sounds like a 'standard' plug and play speaker
and if its got a 'standard' crossover that could cost like 20EUR, and you have replaced it with amps and active crossover that cost more than both speakers, and not to forget the Eq, I guess you would expect some positive effect
Well, damping factor plays little roll in controlling the driver at any frequency other than its resonance. The flat band part of any driver's response is referred to as the mass controlled region for a reason. The effect of damping drops off at 6dB/octave to both sides of the impedance resonance peak. The effect of the suspension compliance is constant below the resonance peak and drops off at 12dB/octave above the peak. I would agree that insertion loss does rob a small amount of power but in a home environment this really isn't of significant concern unless you are trying to drive the system with an under rated power amp. As for additional nasties from inductors that pretty much a stretch. Properly chosen inductor will be extremely linear in their operating range.
I pointed out that actives can make design simpler, particularly digital. But there are other factors with actives. For example the quality and design of the active crossover itself. Active devices introduce distortion. The circuits in most active crossovers are more complex and convoluted than you will find in most preamplifiers and power amps. Digital is another source of error. Certainly in theory an digitally sampled signal can be reconstructed exactly. But in reality digital algorithms use finite math with introduces round off error which limits accuracy. There is an uncertainty kind of relationship between the analog world and the digital one. With analog it is possible to reproduce a signal exactly, but you can never know its exact value. With digital you can set the limit of accuracy so you know the exact digital value within some error. For example, with 2 digits to the right of the decimal point you can have 1.00 or 0.99 but there is no way to express an amplitude of 0.993 or 0.994 or distinguish between them. And when these numbers are multiplied together the errors can grow. A good example here is the banking industry. Money is digital. Taxes, like sales taxes are in % which can lead to fractional values below 1 cent. Everything has to be rounded off to the nearest cent. Adding, subtracting, multiplying, dividing these numerical values can introduce round off errors which have to be accounted for. In digital processing it is not necessary to track these errors. It is only necessary to make sure they remain below a certian value.
I'm not dissing active crossovers. I use active crossover between the woofer and mid/tweeter panels of all my speaker designs, and where practical I offer fully active options, both analog and digital.
What I object to in these threads is these blanket statements that A is better than B. The OP has started at least 3 threads on this topic:
1 this one,
2) What do you think of passive crossovers? I see few projects here that feature very expensive drivers and a lot of time designing and making boxes. Then they use passive crossovers which are convenient but are plagued by wrecking the damping factor of the amplifier thus causing distortion especially at low frequencies that causes that sloppy bass everyone is trying to get rid of here. Some confuse this distortion with loud bass. Overall distortion can be measured in the order of 0.1% and up with passives although a nice couple of op amps are in the 0.0001 region with unity gain could that be a thousand times better? Perhaps theoretically but you have to agree that they do have a lot lower distortion!
3) Passive crossovers... Some hate them and will never use anything other than active, they say due to high distortion brought on by amplifier damping. What do you think? Do you like the sound of active over passive?
These all seem to address the damping issue, which as I have discussed, is pretty much a red herring. In fact, the argument is easily turned upside down. After all, ignoring the VC inductance for simplicity, the relationship between the voltage across the VC and current is V = (If - Ib)/Re where If is the forward current and Ib is the back current (from the back emf). This is expressed at V = I/Zd where V is the applied voltage (by the amp) and Zd is the driver's impedance. Obviously, at resonance Zd is a maximum which is a result of Ib being a maximum. Since it is I that generates the force with influences the driver's motion, and since for any value of V the current flowing through the VC will be a minimum at resonance compared to frequencies to either side of resonance, it seems that the amplifier would actually have the least control over the driver at resonance due to the maximum in Ib. That is, since the musical signal supplied by the amp is a voltage as a function of time, V(t), when V(t) has frequencies near or at resonance the current across the VC will be at a minimum and therefore the amp will have the least influence over the driver's motion. The fact is that most of the arguments on damping address what happens when V(t) goes to zero and the driver must return to its rest state. The point here is that whether an active or passive system, musical instruments are basically highly resonant devices that ring on for many cycles. And in any typical musical recording V(t) does not abruptly drop to zero, but decays over time with some inherent wave form. For example, you could think of a highly damped woofer as crashing a car into a concrete wall. A woofer with lower damping would be like crashing a car into a concrete wall with 10 ft of foam padding on it. The car's velocity is V(t). Music usually ends as if you were putting your foot on the brakes to modulate V(t) so that you stop before you hit the wall. In such a case, it makes little difference what the properties of the wall are. Q of a woofer system, with in reason, has more effect on the amplitude and phase around resonance than it does on how the driver stops compared to an applied musical signal.
And the bottom line is that if there are concerns with, for example, the damping of the woofer in a multi-way speaker with passive crossover, measure it! It's not too hard to do.
Anyway, sorry to take up so much space, but blanket statements by proponents of active or passive crossovers regarding the merits of one over the other just don't fly. Like anything else in life, it's a matter of execution. To get good results you have to do it right.
