New Project: RS-100 and ND20 MTM

[mpotoka]

I would have thought someone else would have done this by now--maybe my internet searching abilities are very lacking. Anyways, I just ordered 4 4-ohm RS100s and 2 ND20FAs. Should arrive on Wednesday, maybe by the end of the week I will get a chance to do some testing. I recently moved to chicagoland and haven't unpacked any of my speaker measurement equipment yet. This should be some good motivation.

[jkrutke]

I would have thought someone else would have done this by now--maybe my internet searching abilities are very lacking. Anyways, I just ordered 4 4-ohm RS100s and 2 ND20FAs. Should arrive on Wednesday, maybe by the end of the week I will get a chance to do some testing. I recently moved to chicagoland and haven't unpacked any of my speaker measurement equipment yet. This should be some good motivation.

I was going to do the same thing with a pair of Aura NS3-193-8A per side and an NT1 tweeter. I picked the Aura woofers because of their performance on the low end and their light weight. (these were to be wall mounted speakers) I got the woofers in and the Fs is 20 Hz higher than the last ones I used, even after breakin. As a result, I think I'm going to try different woofers in that design.

[mpotoka]

Well I had a little time so I ran some simulations. I certainly have some more to learn about XO design, and I will take actual measurements before getting any parts. But it was still instructive. Designed a 6l box tuned to 65hz, did some baffle simulations and used FRC to add those all up.

Here is what I was proposing--effective 4th order LR at 4200 hz

Image not available


Impedance Curve:

Image not available


I had temporarily posted some other info but fixed it.... Any thoughts?

[HareBrained]

I did numerous simulations with the RS100-4/-8 and the ND20FB/FA/ND28 and a couple of other inexpensive tweeters, in MT/MTM/MTMW/MTMWW configurations and really didn't hit upon anything that I felt it was worth building. I say that because I modeled some existing designs and modeled other small drivers with much better (simulated) results.

I used FR and T/S data from Zaph for all of the drivers.

The issues I encountered were related to a complex crossover (9-12 components) and/or raggedness of the RS FR. The RS has a natural rise in the range that is boosted by the baffle step diffraction. The net results in exactly what I see in your simulation, +3db octave at ~1200Hz. Also, the break-up centered at 13kHz is going to be audible (IMO).

I suggest you model with the RS offset in baffle and use a wider 7.5" baffle. You might be able to reduce the level of1200-2400 without a notch filter. Try modeling at 70-75Hz tuning (same volume) and it will offset the droop below 500Hz.

In the end, I decided that the $50+ and net 85db could be bettered by a other driver(s). The TB W4-1337 was one of the drivers that was far easier to use and resulted in higher sensitivity. I'm about to pull the trigger on a MCM 55-3853 MTM with the Vifa DQ25. It'll be 7.5L vs. the 6L of the RS but it still fits within my dimensional requirements. (Thanks for the inspiration Zaph.)

[mpotoka]

I did numerous simulations with the RS100-4/-8 and the ND20FB/FA/ND28 and a couple of other inexpensive tweeters, in MT/MTM/MTMW/MTMWW configurations and really didn't hit upon anything that I felt it was worth building. I say that because I modeled some existing designs and modeled other small drivers with much better (simulated) results.

I used FR and T/S data from Zaph for all of the drivers.

The issues I encountered were related to a complex crossover (9-12 components) and/or raggedness of the RS FR. The RS has a natural rise in the range that is boosted by the baffle step diffraction. The net results in exactly what I see in your simulation, +3db octave at ~1200Hz. Also, the break-up centered at 13kHz is going to be audible (IMO).

I suggest you model with the RS offset in baffle and use a wider 7.5" baffle. You might be able to reduce the level of1200-2400 without a notch filter. Try modeling at 70-75Hz tuning (same volume) and it will offset the droop below 500Hz.

In the end, I decided that the $50+ and net 85db could be bettered by a other driver(s). The TB W4-1337 was one of the drivers that was far easier to use and resulted in higher sensitivity. I'm about to pull the trigger on a MCM 55-3853 MTM with the Vifa DQ25. It'll be 7.5L vs. the 6L of the RS but it still fits within my dimensional requirements. (Thanks for the inspiration Zaph.)

