"The Riahs" Budget 3-way floor standers (under development)

[Blazin]

Thanks for the suggestions. I'm watching the kids again today, so this afternoon I should have an opportunity to try some measurements again.

I am using an ECM8000 mic. I don't know the specific brand of pre-amp, but I'll test it later and see what I'm getting for power.

I am using the claudio cables, and I am not using the amp output as reference. At least I don't think I am. My connections go as follows:

Left output -> Amplifier
Right Output -> Looped to Left Input
Right Input -> Input from Preamp output

I am seriously considering picking up a M-Audio Mobile Pre--I gave my wife my laptop that had a full-duplex soundcard, my new laptop doesn't, so I need something portable to bring to clients homes anyway for RTA work.

I also am thinking of picking up Soundeasy. It seems quite affordable, and perhaps a little more user friendly. I don't want to get so discouraged trying to use speaker workshop, that I decided to quit trying to do any design, and just build instead.

Whoa wait,
Left out -> Amplifier
Right out -> shouldn't be anything, you're not using it.
Left In -> You need a reference, this is your Reference channel(Left) if you have SW set up this way.. so split your Left OUT and loop it back into Left IN
Right In -> Pre-AMP output..

I literally had the same graphs.. 60 Hz with rising freq after the peak.. The way SW works (through my debug knowledge) is it take your Microphone Signal (FR) and divides it by your Reference Signal(Left OUT -> Left IN) that way any noise or irregularities from your output are removed from your measurement.

I would definitely hook that Left OUT into your Left IN.. that was the main issue but it appeared to be a 60 Hz hum..

[cjd]

Good catch blazin.

Yes, right-out should not be used. Left out should go both to the amplifier and to the reference channel you have set (you can pick which is reference, which is live, in SW)

C

[mpotoka]

Hmm--I'll have to double check that I told you the way I was doing it (the wrong way).... Thank for the input

[mpotoka]

Ok--I had a cable issue--the problem was from my from the pre-amp to line in. I figured that all out, and got some graphs that look much more like they should.

M8N



B3Ns



ND16FA



Those are all taken with a 3 ms gate, at accoustic center 1m away. No smoothing has been applied.

My nearfield measurements still show lower amplitude than the farfield, I decided not to splice it in for now, since I don't really need that for crossover desing.

Now trying to design the crossover is proving to be not-so-exciting. I have attempted to use both Speaker Workshop and the Passive Crossover Desinger FRD tool, but for some reason the PCD won't read the FRD files that I export from Speaker Workshop. I can view the FRD files in SPLView, but it cannot load the ZMA files (either in PCD or SPLView). Looking at phase/impedance/on and off axis response, I did decide to go for a 500hz and 4000hz as a crossover point. All the impedance compensation and BSC and the needed attenuation I am not figuring out how to accomplish yet. Maybe its been too many hours today trying to design. Is soundeasy any easier to use for xo design?

[mpotoka]

Not sure why my B3Ns drop off at 1k so much. Would it be

A. Result of size of box
B. Result of 2 drivers and their spacing
C. Result of baffle
D. Result of 3ms gate

Some of each?

I am encouraged that the rest of the graph for the B3N and the ND16FA look quite similar to Zaphs graphs on his site.

[joecarrow]

A- probably not- what's the inside look like? It's possible that they have some kind of internal reflection going on if they're in a tube-like sub-enclosure. Stuffing would help this. Extending your measurement down to 100ish hz should help verify if it's a total drop off or if it's a frequency specific null.

B- Could be- moving the microphone around will have a big influence on this.

C- Not so likely- I think the baffle is wide enough to push the baffle step frequency well below 1 khz. Don't hold me to that, though- "The Edge" can simulate that.

D- Not sure- the measurement software isn't my strong suit.

The B3N can easily play down to 100 hz and below without damage, especially if you're not cranking the volume up too much. Moving the mic around and getting those lower frequencies (this is the gate, I guess) should be a big help.

