Senior Project: Chimera, 3 headed beast

[Dennis H]

You know, when you and I were growing up. :rofl:

Nah, when we were growing up a computer was a room full of vacuum tubes. :rofl:

[Dennis H]

Can you do the measurements at home or do you need to put in the time in the lab?

[Blazin]

Can you do the measurements at home or do you need to put in the time in the lab?

I have no lab time requirement, I just have to get it done; however, at this point with all of my time I spend at school for classes and the other part of my time spent at work, I really don't have much time at home anyway..

I had the school pay for the ECM8000 and the MP13, so I'd have to sneak them out to take them home.. the only desktop I have at my house.. may/may not work....

It's an HTPC, the connection to the reciever is there; however, its TOSLINK.. not even sure if I want to mess with that hahah

I'm going to email my professor and see if he has a receiver(he mentioned something about possibly having one in his basement..) and if he can bring it in for me to use.. he is really helpful and I will be using HIS wood workshop to build my cabinets..

[Blazin]

SD315-88s:




RS150S-8:




Estimates of Free Air..

I realized, it was actually quite simple to hook up the drivers to my receiver at home and use 'B Speakers' channel.. wow, I can't believe I didn't think of that sooner..

Generally sound cards cannot read(or get fuzzy) below 20Hz, my SD315-88 has a spec Fs of 25Hz; however, in test it was approx 30.77Hz meaning an Added mass Vas calculation requires an Fs of approx 23Hz, will this be measurable?


What do you use to add mass to the cone? I saw some people using coins, but I need about 1.5 ounces for my Subwoofer, that's approx 8 quarters, I supposed I could do that, but seems like there is a better way..

[Blazin]

This is all getting very exciting, I'm thinking my mic/preamp will not be in until early November; however, that doesn't mean I'm out of work..

I'll be doing added mass today, and I'll begin using Unibox and BoxyCad to design my enclosures, now this may sound dumb; however, I'm a newb..


Chimera is a MTMW system - When designing enclosures, am I correct in designing three sealed enclosures in each cabinet? ie. 1 Sealed enclosure to house tweeter, 1 sealed to house mids, 1 Sealed to house Woofer, under the assumption I am building a sealed box design..

I fear I keep thinking about this process improperly:

Design sealed enclosure to figure out Vb for a desired Qtc, build the enclosure in any manner to get the desired Vb? ie. Subwoofer sealed enclosure portion of cabinet can be a sealed box below MTM, or it can be more 'L' shaped, as long as the resulting Vb meets calculations for Qtc..

If that's correct, then I am on track, and getting giddy with anticipation.

[ThomasW]

Well you're on track. Might start using a logarithmic scale instead of a linear one for your plots.

Added mass is quite tricky even for the pros, since it's difficult to attach the mass in such a way that it's removable but still securely coupled to the cone.

All in all I'm not sure why your doing Fs and Vas measurements? The factory specs are usually adequate to the task. All the designer really needs to do is in-baffle frequency response and impedance plots.

Closed back tweeters don't need separate enclosures. Yes cone mids need to be in a separate enclosure.

Since the woofer has only 7mm of Xmax I'd run it ported.

[Blazin]

All in all I'm not sure why your doing Fs and Vas measurements? The factory specs are usually adequate to the task. All the designer really needs to do is in-baffle frequency response and impedance plots.

Oh, well, uhm I'm just following the SW manual, haha I mean, I suppose I don't have to, I just figured it would be more accurate, especially since my Fs and Qts readings are higher than the factory specs..

Closed back tweeters don't need separate enclosures. Yes cone mids need to be in a separate enclosure.

I can mount that tweeter in the same enclosure as the two mids??

Since the woofer has only 7mm of Xmax I'd run it ported.

Yea, I was looking at doing a ported enclosure for that portion; however, I'm uncertaining of my ability to do the port/box tuning, I have come a long way, and my risks have been large, I am managing them quite well, perhaps I should push myself to do a ported enclosure as well?

kick my ass, tell me to do it ;x( ;x(

[ThomasW]

Oh, well, uhm I'm just following the SW manual, haha I mean, I suppose I don't have to, I just figured it would be more accurate, especially since my Fs and Qts readings are higher than the factory specs..

