My MTM Project

[Twisted85]

Hi,

This is my second full-DIY speaker project, first with using SoundEasy.

My project is a floorstanding MTM with Peerless 810921 tweeters and Dayton RS180-8 woofers.

I built a single prototype for the measurements and it looks like this:

http://koti.mbnet.fi/twisted-/proto.jpg

The crossover point will be around 2KHz.

Now what I've been wondering is, that should I place the tweeter on the middle like in the picture, or should I move it on the x-axis closer to one of the sides?
The frequency response would be smoother measured from 0 angle, but of course this have it's own effects to horizontal polar plots.

But what do you think? Any suggestions based on your own experiences?

 

[HiFiNutNut]

If you do a search you will find discussions of the topic.

I take what others (S.L., John K) say and believe there is little value of tweeter offsetting. Offsetting the tweeter may produce a flatter on-axis response, but if you move your head left or right a few inches then the response may not be flat any more.

To tackle the diffraction problem, use slim front baffle with a large edge roundover.

 

[Jay_WJ]

I have used both drivers in my projects. I'd cross lower than 2.0 kHz with the RS180s. Its midrange is not as clean as some top performers like the ScanSpeak and the Usher. The Peerless 810921 can go lower and its low-end performance is excellent.

 

[Twisted85]

Ok, I will think about the offset later. (Although I will probably stay with the centered version)

At this point I've measured the responses from the prototype and spent some time with the crossover.

http://koti.mbnet.fi/twisted-/soundeasy/mtm_response.gif

http://koti.mbnet.fi/twisted-/soundeasy/mtm_phase.gif

There's few things about the crossover that I'd like to ask.

1) Does the phase response look good enough as it is now? Can this considered a good match or not?

2) Do you think that the bass response at 150-300 will cause problems in the sound? Is the level too high at this area? Should I make a filter to lower this area few dB?

3) I lowered the xo point a bit from the earlier design, do you think it's now good?

4) Reading the older discussions I found out that people have made filters to lower the Dayton woofer's response at 4-10Khz. How necessary is this? I spent quite a lot of time examining different filters to lower it down but I have to say that my result weren't very good as I have no earlier experience using these type of filters.

 

[Jay_WJ]

Ok, I will think about the offset later. (Although I will probably stay with the centered version)

The tweeter response looks okay in your in-box measurement.

1) Does the phase response look good enough as it is now? Can this considered a good match or not?

They seem to be aligned well.

2) Do you think that the bass response at 150-300 will cause problems in the sound? Is the level too high at this area? Should I make a filter to lower this area few dB?

How did you measure your bass response? Did you merge near field response to the gated far field one?

3) I lowered the xo point a bit from the earlier design, do you think it's now good?

It looks good. It's at about 1.6 kHz, isn't it?

4) Reading the older discussions I found out that people have made filters to lower the Dayton woofer's response at 4-10Khz. How necessary is this? I spent quite a lot of time examining different filters to lower it down but I have to say that my result weren't very good as I have no earlier experience using these type of filters.

In your simple LR4 crossover, the main breakups of the RS180s are suppressed only to 60 dB, which should be audible. Have you seen what people did to suppress the RS180's breakup nodes? Take a look at:
http://www.rjbaudio.com/RS180MTM/rs180-rs28-mtm.html

Also,
http://www.geocities.com/woove99/Spkrbldg/RS180_2way/RS180_2way.htm

The most effective way to suppress the breakups is using the C-E type filter---combination of a 4th order electrical and a notch filter. But in my experience, a simple bottomless notch filter located between the two main breakups works well when combined with a low crossover frequency. In your simulation, try a 0.012 mH inductor in series with your current 33 uF shunted capacitor in the woofer net to form a notch filter, and see how the response changes. Also, try the combination of 30 uF cap and 0.013 mH inductor.

Would you then post the result for me?

 

[Twisted85]

Thanks for your reply!

How did you measure your bass response? Did you merge near field response to the gated far field one?

