Crossover design

[4thSeason]

Just working on my MTM project anyone want to comment on this crossover. Its the first of my own design so be gentle. not sure but the peaks on the midbass drivers between 5k and 10k scare me a bit. I wonder is if that will come through as noise or at least make those areas sound soft or muddy.

Bri

[ThomasW]

the peaks on the midbass drivers between 5k and 10k scare me a bit.

Yep, that's why Jon chooses much steeper slopes for his designs using the RS-180

Modula MTM



NatP

[Jay_WJ]

If you want a simpler XO for an MTM with the RS180s and the Seas TBFCG than in existing designs, benchmark Zaph's L18/TBFCG network topology for the woofer network (just 2 inductors and one capacitor) rather than an electrical 4th order network. It uses a bottomless (i.e. no resistor) series notch filter instead of a parallel notch used in the Modual MTM or the Nat P. The filter performance will not be as good as that in the Modular MTM, but should be sufficiently effective in making the breakup inaudible.

Here's my application to my MT design with the same drivers as yours:




As you can see, breakup nodes are all well controlled under 45dB, which should not be audible. Also, look at the good phase alignment between drivers. The phase responses I used are not actual but extracted through Hilbert transform. But I did simulated SEveral existing designs using this method and all of predicted phase alignments were very close to measured ones---in fact there're some tips I found for an accurate simulation, which are a bit too lengthy to describe here. For example, Zaph's L18/TBFCG design is "in phase" at 2" above tweeter axis and in my simulation the in-phase listening axis occured at about 1.8" woofer offset relative to tweeter, which corresponds to 2" above tweeter axis (about 1.0-1.1" for woofer's recessed acoustic center plus .7" longer distance from mic to woofer than that from mic to tweeter).

I used a driver offset of 1.3 inch in the above simulation. This means that the in-phase listening axis will occur at about 1" below tweeter axis, which I intended. It will not be difficult to modify my design for MTM application. In this case, the breakups will be all reduced to under 50 dBs on axis. A little bit horizontal off axis listening will further reduce the dBs of breakups. I personally think that the 4th order electrical network along with a parallel notch filter in the Modular MTM is an overkill.

[Jay_WJ]

How about start from here?







Frequency response, phase alignment, and impedance all look good.

Ignore responses below 200Hz since I used my small sealed box design for bass modeling.

[4thSeason]

Thanks for the great info guys!
Kudos and many thanks Jay I will play around with your design today looks like it is good to go as is if I want to go this route. People around are so helpful in their guidance. Like I said before this is going to be my first I will be posting pics in a few days the cabinets are cut just haven't pulled the trigger on the order yet.

B

[Jay_WJ]

In case you want to use my design, the value (0.013 mH) of L3 is very small, so it won't be available. You have to unwind coil from a smallest available one. Madisound may be able to do it for you.

BTW, in your XO sim, try different values of L3 (e.g., .012 to .015) for a better result of breakup attenuation you like.

[4thSeason]

Thx

Thanks for the info

I tried a sim in Speakerworks and came up with a bit different curve. I didn't add resistance values but it should have looked close. my midbass response was slightly higher. I will post a pic in abit. Perhaps because it doesn't account for tweeter offset.

[cjd]

What data files are you modeling with? Live measurements from your box? Sim data with summed baffle step and diffraction and minimum phase? Live measurements from somewhere else? Each of these will yield different results and require different approaches...

Impedance of inductors can change a circuit some - it's always good to work with impedance data.

C

[Jay_WJ]

Do you use actual phase responses for midwoofer and tweeter mounted on the actual baffle that are measured from a single mic position? Or did you extract minimum phase using Hilbert transform? If the former is the case, you don't need to set the tweeter offset to a non-zero value.

Also, the twin peaks at 6.5 and 8.5 kHz may be higher than in my sim, but as long as they are under 55 dBs, you'll be alright.

The DCR values I used in the sim for L0, L3, and L7 are 0.2, 0.04, and 0.21, respectively.

[4thSeason]

reply

I took the measured files .frd and .zma off of Romans site. Some he traced and some measured so I cannot say of which these are. I do not have the drivers or setup to measure yet(but I will).

Thanks again Jay for the info. I still haven't pulled the trigger yet I kinda had my heart set on building the RS 3ways when these were done but others have said that once the Modulas are built I will be happy enough not to want them as my surrounds..=O)
I guess we will see.....I was going to start my building projects with the Zaph bargain MTM because they were going to be surrounds but wanted a little bit higher quality that the RS would afford and assumed they would sound similar to the rs 3 ways because they at a minimum used some of the same drivers(I know thats a guess as to voicing).

[Jay_WJ]

I see. The RS180 frd and zma files at Roman's site are from PE measurements, and the TBFCG files, I guess, were traced from the manufacturer's data sheet.

I used Zaph's measurement data for the frd files of both drivers to achieve consistency. And I'm confident that Zaph's infinite baffle measurements are pretty accurate. I have traced many of Zaph's FR measurements (e.g., Usher 8945P, Peerless HDS tweeter, Seas TDFC, Dayton RS28A, etc). I also have zma files for these drivers I traced from other sources. I can share these files if you want. Just let me know.

BTW, I remodeld the above RS180 MTM design using correct box and baffle (PE 1cu ft MTM vented) measures. Accordingly, component values have been changed. Check out post #4 again by refreshing your browser. One option in this design may be to replace R9 (2 Ohms) with a 2.4 or 2.7 Ohm resistor for a bit less bright sound.