Minor variations in these aren't that important to the design process. Also some of the parameters change as the driver breaks in...

I can mount that tweeter in the same enclosure as the two mids??

The tweeter has a sealed back, it can share a space with anything. If needed it can be glued to the top of the box in free-air and still operate.

, perhaps I should push myself to do a ported enclosure as well?

kick my ass, tell me to do it ;x( ;x(

Well then consider yourself kicked. If you can run SW, you can model a ported box..

[Blazin]

Minor variations is these aren't that important to the design process. Also some of the parameters change as the driver breaks in...
The tweeter has a sealed back, it can share a space with anything. If needed it can be glued to the top of the box in free-air and still operate.
Well then consider yourself kicked. If you can run SW, you can model a ported box..

Alright then, I'll port that sucker, the Spec sheet was screaming ported box, the Calculation was saying, either.. I figured I'd take the easy road, but I'll take the experience, all of this is telling me that I will be building many speakers after graduation

[Blazin]

Unibox and BoxyCad done

Sealed MTM:

UniBox -

Vb: 13.4L
Qtc: .709
Fb: 84.32Hz
no fill

BoxyCad -

15x20x5-1/2 (WxHxD)
15.35 L
Front: 1" all other 3/4"

Vented W:

Unibox -

Vb: 85L
Fb: 25Hz
Port Dia: 2.5in
Port Length: 5.12in
Port Type: One end flush
Minimal Fill Vb: 86.5L

BoxyCad- (this is where it got complicated)

Woofer portion:

15x18x16(WxHxD)
49.65 L
1 Shelf 3/4" thickness
Front: 1" all others 3/4"

Woofer added volume portion:

15x20x10-1/2(WxHxD)
36.67 L
2 Shelf 3/4" thickness
Front: 0 all others 3/4"(front is actually back of MTM)

Total Woofer Volume: 86.32L

Total Box Volume: 101.67L

All Assuming BoxyCad does not take the driver displacement into effect.. let me know if it does.


Unless there is something wrong here I guess its time to put these measurements together in the CADlab..

[Blazin]

Preliminary Model:

[Blazin]

Finished design:

[Blazin]

So, I've constructed my first prototype, didn't go as well as planned, over this past break I only had access to a semi decent table saw so upon picking up my prototype material I had HD do the majority of the cuts to save time and be a little bit more precise..

Well obviously that was dumb, they didn't do the best job, and the mixture between under and over cuts with the few cuts I made causing problems( table saw's fence was terrible )

I am left with a prototype box that looks a bit.. funky in some areas..

I am going to end up requiring to use one of the sides of the cabinet for the back as the back doesn't fit (the cabinet is not square now)



This all leads me to:

I need major help when building my finished cabinet, just making the cuts. Assembly is no big deal, I have that down, and routing my speaker holes is something I've gotten down to a science(using a fixed base router)

Is there anyone that would be willing to help me out..?

Thanks a lot

[Brian Bunge]

Let us know where you are located. I'm sure someone can help.

[Blazin]

Oops!

North Shore, MA

But I'd be willing to travel to get this done.. I want my finished cabinets to not look like utter crap.

[Blazin]

UPDATE:

Crossovers are getting designed.. need some help/verification if anyone would be willing..














Impedance

[TacoD]

30 uF for a tweeter is a bit high, I would try to use a bigger coil and a smaller cap (I think it's also your impedance problem).

[cjd]

Woah, you've got a couple pretty big issues in the mid-tweeter end, and some general thing to work on as well.

The data looks like the gating is too wide (long) - narrow it down to just before the first reflection on your impulse response measurement (Claudio explains this pretty well - I find using the tweeter to give the cleanest results for setting this).

Getting this squared away will, I think, help with some other things it looks like you're trying to solve mid-tweet integration-wise.

For one - watch out for nominal response ever being *below* the individual driver response. This will result in poor power response since the flat on-axis is only happening due to phase cancellation (negative summing). Your phase integration is poor - try flipping polarity on the tweeter - I'd expect a peak around 3kHz perhaps. Actually, it looks like a tiny bit of negative sum on the bass end too - perhaps keep flip polarity on the mid(s), not the tweeter.

You've got excessively low impedance at 3kHz, you'll be tripping the protection circuits on your amp(s). But you already know this.

