Frequently Asked Questions:
Is this kit for me? What do I need to
know?.
I want volume control on each output.
What about pots or switches to change the Fc
of the filters?
I want active EQ for my specific
application.
For the equall component Sallen-Key (ecSK) filters
(such as 'bass boost'), what's the difference between 'gain' and Q-gain (peak
dB)?
What about op amps? Are yours any
good?
I want to crossover at *excatly* 2000
Hz.
Are there enough parts in your kit for my specific
application?
I want to copy a highly-acclaimed commercial
unit.
I want to correct the Seas metal woofer
resonance.
I have "professionally-designed" passive schematic
for a crossover that would work for my application. I want to build
the active version.
Automotive Applications
Is this Kit for me? What do I need to know?
This is difficult for me to answer from my perspective. I tried to make the kit such that nearly anyone could build a fourth order L-R crossover. Hence, there are many photos of this ciruit provided. Other, more advanced applications are described on my page and in the docs. Really, there are thousands of applications. That is the power of this kit. If you understand the function of each subcircuit (there are only a few), you can cut, copy, and paste them as you wish into new designs. Of course, you will need to understand the concepts of Q, filter slope, etc. You will also need to know how your speaker behaves in the box you built for it. Look over the intro to active filter theory page. In some ways, maybe you don't need to fully understand the material presented there in order to 'photocopy' a L-R circuit out of my docs.
I want volume control on each output
This makes a lot of sense. If you have fixed-gain amps, and drivers with differing sensitivies, then you will want this. At the very least, you will need more op amps ( 1 op amp / volume control ). For a 2 way cross, you'd only really need 1 attenuator / channel. You'd put this on the more efficient driver (tweeter) to bring it down to the level of the other (woofer). You could use resistor voltage dividers with your excess kit R's to achieve your goal. Pots, however, would be much easier to use, and not much more difficult to include in your circuit (read on).
Some people like to bump up the gain at the input of the crossover to fight the possibility of noise. In this case, you would need to attenutate EVERY channel at the output in order to meet the expecations of your fixed-gain amp. In this case, you would need 1 extra op amp / crossover output. The least efficient driver should be given a voltage divider before the last op amp (constant attenuation). The other channel(s) would be controlled by pots to bring the level of the tweeter (and mid) down to that of the woofer.
I have considered the issue of what type of pot should
be used at great length. I believe that these pots should be cermet
pots placed on your breadboard (or printed circuit board (PCB)), enclosed
in the chassis. This is because the volume of your drivers will need
to be calibrated relative to each other and your room. You will want
to dial them to your listening / measurement preferances. Once they
are set, you will not want anyone to change them. Calibration controls
are typically NOT located on the front of the chassis. If you design
speakers for a living, you might want this control on the outside of your
chassis for easy access. This is not the case if you need to spend
1 and only 1 session setting the volume levels. Your preamp controlls
the volume of the system. If you build a new system, it is no big deal
to change the levels of the pots inside the chassis. Therefore, I bought
the cermet pots, and now have a 6-channel volume control option.
This costs $22, and comes with the extra parts needed (breadboard, op amps,
pots, and R's) to make the volume of each output variable.
What about pots or switches to change the Fc of the filters?
I've spent quite a while thinking about this. Here's the answer: YES - but is it worth it? You could spend a *lot* of time on layout and soldering to get some cool control dials on you chassis. However, most of you will probably adjust these dials only a few times till you find the 'right' setting. Some of you might go to great lengths to find the exact setting. In either case, after this is set, why would you want to change it? The inconveneince of building these functions in many cases outweighs the inconvenience of finding the right alignment by replacing resistors on the breadboard. Do one speaker at a time, compare it to the other, and see which you like better.
If you are a professional designer, who needs a new crossover everyday to evaluate a different speaker system, then it is clearly worth your time to persue these functions.
I want active EQ for my specific application.