Hi,
All starting with the premise active must be better than passive. In an ideal
world it is, but not in a practical world depending on the circumstances,
the real world is cost versus performance benefits, and the downsides
of the extra complication versus the alleged benefits of the arrangement.
Amplifier damping is a complete red herring, nothing is "ruined". Simplistic
active arrangements using preset active functions are nearly always
worse overall than properly designed bespoke passive arrangements.
Its simply not the way to do active properly and is a very poor option.
If you can't design the proper electrical transfer functions to give
the correct acoustic transfer functions, active is near useless.
If you can, then active has some intriguing advantages.
BiAmp (Bi-Amplification - Not Quite Magic, But Close) - Part 1
and Hybrid Design and Hybrid Design
Point to some of them that can be used effectively.
rgds, sreten.
Errors in the digital processing John? ... do you mean around the xo frequency? I understand digital to lose some xo accuracy but the is mostly in the higher frequencies. IE, above where most would even consider placing a xo point to begin with, so, it of no consequence typically. Besides, any error in LP reproduction @ a xo frequency is reproduced on the HP side exactly. Summing is therefore not effected.
I'd imagine that passive components would/can introduce greater/varied error(s) @ any given xo frequency do to the component value ranges.
The audibility of some of the posted reasoning, in this thread, for favoring one side or the other is questionable to me anyway. Pick your poison. Implementing either takes a bit of skill and as stated before .. properly implemented, xo's should be transparent.
I'd imagine that passive components would/can introduce greater/varied error(s) @ any given xo frequency do to the component value ranges.
The audibility of some of the posted reasoning, in this thread, for favoring one side or the other is questionable to me anyway. Pick your poison. Implementing either takes a bit of skill and as stated before .. properly implemented, xo's should be transparent.
When ever you work with numbers of finite length (limited number of significant figures) there are errors. Consider the value of Pi for example. It is defined at the ratio of circumference to diameter, C/D. To a hole bunch of decimals its value is:
3.14159265358979323846264338327950288419716939937510
and that is not exact. But any time you multiply the diameter of a circle by Pi to find the circumference there will be an error if you use that value. If you use fewer decimal points the error will be bigger. We can place bounds on the error and limit the size based on how accurately we specify PI and we can assure that any errors do not grow by insuring that any processing algorithm is stable.
You have the same thing with digital imaging. The more pixels the closer the picture looks like the real thing. It becomes a question of how much resolution is needed before the error can not be observed under some set of conditions.
I am not saying that errors in digital audio processing are audible, just that they are there. Some people will claim 24 bit audio is superior to 16 bit. Other say they can not hear a difference. Passive, active, analog digital 16 bit, 24 bit, 44k sampling, 96k sampling. Pick a topic and someone will argue for and against.
It's audio. It's more divisive than politics. And 99% of it is BS.
This is completely irrelevant in my view, we are discussing which is best in terms of the xover. The amplifiers are assumed to be identical or perfect in that they are effectively removed from the equation. Nevertheless amplifiers tend to have a much easier job driving just a single crossoverless driver in an active system then they would in a full passive one. This almost always improves the amplifiers performance and should therefore sound better.
Again from my point of view this isn't strictly comparing active vs passive, you're tarring active with a brush in the sense that it isn't implemented quite as well as it should be and now we compare this to the passive xover that has been implemented correctly. This isn't a fair comparison and certainly isn't really comparing the best of one with the best of another, which is what this thread should be about, not tolling the downfalls of improperly designed active filters.
Clearly did not read the rest of my post. Obviously you can get phase linear digital filters, I said as much myself, but if we're comparing active to passive filters then it doesn't make sense to include what active can do that passive cannot. Granted this seems somewhat at odds with my above statement, where we should be comparing the best of active vs the best of passive, but in that instance passive will lose flat out because it is more limited.
Evening the playing field, by only comparing an active implementation to its passive counterpart should help to show any inherent flaws in either type. If you include all the extra stuff that one can do that the other cannot, then you've got a situation where people will say, yeah active is better if you need all the extra stuff that it can do, but if you don't, passive will sound better - this is clearly wrong, as it is implying actives are inherently worse and that you should only use them if absolutely necessary.
Then you've not used an active xover that has been designed to allow for full flexibility of the filters that it has been built around. You can do this with variable resistors that cost pence and will give you more design freedom then having lot of boxes filled with passive crossover parts.
Then even in your own words this isn't the fault of the crossover type but in the implementation.
Some digital filters lose accuracy at lower frequencies, IIR biquad based filters for example, and the higher the sampling frequency the worse this becomes. There are different filter types you can use to get around this however.
Exactly there are a lot of subjective opinions being thrown around against active xovers. There are a lot of objective reasons too, but none of them should be used for or against either school if they are done correctly. We should not be comparing poorly or improperly implemented versions of either, but only the best.
I think you've answered your own question there.
You're comparing a second order butterworth closed box response with a 4th order (overdamped?) reflex type response. I think I could hear the difference. It's not a valid comparison.
