Hi folks,

I would like to know how to construct a 2-way, 12-db passive crossover
that is capable of handling up to 2000 watts, and does well at
retaining the quality of sound at that level. Any advice or links
would be greatly appreciated.

thanks,
Sean

In article O6arc.85510$536.14252301@attbi_s03,
(Sean) wrote:

Hi folks,

I would like to know how to construct a 2-way, 12-db passive crossover
that is capable of handling up to 2000 watts, and does well at
retaining the quality of sound at that level. Any advice or links
would be greatly appreciated.

I know a fellow who used to design output filters for 500,000 watt pulse
modulation AM transmitters. He wound his inductors from copper tubes
with water pumped through for cooling. That might work...

Seriously, this sounds like the perfect application for an active X-over
prior to the power amps.

Isaac

I know a fellow who used to design output filters for 500,000 watt pulse
modulation AM transmitters. He wound his inductors from copper tubes
with water pumped through for cooling. That might work...

What a great concept, though probably over the edge for what I'm
trying for.

Seriously, this sounds like the perfect application for an active X-over
prior to the power amps.

The only problem I have with the active crossover configuration is the
inevitable noise that gets introduced. I have gone to great effort to
make my system purely digital, from the lossless WMA compressed music
on my computer, to the Midiman Audiophile 2496 sound card, to the
Midiman external D to A converter, to the pair of bridged Crown XS900
amplifiers (one for each channel). It would be a shame to introduce
powered analog components for the sake of splitting the signal. I
don't think most people have ever truly heard a pure digital setup and
I'll tell you, it's phenomenal. So, I'm willing to accept some loss
of power in order to keep the crossover passive, but my main concern
is overdesigning the components to be able to ramp up into the 2000
watts range.

To be specific:

1) Do the typical crossover calculations for determining component
ratings apply at this level of power?

2) Is there a way to parallelize/serialize a bunch of smaller
components together such that the power load can distributed, but they
still combine in effort to perform the goal of splitting the signal?

3) If #2 is possible, then could someone please show me how to
determine the power capacity of whatever frankenstein crossover I can
mangle together?

thanks again,
Sean

Hi Sean,

The only problem I have with the active crossover
configuration is the
inevitable noise that gets introduced. I have gone
to great effort to
make my system purely digital...

There are digital crossovers available; for example,
Behringer makes them. If you are really interested in
an all-digital system (preamp, crossovers, amps,
speaker correction, room correction, surround), take a
look at Tact Audio components (that's what I use). The
active crossover would be a much better solution than
"developing" your own passive crossover, and the
digital crossover provides the flexibility that you
will need to dial-in the correct slopes and crossover
points.

Regards,
Tip

"Sean" wrote in message
news:RHUrc.13371$JC5.1256243@attbi_s54...
I know a fellow who used to design output filters
for 500,000 watt pulse
modulation AM transmitters. He wound his inductors
from copper tubes
with water pumped through for cooling. That might
work...

What a great concept, though probably over the edge
for what I'm
trying for.

Seriously, this sounds like the perfect application
for an active X-over
prior to the power amps.

The only problem I have with the active crossover
configuration is the
inevitable noise that gets introduced. I have gone
to great effort to
make my system purely digital, from the lossless WMA
compressed music
on my computer, to the Midiman Audiophile 2496 sound
card, to the
Midiman external D to A converter, to the pair of
bridged Crown XS900
amplifiers (one for each channel). It would be a
shame to introduce
powered analog components for the sake of splitting
the signal. I
don't think most people have ever truly heard a pure
digital setup and
I'll tell you, it's phenomenal. So, I'm willing to
accept some loss
of power in order to keep the crossover passive, but
my main concern
is overdesigning the components to be able to ramp up
into the 2000
watts range.

To be specific:

1) Do the typical crossover calculations for
determining component
ratings apply at this level of power?

2) Is there a way to parallelize/serialize a bunch of
smaller
components together such that the power load can
distributed, but they
still combine in effort to perform the goal of
splitting the signal?