You'd most likely want this for your woofer. In many cases, you can extend the F3 down one octave. For sealed systems, the usual way to do this is via a high-Q 2nd order low freq highpass filter. This transforms your acoustic low-end response from a 2nd order rolloff to a 4th order rolloff (depending on your application). This modification comes at a penalty. Your amp and woofer will need to handle higher currents (wattage). Furthermore, the group delay and transient response will be somewhat degraded. A similar filter can be used in a vented woofer system (4th order) to transform it to a 6th order system. The same pentalties apply. If you want to add this to an existing box, you will need to re-tune the box to a lower freq (make the vent longer or add weight to the PR). One advantage of using this type of filter with vented systems is that it protects the woofer from potentially damaging very-low freq's. Generally, the lower the Fc and Q of the added active filter, the less likely you will notice the group delay problems.
For either of these applications, I recommend that you use software to determine the Fc and Q of the filter you require for your alignment (alternatively, you might get these parameters from the manufacturer's data sheet or white paper under "typical applications" heading). I believe that many basic and inexpensive (maybe even freeware) loadspeaker design programs can do this (please advise me of such exisitng freeware / demos). Once you know the Q and Fc, you can easily design a ecSK filter to meet your requirement. See the Shiva Active EQ schematic and relevant discussion.
For the equall component Sallen-Key (ecSK) filters (such as 'bass boost'), what's the difference between gain and Q-gain (peak dB)?
See: Second Order Equal Component Sallen-Key Filters (ecSK filters). The definition of gain is: voltage out / voltage in. Note this definition is *independant* of frequency. The volume knob on your preamp controls the gain of your preamp. 'Turning up' your preamp boosts the voltage (gain) of all audio frequencies equally. For purposes of clarity, I'll refer to this type of gain as 'classical gain'.
When implementing a woofer 'active EQ' circuit, there is another type of gain you should also be interested in. I'll refer to this as Q-gain (Q-dependant gain, peak dB, narrowband gain, etc). The amount of the Q-gain depends upon the order and Q of your active filter, and is usually given in terms of 'Peak dB'. A bass boost circuit is often implemented as a high-Q, lowcut 2cd order ecSK filter. These filters can be modeled (thought of) as a sealed-box woofer system (both are 2cd order filters). I draw this analogy because many of you are probably intuitively familiar with sealed box alignments. (Actually, it is ironic that I draw this analogy, because sealed-box woofer theory was originally derived from analogies in *elecrtical* theory!) You know that a sealed system with Q=1.9 will have a large 'bump' near the low end of it's response. Same is true of an ecSK active filter. The size of this bump, as you know, depends on the Q of the system. The maximum height of this bump could be referred to as 'Q-gain', or 'Peak dB'. Note that this Q-gain occurs across a narrow band of frequencies.
Some confusion may arise over the fact that the same pair of resistors in the ecSK filter simultaneously control the Q AND the classical gain of the filter. This means that Q (and thus Peak dB), and classical gain (volume), are all dependant on just 2 resistors in the circuit. We don't really care too much about the classical gain; our volume control downstream (be it an integrated amp volume knob, or a pot in your active crossover) will take care of that. We need a way to control volume to each driver because they are likely to have differing effeciencies anyways. However, we DO care about Q (and the attendant Peak dB), because we need to know how much more power will be needed from our amp to drive the woofer. We also want to know the max SPL our woofer can handle with all this power its being fed. The higher the Q, the higher the peak dB, and thus the more power you'll need.
What about op amps? Are yours any good?
I have done nearly zero op amp A/B testing. However, as far as I know, the BB2134's are the very newest high-end audio op amps designed by Burr Brown. I don't feel I need to worry about second-guessing the full-time engineers at BB who developed the latest op amp designed *specifically* for high-end audio. There are other excellent audio op amps, no doubt. The BB2604's are often cited. Although both of these op amps operate flawlessly at frequencies a few *orders of magnitude* (400 X) beyond the audio range, the 2604 does have a better bandwidth and slew rate. But, the BB2134 has better DC performance, and (inaudable) better distortion specs. Analog devices has great products as well (AD OP275, etc). National Semiconductor IC's get mixed reviews. One noteworthy product of theirs is the LM6172 video opamp.
