PDA

View Full Version : Crossover networks 101



penguin
04-04-2005, 05:16 PM
Crossover networks 101
John,
thanks for the hopeful note on the 2461s. I have always found them to sound pretty good in the slant monitors that they are in, albiet with quite a bit of graphic EQ help.

(Zilch, Thanks again for pointing out the link for the network schematics.)

I have been going over the schematic for the 3133 network used in the 4333 monitors. It all seems pretty straight forward. However, there were a few questions.

a) In the band pass section there is an inductor that is tapped in the middle. The three leads are labeled Rd, Grn, and Blk, so if I can buy an inductor that is tapped and has the leads color coded I be in great shape. However, looking in the PartsExpress catalog I don't see anything like that. Should I be able to find something like that? Can I tap it myself?

b) The values for the inductors are taken to the 100th of a micro farrad in some cases. In many cases there are not equivalent value inductors in the PartsX catalog. How much tolerance is there for these values and/or where can I find inductors with these values?

c) I understand that coils are often mounted on network boards in alignments that prevent them coupling and interfering with one another. What are the rules of thumb there?

d) Finally, the adjustable L-pads are labeled 1,2,3, and blk, wht, red on the schematic. I want to use an attenuator out of a trashed VOTT network for the mids and buy one for the highs. I am assuming that 1=gnd and that 2 and three should be interchangable. ?? Is that right?

Thanks to anyone willing to wade through all my questions,:blah:
Penguin
http://www.audioheritage.org/vbulletin/images/statusicon/user_online.gif

Zilch
04-04-2005, 05:33 PM
a) The tapped inductor will have to be custom wound, or assembled from two inductors of the correct values. I believe if you search the forum, you may find some information on this, as others here have done it.

b) It's milli-Henries (mH) for inductors. Sometimes you have to purchase the next higher value and remove windings down to the correct value. This requires having an inductance measuring device. More importantly, you have to also know the DC resistance of the inductor in order to choose the proper wire size. Again, search the forum for any available information on that particular crossover.

c) Rule of thumb is space them out and orient their axes at right angles to each other.

d) Wrong. L-Pad is composed of two separate resistance elements of different values, one in series with the driver, and one in parallel with it. As the resistance goes up on one, it concurrently goes down on the other to maintain the constant impedance. Pin 3 is the input, 2 is the output, and 1 is the common.

Note: Bumping your threads is better than starting repetitive new ones, so folks can respond in context. You asked this just yesterday, tho....

aust-ted
04-05-2005, 05:53 AM
I am also very interested in this topic, and am trying to get a handle on practical issues involved in rolling your own equivalent xover circuits for using with JBL drivers. Just did a search on "tapped inductors" on the search function on this forum and found much interesting reading, much of it contributions from Giskard and Zilch. Also been following the excellent "Quick and Dirty 4430s" project and picked up some practical hints from Zilch's work.

There are a couple of practical issues I would appreciate clarification on.

1. I gather that JBL inductors are typically iron cored which allows for a smaller inductor. I have noted that most of experienced DIY projects on this forum use air cored unless using original JBL inductors. Is there any benefit in using iron core inductors (in a DIY replication), ignoring the resistance question which can be adjusted out as discussed in "Quick and Dirty 4430s"? The reason I ask this that I have a collection of salvaged inductors and think I can wind some myself if required. It is also now fairly easy to measure coil inductance given the plethora of cheap test equipment around. Iron cores are more problematic but I suppose could be replicated using salvaged transformer cores at a pinch.

2. Zilch above writes in part -"The tapped inductor will have to be custom wound, or assembled from two inductors of the correct values." Now from theory assembling a tapped inductor from two separate coils should work just as it does for resistors. I ask has anyone constructed a tapped inductor by joining two aircored coils and found any practical problems in comparison to a JBL iron cored tapped inductor (autoformer)? Again I am assuming that the difference in wire resistance was compensated for with series resistors.

Regards
Ted

Earl K
04-05-2005, 06:22 AM
Ted.

I ask has anyone constructed a tapped inductor by joining two aircored coils and found any practical problems in comparison to a JBL iron cored tapped inductor (autoformer)?

Not I, but ;

Giskard posted a bunch of working formulas in a few charts. These charts had the necessary inductance ratios that one would need to maintain to create autoformers. Wiring inductors in series & using them as a tapped autoformer is certainly doable - it's just not as intuitive & straightforward as you might think . Finding the 3 db & 6 db down-points within an autoformer means applying a bit of math. It is an interesting alternative to a resistive Pad .

Giskards post was located inside a thread that was started by one of our forum friends from France. It was maybe Herve M, Bruno Ginard or Stephane ( perhaps someone else ).

<> :cheers:

Ian Mackenzie
04-05-2005, 06:27 AM
Ted,

1. Air core inductors are generally better as they offer best performance for mid / high pass filters as they will not saturate. Iron core have a benefit of low dcr for sub woofers and tend to be ower cost. I think JBL used iron core for low production cost given even their iron core had relatively high DCR.

2. I have made split air cores for attenuation without any problems, Giskard posted a spead sheet for calculations some time back. The benefit is attentuation of the voltage via the coils rather than wasted power dissipated in voltage drop across the resistors.

I still think an L pad is very useful for fine adjustment of attentuation while the split coils provide the bulk of the attenuation (see 3145 network).

Once you find the exact settings, the L pad can be replaced by high quality Mills power resistors for the best sonic performance.

While fun to make, passive crossovers are at best a wet rag attempt at filter design and better results can be attained at lower cost with a hi performance fully adjustable active crossover.* They don't waste power and you have more control over the crossover parameters.

*see hi end active crossover project.