IMO, when PRoperly used, the modeling method using FRDC tools can be sufficiently accurate in predicting the end result. For example, the way of tapering either end (especially top end) of frequency response can affect the result of minimum phase extraction in a critical range near the intended XO point. For just one-time DIY speaker builder (like me!), having a measurement setup is costly in terms of both time and money. Instead, I chose to refine my modeling skill by trying to accurately reproduce many exsiting designs (not only FR but also phase alignment (or reverse null and in-phase listening axis if phase measurement is not available for the reproduced design)) via a CONsistent method. I learned a lot by doing this. Somebody may say that I still need a measurement setup for an accurate result. I agree. But now I'm convinced that correctly implemented simulation can be used to achieve a good enough result.

-Jay

[4thSeason]

above

Thanks Jay for all the help withthe sims. I surmise that mine may be falling short somewhere I am just so new to this. I have been doing car ausio stuff for 20 years and been into theatre stuff for a few but building home equipment with passive crossovers is a whole new game. I will show my results so you can compare them to yours and will probably orer tonight or tomorrow as PE shipping deal expires soon

Bri

[cjd]

Sims *can* get good results, but they can fail too...

C

[Jay_WJ]

Sims *can* get good results, but they can fail too...

C

Sure. Speakers built via measurements *can* also fail if the work is sloppy. But I agree that sims are more *likely* to fail. That's why I emphasized enough above how important a carefully and meticulously designed sim is. For example, we can think of many possible cases of ill prediction to avoid worst-case scenarios and choose a safe design and component values.

[4thSeason]

I would certainly agree that sims can fail or fall short of expectations or stink or whatever but the truth is that if simulating in Speaker workshop didn't work then no one would use it- I am not saying it's magic but Jay was trying to help me with a crossover design issue because I am trying to build a MTM using the same drivers as the Modula. I havent really had anyone else chime in with helpful advice other than use what's fed to you or your on your own> I like to build, I like to experiment and I am here for sound advice regarding proven designs. Truthfully I have respected and enjoyed learning about everyones projects on this board and the only person who has offered constuctive advice is Jay. It sort of feels like if I don't clone the projects you guys build then I am not going to get advice or not welcome here. The first time I asked for help withthis my first project one person I believe Thomas just said "build Jons crossover"...Thats terrific advice and I am sure an amazing speaker but I wanted somthing a bit more resonable in the cost area, perhaps not quite as amazing sonically but somthing solid and to build my experience fabricating and designing with

[cjd]

Wow, you're reading way into the comments...

Roman has outlined quite well the process to go through when using data you get from others. Using PE's data (or other "infinite baffle" measurements) makes it easy, as long as you keep your eye on measurement distance and normalize for that. Sim baffle, sum, extract minimum phase, enter offset on mid-woofers, and run.

Thomas pointed out why Jon uses steep filters... I didn't see anything else.

Jay offered a crossover topology... how's that different from Thomas pointing to other designs? :P Yeah, some other good info though.

The TMWW 3-ways will have a very different character from the Modula or Natalie P I think, based on some comments and observations I've heard.

I was tossing out ideas between things at work. One to get some (needed) background so we could answer better, the other simply a caveat. I've done a lot of work with simulated data, and I've failed quite a bit till I understood it all. Never had luck with 3-ways and sim data yet. Had some very good results with 2-ways, and some only OK results.

If you want more specific feedback, make sure your response graph shows individual driver phase, and post the network.

FWIW, some people like having breakup only a little down, it pushes vocals (particularly female) forward and gives some semblance of being "detailed"

C

[4thSeason]

I wasn't trying to be overly whiny- but like I had said : in a prior post I only got one response and it said the build Jon's crossover comment.

I am so new to this unlike mobile audio and they are totally different worlds.. :P

I would agree that a little knowledge can be dangerous..


I will take your advice for posting graphs I actually did in my first post here I will post more data next time

Blessings

B

[ThomasW]

I wasn't trying to be overly whiny- but like I had said : in a prior post I only got one response and it said the build Jon's crossover comment.

No where I did say build Jon's designs. I simply posted his design to show you what happens in the 5kHz-10kHz range with the high order topologies.

FYI, Jon's a twice published member of the AES, so he knows this stuff inside and out. If you haven't done so you might want to download and read his speaker design article. It's a particularily good read for a n00b ....

http://www.htguide.com/forum/showthread.php4?t=22862

[Rick Craig]

No where I did say build Jon's designs. I simply posted his design to show you what happens in the 5kHz-10kHz range with the high order topologies.

FYI, Jon's a twice published member of the AES, so he knows this stuff inside and out. If you haven't done so you might want to download and read his speaker design article. It's a particularily good read for a n00b ....
http://www.htguide.com/forum/showthread.php4?t=22862

Sean Olive from Harman did some blind tests on breakup audibility that were published in the AES. If I remember correctly they found that once the breakup was 20db down from the reference level it wasn't audible.

[4thSeason]

I was indeed incorrect Thomas you told me to build the Natalie P crossover in my other post and I said another post not this one. I'm not trying to instigate an argument, only clarify

The stock NatP crossover is what you should build.

My appologies for the mistake indeed you did not say "build Jons design"

The point is I have done alot of reading and I will continue to do so. I will post my results as my build progresses and I hope that others will find use from them. And everyone is a Noob at somthing somtime.

Blessings

Bri

[rj45]

You will find several different crossovers for the drivers in an MTM and TMM at this site. Roman does good work.
I've found it very useful to use .frd and .zma files in SpeakerWorkshop, and build these crossovers to examine breakup node attenuation and phase response.

http://www.rjbaudio.com/RS180MTM/rs180-rs28-mtm.html

[4thSeason]

Thanks for the link but like I stated above I got my files from Romans site and his MTM info uses the Rs28a not the H1212 tweet. I have othewise speant plenty of time over there especially the tutorial for SW

Bri