I measured the farfield with 5.28ms gate, then made the nearfield measurement and port measurement and added the diffraction to the nearfield data. The Merge point is 350Hz.
What actually made me wonder in this, that when I added the diffraction the lowest frequencys droppep lower level.

I have a screencap from adding the diffraction:

http://koti.mbnet.fi/twisted-/soundeasy/diffraction.gif

In JohnK's SoundEasy guide is a picture of adding the diffraction, and in this pic the response only starts to rise from 100Hz, but in my case the response also lowered after under 70Hz. Is this just because the simulated front panels are different or why does this happen?

It looks good. It's at about 1.6 kHz, isn't it?

Yes.

Would you then post the result for me?

Sure!

http://koti.mbnet.fi/twisted-/soundeasy/mtm_response_withinductor.gif
http://koti.mbnet.fi/twisted-/soundeasy/mtm_response_withinductor2.gif

Both has the response without the inductor for reference. It seems to do the job pretty good, but where can I get inductors with so low value? In Finland I only know dealers for Intertechnik and Mundorf components, and couldn't find so small values from their inductors. The lowest I could find was 0.05mH

I also tried to do something to the bass response.

http://koti.mbnet.fi/twisted-/soundeasy/mtm_response_basscorrection.gif

What are your thoughts about this? Is this necessary? I don't want the speaker to sound boomy as it will be placed in a middle sized room with hard stone walls.

In this box the tuning frequency is ~33-34Hz, I've been thinking to use a bit higher 35-36Hz tuning in the final speaker.

 

[Jay_WJ]

Actually the bump is located a bit higher than the upper bass region that makes speakers sound "boomy." So, even if the bump is real, not measurement artifact, it may make speakers sound a bit warm rather than boomy. Anyway, I'd worry about this later. You can judge it after you hear the speakers, and if necessary you can make a change at that time.

Yeah, the small inductor should be a problem if you can't buy it. Can you ask your parts dealer to unwind the 0.05 mH one to reach that value? They should have an inductance meter. I think its accuracy is important because it's used in a notch filter.

 

[Twisted85]

Yeah, the small inductor should be a problem if you can't buy it. Can you ask your parts dealer to unwind the 0.05 mH one to reach that value? They should have an inductance meter. I think its accuracy is important because it's used in a notch filter.

Ordering from overseas isn't a problem if the shipping cost is reasonable. The Dayton drivers are also from Partsexpress as those are not sold in here.
Do you know any dealer that would sell those kind of inductors? I haven't been able to find any.

 

[spock]

Janzen makes a .025 mH inductor. I got mine from partsexpress, but I believe that they are a Europian company.

 

[Jay_WJ]

Madisound winds/unwinds an inductor to a custom value for a small charge. But if you have anyone who has an inductance meter around you, it's very easy to unwind any smallest one to get the value you want.

 

[Twisted85]

I ended up making the inductor by myself. Measured the values with SoundEasy and tested with the whole woofer crossover parts with good results.

I also made the first listening tests and I'm definitely going to try the bass response correction I mentioned before.

http://koti.mbnet.fi/twisted-/soundeasy/mtm_response_basscorrection.gif

But, should I pay any attention to the inductor resistance in this filter? Should I use resistor value 1.0ohm like in the screencap, or should I subtract the inductor resistanse from the resistor value that will be used?

 

[Jay_WJ]

I'd definitely not (and people don't) use a parallel notch filter in series with woofers at that low frequency. Most of the signal above that frequency will go through the capacitor, and you will certainly use an electrolytic cap for that large capacitance. I think the use of a series notch filter in parallel with the woofers is also problematic in your case. It will lower the impedance around the target frequency, which is already very low due to your MTM use.

If I were you and really heard somewhat muddy midrange, I'd experiment with reduced BSC by unwinding the primary inductor a little bit (e.g., changing 1.2 mH to 1.1 mH)---for the experiment, you don't need to cut the unwound coil until you find a value that suits you since it doesn't contribute to the inductance. Of course, by reducing BSC you'll lose some bass impact, but that's a design compromise---as you know, every loudspeaker design has some sort of compromise. For BSC adjustment, refer to the "Voicing your speakers" section in my crossover design page. And as you know, if you decide to reduce BSC, you'll also need to adjust the tweeter level.