Ian

spkrman57
04-05-2005, 06:35 AM
What is the interaction(mutual induction) when coils are placed in series? How far from each other must they be placed in order to avoid the interaction?
Ron

Earl K
04-05-2005, 06:51 AM
Ted,

Here's the Tapped Inductor Thread . (http://audioheritage.csdco.com/vbulletin/showthread.php?t=846&highlight=Tapped+Inductor)

It was member "delahais" who started the thread.

Oh well , bad brain day on my part ( but at least Bruno did make an appearance :p ) .



:cheers:

aust-ted
04-05-2005, 07:14 AM
Ted,

1. Air core inductors are generally better as they offer best performance for mid / high pass filters as they will not saturate. Iron core have a benefit of low dcr for sub woofers and tend to be ower cost. I think JBL used iron core for low production cost given even their iron core had relatively high DCR.

2. I have made split air cores for attenuation without any problems, Giskard posted a spead sheet for calculations some time back. The benefit is attentuation of the voltage via the coils rather than wasted power dissipated in voltage drop across the resistors.

I still think an L pad is very useful for fine adjustment of attentuation while the split coils provide the bulk of the attenuation (see 3145 network).

Once you find the exact settings, the L pad can be replaced by high quality Mills power resistors for the best sonic performance.

While fun to make, passive crossovers are at best a wet rag attempt at filter design and better results can be attained at lower cost with a hi performance fully adjustable active crossover.* They don't waste power and you have more control over the crossover parameters.

*see hi end active crossover project.

Ian

Ian Thanks for the reply. I must admit it confirms my understanding. I was really asking if there was more to it than JBL trying to save costs by using iron cores. Interestingly some of the UK domestic speaker manufacturers in the 60s (eg Leak) used large aircores but I must admit these had less components in their xovers whereas Tannoy used iron cores in their autoformers. JBL components, at least the pro ones, do not give an impression of cost cutting principles being very much applied.

Also could not argue with your points about active xovers. In fact I am currently using one, a JBL M553 and I read your very interesting active xover project post some time ago. I have been thinking of building one when time permits. My current interest in passives is a pair of 4331s I am thinking of converting to 4333s by adding a pair of 2405s and mods to the xovers. Also I would find it harder to justify going down the active route for rear speakers in a theatre setup. Just gets too complicated and requires too many components.

Regards
Ted

4313B
04-05-2005, 07:28 AM
I believe iron core inductors were popular because they fit in the cans. I believe it's that simple. Size was an issue.

Ian Mackenzie
04-05-2005, 11:45 AM
Yeah,

Well there's not much room in those cans is there.

About the attenuation, obviously there will be power dissipated across the 19-20 ohm resistor in parrellel with the tapped attenuator and may be in the order of 30% of the power depending on the reactance of the circuit.

But depending on the circumstances, there maybe more or less power losses there than across the more typical series/shunt resistor attenuator, particularly if you need say 10-13 db attenuation on a horn (13 db - 107 db to 94 @ 500 hertz). In this instance if a 100 watt sine wave (28.3 volts) is present 95 % of the power is lost in attenuation. The degree of accuracy with high orders of attenuation is perhaps more accurate and reliable with tapped inductors. In high power networks this would be a major issue.

Of course active filters solved these concerns.

Ian

4313B
04-05-2005, 11:53 AM
Yeah, that is a good point Ian. I always thought it was kind of goofy pairing 108 dB HL92's with 93 dB 136A's... going active really helps in situations like that.

I liked the way the 4430/4435 handled attenuation.

Ian Mackenzie
04-05-2005, 12:13 PM
The 4430/35, oh yeah that was smart.

I think it was adaptive marketing on JBL's parts in that domestic component brochure. I still remember buying that builder kit back in the 70's.

I admire the simplicity of the JBL networks although the 4 ways are a bastard to drive and I am not sure why. The 4345 has abroad dip to about 4.5 ohms from 400-900 or so and I think the load is some what reactive. I reason this is why the system sounds a bit compressed with all but the most hi current amps using full passive operation.

Bi amping the system the impedance of the mid range cone measures more like 8-9 ohms over this range as seen by the amp.

Ian



Ian

Zilch
04-05-2005, 12:19 PM
I liked the way the 4430/4435 handled attenuation.Howzzat? The 7.5 Ohms across the driver and 20 Ohms in series? (Ignoring the HF lift part.)

Itza variable L-Pad followed by a fixed one, no? Something I'm missing?

[Most parts are in. Time to build soon.... :D ]

4313B
04-05-2005, 02:03 PM
Howzzat?No tapped autotransformers anymore Zilch. The previous 2-ways were 4331's with tapped autotransformers.

penguin
04-05-2005, 08:24 PM
The autotransformers seem to be a bit more of a puzzle than I need right now. I found the schematic for the Giskard-modified 3133 crossover where the bandpass section is revised to avoid the autotransformer. Is this a cool forum or what?



A much more manageable problem that I have now:



1) The schematic calls for 2 20-watt resistors and 2 40-watt resistors. I can’t find non-inductive resistors of that wattage anywhere. I’m guessing that if I put two resistors in series—splitting the ohms required between the two—that the wattage rating will be essentially the sum of the two ratings, or rather, with two resistors combining to accomplish the load the heat would be half for each one. (?)

2) Second idea related to the same problem. Are those wattage values overkill for home use. For example, can I use 3 Mills 12 watt rated resistors that sum to the resistance value I need but come in a little short of 40 watts?

Thanks for the continuing wealth of info,
Penguin :)

Zilch
04-05-2005, 08:32 PM
Yes and yes, but Michael Percy has the Vishay Dale (or equivalent) non-inductive aluminum heat-sink clad resistors:

http://www.percyaudio.com/

They're rated 50 W themselves mounted on an adequate heat sink, or 20 W, if not. $5.95 each, cheaper than multiple Mills N/I 12 W.