Take a look at the Khanspire crossover I did for the RS150's and the RS28A - different tweeter of course, but it may help you.

I think the transfer functions you're using are going to be peaky which will translate to harshness - particularly that tweet network.

C

[Blazin]

Woah, you've got a couple pretty big issues in the mid-tweeter end, and some general thing to work on as well.

The data looks like the gating is too wide (long) - narrow it down to just before the first reflection on your impulse response measurement (Claudio explains this pretty well - I find using the tweeter to give the cleanest results for setting this).

Getting this squared away will, I think, help with some other things it looks like you're trying to solve mid-tweet integration-wise.
C

Since you can't tell thats.. 2.26mS to 5.521mS I should shorten it? I'm almost positive thats the gate I used in the lab... I'm on a diff setup so I don't have the settings.. however, I can redo the measurements now that I know how to get it working I also may do these measurements again in a different location, I did this in the lab at school so I could debug.. but the HVAC system probably added in some unwanted noise and the general concrete super reflective nature of the room probably didnt help.

I will try the measurements again and see what I can do..

Thanks for the help, I'll take a look at the khanspire network to see if I can pick up any tips there..

[cjd]

Set the start gate back to zero.

Try setting the second to like 3.3

That's insanely low gate time. You need to get the measuring environment optimized some to get longer first reflection times. I don't think it's usable for a 3-way at all unless you're crossing up around 1kHz woofers to mid (so, dome mid or something).

[Blazin]

Quick question..

This has been confusing me.. with the limitations of SW

I have 2 midwoofers.. how do I measure them?

Seperately or Together.. I've done together now because I can't model a mid range crossover to begin with.. and I certainly cannot use 2 drivers in the network model..

Together they are wired in parallel..

Also measuring, nearfield I was placing the mic about 6" from the midpoint of the two drivers.. probably not the best method, but I'm very limited and can't seem to find a better solution.. maybe someone can provide some solutions

Thanks again

[Blazin]

Did some searching on the old SW forum.. it appears the recommendation is wire them both up, as that is how the box was modeled.. however perform nearfield measurements with each driver individually.. then combine their responses (addition) to obtain the actual response.. which should be +6dB or so due to the nature of 2-8ohms in parallel

Update:

I will be making nearfield, and gated measurements again this friday in a new location with less reflective nature.. only problem is the room is quite small or the area probably 2.5-3 ft to each wall if I place the speaker in the middle of the room.. if I do this.. I would have to move the microphone to the doorway of the room and I feel that due to the nature of the 6x6 area the doorway will be acting as a very large PORT.. lol

Any suggestions on mic placement? Opposite side of the room so the door is behind the speaker?

[Blazin]

So why dont I have any resolution when I combine a full system crossover?

or apparently when I try to sum 3 individual crossover network responses .. this is the gating?

[Blazin]

Sooo






Impedance still diping down to 2 Ohms.. I need some help here..


One thing I'm confused about..


Gate + Nearfield = Gate.Splice..

Nearfield must be scaled down to match gated resposne; however, if you're scaling each response a different amount to meet their respective gates.. won't the Amplitudes be off?

Are you supposed to normalize them back to 0?

I got measurements down now, my new location is perfect! Just a bit confused about that procedure.. amplitude issues would obviously be causing issues with L-Pads, and various other crossover components.


The high is down ~ 6dB.. ideal BSC.. sure.. whatever.. I can't seem to match the levels.. this leads back to my prior question

[Dave Bullet]

Hi Blazin,

First - if you are modeling the combined network - use one source not three. In my simulations - 3 different sources for a 3 way crossover gets the woofer/mid out of phase. This could lead you to think the polarity might need to be inverted (and if your target is LR4 for all Fc then drivers will have the same polarity - since 360 degrees phase wrap).

For a 3 way design, you typically need to take your gated measurements further than the standard 39" (1 metre) used for a 2 way design. This is so all drivers resemble more of a point source. You probably would benefit from a larger room. If that isn't an option - you can nearfield all drivers (maintaining the same amp output level and mic pre-amp level) then HBT transform to get phase data. Import into SW then enter acoustic offsets for each driver in the combined XO.

Maybe if you can post your raw driver FRD and ZMA files here - I can validate your use of SW (or help you troubleshoot SW problems with the measurements).