Thanks for the feedback. When I spent some more time with it last night--I see what you mean. I hadn't looked very closely, and the scale I was using was pretty wide--making it appear less peaky than it actually was. I have a question about Baffle Diffraction--I loaded the 2 drivers in an mtm configuration, and wasn't sure if I needed to combine the gain from both woofers, giving me not a 7.5 dB baffle gain at peak but with both drivers running a summed 16 dB of gain?

I ran a lot of options, and still have not seen anything I really like. I tossed around the idea of a horizontal center using all 4 drivers in series with 1 tweeter in the middle....

[Undefinition]

I recently moved to chicagoland and haven't unpacked any of my speaker measurement equipment yet. This should be some good motivation.

Where in Chicagoland, can I ask? There are a few of us around here, and it might be nice to get together from time to time to listen and talk shop.

[HareBrained]

Thanks for the feedback. When I spent some more time with it last night--I see what you mean. I hadn't looked very closely, and the scale I was using was pretty wide--making it appear less peaky than it actually was. I have a question about Baffle Diffraction--I loaded the 2 drivers in an mtm configuration, and wasn't sure if I needed to combine the gain from both woofers, giving me not a 7.5 dB baffle gain at peak but with both drivers running a summed 16 dB of gain?

I ran a lot of options, and still have not seen anything I really like. I tossed around the idea of a horizontal center using all 4 drivers in series with 1 tweeter in the middle....

I don't quite see how you could "gain 7.5db" as I've always thought of the baffle step diffraction as decreasing the FR output. And summing to 16db would be really odd. I use Response Modeler to apply the BSD to a single driver (modeled in half volume). When exported and used in Passive Crossover Designer (PCD), the driver FR is summed across the entire range resulting the in only the 7.5db loss.

Oh, are you referring to the xover transfer function? In that case, you should be modeling the drivers as one (which would be correct as a MTM). If your tools don't model it properly, you either have to alter the T/S parameters for the drivers wired in series or you need to switch tools. If it's matters, I was able to use RM and PCD to model Zaph's ZMV5 with great success. (Which I did to understand whether I was using the tools correctly and to understand how to decrease the effects of the BSC.)

Either way, the compensation for the baffle step should be 7.5db for the system but that's relative to the upper frequencies (>1000Hz) as they aren't affected. Some to consider may be going with a 2.5-way alignment since you've selected the 4-ohm drivers. An MTM won't have the 6db advantage since your wired in series so, if you use 2.5-way you simply have to model as a TM and then model the other driver with the BSD to a net flat response.

[mpotoka]

Where in Chicagoland, can I ask? There are a few of us around here, and it might be nice to get together from time to time to listen and talk shop.

I'm just south of downtown--16th and Indiana. I can hear the Bears games if I open my window....

[mpotoka]

Ok I removed my sim pictures for the moment.... here is my original SPL trace (black) and then combined with my box volume and the output from BDS (frd tools) and combined using FRC (frd tools). Looking apropriate?





Oh. And how do I tame that beast at 13000 hz???? Or do I use a 6th order filter and how many octaves down do I need to get?

[Jed]

Oh. And how do I tame that beast at 13000 hz???? Or do I use a 6th order filter and how many octaves down do I need to get?

Just a simple notch filter will take care of the peak. No need to go 6th order.

[mpotoka]

Just a simple notch filter will take care of the peak. No need to go 6th order.

For the love--I can't get a "simple notch filter" to work! Maybe its a Speaker Workshop thing.... I realized I didn't have proper impedance data coming from Unibox--at least it didn't have any information beyond 800hz. I extended it with FRC and thought that might help my filter work, but alas it still does nothing (at least at the target frequency) I was using the ParallelNotch utility at diyaudioandvideo.com.

Darn this stuff can be humbling sometimes... and to think all this time just for sims.


Edit: The values I was getting were worthless. I just "borrowed" one from another place, then started adjusting values to get it where I wanted. Life is looking better now.

[HareBrained]

Your simulation is close but (including the ZMV5 diffraction)...



Notice the boost (+2db) centered at 1k. You don't have that in your simulation. Compare the infinite baffle and box measurements for the 55-3870 on Zaph's site. You'll see how this diffraction applies.