3) If #2 is possible, then could someone please show
me how to
determine the power capacity of whatever frankenstein
crossover I can
mangle together?

thanks again,
Sean

(Sean) writes:

I know a fellow who used to design output filters for 500,000 watt pulse
modulation AM transmitters. He wound his inductors from copper tubes
with water pumped through for cooling. That might work...

What a great concept, though probably over the edge for what I'm
trying for.

Seriously, this sounds like the perfect application for an active X-over
prior to the power amps.

The only problem I have with the active crossover configuration is the
inevitable noise that gets introduced. I have gone to great effort to
make my system purely digital,

In that case, why not do the crossover digitally?

--
Jón Fairbairn

Sean,

I assume that you do not currently have a particular loudspeaker in
mind for this application since most loudspeakers come with passive
crossovers to begin with.

Your loudspeaker and listening room parameters will have more to do
with crossover power disipation than amplifier power output (i.e., if
the loudspeaker driver fails first, it doesn't really how much power
the crossover can dissipate).

Sean wrote:

1) Do the typical crossover calculations for determining component
ratings apply at this level of power?


Yes - but at the associate current levels, crossover layout is very
important to avoid current induced voltage effects.

2) Is there a way to parallelize/serialize a bunch of smaller
components together such that the power load can distributed, but they
still combine in effort to perform the goal of splitting the signal?


Yes

3) If #2 is possible, then could someone please show me how to
determine the power capacity of whatever frankenstein crossover I can
mangle together?


Resistors can be paralleled for increased power handling - i.e., 2
equal value 10-watt power resistor in parallel can handle 20-watts.

Capacitors can be paralleled to increase power handling, but pay
attention to capacitor maximum voltage rating (higher is better) and
dissipation factor (lower is better).

Inductors can be paralleled to increase power handling, but pay
attention to DC resistance (lower is better) and saturation power
levels (higher is better).

Standard formulas for parallel connection
resistors/capacitors/inductors apply when paralleling components.
When paralleling non-equal component values, you will also need to pay
attention to individual component power dissipation requirements to
avoid stressing a sinlge component (e.g., a 1 Ohm and 10 Ohm resistor
in parallel will have the power largely being dissipated in the
smaller value resistor, defeating the purpose of paralleling components).

As to what it will actually need to dissipate, that depends not only
on the particular crossover design employed, but also on the load and
the intended use, not the amplifier's output capability -

Loudspeaker driver sensitivity? Higher sensitivity means less
crossover stress

Type of music will be played (rock, jazz, classical)? Higher peak to
average level means less crossover stress

Maximum desired sound pressure level at the listening position? Higher
maximum level means more crossover stress

Distance from loudspeaker to listening position? Closer means less
crossover stress.

Without having detailed driver specs and measurement tools, a do-it
yourself passive crossover with off-the-shelf drivers may actually
turn out significantly less than optimal. Truth be told, I can't tell
from your posts what it is you are ultimately trying to achieve with
this approach, nor can I figure out why you'd want to do this if you
don't already know the answers to your original questions.

Best regards,

Terry

thanks again,
Sean

Hello Terry

I assume that you do not currently have a particular loudspeaker in
mind for this application since most loudspeakers come with passive
crossovers to begin with.

I'm building a pair of dual-12 cabinets. The 12" Oz Audio Elite's
each handle 500 Watts RMS so it is my intention to drive each cabinet
with peaks of over 2000 watts, which I should be able to accomplish
with a bridged Crown XS900.

Your loudspeaker and listening room parameters will have more to do
with crossover power disipation than amplifier power output (i.e., if

These speakers will hopefully perform for a medium size dance floor,
as well as outside. It's not for any DJ'ing purpose though, more for
my own gluttony to have speakers that are just ridiculously powerful.
All genre's of music must play satisfactorily though I find the techno
genre to be the most challenging for the faithful reproduction of
bass, and this is my goal, to have thunderous bass.

1) Do the typical crossover calculations for determining component
ratings apply at this level of power?


Yes - but at the associate current levels, crossover layout is very
important to avoid current induced voltage effects.

Noted

Standard formulas for parallel connection
resistors/capacitors/inductors apply when paralleling components.

