Restoring an awful DIY speaker

[ChrisABC]

Audio Excite- Kensho 2
SB15NRXC30-4 & SB26STAC-C000-4 (vs Morel MDT29)

http://www.audioexcite.com/?page_id=6373

http://www.audioexcite.com/?page_id=4121

Madisound Leap Crossover
Morel MW144/8 & MDT29/4

https://www.madisound.com/library/stacks/MW1448-MDT29-4.pdf

Both crossovers 2000 hz. No waveguide.

How about using the Audio Excite SB15 woofer crossover components and the Madisound MDT29 tweeter crossover components.
See how they simulate.

 

[Mister Audio]

Here is a crossover design I put together:

View attachment 1312123 .

I am getting quite a null in reverse polarity (should it be better?)


View attachment 1312124


The overall response looks kind of satisfying, but I am concerned about the woofer's persistent break up. I was able to knock it down with a 3rd order filter but then I had trouble getting the reverse null. Do you think it's fine if I leave it like this or is it going to be audible?

From what I understand about your woofers upper frequency response > a R/C Zobel network should be considered essential.
Along with tweeter 'polarity wiring', the listening height of the speakers can make huge differences to CO frequency nulls.
The difference between tweeters below ear level compared to tweeters above ear level can even remove a null.

 

[hautparlurker]

You need to place those LPAD resistors AFTER the xover. They are across the input and will get hot.

Thank you for pointing that out, I knew something was sketchy about it. At first I've placed them to the tweeter side but in that case the impedance curve was off the charts (literally)

I am not very good at interpreting the impedance charts but I don't think this looks good does it? Can anybody explain what is happening here?

 

[hautparlurker]

How about using the Audio Excite SB15 woofer crossover components and the Madisound MDT29 tweeter crossover components.
See how they simulate.

I gave it a try but the tweeter side doesn't really work for me since my tweeter is in a waveguide. However that paralel CLR on the woofer helped a lot in knocking down the energy around 700-1.5k.

From what I understand about your woofers upper frequency response > a R/C Zobel network should be considered essential.
Along with tweeter 'polarity wiring', the listening height of the speakers can make huge differences to CO frequency nulls.
The difference between tweeters below ear level compared to tweeters above ear level can even remove a null.

Are you most definitely certain that the break up frequencies gonna be a problem at -20db's? I am asking because I am finding it really difficult to keep my parts count at a moderate level. I am already at 4 caps 3 resistors and 3 inductors without the zobel network.

By the way I am attaching the .frd and .zma files in case anybody is interested...

 

[wolf_teeth]

You can try the series resistor before the xover, and the parallel resistor after and see if that improves things. The parallel resistor before the xover is what creates the previous issue.

 

[Mister Audio]

I gave it a try but the tweeter side doesn't really work for me since my tweeter is in a waveguide. However that paralel CLR on the woofer helped a lot in knocking down the energy around 700-1.5k.

View attachment 1312458




Are you most definitely certain that the break up frequencies gonna be a problem at -20db's? I am asking because I am finding it really difficult to keep my parts count at a moderate level. I am already at 4 caps 3 resistors and 3 inductors without the zobel network.

By the way I am attaching the .frd and .zma files in case anybody is interested...

I seem to be loosing track of which circuit diagram is your current/new/final version.
The circuit just above includes a CLR network across the woofer, which is effectively what I was referring to re. woofers impedance & response.

 

[ChrisABC]

Try your tweeter crossover in place of the Madisound tweeter crossover. With Audio Excite woofer crossover.

Which woofer are you using? 4 ohm or 8 ohm?

 

[abstract]

The shape of the EQ is different for baffle step than in a graphic equalizer, but yes, either way the baffle step loss is compensated by boosting power to the bass. By doing this at line level, you are not throwing away the midrange power of the amp into a resistor, since you can amplify a little to compensate for the loss of midrange I think you would end up with slightly higher SPL at clipping with the line level solution and of course have a simpler crossover with fewer components.
Either will give you the same ultimate response with adequate components, but the components for the line level solution may be rather cheaper.

Agree. The Pass forum has a line level crossover, using J113 JFETs if I recall.