There are a couple of ways to tame the peak. One is to use a lower xover (2kHz) but your tweeter won't reach that low. Another is to use a 3rd order xover. This gives you more options but you'll be dealing with a bump from 1k-3k, once you have the BSD correct. Lastly, you can use a 1st order with LC in parallel (0.05-0.1uf shunt). This unique topology nets a very nice looking woofer FR but getting the phase correct and tweeter to mesh is difficult. A notch filter can be used but really only necessary if you're going up to 5k+. Below that, it can be controlled by the LP xo.

It may be simpler to start with the tweeter xover from Zaph's ZBM4 (Bargain Mini) and play with the RS xo until they mesh.

[mpotoka]

Ohhh--I am guessing our diffraction simualtions don't match because I am not using a rectangular box. Lets see if I can output something from the BDS......

Ok this is what I'm proposing. Its a chamfer, not a roundover.

[mpotoka]

Ok---so I have a Notch Filter on the Left, and a Impedance Compensation circuit on the right. How do I go about adding in Inductors and capacitors to increase my order?

[dlneubec]

If it were me, I would probably look for a more robust neo tweeter that I could cross at about 2500 hz, 4th order. I think the Dayton ND28 or the Vifa D26NC55 could work, for example. This would allow me to more easily tame that peak in the woofer at 1500-2000 hz, without a notch filter. IMHO that peak, unless tamed, will sound pretty bad. The lower crossover point could very well reduce or eliminate the need for the notch filter for the peak at 13000hz, depending on the low pass slope you use on the woofer and how your phase match or tracking is between the woofer and tweeter. If the slope is at least 4th order, it should drop that peak in the range of 15-20db, compared to the 4000hz xover point.

In addition, I've heard that Dayton neo in a number of systems at Iowa this year and it was not the most pleasant sounding tweeter. The Aura Nt1 is a better option, IMO and should be able to be pushed lower as well, though probably not below 3khz. I noticed that it is back in stock at Madisound recently. Still, I'd choose the D26 first, the Dayton ND28 second and then the Aura. Just one man's opinion.

[Blktre]

In addition, I've heard that Dayton neo in a number of systems at Iowa this year and it was not the most pleasant sounding tweeter. The Aura Nt1 is a better option, IMO and should be able to be pushed lower as well, though probably not below 3khz. I noticed that it is back in stock at Madisound recently. Still, I'd choose the D26 first, the Dayton ND28 second and then the Aura. Just one man's opinion.

I ran the Dayton neo tweeter in Iowa. I chose to run it a little hot as a gamble for the large auditorium. Obviously it didn't work well, such is life. The neo tweeter measured exactly 1.5db higher than the Aura. As your ears heard. R4 padded down to 2ohms, both tweeters measure pretty identical. So if your referring strictly to what you heard in Iowa, id definitely take that into consideration.

[HareBrained]

mpotoka IS using the Aura tweeter and he's already purchased them. That's why I suggested he use the ZBM4 xover as a baseline.

I don't think the shape or edge treatment will negate the rise, just simply change it's shape and position as it's due to the fact that the edge distance is less than a wave length away. If you cabinet was wider, I might agree but can't given your design. Again, if you compare the woofer measurements of IB and in-box of the ZMV5 or ZBM4, you'd see the boost. Even RM has a "simple" BSC that applies a slope below 1000Hz without the edge boost but it's only marginally accurate. And given the natural rise in that range of the RS, I would model a reduced midrange.

The notch filters need to be inserted first in the xover, but it's an expensive solution in both cost and SQ. But you can listen to it yourself. The impedance compensation (Zobal) is a good tool for getting the woofer to model well but does add cost. (I'm doing that in my MTM.)

I don't use Speaker Workshop to do modeling and so can't help you with adding the xover components.

[mpotoka]

mpotoka IS using the Aura tweeter and he's already purchased them. That's why I suggested he use the ZBM4 xover as a baseline.

Actually no--I'm using the ND20FA 3/4" Dayton Tweeter. I am not going to worry too much about getting my sims right--it was all just in an effort to get excited about working on them. Instead for right now it became a bit of a frustration.

I should have the drivers sometime tomorrow--I'll break them in and then get out my gear and take some impedance measurements. I should easily be able to build a test box once I get to that point.