This is where my electrical theory is a little weak (I'm on a need to
know basis). On your advice, I intend to use equal-value components
in parallel but I'm not exactly sure how they would combine in effect.
For example, if I have two 10-watt, 50-ohm resistors in parallel, is
this the same as a single 20-watt 50-ohm resistor? Or does the
resistance change for a parallel setup? How about for capacitance and
inductance?

As to what it will actually need to dissipate, that depends not only
on the particular crossover design employed, but also on the load and
the intended use, not the amplifier's output capability -

I can see the speakers being driven at high loads with the subs taking
a pounding. I wouldn't know where to guesstimate the average load
that the crossover would bear. Is there a working ratio that makes
sense? (e.g. design the crossover to handle 50% of amplifier's
maximum output?)

thanks for you help,
Sean

Sean wrote:
This is where my electrical theory is a little weak (I'm on a need to
know basis). On your advice, I intend to use equal-value components
in parallel but I'm not exactly sure how they would combine in effect.
For example, if I have two 10-watt, 50-ohm resistors in parallel, is
this the same as a single 20-watt 50-ohm resistor? Or does the
resistance change for a parallel setup? How about for capacitance and
inductance?


The resistance (R) changes in proportion to the number (n) of
resistors used: R/n. For n=1, it's just the resistance of the one
resistor (Req=R/1=R). Two 50 Ohm, 10 Watt resistors in parallel
combine to form an equivalent 25 Ohm, 20 Watt resistor (Req=R/n=50/2).

n equal value inductors (L) in parallel combine the same way as
resistors (Leq=L/n).

n equal value capacitors (C) in parallel combine as Ceq=n x C. For
instance, three 10 microFarad capacitors combine to form an equivalent
single capacitor of 30 microFarads (Ceq = 10 + 10 + 10 = 3 x 10 = 30).

You can do a multil-term Goggle search on the terms: parallel resistor
capacitor inductor formula to find the general equations for non-equal
component values.

I can see the speakers being driven at high loads with the subs taking
a pounding. I wouldn't know where to guesstimate the average load
that the crossover would bear. Is there a working ratio that makes
sense? (e.g. design the crossover to handle 50% of amplifier's
maximum output?)


Depends on how you arrange the loudspeaker drivers (parallel, series,
or independently driven) and what kind of crossover configuration you
plan to use (LC, dual RL or RC, driver impedance correction, driver
resonance correction, etc.). In general, those components in series
with the drivers will see all of the current into the drivers, and
those in parallel with the drivers will see all of the voltage into
the drivers. For components in series with the driver (in general),
the higher its resistance to the current flow, the greater the power
dissipation in those components. For inductors and capacitors, this
resistance should be low - say on the order of 0.2 to 0.5 Ohms worst
case in the audio band (inductors in parallel also decrease the
equivalent DCR, which is a good thing; similarly, parallel capacitors
reduce the equivalent internal resistance). Put 20 Amps through 0.2
to 0.5 Ohms, that's equivalent to 80 to 200 Watts dissipation (i x i /
r, double if you include a safety margin). For components parallel to
the driver (again in general), it's the voltage across the component
and its equivalent resistance that determines power dissipation (v x v
/ r).

For the equivalent load presented by your drivers, you can calculate
the worst case current and voltage that your amp can deliver. For
example, for an amplifier with a 900 Watt, 8 Ohm rating, that's 85
Volts max voltage. Let's say the driver load is 4 Ohms, then the
maximum current through the driver is 30 Amps. That's an equivalent
peak output power of 3600 Watts and an average output power of 1800
Watts (assuing both the amp ad the wall outpet can deliver these
levels). With bridged amplifiers, the voltage swing available
doubles (85 becomes 170 Volts), but current is limited to what a
single channel can deliver (probably still on the order of 30 Amps),
and the output noise and distortion doubles.

Best regards,

Terry

"Tip" wrote in message ...

There are digital crossovers available; for example,
Behringer makes them. If you are really interested in
an all-digital system (preamp, crossovers, amps,
speaker correction, room correction, surround), take a
look at Tact Audio components (that's what I use). The

Three things steer me away from this approach,

1) I would need two more amplifiers and of course the digital
crossover ($$$)

2) I have little faith in today's standard of what 'digital' means. I
have gone through a plethora of purportedly 'digital' devices, only to
have them spew volumes of noise into my signal. No sir, it's been a
precarious path to get where I am and I'm most reluctant to further
add any form of electronic signal processing.