[mpotoka]

If I remember right, the 2 B3Ns are in a small sealed enlosure seperate from the sub. It is 9.5" wide, 10" tall, and about 4" deep. Doing a room mode calculator with those dimensions, I'm getting a 680 and 715, plus a 1361 and 1430 (at least in that range) I don't think I put anything in it, however. I will put some stuffing in it and try again tomorrow.

I didn't think of moving the microphone at all, nor did I take any measurements of the drivers seperately. The only reason I put 2 in was to increase the sensitivity, so I just wired them in parallel and am testing both at the same time (is that a nono?)

What is this "Edge" you speak of? I have the FRD baffle simulator.

I was hoping to use SW for measurement and crossover desing, since I belive it calculates driver offset automatically. Someone needs to write a "Speaker Workshop for Dummies" because I feel like one sometimes

[joecarrow]

If it's an issue of how the two speakers' response combines, it will be a very directional effect. If you moved the microphone around and graphed the SPL versus angle at a given frequency, you would see distinct lobes in space where there are high and low intensity areas. This is created by constructive and destructive interference, and tends to get a little better as the distance from the speaker increases.

The program I'm talking about is located here:

http://www.tolvan.com/edge/

I can't say if it's any better than the FRD tools, but it does seem nice on its own.

Regarding the stuffing- a cone midrange really does a lot better with its rear wave managed. This can be done with an open back (like the Statements), as a dipole (like the Arvos, or Linkwitz Orion, etc), or with a terminated stuffed tube (like the B&W Nautilus), or just a reasonably well stuffed cavity. The good news is that it doesn't look like you're planning on running the B3N's very low, so most stuffing materials should be fairly effective.

[mpotoka]

Well I was out of town again... finally got around to trying to make some more progress. I bought a project box, and built myself a modified jig. I have 2 banana plugs, 2 phono jacks, and 2 switches. Then I have a couple cables coming out with stereo minis for the soundcard. No more messing around with the cables, I'm pretty happy with it.

Anyways, after recalibrating everything, I started over taking measurements. Starting with free air, onwards to mounted in box, and then finally some FR sweeps. Whatever I did differently, I'm happy. My nearfield + port measurements looked much better, and they came out higher in amplitude than farfields for once. Spliced them together and I'm thrilled.

My B3Ns, on the other hand, are still suffering from major issues. When I took them out to make sure I had stuffed the cavity, I think I discovered the problem. I have an 1.5" baffle, and even though the drivers are flush mounted, and I chamferd out the backside, the drivers are still basically stuck in a tube. There is probably hardly any room for airflow at all. So my graph still looks like garbage. Tomorrow I'll have to get out the jigsaw and do some major enlarging. I have started glueing my second box together, this time I'll make sure and route out the baffle so its only 3/4" thick in the area around the B3Ns. So for now, here's what my 3 response curves look like (as I start to think about crossover design)

[mpotoka]

Well--I got out a chisel, chipped away another 5/8" or so of the baffle. What a difference. Now I think I might have something to work with.

[Mazeroth]

Yeah, I'd say that looks a little better. :B

[mpotoka]

Now, after doing a bunch of xover design, I've got a 2.4 impedance dip at about 2200 Hz. Running my B3Ns in parallel is pusing it too low. I thought i read somewhere that if I wire them in series, theoretically I will still gain +3 db? I am going to wire them in series, and start over with measurements (again). Unless, of course, some tells me not to worry about it

[69Stingray]

Wiring them in series will double your impedance, R1+R2=Rtotal.

You will get +3 dB for doubling your cone area, BUT, you will get -3 dB for having higher sensitivity, less output for the same amount of Vin. Net 0 dB gain/loss.

[mpotoka]

That is what made sense to me. Anyway, here is what I've come up with so far. I gave up on trying to do anything typical and just started trying to make curves that worked together. I have a feeling that is what is causing my breakup in the 700-1500 range. If anyone can give me any direction that would be great!

