Filter recommendation (using Hypex Software Design)

Geertidow

Member
2019-12-10 4:03 pm
Dear forum members,

As I am pretty far outside my comfort zone, I hope to offend no one by asking. A forum search did not give me the answer I was looking for (if there is any).

My father-in-law and I have converted my B&W CM8 S2 speakers to active monitors by removing the filters, replacing cables and using two (one per speaker) Hypex FA123 modules.

B&W CM8 are three-way speakers, 4 Ohm (in contrast to what they market), have crossover frequencies at 350 and 4000 Hz and have two identical, parallel mounted bass drivers.

I've spend a whole day measuring using REW and and UMIK-1 microphone and building filters to get the hang of the REW software. The first results are promising. As I will embark on another day of measuring together with my father in law (who has deep knowledge on the physics of sound and the working of audio electronics but no experience in the software itself) I hope you can provide me with some tips.

My questions would be:
1. What would be the most logical order of building the filters/adjusting the timing? I have used this approach so far: Using REW to Determine Time Delays Between Drivers - Technical/Modifications - The Klipsch Audio Community.
2. What would be, given the type of speaker (B&W CM8 S2) be the most logial filter design. I have started with a fourth order Linkwitz-Riley filter but I had to delay th tweeter 40 microseconds for a proper frequency response curve.
3. What is the optimal microphone distance for designing the filters?
4. Does anyone have any general recommendations for this project?

Thanks in advance,

Geert
 

YSDR

Member
2013-11-24 8:13 pm
As a general advice for DSP crossovers is that, you can relatively easily adjust the individual driver responses to be flat over a wide bandwidth before applying the crossover filters. This way it's easy to experiment with different crossover points and slopes, because the natural response of the drivers are less dictating the result. This way, timing adjust is the last thing in the row.
But attention to tweeter response flattening, don't do it loud.
 

AllenB

Moderator
Paid Member
2008-10-18 11:31 am
Measuring and simulating are a natural step regardless of whether you are doing an active or passive network, so I'd have the crossover designed first. This is important so that you can take the necessary steps in designing a crossover. Then translate it into Hypex as a separate step.
 

Geertidow

Member
2019-12-10 4:03 pm
Thanks for your reply.

So far, I've tried an measured a number of possible setups.
In general, it seems that 2nd order LR filters both measure (REW, nice freq response cruve) and sound better ('coherent') than 4th order LR filters.

However, I've come to the discovery that the best sounding setup seems to be:
Bass: LR2nd order lowpass at 240 Hz.
Mid: LR2 240 Hz (highpass) + LR2 2000 Hz (lowpass).
Tweet: LR2 2000 Hz (highpass).

Due to safety of the tweeter I've raised the 2000 Hz crossover to 2400 Hz.
Would it be best to raise the tweeter crossover to +/- 3000-4000 Hz so that the segment of 1000 - 2000 Hz is done by the mid speaker?
Based on the measurements (as YSDR suggested) the mid speaker would be perfectly capable of doing this, but I'm a little lost since the 2000 Hz crossover tweeter sounded so nice. But I don't want to ruin my tweeter at the first party.
 

AllenB

Moderator
Paid Member
2008-10-18 11:31 am
The default crossover seems to be at 4000Hz. This may be a little on the high side for the 5 1/4" mid, guessing it may explain your preference. Have you considered making two crossovers, one just for you and one on party mode.
 

wintermute

Administrator
Paid Member
2003-08-03 11:43 am
Sydney
One question, the crossovers you have tested, were they 2nd order acoustic crossovers, or just the default 2nd order filter available in the package you are using?

a 4th order electrical filter will generally not give you a fourth order acoustic result unless you have very wide bandwidth speakers being crossed over well inside of their bandwidth. However a second order electrical filter will often give you a 4th order acoustic roll off.

Regardless of what technology you are using to create your filters, passive, analog active, or digital active, you really want to be aiming for the acoustic targets that create the best result (and this needs to take into account what happens off axis as well), not worrying about which electrical implementation to use. Almost always you will end up with asymmetry in your filters to achieve symmetry in your driver summing, and a coherent result.

Tony.
 

Geertidow

Member
2019-12-10 4:03 pm
Thanks Tony for your reply. I think I create active digital filters using Hypex software.

