3-Way Crossover Assistance Request.

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Interesting circuit Pico, rearranged it looks like the attached

perhaps if it was 2nd order the extra efficiency of the mid might not be so important?
 

Attachments

  • 3way series.JPG
    3way series.JPG
    3.3 KB · Views: 150
Don't worry Sreten, I really know what i'am saying ;). I just do his crossover works ! When you don't have tools, you are happy to receive some consistent helps.

I think Istoc is very lucky, in this case his crossover works !
Istoc, play a lottery ticket :D
I think you don't need to spend more money in components, you just need to equalize the level of the drivers. You can buy resistors instead LPAD but I think the LPAD will make you to tune easily to your ears. You can put the LDAP near the resistor values as a starting point.
I also recommend to suppress one component (L4) : the inductor of the tweeter !
The only bad thing, sensitivity of this speaker is low... 82dB.
I also recommend to damped well the transmission line because if not well done, some bad waves from the line can disturb the low mid.

Have fun.

Hi,

No. If you don't model it properly you are simply guessing, as this is.

It won't work up to capabilities of the drivers and is wrong in many respects.
For a start you've assumed an 8 ohm mid driver when it is actually 4 ohms.

It is not the way to design a speaker with expensive drivers, I've posted
links with details of how to do that, and there is no other way of doing
it properly, the acoustic effects of the cabinets must be included.

rgds, sreten.
 
Just to clear this up real quick, he assumed correctly my midrange is infact 8ohm not 4ohm.
I'm thankful for all the input you guys are giving me. I'm not entirely sure what Sreten means by modeling though? I've never dealt with LPADs before how much of a decrease should I be aiming for with them?
 
Last edited:
Just another Moderator
Joined 2003
Paid Member
Just looking at the manufacturers curves around 6db of attenuation on the tweeter, though that may be too much.

Modelling is putting the spl and impedance curves (grabbed by something like spltrace) into a crossover modelling program, and simulating what will happen with your various crossover components (I suggested earlier as well). It will give you a much better chance of getting things right (though not as good a chance as if you have real measurements taken on your actual speakers).

Tony.
 
diyAudio Moderator
Joined 2008
Paid Member
Modelling is where you simulate all of the major effects on your computer and you come up with something similar to reality. Doing so can be better than hand tweaking, but it is intermediate level stuff and there is so much to learn first.

So, if I might assume you want something you can achieve now that will make a significant difference, it would be two things. Set the levels, and flatten the impedances.

Fortunately you can virtually achieve both of these steps in one go, and that is to put an L-pad on each tweeter and mid, as some here have been saying. I am assuming that this will probably do more for you than any other single step at this point. Here's how...

L-pads are volume controls for a single driver. They have a parallel and a series resistance which can lower the level at a driver but will still maintain a total 8 ohms, and as it happens the more you attenuate the driver, the more its impedance looks just like an 8 ohm resistor, which is easier to work with.

Wire it between the drivers and their crossover. Then turn the tweeters all the way down. Dial in the midrange until it sounds right, and then do the same for the tweeter.
 
So when getting the resistors for the LPADs, Which I believe are two wired together resistors of different ohm loads correct, where is the best place to start with them? When you say turn the tweeters all the way down what do you mean? Set a really high resistance LPAD in like a 24db or what?
 
MMmmhh leave it outside at all ??!! :)
I believe the transient response that goes from DC to infinite , needs to be practically developed step by step and gradually ...maybe also a 24 dB attenuation would bring alteration to the musical envelope

@Pete : yes ,for sure ,but the minimalistic approach also facilitates the hit'n'run way of doing that . There are already too many variables with first order slopes :)
 
Last edited:
diyAudio Moderator
Joined 2008
Paid Member
I was thinking about the L-pad volume control thingies that I know they used to sell at RS. They were similar to a standard volume control but used coiled resistor wire. I don't require them anymore but in your position I feel they'd be great. Even if you could only get one, use it on one mid then measure the resistance between the pins and replace with resistors...etc
 
I was thinking about the L-pad volume control thingies that I know they used to sell at RS. They were similar to a standard volume control but used coiled resistor wire. I don't require them anymore but in your position I feel they'd be great. Even if you could only get one, use it on one mid then measure the resistance between the pins and replace with resistors...etc

Yes it is a good advice :)

Just a little more about the crossover ...

Istoc, i calculate your point of crossover are 500Hz and 2kHz. Don't worry about this because the slopes are low (12dB acoustic) and there are huge overlap between the drivers. I think you don't need to change the points of crossover. But you should listen before ;)

The contribution of the midrange is lower. Take a look on one of my design with 12dB acoustic slopes :
An externally hosted image should be here but it was not working when we last tested it.

The sensitivity is 85dB but the mid contributes only 82dB, 3dB lower. In you case the mid will contribute to 79dB.
 
Just to clear this up real quick, he assumed correctly my midrange is infact 8ohm not 4ohm.
I'm thankful for all the input you guys are giving me. I'm not entirely sure what Sreten means by modeling though? I've never dealt with LPADs before how much of a decrease should I be aiming for with them?

Hi,

My mistake regarding the midrange impedance, must of misread something ....

By modelling I mean the process outlined in the links of post 12.

SPL trace the curves of each driver. These are into half space.
Then model the effects on these curves of your boxes using a baffle
simulator to model the actual response of each driver in the boxes.
(model the mid baffle as tall and thin to reflect its actual placement).