I had budget in mind big time in designing this xover, I think I am at about $140 in parts (for both speakers) I know I'm skimping here, but I don't want to invest too much if I can help it.

Also I have the impedance dip at 2500 or so, is there any great way to deal with this, or will a 4 ohm amp have no problems here? What about a 6 or 8 ohm rated amp?

[69Stingray]

Why don't you show two BSN's in your SW crossover file?

[cjd]

really really should try to avoid padding the woofers down. Kinda weird that it's needed, but only kinda. Try covering all of baffle step loss with the woofer, bring the net sensitivity up some If you can.

C

[mpotoka]

Why don't you show two BSN's in your SW crossover file?

Well I have them wired in parallel and measured them as a single unit--both impedance and FR sweeps. Should I not do this?

[mpotoka]

really really should try to avoid padding the woofers down. Kinda weird that it's needed, but only kinda. Try covering all of baffle step loss with the woofer, bring the net sensitivity up some If you can.

C

I undertand I think what you are saying, but not sure I know how to implement... I know I padded down the woofer, the B3Ns are rated at 81 dB, so even with 2 of them, I am only counting on 84. The tweet is at 90, and the M8N is rated at 86. I thought I would have to pad down the woof and tweet.

The woofer seems to have sort of a hill from 20-90 hz, then drops a little in volume. Is that the baffle step loss I am seeing? I am still trying to figure out how to use the Edge.

[cjd]

First, to answer the previous question: when using live measurements, you WILL only see "one" driver in SW even if there are two because measuring them together IS the right way to go.

At 86dB sensitivity you should be looking at 80-82dB net sensitivity on the woofer after baffle step loss. Did you measure output voltage of your amplifier? The mids will be 81dB give or take. Are you using merged nearfield/farfield data here? I'm not sure how you're getting the 20Hz data, but I am highly skeptical. The data you have 200Hz and down does not look accurate. Above, it is as I would expect: sensitivity on par with the mids. I think this is accurate.

C

[Blazin]

Well I have them wired in parallel and measured them as a single unit--both impedance and FR sweeps. Should I not do this?

That's correct. Some people dont use SW and don't know you can't model 2 drivers in a network.

[mpotoka]

First, to answer the previous question: when using live measurements, you WILL only see "one" driver in SW even if there are two because measuring them together IS the right way to go.

At 86dB sensitivity you should be looking at 80-82dB net sensitivity on the woofer after baffle step loss. Did you measure output voltage of your amplifier? The mids will be 81dB give or take. Are you using merged nearfield/farfield data here? I'm not sure how you're getting the 20Hz data, but I am highly skeptical. The data you have 200Hz and down does not look accurate. Above, it is as I would expect: sensitivity on par with the mids. I think this is accurate.

C

I did use spliced data actually on the woofer--hence the data in the 20s. Fixed that (and I even remember reading not to use it) I tried to measure output voltage on my amp. It didn't seem to register anything, which didn't make much sense to me. It would be DC not AC correct? It is a Class T amp, would that impact voltage output? I know I really need to do some more measurements. My question is this--In the SW manual it says to make sure that the VU meter reads the same for Left and Right before taking measurements. If I do this (either increase/decrease amp gain, or increase/decrease mic pre-amp gain) then my measurements wont give me any useful sensitivity gain correct? But at the same time, if it is using the Left input as reference, does it mess up the measurements if the right input is not at the same level?

I am still not sure how to add some BSC either. I tried putting in a network as shown on the Edge, but it didn't seem to alter response at all

[mpotoka]

I am starting the question the worth of having an MTM type configuration with the B3Ns. The low impedance dips dont seem to be worth it--should I build a new baffle with just a single B3N, and use the other B3Ns for a different project? I realize they are going to be pretty ineffecient. If I implement my crossover correctly, I should be able to pick up at least 2 db of bandpass gain--bringing me up to maybe 83.5 dB, but I should have a much more amp-friendly speaker. Would I even need to add any BSC then, or would the woofers baffle loss not be even? I really need to read and gain a better understanding of baffle loss.