It's interesting to see what the simple 1.2mH bass coil does with that Visaton W200S 8" driver I looked at earlier. It slopes the driver response for a standmounter to be approximately flat in the 500-5000Hz region with a 6dB rolloff. The very deepest bass response is largely unaffected. It's worth mentioning that some drivers like the polycone Vifa P13Wh are approximately flat in the right cabinet and room placement without a bass coil at all. A question of loudspeaker voice coil inductance.
Attachments
I dont know if its just as hard. Making changes is far ieasier with the turn of a knob and pressing a few buttons. Making changes in high order crosses from sub to sats is pretty expensive and cumbersome if you dont have quite the selection of parts on your shelf. While I completely agree that a well implemented John K design will beat any crossover that I can design, a high order sub crossover is just rediculous in signal loss and cost over active.
In the end, time alignment, notch filters, and high order crossovers are just so much easier to tweak with buttons and knobs. Nothing I have made with passive comes anywhere near my active systems. I guess for me, my lack of skill gets rescued by active which for me is easier to impliment. My hats off to you guys who can build great passives but I'm not there myself.
I know the thread is about one path or the other but the builder has to part of the real world result in that for the average guy with minimal equipment for testing which is better for him, is different from which is better for a prodigy builder.
Last edited:
Originally Posted by puppet http://www.diyaudio.com/forums/multi-way/221643-active-vrs-passive-4.html#post3204819
Errors in the digital processing John? ... do you mean around the xo frequency? I understand digital to lose some xo accuracy but the is mostly in the higher frequencies. IE, above where most would even consider placing a xo point to begin with, so, it of no consequence typically.
Errors in the digital processing John? ... do you mean around the xo frequency? I understand digital to lose some xo accuracy but the is mostly in the higher frequencies. IE, above where most would even consider placing a xo point to begin with, so, it of no consequence typically.
... you left that part out while quoting my post. The errors aren't important/audible. They have no effect on the magnitude response ... whatever the order/alignment design goal. They remain symmetrical.puppet said:
Last edited:
No I didn't, the areas I was talking about are usually out of band as the actual xover frequencies are typically significantly above where the filters start to have issues. If however you do crossover lower then is recommended for a certain filter type you can see large errors occurring in the amplitude as well as the phase, these are not in any way acceptable and do require the use of a different type of filter if you want the filters to do what you want them to.
There is nothing saying that these are going to be symmetrical either. If the filter type used in the high pass is identical to the low pass then that might occur, but this never happens in the real world. This kind of thing would only be applicable I would assume with text book perfect filters and you almost ever end up using them. Most of the time the filter Q and fcs are completely different for the drivers on either side of the highpass/lowpasses. The acoustic results might be very similar, but the filters used to get them there will not be.
There is nothing saying that these are going to be symmetrical either. If the filter type used in the high pass is identical to the low pass then that might occur, but this never happens in the real world. This kind of thing would only be applicable I would assume with text book perfect filters and you almost ever end up using them. Most of the time the filter Q and fcs are completely different for the drivers on either side of the highpass/lowpasses. The acoustic results might be very similar, but the filters used to get them there will not be.
I realize this ... but it speaks to specific implementation rather than the differences between the two strategies.
OK, VC temperature effects are largely responsible for changes in the transfer function with a passive filter. But it doesn't completely let the passive crossover off the hook. It's still a matter of putting the cart before the horse. Passive C & L components which are almost purely reactive, Zobel resistors that track VC resistance plus the time invested in design work and testing can be cost prohibitive. Economic concerns will drive the decision between the two.
That is one area where active filters do easily make headway over the passive equivalent as they are far easier to tweak. I don't think I've ever claimed otherwise though.
And this is one of the reasons why active filters are preferred by many as they do allow you to arrive at a better sounding design if you don't have years of experience and lots of measurement tools/CAD software at your disposal. I was just trying to point out that this doesn't automatically mean that active is better, it's just different.
Absolutely I'm active 100% of the way too as the design flexibility and it's numerous advantages over passive designs are well worth having. This is especially so if your main system is subject to change. But I think what I was trying to say is that active systems are no panacea and if you want to get the most out of them they essentially require the same tools and skill set to get the most out of a passive design.
Sound Easy for example will design active filters along with passive ones. Now if Zaph were inclined he could do all his designs active instead, but the measurement and design process for both would be roughly the same, only the type of crossover used would be different. And to arrive an excellent end point, you need to have done the same ground work for both.
Indeed, I was just pointing out what some of the problems can be, which is what I thought you were asking in post 36.
Sure, I have no doubt that your system sounds great, and you use that as your means of comparison. However, if you're comparing to passive speakers, are you sure you are getting the same frequency response? It's very easy to set levels higher for the bass unit and come to the conclusion that active has better bass (or treble).
I think you've answered your own question there.
You're comparing a second order butterworth closed box response with a 4th order (overdamped?) reflex type response. I think I could hear the difference. It's not a valid comparison.
To avoid it getting lost due to the merger I'll have to ask again:
Where do you get the closed box response from?
It was still the same ported box in exactly the same position within the same room.
Are you saying that a 2nd order (overdamped according to Tannoy) crossover at 1.2kHz changes the bass response at 50Hz in a ported box to the bass response in a closed box when compared to a 4th order L-R at 1.2k?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.