Cheers,
David.

[Blazin]

Hi Blazin,

First - if you are modeling the combined network - use one source not three. In my simulations - 3 different sources for a 3 way crossover gets the woofer/mid out of phase. This could lead you to think the polarity might need to be inverted (and if your target is LR4 for all Fc then drivers will have the same polarity - since 360 degrees phase wrap).

One source? all drivers wired in parallel?

For a 3 way design, you typically need to take your gated measurements further than the standard 39" (1 metre) used for a 2 way design. This is so all drivers resemble more of a point source. You probably would benefit from a larger room. If that isn't an option - you can nearfield all drivers (maintaining the same amp output level and mic pre-amp level) then HBT transform to get phase data. Import into SW then enter acoustic offsets for each driver in the combined XO.

Maybe if you can post your raw driver FRD and ZMA files here - I can validate your use of SW (or help you troubleshoot SW problems with the measurements).

Cheers,
David.

Alright, would you like me to upload the Nearfield and Gated measurements?

[Dave Bullet]

Yes to both your questions.

[Blazin]

I'm confused, all three in parallel... after using individual driver responses to create crossovers? then with total speaker response put into the 3 crossovers and model the total system response?


It seems there are multiple different ways to do this process, and its all confusing me.

Can you you lay it out step by step?

I am running out of time to be making these measurements 10 different times to satisfy 10 different modeling processes.

Thanks a lot.

[Blazin]

Alright lab later this week..

I'm gonna start my measurements over.. making copies of everything.

Task list:

1.) Nearfield of Tweeter, Mids, Woofer, Port - Measure the individual mids (with both wired) then sum together to achieve accurate response.

2.) Farfield Gated of Tweeter, Mids, Woofer - I will wire all 3 drivers in parallel (Mids in parallel act as 1 source) and make a total measurement. Measure each driver seperately for individual driver response. All measurements made at a little over 1M on axis with tweeter.

3.) Post all inital measurements for review.

4.) Splice Nearfield + Farfield measurements - Claudio's technique using Baffle width for Nearfield useful range up to 287Hz (4311/15in width of baffle)

[cjd]

Here's what is important for you to get a good crossover:

1. Wire the drivers up as you expect they'll need to be wired for sensitivity purposes. I would be very surprised if wiring the mids in parallel is the way to go. If you're not sure, make sure you save one set of measurements with the mids in parallel, one set with them in series.

2. Room setup. The bigger the room, the better. Try to get the speaker up on a stand so tweeter to ceiling/floor distance is maximised. Keep it as far away from walls as you can. This is very important, as it is what determines how long your gating can be.

3. Gating: ALWAYS leave your start gate at zero, unless you're actually measuring in an anechoic chamber and don't have reflections to gate out. This doesn't gain any relevant data, but it DOES increase the total gate time which is what drives RESOLUTION. Hopefully you can get at least an 8ms gate. But any "bouncy" response means your gate is too long.

4. Farfield measurements - try these at 2 meters if you can. Do not move the speaker or the mic for ANY of these measurements. 100% exactly the same. (line the mic up with the tweeter normally). If you have to move speaker or mic, you're starting these all over.

Make sure you get impedance measurements of course.

That's it. Don't mess with nearfield response or with splicing. You certainly CAN do these things, but for pulling together a crossover you only need the farfield measurements with enough resolution to be relevant. Under 5ms gate and you won't have the data to do a 3-way with these drivers, period. I'd be really wary if you can't get above 8ms. It'll be tricky.

The comment about "one source" was a reference to your crossover network setup in Speaker Workshop only. There are three items named "source" - get rid of two, and use only one. (again, go peek at the networks I post as I use SW - it'll show you how this works). Speaker workshop does not maintain polarity on the source, so you don't actually know which way to wire the speakers up to the real crossover! Only way around this is to use only one "source" element in the diagram.

C

[Blazin]

Here's what is important for you to get a good crossover:

1. Wire the drivers up as you expect they'll need to be wired for sensitivity purposes. I would be very surprised if wiring the mids in parallel is the way to go. If you're not sure, make sure you save one set of measurements with the mids in parallel, one set with them in series.