3) I cannot see the advantage of active digital over passive
crossovers, assuming of course that I would never have reason to
dynamically adjust the frequency separation characteristics, which I
do not anticipate needing. You won't convince me that I cannot
acceptably digitally equalize my original signal (in the computer
before it even gets to the sound card) to compensate for any passive
crossover, speaker, or acoustical aberrations.

I sincerely appreciate the post though, thanks,
Sean

Sean wrote:
I'm building a pair of dual-12 cabinets. The 12" Oz Audio Elite's
each handle 500 Watts RMS so it is my intention to drive each cabinet
with peaks of over 2000 watts, which I should be able to accomplish
with a bridged Crown XS900.

Are you familiar with Meridian products ? They make active speakers
which have a digital input, then digital crossovers, and then separate
DACs and amplifiers for each driver.

http://www.meridian-audio.com/

If you read their white paper:

http://www.meridian-audio.com/w_pape...eakers_prt.pdf
(ok, it has a bit too much propaganda and too little information
to be a true white paper):

you might get some useful ideas for your project. In particular they
explain their reasoning for having line-level crossovers and separate
amplifiers for each driver. * With such a setup you wouldn't have any
worry about power dissipation on the crossover.

* Although I suspect their calculations in which they shown that
you need much less total power with separate amplifiers don't really
apply to RMS power, only to peak power.

The white papers at harman might also be useful:
http://www.harman.com/wp/index.jsp?articleId=122

Standard formulas for parallel connection
resistors/capacitors/inductors apply when paralleling components.

This is where my electrical theory is a little weak (I'm on a need to
know basis).

There are probably on the web introductory texts on electricity and
electronics which explian the basics (see below).

On your advice, I intend to use equal-value components
in parallel but I'm not exactly sure how they would combine in effect.
For example, if I have two 10-watt, 50-ohm resistors in parallel, is
this the same as a single 20-watt 50-ohm resistor?

No, it would be equivalent to a 20-watt 25-ohm resistor. But if you used
four 10-watt, 50-ohm resistors in serie-parallel, you would get a 40-watt
50-ohm resistor.

Or does the resistance change for a parallel setup?

Yes. The resistance of two resistors in parallel obeys the equation:
1 / R = 1 / R1 + 1 / R2
In the case that both resistors are equal (R1 = R2) this gives:
R = R1 / 2

In series the resistance is simply added, so:
R = R1 + R2

How about for capacitance

It is reversed: in parallel it sums, in series it divides. See for instance:
http://www.tpub.com/neets/book2/3e.htm
http://www.allaboutcircuits.com/vol_1/chpt_13/4.html

and inductance?

Same as resistance:
http://www.aikenamps.com/AddingComponents.htm
http://www.allaboutcircuits.com/vol_1/chpt_15/4.html

--
http://www.mat.uc.pt/~rps/

..pt is Portugal| `Whom the gods love die young'-Menander (342-292 BC)
Europe | Villeneuve 50-82, Toivonen 56-86, Senna 60-94

explain their reasoning for having line-level crossovers and separate
amplifiers for each driver. * With such a setup you wouldn't have any
worry about power dissipation on the crossover.

I understand why you suggest this route but you have to agree that any
addition of powered components at any point along a signal's journey
will invariably add noise. My ultimate goal is to minimize the number
of these noise-generators. If the price for that is crossover
dissipation, then I'll just have to drive more power.

It is reversed: in parallel it sums, in series it divides. See for instance:
http://www.tpub.com/neets/book2/3e.htm
http://www.allaboutcircuits.com/vol_1/chpt_13/4.html

and inductance?

Same as resistance:
http://www.aikenamps.com/AddingComponents.htm
http://www.allaboutcircuits.com/vol_1/chpt_15/4.html

Thanks for the links! Between you and Terry, I think I'm about ready
to build the monster speakers of my dreams!

thanks,
Sean

In article mJ1tc.115688$536.21522461@attbi_s03,
Sean wrote:
explain their reasoning for having line-level crossovers and separate
amplifiers for each driver. * With such a setup you wouldn't have any
worry about power dissipation on the crossover.