Here wouls be my summary of remaining questions so far:

1. Can anyone guess if a 2nd order filter at 2000 Hz is indeed to low for a tweeter from B&W in general? I can't find any data on the CM8 S2 models.
2. Should I reverse polarity on my mids speakers when using 2nd order filters on my 3 way speakers when building digital filters?

Thanks
 

AllenB

Moderator
Paid Member
2008-10-18 11:31 am
The hypex software wasn't necessarily made for crossover design. You use it to set up your filters. You ask here some very important questions, which we may or may not take a guess at, yet this is the very reason we use crossover simulation software and a systematic approach.

I don't mind helping you cut a few corners but the quality of the information may be limited. For example, I would guess that a higher order filter will do the most to protect your tweeter. Try something similar to the original crossover slope at 4k, and compare that filter response to a higher order at 2k. Notice whether the sub 1kHz region is no higher than before, preferrably lower.

With regards to your second question, measure it afterward, each driver separately and then together. Overlay them if possible and post here.
 

Geertidow

Member
2019-12-10 4:03 pm
Dear Allen,

Thanks for your help.
I did take all the steps posted so far. Unfortunately, I was dumb enough to accidentally delete my REW measurements. I'd happily make them again when I'm back home (abroad for a week). I've tested all drivers separately with a sweep measurement in REW. The bass driver performed well up to 2 Khz, the mis from 100 (I think) up to 8000 Hz or so.

The frequency response for the tweeter was flat up to about 1 Khz and then started to decrease in a lineae fashion with lower frequencies stopping at 500 Hz. This led to my assumption that a crossover at 2000 Hz with 12 db per octave was not a big problem but I lack sufficient knowledge to make a proper conclusion.

I have tested 2nd and 4th order LR filters at 2000, 3000 and 4000 Hz with different bass/mid crossovers and in terms of frequency response measurements and subjective listening the 2nd order filters gave more satisfactory results. I did not take phase and timing in consideration yet since I lack understanding of the first and wait until the crossovers are done for the second).

In terms of measurement I think there was no clear winner in terms of frequency response at 2000, 3000 and 4000 Hz for the mid/tweeter crossover (with a 2nd order LR). However, the 2000 Hz crossover just sounded the best. Even when rising the crossover to 2400 Hz for tweeter safety I felt a small decrease in 'coherence'. However, I do not want to ruin my tweeters.

I do not mind remeasuring my unsaved measurements or doing the work. But I do have to wait until Christmas before I can measure again.
 

Geertidow

Member
2019-12-10 4:03 pm
I did find a photo of the tweeter response! Screenshot_20191218-094428.jpg
 

AllenB

Moderator
Paid Member
2008-10-18 11:31 am
The CM8 (maybe not the 2) have a very simple filter on the tweeter. Please check the crossover board and confirm what components are working with the tweeter.

If this is true, you have much freedom of choice in this matter, as long as you use a higher order filter going low.

I suspect the mid is the reason that crossing at 2 is better. Not because it is a bad mid, simply because of its size.
 

Geertidow

Member
2019-12-10 4:03 pm
The CM8 (maybe not the 2) have a very simple filter on the tweeter. Please check the crossover board and confirm what components are working with the tweeter.

If this is true, you have much freedom of choice in this matter, as long as you use a higher order filter going low.

I suspect the mid is the reason that crossing at 2 is better. Not because it is a bad mid, simply because of its size.

Dear Allen,

All tweeter filters from the B&W's have been removed, so there is no analogue filtering anymore.

The attached photo was the hf filter (tweeter and mid). Based on the frequency response of the tweeter, would LR2 on 2000 Hz be to extreme?

IMG-20191209-WA0026.jpg
 

YSDR

Member
2013-11-24 8:13 pm
Can you make a photo of the front and back of mid/hi board?

I say carefully that, a LR2 at 2kHz is fine if you don't play excessively loud. But this speakers is not meant to play too loud anyway.

Btw, I think a properly designed LR4 filter at 2kHz is not worse to the ears than an LR2. IMO. So I would not remove that from the options.
 

wintermute

Administrator
Paid Member
2003-08-03 11:43 am
Sydney
YSDR I agree completely! A well implemented LR4 will sound much better than a poorly implemented LR2 and vice versa :) If there are issues with breakup on the mids, or distortion on the tweeter, then a well implemented LR4 may actually sound better than a well implemented LR2 as well.

I think I need to post some sims for Geertidow to properly understand what I asked before. I'll see what I can come up with.

Tony.