2. Room setup. The bigger the room, the better. Try to get the speaker up on a stand so tweeter to ceiling/floor distance is maximised. Keep it as far away from walls as you can. This is very important, as it is what determines how long your gating can be.

3. Gating: ALWAYS leave your start gate at zero, unless you're actually measuring in an anechoic chamber and don't have reflections to gate out. This doesn't gain any relevant data, but it DOES increase the total gate time which is what drives RESOLUTION. Hopefully you can get at least an 8ms gate. But any "bouncy" response means your gate is too long.

4. Farfield measurements - try these at 2 meters if you can. Do not move the speaker or the mic for ANY of these measurements. 100% exactly the same. (line the mic up with the tweeter normally). If you have to move speaker or mic, you're starting these all over.

Make sure you get impedance measurements of course.

That's it. Don't mess with nearfield response or with splicing. You certainly CAN do these things, but for pulling together a crossover you only need the farfield measurements with enough resolution to be relevant. Under 5ms gate and you won't have the data to do a 3-way with these drivers, period. I'd be really wary if you can't get above 8ms. It'll be tricky.

The comment about "one source" was a reference to your crossover network setup in Speaker Workshop only. There are three items named "source" - get rid of two, and use only one. (again, go peek at the networks I post as I use SW - it'll show you how this works). Speaker workshop does not maintain polarity on the source, so you don't actually know which way to wire the speakers up to the real crossover! Only way around this is to use only one "source" element in the diagram.

C

Ahh mids in parallel was chosen for the what if purpose.. what if one dies. lol

I'll do the measurements again in series/parallel to compare.

The room i'm using is large but divided in two smaller portions wtih a big test bench int he middle.. I mayyy be able to get 2 meters but at the sacrafice of nearly placing the speaker against the back wall -which is not good-. I had put my speaker up on a stool that was probably about 24-30" in height so thats good.

My gate will definitely be large enough in that new room setup i forget exactly what my gate length was but I'm nearly positive it didn't even start til 8 ms.

Ahh farfield gated is only the measurement I need, I figured as much which is why my crossover (one source object) method was producing such crappy resolution at low frequency.

I'm definitely going to need to splice a nearfield with my woofer, the port response is ~30Hz I doubt I can get any data for that frequency using farfield.


Thanks a lot cjd you've been very helpful.

[cjd]

Splicing in nearfield will give you an overall picture, but it doesn't help you design the crossover. Port response is irrelevant for a crossover at 250Hz or 300hz or wherever you put it.

If one of your mids goes dead, you have a problem whether they're in series or parallel. Doesn't matter. Remember, a crossover expects a very specific impedance - one missing driver makes that impedance nonexistent. If you measure both ways, remember to also get impedance for both!

Back wall, noo. That definitely won't help. Balance distances out, but get the mic as far back as you can.

Your poor bass response measurements were due to having a VERY short gate time.

C

[Blazin]

Alright, so looks like I'm getting pretty good resolution..

0-7.5ms on my gating.. can't get anymore.. im at around 4.5-5 feet. I'm getting wall reflections.. but not exactly wall.. more like.. benches on the sides of the room..

So I guess I understand this as.. Crossover points are all I care about.. I can't control whats going on under lets say 100Hz on my sub.. I just need to make sure it crosses over at my selected freq (I think 210 is too low.. 250 may be in order).. and on the high end.. i'll probably select 2500, well within range on both mid-tweeter

CJD - Far field gated measurement of Mids in series.. about.. -10dB.. parallel -5dB

[cjd]

it doesn't matter, the question is which works better with the woofer SPL (and tweeter, though that's usually not the bottleneck). In parallel I suspect would require padding down, which is to be avoided where possible (hence, wiring in series to knock out 5dB). The Khans run the RS150's in series and an unpadded RS28A tweet... and themids might still be 1/2 to 1dB hot.

250 is good, I'd not recommend 2500. 2100 is listenable but my preference for the RS150 is 1800. (I don't prefer it enough to have gone and updated the crossovers in my original RS150's, which still have the 2100Hz 4th order crossovers)

7.5ms should work.

Getting measurements is THE pain. Once you master this, you're golden. It gets easier for me every time.