I haven't cut any holes on my 2nd box yet.... I am not too concerned with high SPL, this was intended to be a learning process and thats about it. I would like something listenable at the end though

I think I should get a bigger roundover for the front also. I bet that is why my tweeter looks pretty ragged. I thought I had a 3/4" but its only 1/2". Perhaps a 1.25" or so is in order..

[cjd]

MTM is fine running the B3N's in series. I've ended up wiring the mids in series on both the big 3-ways and the Khanspires.

Music signal is always A/C... DC = 0Hz...

Baffle step is unavoidable unless you build in-walls. The larger your baffle, the lower the step frequency (with the un-ending baffle of an in-wall netting that down to zero)

[mpotoka]

WTF--my DMM will measure DC voltage fine coming out of my laptops transformer, but I hook it up to +/- terminals on the amp, play a MLS signal, and nothing. I'm getting sound out of a speaker, so thats not the issue either. Does a class T amp not output "normal" dc voltage?? WTF....

[dawaro]

I think you might want to take a look a couple of designs using the B3N on Curt's site, it might give you a couple of ideas how to handle things.
Scrappys
Cinderella's

[mpotoka]

Curt? Sorry I'm not familiar..

[cjd]

WTF--my DMM will measure DC voltage fine coming out of my laptops transformer, but I hook it up to +/- terminals on the amp, play a MLS signal, and nothing. I'm getting sound out of a speaker, so thats not the issue either. Does a class T amp not output "normal" dc voltage?? WTF....

I should hope not!

Sound is A/C

C

[dawaro]

Curt? Sorry I'm not familiar..

Curt Campbell, he is one of the designers that frequents the site. He designed one of the versions of the Dayton RS WWMT & WMTW's as well as the Statements in the Mission Accomplished threads.

[mpotoka]

I should hope not!

Sound is A/C

C

Whoops! You did say that earlier.... you said "music is always A/C.....DC..." and I just saw the DC. Need to check my eyes I guess 8O

[mpotoka]

Well I'm sitting here listening to one of the speakers. My test crossover is assembled loosely on the couch, wires running everywhere.... I think I got some decent measurements (although I have about no data below 200, making the low end difficult to "see")

Anyways, I just picked something and went with it. I bought electrolytic caps for the big values, so that it wouldn't be an expensive experiment, and I could easily try different things out. The Inductors, unfortunately, aren't so cheap Even if they don't end up in this project, I'm sure I'll find some use for them. So here is what I am running right now... my impedance/phase curve is pretty ugly, I think--hopefully someone can tell from the graph what I did (wrong).

The crossover schematic does not show an L-pad I have on the tweeter, and I also have a 8 ohm resister in parallel with the 1.8 inductor on the M8N.

It doesn't sound bad--but I don't really feel like it sounds great. The bass is there, but doesn't seem to have that much oomph. I'll include the WinISD model based on the measured TS parameters of the woofer I have. I double checked that there is adequate air flow to the ports... I am guessing the issue is somewhere else. Any suggestions?

[mpotoka]

I just noticed this--my winISD model looks quite similar to what I had posted earlier--my nearfield/farfield spliced graph





And the slope downwards on my measured response--would that be my baffle loss?

It makes me think I got some good measurements, maybe I just got lucky, I dunno. I probably have something funny with the phase causing my bass to sound shallow

[mpotoka]

Well, I might have done sort of OK first time around... Here is my on on axis (blue) 30 degree (red) and 60 degree (black)

I'm going to work on that 5-8k range a bit... I am hoping that the baffle is causing some of the garbage above 7k... I have a 1" roundover bit on the way, currently the baffle only has a 1/2" roundover, and only on the 2 sides, not the top. If nothing else it will be an interesting experiment.

