, 08-30-2010 at 01:32 AM (2332 Views)
'Tips! Volume 4', by Ben "Wolf" Shaffer, 8-30-10.
This is to tweak and minimize xovers, or to eliminate redundancy.
Let's look at what you should have in the way of ranges for parts....
-For resistors in speakers, anything over 50 ohms seems to be unnecessary, as minimal changes are resulting above that. If you have a very small resistor below an ohm, it can sometimes be completely omitted without any resultant changes. I will say that some tweeters benefit from that last half-ohm in attenuation to balance the system, and sometimes DCR increases are required to smooth out a broad hill in response. Still- it's best to see how it looks without the resistors there as a reflection of the whole.
-For Capacitors, unless you are using a very large NPE for a Passive Assist alignment in the bass on the woofer, then there is even a range they can be employed. I have never used a capacitor smaller than a 0.1 uF value- but once. I used a set of .01 uF caps to see if a minimal bypass for lower ESR made a difference, and I don't think I can hear it. Anyway- Stay away from nano and picos, as they are typically too small to do anything in speaker passive networks.
As for the larger value end of the scale, and only in 3-ways or larger/lower-xovers, I have used a maximum of 200uF. There are exceptions in the form of notches on tweeters' Fs that might require a larger cap, or for reducing a broadband hill on a woofer's passband. Typically, over 200uF gets to be too large for me, or excessive due to not a big enough difference. In Passive Assists, the value is typically between 500 uF and 1500 uF.
-Coils- Yep, some of them are kinda unnecessary too. In a notch, if you get to about 0.01 mH, just remove it completely. Typically there are other methods of notching a high frequency breakup or issue that won't require such a small coil. I have not ever used a 0.025 mH value, but the 0.05 mH parts do come in quite handy sometimes.
Large coils, unless used passively on a subwoofer have a tendency to cause peaking in the response due to the average inductive rise and/or the Fs spike. I know of designs that have used up to a 17 mH steel-laminate coil, but personally, I've not ever required the use of more than a 10mH steel-core in a low xover point.
Now, say you want to know the average bracket-range for the value of the parts in certain applications....
For caps, I rarely use more than a 25 uF cap in series on a tweeter circuit for a low-crossover 2-way; read: below 2 kHz. Most of the time it's <20 uF. Notches on tweeters for Fs-comp generally are 50-200 uF range, and I will use NPE's there. 3-ways are normally between 4.7uF and 12uF.
On midranges, it really depends on your objectives, but I've never exceeded 200uF there, with an average of 20-60 uF range.
On woofer shunts- I can hear the difference in poly vs NPE, but back on topic- I've not exceeded 200 uF in the shunt there, with a large range of possibilities being used. Anywhere from 6.8 uF to 200 uF have been employed.
For coils on tweeters' HP, I have rarely exceeded 0.5 mH, with the average being between 0.1 mH and 0.35 mH. 20AWG here folks, as the added DCR can help the damping of the circuit. Remember the smaller the coil, the more protection for the tweeter it provides on even-order xovers.
On mids, it really depends on my angle or xover freq. If I'm using Foils, I really could have to add more resistance to attenuate, but using an 18AWG or 20AWG air core can be beneficial for less added resistance afterwards. For the HP on mids, the coil is sometimes 'what is less expensive?' for the value required. I have had to use 2.7 mH P-cores for LCR circuits on the Fs of mids due to cost, and usually air-cores of 18-20AWG for the rest. Generally 0.3 mH to 2.5 mH is my range here.
For coils on woofers, the DCR is the king of the deciding factor. Since attenuating them is not advised, the coil is virtually the only way to droop a certain range with the DCR on the unit. Otherwise, use what you want! 0.75 mH to 10mH range. It's a lot of distance to cover, but respectively, the 2-ways are smaller values, and the 3-ways are larger.
Now let's say you have a ton of parts on your xover, how to make that less?? or less costly???
A lot of times, IMO, shunts that are directly in parallel can be combined. If you have a cap and a zobel, you can probably eliminate one part there, the extra cap. Typically LCR's or other LR shunts are there to stay if necessary. If there are more than one CR filter in the shunt, it is possible they are serving 2 different functions and cannot be combined.
One tweak I found is to place a zobel between the HP and LP on the mid, and you can really tailor quite a bit on those values to balance it out.
If you have to attenuate the mid or tweeter at all, then you can place resistors in certain places to reduce copper cost for your coils, or even reduce cap-cost in value. For example, you actually require a larger cap for a lower impedance. If you add resistance between the HP and the driver, you can reduce cap size. The opposite is true if you want to cut copper cost, which is more costly. Decrease the driver impedance if you want a bit larger caps, and a smaller coil set. In this applcation a parallel resistor could also damp the Fs, so you get less cost, more performance, and one additional inexpensive part. This can lessen the cost if you had a full LCR before.
Remember that you can save some coin on LP's by slightly reducing the coil size and using a slightly larger cap with a resistor in series with the cap. This can allow you to tweak a lot of things simply, and get a decent response for not too much money a lot of times. The tank-cap is a further tweak to save money, as copper is expensive. The shunt-resistor can also dial in your phase to match the adjacent driver.
Less copper = less costly.
Using a higher added resistance (DCR) in a HP shunt with the coil can lessen its mH value. Remember resistors are typically <$1!
Use only what it takes to first match your target, and then remove parts if necessary when you optimize the blending between drivers. Sometimes you can get by with much fewer parts or even lesser slopes. Take the driver into account with its comfort zone, and allow it to not exceed it. Remember that -25dB from reference-level is the minimum recommended attenuation for preoblem areas in designs. Sometimes increasing an order/slope by one step will decrease the problem area enough to avoid using other conjugate filters, or even altogether. Watch the Fs on tweeters and dome mids, and the breakups on mids and woofers, if any/all are of significant magnitude. You can xover closer than one octave to an Fs of a dome driver, IF the Fs is damped/attenuated far enough below reference level, and it does not sound bad xover'd lower.
There are several methods to approach to dealing with issues, some that I've already stated in the past blogs, but a lot of it is experience-based to know what to use where. That part is the 'art', and the drivers will often tell you where to place the xover as you are modeling the simulation. Think of it as a stone block wanting to be a sculpture, and you have to figure out what it wants to be.
My modus operandi is: "minimalism with regards to necessity- if at all possible" UNLESS You want to experiment and see what something does or sounds like.
I know this is a simple primer for value-ranges, or how to minimize (reduce cost) parts count, so I hope this helps.
Your #1 Source for Audio, Video & Speaker Building Components
Deal of the Day
View Our latest
Order our FREE 336 Page Full Color Catalog
Speaker Component Categories
Home Audio Speakers
Professional Audio & Guitar Speakers
Car Audio Speakers
Measurement & Design Tools
Subwoofer Plate Amplifiers
Full-Range Plate Amplifiers
Cabinet Hardware & Speaker
Subwoofer System Kits
Speaker Repair Parts