C

[Blazin]

it doesn't matter, the question is which works better with the woofer SPL (and tweeter, though that's usually not the bottleneck). In parallel I suspect would require padding down, which is to be avoided where possible (hence, wiring in series to knock out 5dB). The Khans run the RS150's in series and an unpadded RS28A tweet... and themids might still be 1/2 to 1dB hot.

250 is good, I'd not recommend 2500. 2100 is listenable but my preference for the RS150 is 1800. (I don't prefer it enough to have gone and updated the crossovers in my original RS150's, which still have the 2100Hz 4th order crossovers)

7.5ms should work.

Getting measurements is THE pain. Once you master this, you're golden. It gets easier for me every time.

C

You're right, I won't need an L-pad in series.. ok I'll shoot for 2100 and 250.. the Tweeter says it can do 1800.. through measurement appears ok..

UGH i cannot get the impedance right for the tweeter.. I can't not get 1.25 Ohms..

Is this why nobody uses the DC28? lol

[Blazin]

FRDs attached below..

new annoyance some 10KHz peak.. :M


Anyone wanna try and help me out.. I don't think there is a way to cross the DC28 and RS150 well.

[Dennis H]

You need to include impedance files for anyone to help you. Just glancing at the tweeter FRD, that's one ugly response, but I suspect that 10K peaking may be baffle effects. Are the drivers flush mounted? 10K would be about right for reflections off the edge of the tweeter or maybe the mids.

[Blazin]

You need to include impedance files for anyone to help you. Just glancing at the tweeter FRD, that's one ugly response, but I suspect that 10K peaking may be baffle effects. Is the tweeter flush mounted? 10K would be about right for reflections off the edge of the tweeter.

Yea it is.. but look 10K is on all responses.


I can't upload ZMAs.. too large for the forum.. 100k max and my files are like 300k

[Dennis H]

Even zipped? Why are they so big?

[Blazin]

Even zipped? Why are they so big?

haha i dunno a LOT more data points probably. and yes thats zipped

[Blazin]

Weird note..

The Nearfield response of my DC28FS looks VERY much like the RS28A response..

aka it starts to roll off at 1KHz..


Odd.. I think my farfield measurements are off.. UGH

Alright I'm gonna need to steal a PS and Computer from my School and make these effing measurements in my garage.. that is clearly large enough.. I should be able to get a gate of 12 mS and I'll be able to limit floor reflections with towels and such


Alright I'll update this friday I suppose with new measurements.

[Dennis H]

Taco gave you good advice way back. Make the tweeter cap much smaller to keep the impedance up. Using your FRD and PE's ZMA, here's a quicky pass in LspCAD using 10uF and optimizing the coil for a 1.8K LR4 response. I actually got a slightly better SPL fit with 12uF but the impedance was dropping below 3 ohms. 10uF keeps it above 4 and the fit isn't much worse. That resistor is just the nominal ESR of the coil.

[cjd]

I had an issue at 12kHz with a faulty solder joint in my cables way back when. You can see it in some of the original MTM measurements I put out there. This stuff *really* is a headache sometimes.

On the other hand, if you're confident the data is good it won't prevent you from working with it, anomalies and all.

C

[Blazin]

I had an issue at 12kHz with a faulty solder joint in my cables way back when. You can see it in some of the original MTM measurements I put out there. This stuff *really* is a headache sometimes.

On the other hand, if you're confident the data is good it won't prevent you from working with it, anomalies and all.

C

Hmm I was using a clip today to make it easy to reach.. that was probably it..

thanks for that hint..

and Dennis.. I see why I was failing.. I was trying to crossover too high!




OK!











Lowest Dip is 2.6 Ohms.. My Receiver does 3.. ok?

I see there is some of the total response falling underneath the individual response.. I recall someone saying this is bad..

How could I fix this.. 4th order filter on the Lowpass @ 1.8K on the Mid?

I reversed the polarity on the tweeter, because there was a huge dip at 1.8K

I think I'm good to go!

Am i?

[Dennis H]

You have to start looking at the phase to get a good blend. In a perfect world they would be in phase at all frequencies or at least through the crossover region. You might need to adjust the XO to compensate for acoustic center offsets -- the AC of a cone driver is behind the baffle. And that's assuming you measured phase right in the first place which is a BIG assumption. It's a delicate balancing act -- SPL, phase, impedance, how they all sum.....

Are we having fun yet?