 

[Twisted85]

It is hard to describe what is wrong with the sound (also because english isn't my native language 🙂 ), but the sound isn't as open/wide as I'd like. The lowest midrange is most definitely the problem area. I played the speakers to my dad without mentioning anything beforehand about my own views or measurements, and he described the problem area just as I had been thought. I've listened for several different speakers in the same room so the acoustics isn't the problem.

I'd definitely not (and people don't) use a parallel notch filter in series with woofers at that low frequency. Most of the signal above that frequency will go through the capacitor, and you will certainly use an electrolytic cap for that large capacitance. I think the use of a series notch filter in parallel with the woofers is also problematic in your case. It will lower the impedance around the target frequency, which is already very low due to your MTM use.

Electrolytic caps are problematic at high frequencys, right? Is there any rule of thumb which frequencys shouldn't go through this type of capacitor?
One of my speaker design books (written in Finnish) presents the example usage of the parallel notch filter exactly as I was intending to do, and the capacitor in this example is around 200-400uF.
But if the electrolytic cap really is a problem, does it help if a lower value polyester capacitor is used parallel with the higher value cap? For example 470uF +33uF.

Here's a screencap of the impedance curves:

http://koti.mbnet.fi/twisted-/soundeasy/filter_impedance.gif

The red is without the filter, and the green is with the filter described before. The impedance actually seems to rise.

If I were you and really heard somewhat muddy midrange, I'd experiment with reduced BSC by unwinding the primary inductor a little bit (e.g., changing 1.2 mH to 1.1 mH)---for the experiment, you don't need to cut the unwound coil until you find a value that suits you since it doesn't contribute to the inductance. Of course, by reducing BSC you'll lose some bass impact, but that's a design compromise---as you know, every loudspeaker design has some sort of compromise. For BSC adjustment, refer to the "Voicing your speakers" section in my crossover design page. And as you know, if you decide to reduce BSC, you'll also need to adjust the tweeter level.

The baffle step might be compensated a bit too much, you are right about that. I could also try to lower the inductor value a bit, but I don't think it will solve the whole problem as all frequencys under 600Hz are left untouch and the bump around 200Hz is still there.

 

[Jay_WJ]

It is hard to describe what is wrong with the sound (also because english isn't my native language 🙂 ), but the sound isn't as open/wide as I'd like. The lowest midrange is most definitely the problem area. I played the speakers to my dad without mentioning anything beforehand about my own views or measurements, and he described the problem area just as I had been thought. I've listened for several different speakers in the same room so the acoustics isn't the problem.

Electrolytic caps are problematic at high frequencys, right? Is there any rule of thumb which frequencys shouldn't go through this type of capacitor?
One of my speaker design books (written in Finnish) presents the example usage of the parallel notch filter exactly as I was intending to do, and the capacitor in this example is around 200-400uF.
But if the electrolytic cap really is a problem, does it help if a lower value polyester capacitor is used parallel with the higher value cap? For example 470uF +33uF.

Here's a screencap of the impedance curves:

http://koti.mbnet.fi/twisted-/soundeasy/filter_impedance.gif

The red is without the filter, and the green is with the filter described before. The impedance actually seems to rise.

I don't think electrolytics are bad for high frequencies. The reason why I said against the use of a parallel notch at that low frequency is that usually an electrolytic cap has lower power handling than a poly cap, and you may not want to use an additional series component in the network.

Have you tried a series LCR notch filter in parallel with the woofer? This is what I meant would lower the impedance.

The baffle step might be compensated a bit too much, you are right about that. I could also try to lower the inductor value a bit, but I don't think it will solve the whole problem as all frequencys under 600Hz are left untouch and the bump around 200Hz is still there.

A downward tilt of of a woofer filter transfer function usually starts from as low as 300 Hz. If you reduce BSC, it will tilt up the SPL in the 300 Hz to 1 kHz range. Thus, the SPL in the 100 to 300 Hz range will be lowered relatively. Try this in your simulation. If I were you, I'd try this approach instead of using a notch filter.