I understand why you suggest this route but you have to agree that any
addition of powered components at any point along a signal's journey
will invariably add noise.

In any sort of reasonable implementation the noise is irrelevant.

The noise floor of my 3-way active cross-over with 13 op-amps the tweeter
signal path is no higher than my solid state power amplifers' with the inputs
shorted - completely inaudible a bit past a foot from the tweeter (listening
position is 8' away).

--
a href="http://www.poohsticks.org/drew/"Home Page/a
Life is a terminal sexually transmitted disease.

Sent this a few days ago but never appeared.

"Sean" wrote in message
news:O6arc.85510$536.14252301@attbi_s03...
Hi folks,

I would like to know how to construct a 2-way, 12-db passive crossover
that is capable of handling up to 2000 watts, and does well at
retaining the quality of sound at that level. Any advice or links
would be greatly appreciated.

I've designed crossovers for 1,000 W systems but, as Isaac says, much higher
than that and you really should go active.

With a 1,000 W crossover, the broad guidelines would be:
Air-core coils only
400 -600 V polyprop caps
Thick and wide tracks on the PCB
Feet under the PCB to space it away from the cabinet
Polyswitch HF unit protection
Silicon sealant glue under the components (and tiewrapped)

The critical problem is cooling the resistors. I used 50 W copper-clad types
on heat sinks off the PCB, with plastic spacers under the heatsinks so that
they did not touch the cabinet.

It should not be too difficult to build a 2,000 W crossover on these
principles. You can always double up on the caps, use 100 W resistors and
hardwire the components. I'd be tempted to mount the heatsinks on the
outside of the cabinets, too.

Better go active, though.

Stephen

Sean wrote:
explain their reasoning for having line-level crossovers and separate
amplifiers for each driver. * With such a setup you wouldn't have any
worry about power dissipation on the crossover.

I understand why you suggest this route but you have to agree that any
addition of powered components at any point along a signal's journey
will invariably add noise. My ultimate goal is to minimize the number
of these noise-generators. If the price for that is crossover
dissipation, then I'll just have to drive more power.


Sean I think you overestimate the audibility of the added noise. When just
normal TL074 opamps with 15nV/Hz^-2 are taken a 4th order L-R highpass would
have around 8uVrms noise, which is around 100dB below 1Vrms maximum level
and less than the noise floor on a CD. The power amp will probably have
almost 10 times that amount. :-(
To use coils and caps in this power range will cost and weight a lot,
especially for low xover frequencies. You also have to compensate the
loudspeaker pole frequency if this approach should work.
It is also quite difficult to design passive filters and requires a
specialized software, where as simple active filters are a piece of cake.
good luck
--
ciao Ban
Bordighera, Italy

Hi Sean,

There are digital crossovers available...

Three things steer me away from this approach,

1) I would need two more amplifiers and of course the
digital
crossover ($$$)

Yes, that is a hindrence! However, the two additional
amps do not need to be nearly as powerful if used for
the tweeters. And high-quality crossover components
that can handle the high power you are talking about
aren't particularly cheap either, although an order of
magnitude cheaper than the amps.

2) I have little faith in today's standard of what
'digital' means. I
have gone through a plethora of purportedly 'digital'
devices, only to
have them spew volumes of noise into my signal...

In the Tact system, the digital output of the transport
(PCM) goes directly to the digital preamp/crossover,
which processes the data in the digital domain (24
bits) and outputs the data to the digital amps (PCM).
The digital amps process the data in the digital
domain, directly converting PCM to PWM without an
analog conversion in between. The digtial signal is
finally converted to analog by a filter just before the
output terminals. The system is dead quiet. The
volume is controlled by changing the output of the
power supply in the power amps so that all 24 bits of
data are always sent between the preamp and power amps.
There is no distortion added to the signal by the
digital processing, which cannot be said of analog
domain processing.

