passive and active, and those who tell us that only active are worthwhile. That seems like a purely technical for technique's sake argument. (see above). Reality isn't so neat and tidy.
Who has referred to passives as holy? Is it in this thread somewhere? If it is, I have missed it.
What I see is basically those who understand the advantages of crossovers both passive and active, and those who tell us that only active are worthwhile. That seems like a purely technical for technique's sake argument. (see above). Reality isn't so neat and tidy.
I apologyize, sometimes i try to be funny but come across as looking rude. My point was that sometimes one needs to find what the big advantage (and big disadvantage) would reside in, before getting too involved in a minor subtlety.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.