-Jay

 

[Twisted85]

I don't think electrolytics are bad for high frequencies. The reason why I said against the use of a parallel notch at that low frequency is that usually an electrolytic cap has lower power handling than a poly cap, and you may not want to use an additional series component in the network.

I read somewhere that electrolytics are bad for higher frequencys, so I thought this would be the problem.

Have you tried a series LCR notch filter in parallel with the woofer? This is what I meant would lower the impedance.

My bad. I didn't read your text carefully enough and didn't notice that you spoke about a different filter.

A downward tilt of of a woofer filter transfer function usually starts from as low as 300 Hz. If you reduce BSC, it will tilt up the SPL in the 300 Hz to 1 kHz range. Thus, the SPL in the 100 to 300 Hz range will be lowered relatively. Try this in your simulation. If I were you, I'd try this approach instead of using a notch filter.

-Jay

I tried this before, but to me it doesn't seem to affect low enough.

http://koti.mbnet.fi/twisted-/soundeasy/bsc.gif

Red 1.2mH, Green 1.1mH, Pink 1.0mH

Lowering the inductor value is most probably a good idea, but I still don't like how the response looks and I probably won't be satisfied with the result (although I will try it).

I don't understand why the response even ended up looking like that in the first place. The small bump around 180Hz, and then the response drops 6dB lower around 100-150Hz. I made the measurements twice, on two different times and got the same results. The diffraction is added with correct driver diameter and placement in the simulation.
When I was measuring the nearfield measurement, the other woofer was also playing, is that a problem?

 

[Jay_WJ]

I tried this before, but to me it doesn't seem to affect low enough.

Of course, it doesn't affect the low freq response. It will only tilt up the midrange.

Lowering the inductor value is most probably a good idea, but I still don't like how the response looks and I probably won't be satisfied with the result (although I will try it).

I don't understand why the response even ended up looking like that in the first place. The small bump around 180Hz, and then the response drops 6dB lower around 100-150Hz. I made the measurements twice, on two different times and got the same results. The diffraction is added with correct driver diameter and placement in the simulation.
When I was measuring the nearfield measurement, the other woofer was also playing, is that a problem?

I think you need to re-check your baffle diffraction simulation you added to your near field measurement. Can you post the diffraction effect itself, not the summed response?

 

[Twisted85]

I will probably try to make new measurements this week, but I have one screencap from the old measurements:

http://koti.mbnet.fi/twisted-/soundeasy/diffraction.gif

This is the nearfield measurement from the lower woofer. (both playing when measured) Port measurement added to the response. The one without the arrows is without diffraction added, and the one with arrows is after the diffraction was added.

 

[Jay_WJ]

That looks strange. Accrording to my BDS sim, diffaction loss should be at least 3.5 dB at 200 Hz. But in your graph, it is at most 1 dB. I think there's something wrong with your diffraction simulation.

 

[Twisted85]

Ok, I made new measurements today.

Here are the nearfield measurements from both woofers separately, the one with a bit higher output is the upper woofer. Port output is added to both of these.

http://koti.mbnet.fi/twisted-/soundeasy/nearfields.gif

And here is the diffraction simulation:

http://koti.mbnet.fi/twisted-/soundeasy/diffraction_simu.gif

The effective diameter of the driver is set to 12.8cm.

And here is the result adding this diffraction to the lower woofer data :

http://koti.mbnet.fi/twisted-/soundeasy/diffraction2.gif

It doesn't differ at all from my first measurements, and I believe I am doing the simulations correctly 😕

 

[Jay_WJ]

Now I can see. This is mainly due to the RS180's unique FR. And this is a bit exaggerated in the case of your bottom woofer because the tall baffle makes it less affected by baffle step loss. The top woofer's response will be a bit better if you add its diffaction effect.

As far as I know, no existing RS180 MTM designs use a notch filter to remedy this FR rise around 180 Hz. But you can use it for your design. Try the series LCR notch in parallel with the woofers and see how much the impedance is lowered. If it's okay with your amp, go with it.