3) I cannot see the advantage of active digital over
passive
crossovers, assuming of course that I would never
have reason to
dynamically adjust the frequency separation
characteristics, which I
do not anticipate needing. You won't convince me
that I cannot
acceptably digitally equalize my original signal (in
the computer
before it even gets to the sound card) to compensate
for any passive
crossover, speaker, or acoustical aberrations.

That's because you are just learning about crossovers
at this point. I don't want to discourage you from
the educational experience of designing your own
passive analog crossovers, but it is much simpler doing
it in the digital domain where you don't have to deal
with the negative side effects (e.g., phase shift).
The ability to easily change the parameters of the
crossover will be needed to fine-tune the crossover
design. I doubt that you can compensate for side
effects like phase shift using digital EQ in your
computer, unless you have some kind of time-domain
based digital EQ software. Changing the parameters of
the passive crossover means buying more parts, and the
parts probably have a tolerance of 10% which will make
it harder to get the exact values you want.

Good luck and keep us posted on your progress!

Regards,
Tip

Terry,

Thank you very much for your time and explanations, I really
appreciate it. My speakers are going to be phenomenal.

Sean

Sean,

de nada ... best of luck in your quest (but, along the way, I would
recommend highly that you try to avoid permanent hearing damage

Terry

Sean wrote:

Terry,

Thank you very much for your time and explanations, I really
appreciate it. My speakers are going to be phenomenal.

Sean

I've designed crossovers for 1,000 W systems but, as Isaac says, much higher
than that and you really should go active.

This is why I love the newsgroups

With a 1,000 W crossover, the broad guidelines would be:
Air-core coils only
400 -600 V polyprop caps
Thick and wide tracks on the PCB
Feet under the PCB to space it away from the cabinet
Polyswitch HF unit protection
Silicon sealant glue under the components (and tiewrapped)

Excellent, thanks for sharing this. I have a couple related questions
if you don't mind:

1) Was it a two-way or three-way?
2) Where were your frequency splits (1600 Hz?)?
3) Was it a 6 dB or 12dB rolloff (or more)?
4) How much power did you design into each frequency split?
5) What gauge wire did you use

hardwire the components. I'd be tempted to mount the heatsinks on the
outside of the cabinets, too.

I think the execution of an exterior heatsink design would prove
challenging. I'm more inclined to mount the heatsink interiorly but
in the near vicinity of the port's airflow.

Better go active, though.

I have considered this many times, especially in light of multiple
postings pushing me this direction, however, I still maintain that
adding another noise-producing amplifier (even if the crossover is
digital) to my setup will demote it (albeit, perhaps only
psychologically) away from my 'dream' noiseless system that I have
been pursuing forever.

There was a post earlier that describes the noise floor as being
irrelevant from eight feet away...well, my goal is to not hear a thing
from less than a foot away. Why? For the same reason we do not all
drive Honda's, because my system is better! Perfectionism is a
crippling disease.

"Sean" wrote in message Excellent, thanks
for sharing this. I have a couple related questions
if you don't mind:

1) Was it a two-way or three-way?

Two-ways with 12s and 15s and 1 inch compression drivers, 3-way with 2-inch
compression drivers and bullet HF. These were usually coupled to a sub-bass
cabinet crossed over actively.

2) Where were your frequency splits (1600 Hz?)?

Various - it depends on the compression driver and horn, but 1600 Hz is a
good point to cross over from a 12 inch, although not every 1" compression
driver is happy with that. I used the big RCF 1" (don't think they make it
any more) and crossed it over at 1k5, which was fine.

3) Was it a 6 dB or 12dB rolloff (or more)?

I've always favoured 24 dB Linkwitz-Reilly. You really need at least 18 dB
to protect the HF unit.

4) How much power did you design into each frequency split?

At 1500 to 2000 Hz crossover frequency you should reckon that the HF will
take about 20 - 25 percent of the total power.

5) What gauge wire did you use

I used 1.5 mm for the LF and .8 mm for the HF.


hardwire the components. I'd be tempted to mount the heatsinks on the
outside of the cabinets, too.

I think the execution of an exterior heatsink design would prove
challenging. I'm more inclined to mount the heatsink interiorly but
in the near vicinity of the port's airflow.

You should be OK with that. Just don't skimp on the heatsinks.

Stephen