The NE5532 is the dual op amp I offer standard with the kit. It seems to be amoung the favorite op amps before the latest generation of BB's and AD's. Some people who are very particular about op amps say they would have no problem using the NE5532's to drive woofers or midranges. The difference at the tweeter is most likely audible - as long as the rest of your system can keep up. I also offer the BB2134's as an optional upgrade. Many people would argue that this upgrade is well worth it.
Audio frequencies are no problem for modern IC's. The real action is now approaching gigahertz freq's. If you want to go absolutely nuts on op amps, I hear that Analog Devices' video op amps (or the LM6172) are one avenue to explore - maybe not for your woofer though ;-) Ask yourself what kind of op amps are likely to be inside your CD player, pre-amp, etc. One person told me his NAD CD player used NE5532 op amps. He replaced them with BB2604's, and has convinced me that he can hear a difference in his system. The better the resolution of your overall system, the greater the liklihood that you may discern a difference between op amps. If in doubt, I suggest you upgrade to the BB2134's I offer.
I want to crossover at *exacty* 2000 Hz.
I can understand this desire. You can get freq's other than those I list by using R's in series or parallel. But ask yourself if it would really make any difference. Octave-wise, the cross freq's avialable in the standard kit are not very far apart. Furthermore, the audibility of shifting your cross point 1/6 of an octave away from your desire is arguably zero.
Are there enough parts in your kit for my specific application?
Most likely - YES. If you need more than the parts provided for stereo, 3-way, 4th order L-R cross, then you are building one heck of a system. If so, your application is probably very unique, and there are infinite numbers of applications out there. This question can be analyzed in the following categories: How many opamps do I need; How many caps do I need; how many breadboards do I need; etc.
How many op amps do I need?
If you have an advanced, or high-order application, you may need more op amps than what's provided. I supply dual op amps, so lets just count op amps for the left channel of your stereo system. This number will be the same as the number of dual op amps needed for a stereo system:
Input buffer = 1 op amp
A 4th order filter = 2 op amps; a 3rd order filter = 2 op amps; a
2cd order filter = 1 op amp; a 1st order filter = 1 op amp
Each bandpass = two filters
(a 3-way system has 1 bandpass + 1 highcut filter (woofer) +
1 lowcut filter (tweeter) = 4 filters
(a four way system has 2 bandpass filters + 1 high + 1 low = 6
filters)
Each volume control = 1 op amp (3-way system = 3 op amps)
Each 2cd order ecSK filter (Bass boost, or active EQ) = 2 op
amps
Total Op amps Needed = 1+(# req'd for the order of you filter)*(# of filters) + # of volume controls + 2*(# of ecSK filters)
How many caps do I need?
Look at my 'Filters' schematic. For ugSK filters such as this, you need 2 caps per 2cd order lowcut, and 3 caps per 2cd order highcut. For a fourth L-R, this means 4 caps per lowcut, and 6 per highcut. A band pass (for the mid) is a highcut cascaded with a lowcut = 4+6 = 10 caps per bandpass. For a 3 way system: high + bandpass + low = 4+10+6 = 20 caps. For a stereo system, this means 40 caps. Note that 10 of these will be a lower value (uF) than the other 10 (one value for lowcut, one for highcut).
A ecSK filter (bass boost, etc) will require 2 caps per second order filter. (see Shiva Acitve EQ)
These rules apply for the cross freq's listed on my page. If you want to go lower for active EQ, you can use 2 of my biggest R's in series to cut the lowest listed 'Butterworth' freq in half. You could also use 2 caps in paralell for each C value to cut the Fc in half yet again (extra 2 'big' caps/ecSK filter).
How many breadboards do I need?
One is enough for 9 dual op amps (with adequate spacing between) and the voltage regulators. Actually, you might be able to squeeze the tenth dual op amp on this same board. If you need a stereo, 3-way, 4th L-R crossover AND volume control or bass boost, you'll need another board. Alternatively, you could just solder the volume control circuit, etc. onto a printed circuit board (PCB) available at Radio Shack. Keep in mind that breadboard testing is very nice.