Then the fun begins designing the crossover. Standard textbook
crossover values may or may not work well with these responses.
The woofer peak looks tricky to me with low order slopes. Secret
is good phase matching so you get good reverse nulls with the
midrange polarity reversed. You just need to play with it a lot.

Its the phase that tricky, the units peak that tricky, the mid
units on axis rise needs to be accounted for, treble unit
seems straight forward, and as it can go low c/o mid to
treble quite high doesn't seem a great idea.

With an 89dB bass unit, padding mid/treble down to 82dB
seems excessive, 5dB of BSC means mid treble is ~ 84dB.

rgds, sreten.
 
Where the 82dB comes ?
If you look the used values 7mH/27uF, you have 6dB loss... it is a fact ! a good thing here because of the transmission line.
If you look the manufacturer curve, i read 88dB, although the specification says an other thing, the measurement says something else, i prefer to trust the measurement.
22W_8534G_00-new-freq.gif


I try to think before saying something ...
That all !
 
So this L-PAD: L-Pad 50W Mono 3/8" Shaft 8 Ohm 260-252
Would work well in helping me diagnose how much resistance I'll need to add to my crossover layout? Should I get 2 or would 1 be good enough?
You should buy at least two and after buy the adequate resistor.

I hate to say this but I'm getting lost with whats being said in the thread at the moment. Can you two clarify? I'm a quick learner its just not registering in my brain bank.

Don't worry... ;) We say the same thing, if you don't have reliable measurement it is difficult to do something precise. But do measurements, build a good crossover could take years.
What I say, with my experience, you can keep the crossover as it and with a little among of money and time, you can do something good ;)
Again you are lucky because your values of crossover seems to be good and work !
 
One for each driver in the speaker as I dial it in correct? Connect them turn them way down and dial it up till each sounds meshed with my bass woofer output, then how do I measure the resistance on the LPADs? Do I use a voltmeter?

Yes you do the test in stereo. It's better. I don't think it is easy to tune in mono. For a beginner, tune the level of a 3 ways speaker is a nightmare.
Put the level of the tweeter with fixed resistor (as i suggest) and vary the level of the mid with the LPAD in order to have a well balanced sound. After if the extreme treble seems missing rise the treble and replay with the LPAD.
If you have two LPAD more you play with the level of the tweeter and adjust the mid in consequence.

Yes you measure a LPAD with a multimeter in the ohmmeter position.
 
Hi,

Im being browbeaten by the appeal of simplistic conjecture that does not take
properly into account driver responses, the actual baffle ripple, which will be
significant for the midrange and generally the suggestion you can simply do
this without any real modelling or checking, I personally don't think so.

You can do it and it will be better than what you currently have, and you
can reuse your c/o components, but it won't be optimum and you won't
know exactly what is going on, I personally don't like the approach.

Without measurements simulation can never be confirmed, but the idea
you can't simulate if you can't measure is just plain wrong, all proper
designs are simulated and confirmed by measurement, nearly all the
design is in the simulation domain, with measurements to check.

I'm not going to tell you what to do, I haven't the time to properly
model it, or the time to model what is being suggested properly.

Or the time to argue subtle technical details, a question of valid
interpretation, not lack of thinking, which is a tedious suggestion.

rgds, sreten.
 
Last edited:
Just another Moderator
Joined 2003
Paid Member
I agree 100% with what sreten said in post #32 there are quite a few aspects to these drivers that make them tricky to work with.

Also don't underestimate the effects of the baffle! It really does make a big difference compared to the manufacturers curves. BDS (baffle difraction simulator) does work very well for simulating these effects, I've verified through sim + actual measurements.

Allen, I think you've mixed up the mid and the bass unit??? the mid has the lowest sensitivity (unless they weren't all 2.93V/1M and I've missed that). The mid shouldn't need any attenuation I don't think.

As sreten said earlier once baffle step is taken into consideration, the Bass unit will probably be fine without any attenuation. it is 4db more efficient than the mid, and 4db is not a bad figure to be using for BSC. could work out quite nicely.

There are two different approaches being presented in this thread. Allen is giving good advice for improving what you have with the minimum effort required, l-pad + some impedance compensation should improve things! Sreten is giving advce on how to give yourself the best chance of getting the best possible result without doing any measurements.

Istoc, ultimately what you are trying to achieve is to make each drivers response meet a certain target response. When you achieve that the drivers sum together to give a flat overall frequency response. Your filters need to make this happen. The drivers have characteristics that are probably going to mean that standard filters won't achieve that.

I'll do another post with a few screenshots of my own two way crossover design, to give you an idea what to aim for.

Tony.
 
diyAudio Moderator
Joined 2008
Paid Member
You can do it and it will be better than what you currently have, and you can reuse your c/o components, but it won't be optimum and you won't know exactly what is going on, I personally don't like the approach.
I agree with this, what I am suggesting is not optimum.

It is where I started with speakers though. It felt positive for me at the time and set me off on the right foot.

Besides, my suggestion thus far can be done on one day and gives Istoc something better to listen on and think about while he spends time going deeper into speaker building.

The woofer peak looks tricky to me with low order slopes.
The peak is only on axis, the power in the breakup region doesn't seem too drastic especially considering Istoc's crossover is more than an octave back.

Allen, I think you've mixed up the mid and the bass unit??? the mid has the lowest sensitivity (unless they weren't all 2.93V/1M and I've missed that). The mid shouldn't need any attenuation I don't think.
Keeping in mind the approach I had been taking, it's about the bass/mid crossover approximately coinciding with the baffle step.

By the way, where do you get your sensitivity difference of 4dB? (I'm only going off the Madisound pages.)
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.