I do think that maybe I should use a different tweeter that can go a bit lower. Another design note I should add is that I do have a couple of HiVi M8Ns that I was going to use as bass bins. I've got a behringer sitting here for some active XO duty.

Shame on me for not doing more sims before the purchase but that is just how I operate. I will always find a use for the tweets somewhere else. I know Curt used the ND20FA at 3450 4th order, but I think I need to get a bit lower than that.

[mpotoka]

So I think a new tweeter might be in order.... any strong opposition to the ND28. It appears to be a good fit--the lower sensitivity doesn't hurt me much since I can only expect about 87 for the 2 RS100s in series.

[alias2]

According to Zaphs data the 4 ohm RS 100 is ~ 84dB, two in series is the same.
The 8 ohm is ~ 82dB and two in parallel for 4 ohms will give you ~ 88dB nominal.
The actual speaker sensitivity midband then depends on the amount of BSC.

[mpotoka]

According to Zaphs data the 4 ohm RS 100 is ~ 84dB, two in series is the same.
The 8 ohm is ~ 82dB and two in parallel for 4 ohms will give you ~ 88dB nominal.
The actual speaker sensitivity midband then depends on the amount of BSC.

If this is true--why is Unibox telling me that the SPL at 1 W at 1 m changes from 84.2 dB to 87.4 when I change the Drive Unit Configuration to "2 drive units in series" ?

I do see that the "SPL at 2.83 Vrms 1m" does not change--what is the difference between these two parameters?

[HareBrained]

If this is true--why is Unibox telling me that the SPL at 1 W at 1 m changes from 84.2 dB to 87.4 when I change the Drive Unit Configuration to "2 drive units in series" ?

I do see that the "SPL at 2.83 Vrms 1m" does not change--what is the difference between these two parameters?

Unibox automajically doubles the power because it takes the power stat as the amount applied to that driver. Double the drivers, double to power.

Also, pay attention to the excursion sim within Unibox. The RS100 hits the limits with very little power.

Actually no--I'm using the ND20FA 3/4" Dayton Tweeter. I am not going to worry too much about getting my sims right--it was all just in an effort to get excited about working on them. Instead for right now it became a bit of a frustration.

I have no idea how I got confused, especially since my simulations were using the same tweeter. Sorry about the confusion. ops:

For MTMs, I found the ND20FB to work much better with the RS100-4 while the ND20FA and ND28FA worked much better with the RS100S-8. I think it had to do with the impedances (4 vs. 6) but I don't know why for certain. Try a TM with the -4 and FA. IIRC, they meshed well. One of the reasons I started considering other drivers was the strange impedance I got from the -4 in series and the FB. I spent far more time with the S-8 in parallel and the ND28.

I believe the Cinderellas used the ND20FA. That may give you something to consider regarding Fc and topology.

[mpotoka]

I think I've got a game plan. I have a simulation that looks doable using what I have ordered. The sim puts the XO at 3500hz 4th order LR. I actually will be using all 4 components for the woofers, probably a 3.3 mH and 1-1.2 mH inductor, so the cost won't be as low as I hoped but so goes life--it eliminates the zobel and notch filter and keeps impedance high.

I did get the drivers today, and to break them in a bit I hooked them and listened to them at my desk for the day--they were at least listenable as background music.

Maybe by the weekend I'll get my TS parameters and decide what I'm going to do with a box for sure. It will realistically be a week before I get a test box made, could be another week at least before I can take some measurements. It will be nice to not measure at my house anymore--now I've got space to measure where the walls/ceiling are 30' + apart, so I can elevate the speaker and microphone and should be able to get some nice measurements without reflections.

[mpotoka]

Update

Time for an update: I have gotten my measurement rig mostly put back together. I still need to find my microphone stand and figure out a time for the FR measurements. I broke some drivers in and took some impedance measurements at least.

So I built a box--here is a picture of the test box next to a pop can for scale:




It isn't a big guy by any means--7.5" wide, 12" tall, and 9.25" deep. The driver should see a volume right about 8l. There is a 2" port 4" long. It is a bit of a hump but should help offset some of the baffle loss. Tuning should be about 75 hz (have not yet confirmed with impedance measurement)



On the front you can see the woofers are flush mounted (and gasketed). I accidently got a little too deep, once the veneer is on it might be a bit more obvious. You can also see the bevel I chose to go with for the front edge. I wanted to do something differently from a roundover, and was suprised it actually modeled better than a roundover in the baffle diffraction simulator. I will most likely go bigger with the chamfer--this was the size I was limited to by the bit I had in the table at the time--I can always cut more off (and will be interested in the changes in FR) Right now the face of the bevel is 9/16" leaving me a short little 3/8" flat edge on my removable front baffle. If I don't go bigger I may end up painting the front baffle because I really do not want to try and veneer that little lip.