[Blazin]

Just a quick glance, you're at like +/- 6 dB on that on axis response.. and it looks like at 7KHz you have some phase deconstruction.. but I'm just a newb haha

CJD?

Why wouldn't you cross the mid/tweeter lower.. like.. 2KHz.. the tweeter apparently has the ability and the mid looks like it's dipping down there as well, perhaps better phase matching and a flatter response..

[cjd]

The tweeter doesn't really have the ability to cross 2k. 3.5kHz is a great target for this, 4k at the outside. Though these drivers can reallly be snuged up so you could go pretty high (or shallower slopes) I think the 7k is baffle or box induced. 1.5" to 2" distance somewhere? Perhaps tweeter to 3 baffle edges?

I will say, that box alignment looks terrible. It'll deliver the boomy nicely tho.

C

[mpotoka]

I will say, that box alignment looks terrible. It'll deliver the boomy nicely tho.

C

Just gotta ask--are you refering to the asthetics or performance???




The B3Ns are 5.25" center to center, the tweeter is in the center. There is 1.25" from the center of the tweeter to the surrounds on the B3Ns. Would that lip be causing issues? If so, can I recess the mids any more, or do I create new issues then?


I did order a bigger roundover bit, The top of the baffle is not rounded over (does that matter as much as the sides?) so I can to see if that helps.

Here are the Mids and Tweeters pre-crossover FR graphs, they both show a null around 7k. Looking at Zaphs graphs of the tweeter, I don't see any of that raggedness from 5k-10k that I have, so it must be baffle induced? The B3Ns look much more what would be expected.



I should try and take some individual driver tests today with crossover in place.

[cjd]

box tuning seems a little peaky for my taste is all. i'd tune the port lower.

C

[mpotoka]

box tuning seems a little peaky for my taste is all. i'd tune the port lower.

C

Heh--I just shortened my ports, not lengthened--but after comparing to my Modulas they now seem quite bassy. I used to have a 10" sealed sub running with my modulas, but I got rid of that just recently, so I was probably too used to bassy music. I actually think I was almost maxed out in lenght on my ports, they are adjustable, and I'll have to play more with length till I pick something.

[mpotoka]

Well it as been awhile. I have gone through four different orders from PE for different crossover parts. I found it much more rewarding to put the xo together, measure it, compare it with the model, change and repeat. The tweeter modeled vs actual was right on. For some reason though my midranges didn't seem to model properly. This caused me a bit of guesswork on what xo parts to use. Anyway I finally settled on something. I'll get my other laptop out that has the measured response and xo schematic. Here are a few pics of the XO assembled, installed (my kids artwork too that they get to put inside the speaker).







And no--the insulation is not solid--it forms a chute so that the air can easily travel down to the ports.

And yes--my lack of bracing is shameful. I'll do that when there isn't 4" of snow on the ground, and I take them apart to paint them (going for piano gloss black)

[mpotoka]

Well here's the XO schematic. I am having issues with my other laptop, the screen is going defunct (guess that is to be exptected when you let kids use them)

Guess I'll invest in a good USB soundcard for doing measurements, since that was the laptop I had been using.

I thought I saved the response file for what I measured in room, but I guess I saved it local, so I will have to pull it off later.

[69Stingray]

Great looking speaker. I must be nice having the crossover finished!

You may want to trim the eggcrate foam around the ports, just to ensure smooth airflow. Once turbulence starts in the port, it doesn't stop.

[mpotoka]

It is very nice to have something to listen to. Tweeter is probably rolled off a little too steeply. But this is my measurement at listening position (one driver, never took a test with both running)

It got me thinking... we do baffle compesation in driver design... what can you do about room measurements and incorporating them into your design? I guess you place the test baffle at placement position, and take measurements, and then design your XO around those measurements?