It could be a matter of adapting that to a simplified version using just one filtering stage nestled between 2 buffers. The filtering stage would be an R-C-R low-pass shelf. It's like a cross between an RC low pass filter and a resistor divider.

Once you get your head around it, it becomes quite simple. At v.low frequencies, the capacitor blocks the shunt resistor, so the gain is close to 1. At high frequencies it's a resistor divider, so you can set the exact amount of attenuation by controlling the resistor ratio. At intermediate frequencies, you adjust the 2 cut-off points by controlling the R1:C ratio and R2:C ratio.

One advantage of this approach is that you don't feed the amplifier with too much signal, causing distortion, only to be forced to reduce it again before it reaches the speaker.

 

[ChrisABC]

more -
Madisound Leap SB15NRXC30-4 & SB29RDNC
https://www.madisound.com/library/stacks/SB15NRXC30-4_SB29RDNC-2-way.pdf

 

[hautparlurker]

You can try the series resistor before the xover, and the parallel resistor after and see if that improves things. The parallel resistor before the xover is what creates the previous issue.

Nope, didn't help.

I seem to be loosing track of which circuit diagram is your current/new/final version.
The circuit just above includes a CLR network across the woofer, which is effectively what I was referring to re. woofers impedance & response.

That CLR you are referring is for taming the area roughly around 700-1.5k. I needed to add another CLR network to knock down the cone breakup as you can see on the graphs below.

before:

after:

My question was if I must include this additional network for the higher frequencies. Is that breakup going to be a problem when it is -20db's lower than the overall response. (bare in mind this is a paper woofer not aluminum or something like that)

 

[hautparlurker]

Agree. The Pass forum has a line level crossover, using J113 JFETs if I recall.

It could be a matter of adapting that to a simplified version using just one filtering stage nestled between 2 buffers. The filtering stage would be an R-C-R low-pass shelf. It's like a cross between an RC low pass filter and a resistor divider.

Once you get your head around it, it becomes quite simple. At v.low frequencies, the capacitor blocks the shunt resistor, so the gain is close to 1. At high frequencies it's a resistor divider, so you can set the exact amount of attenuation by controlling the resistor ratio. At intermediate frequencies, you adjust the 2 cut-off points by controlling the R1:C ratio and R2:C ratio.

One advantage of this approach is that you don't feed the amplifier with too much signal, causing distortion, only to be forced to reduce it again before it reaches the speaker.

Thank you for your comments but I've decided to fix the problems at the crossover. I don't really want to make it dependent on external adjustments, and quite frankly, I am not sure if I can wrap my head around the electronics involved in this right now.

more -
Madisound Leap SB15NRXC30-4 & SB29RDNC
https://www.madisound.com/library/stacks/SB15NRXC30-4_SB29RDNC-2-way.pdf

I have given it a try, doesn't really work in my case

 

[hautparlurker]

here is the final crossover I came up with in xsim. The response looks very good, but the impedance? does this make any sense?

I am a total noob when it comes to interpreting the impedance charts, so I definitely need your inputs.

Thank you very much!

 

[AllenB]

Does this sound light in the bass?

Two reasons. Your chosen crossed sensitivity level may challenge the lower hundreds. You say this is baffle step but it may be anomolous data from your gating.

Secondly, you have used the tails capability in the Xsim Hilbert function to modify that region.

 

[hautparlurker]

Please ignore the last post, it was too late to edit so I have to post again. (sorry for flooding)
this is the actual, latest, final, utmost, definitive design I made:

 

[hautparlurker]

Does this sound light in the bass?

Two reasons. Your chosen crossed sensitivity level may challenge the lower hundreds. You say this is baffle step but it may be anomolous data from your gating.

Secondly, you have used the tails capability in the Xsim Hilbert function to modify that region.

View attachment 1313022

My gated measurements are from 250hz on, below that is filled in xsim using the hilbert bode function. My concern was the difference between the 250-500hz region compared to 1k onwards.

this final crossover design I made flattens that out but yeah lots of sensitivity is sacrificed... I don' know what's gonna happen around 100hz but I believe I should be fine given that area is almost always all over the place (in room), and I always find my self applying some eq for room correction at that region (100hz and below)