[mpotoka]

Took a couple more pictures after I opened them up to solder on a few wires:

First you can see the backside of the woofer cutouts have been relieved for improved airflow




And I have lightly stuffed the cabinet with Dacron I had leftover from another project:



Perhaps by the weekend I have a chance to get my FR measurements done.

[HareBrained]

Thanks for the update. I look forward to the measurements and listening impressions.

[JonP]

Looking good!!

Yeah, that back baffle shot is a prime illustration on small drivers NEEDING some breathing room. Very much like the B3S/N's, they'd be down in a tube with unchamfered driver holes.

A bit deep on the driver rabbets... you might get by with another layer of gasket, or thicker gasket. I used some craft store 1/16" foam to cut gaskets on a tweeter, and had cut my holes before veneering, oops... added a thin felt gasket with the foam, raised it up nicely.

On the front chamfers, the wider the better. Same as increasing the radius of a roundover. I'd guess that 45 deg would be the most optimum angle, but I'd just be guessing. Hmmm... wonder if a wider chamfer detracts from the baffle width. Might want to simulate it and see, but 1/4" isn't going to be a big change..

[eyekode]

Cool little speakers . Another suggestion for the too deep flanges is to use multiple layers of closed cell foam tape. Just keep stacking them! I have had to do this before and it works great.

[mpotoka]

Well I actually stayed at work a bit late and took a couple of measurements. I will move into a big room either tomorrow or saturday so I can some better lower-frequency data. It still gave me something to play with crossover wise. I have a pretty good simulation going for a 2nd order butterworth at 5000hz

Anyway what really suprised me was the lack of the monstrous peak I was expecting in the 12k-13k range. This is what my gated farfield measurement looked like for the two RS100s.



It doesn't look near as bad as either the specs from PE or what Zaph measured. I will have to certainly do some more measurements to confirm this.

[mpotoka]

I can't believe its been three weeks. I've been listening to my test box at my desk--I finally got the tools out and took some measurements. The Tweeter center was about 7' off the floor--no side walls for about 10, and the ceiling is another 20' up.

Here is the On axis, 30 deg and 60 deg graph.



The next picture shows the on axis along with the 4th order bessel target curves. It also has a line showing a measurement with the woofer polarity reversed--Does this confirm phase alignment--or not in a bessel alignment?




Now here is where I got a little confused. I have a tweeter offset from the center. When I took my first 30 deg off axis measurement I turned the speaker so that the tweeter was the closest driver to the microphone. I had envisioned that the tweeters would point inwards to the listening position so that is what I measured. However, my measurement ended up looking like this!



That is on axis along with the (I'm calling it inside 30 deg) measurement. Not very pretty. I turned the speaker so the tweeter would be the furthest driver from the mic (tweeter on the outside of the listening position) and got the resulting graph (first one I posted). Is this a red flag that something isn't right? I am thinking it is all baffle induced--but I wanted to run it by everyone and see what they thought.

My predicted response graph was very close to my final result--that was encouraging!

I also forgot to take an impedance measurement through the XO.

I also have had my veneer here for a couple weeks now--if I can finalize that I at least like the box and alignment I can get started on finishing the boxes.

[Dennis H]

I turned the speaker so the tweeter would be the furthest driver from the mic (tweeter on the outside of the listening position) and got the resulting graph (first one I posted). Is this a red flag that something isn't right?

I don't think there's anything wrong. Probably just means you should use them with the tweeter on the outside.

[rhubarb9999]

Any update on this project?

[mpotoka]

I'm in the slow patient process of finishing. I've got the boxes veneered and the front baffle is into its final stage of painting. I was hoping to have them done for inDIYana but didn't make it. Another week or two should get it done.

Oh--and I lost my crossover files and my one printed copy so I may be needing to re-invent the wheel a bit