-----------------------------------------------------------------------------------------------------------------------
I want to copy a highly-acclaimed commercial unit.
First, you must figure out what drivers they used, and if they made any modification to the drivers. Next, I'd recommend that you copy their baffle as closely as possible. Generally, it is better to minimize most inherant reproduction problems using baffle layout rather than passive crossover 'fixes'. Therefore, you might assume that they chose a good, or maybe 'easy to correct for in the crossover' baffle layout. Next, correct for any (if any) well-know problem of the driver you are using (maybe passively - see the Seas metal driver discussion below). After this point, you may wish to design the rest according to the standard DIY principles and considerations. Hopefully, any anomolies you face during the audition of your final product will be well-defined and fixable.
If instead, you wish to follow thier design further, you might consider the mechanical transfer function of thier midrange box. Is is sealed (2cd order) or vented (4th order)? Considering the driver's specs, and the volume of the enclosure for the midrange, what is the mechanical Q (Butterworth = .707, etc.) and rolloff of the box-driver combination? What is the Fc? Remember that the midrange enclosure operates *exactly* like a highpass filter. Knowing the midrange Q and order of rolloff, cascade it with the *appropriate* active highpass (relatively easy to do - see: http://members.aol.com/johnpomann/speakers/theory.htm#Cascading ), and you will get a perfect cross to your active woofer response. Advanced software would clearly help, but many simple and elegant (sealed midrange) commercial designs exist where this might not be required.
I recommend you make your intitial active cross *without* the baffle step / diffraction correction. Remember that much of basic speaker design efforts fail to consider the nature of your listening room and the placement of your speakers. Add the baffle correction later, and see how much of an overall improvement it offers.
I want to correct the Seas metal woofer resonance (etc).
If you want to use a driver with a well-know 'flaw', you might choose to get rid of these problems passivley, simply because the solution might be well-documented. In this case, the driver plus the standard 'passive fix' can be simply treated as an 'ideal driver' by your active crossover network. You'd never need to change the passive fix as long as the solution really does correct the problem. For the Seas drivers, please see http://www.speakerbuilding.com/drivers/seasnote.html for the passive fix. Note that because the rest of your crossover is active, you need not worry about how these passive components will interact with the rest of your crossover.
It is also possible to correct these problems using active notch filters. So far, the simplest (capacitior and calculation - wise) circuits I have found require 3 op amps. Using these filters, I remain uncertain as to how much the passband is affected. Advice / referances on active notch filters would be greatly appreciated.
A recent discussion on the Bass Lists seems to suggest that using the Seas driver with a 4th order crossover at 2kHz might be enough to solve the ringing without requiring a notch filter.
I have a "professionally-designed" passive schematic for a crossover that would work for my application. I want to build the active version.
This is not as easy to do as you might hope. It is clearly an advanced topic. The passive crossover is connected to a driver, whose impedance changes with frequency. Therefore, the load that any given passive element 'sees' is not necessarily intuitive. If you had all of the driver's specs, and advanced software, you could figure out the total transfer function of the passive crossover and driver. Then, you could subtract the driver's transfer function, and build the rest using active components. A fully optimized , computer-designed crossover might be the toughest to reverse engineer by hand. Classic 'cook book' filters, where you can ID zobel ciruits, etc. would be easier. Essentially, this requires you to have a very strong understanding of *both* passive and active filters.
It might be easier to redesign the crossover from scratch using all the info you know about the crossover and drivers. If you can find out the properties of the cabinet, and the order and basic type of filter (Butterworth, Bessel, L-R, etc.) your passive unit uses, you could start by copying these with the active cross. Next, consider if there are any 'known flaws' inherant to the driver you are using. Often, there is a published passive 'fix' to bring the driver toward ideal (see both discussions immediately above).
After recieving enough interest, I have now started work on this. Please see Auto Applications Page