Back and forth on DIY speakers

I am trying to decide what to do. It doesn't look like i'll have enough money to buy a reasonable set of mid-hi fi speakers. I was very seriously considering the martin logain aerius i's.

The reason I had intended to go the retail route was I think they have a lot more time and money to work out the various enclosure/cone resonances, and crossover designs that I could never accomplish.

But, since I dont think i'll be able to swing the 2 grand anytime soon, i've been thinking about putting a diy set together for about $5-600. I am looking to do a 2-way floorstanding setup with the -3dB point around 40-50hz. I may or may not add a sub at a later date, I do listen to a lot of trance and techno with substantial low frequency material.

I've read through the LDSG a couple times and downloaded winISD to do some preliminary research. I love the way the focal polykevlar drivers look and they are reasonably priced, but when I modeled the 7K4211D in winISD the lowest -3dB point I could reasonably achieve was around 55-60hz. This is getting a little too high. The dual 8 ohm coils are somewhat worrying me because I do not have much experience building crossovers with unequal load impedances. Has anybody been able to get good low frequency response out of the focal's ? The focal 8K5412 might be a possibility, but I really didn't want to move to an 8 inch driver. I was planning on crossing over at about 1.8-2khz. Is that a reasonable frequency for an 8" driver? Focal's frequency response graph for this driver shows a lot of undulation from about 1.2khz and up.

I have gotten good winISD responses from the Scan Speak 18W/8546 7" Kevlar coned driver. More expensive than the 7" focal's, but cheaper than the 8" focal. I think it showed a lower 3dB frequency of 45 hz as well. Has anybody used these? Results?

Also, the peerless 850122 shows a -3dB frequency of 33 hz. But somehow I doubt that is accurate. Anybody have experience with this driver ?

I fully intend to model the final design in spice so I can get reasonable high frequency rolloff information and observe the effects of my crossovers before I spend any money. The problem with this is that it takes me about 4 hours to calculate all of the required values to use in spice and I'd rather not go through it more than once or twice.

Are there any drivers designed for sealed applications you might have used and would recommend for my situation?

The dynaudio 17 W 75 looks like it might work well. any experience here? What is the difference between the 17W75's, XL's, and EXT's ? Will i even be able to find them anymore ?

I've seen a lot of commercial and even diy projects use two 6.5-7" drivers and a tweeter. Doing this in winISD I was able to either get more efficiency, or a lower frequency response by 7-10hz at the expense of efficiency and non traditional transfer function. Is this about right ? It would solve the earlier problems with the focal drivers. When you do multiple driver simulations in winISD I assume it calculates the response based on the two drivers sharing the one common enclosure and port ?

Are the phase interactions in the crossover much to worry about? I can design a 1st order xover where everything is exactly in phase, but my acoustics and audio professors mentioned that this is impossible with higher order networks. I have seen plenty of people advertise "phase-perfect" and whatnot on higher order crossovers and always wondered if there was a way to compensate, or if they were just marketing. If there is could someone point me to a book or paper that covers it ?

Last question: what is group delay. winISD gives plots for it and I have heard the term used and was never paying enough attention to get its meaning. I am guessing it would be better to have a system with a lower and smoother group delay?

thanks a lot

jt
 

Conrad

Member
2001-12-20 1:03 am
Hi jt,

I have just recently finished a project of designing and building a set of 5 bookshelf speakers for a HT system using the Peerless 850122 driver.

You are correct in been not that beleiving about the manufacturers specs for the 850122 as my measurements and resultant T/S of them was a significant deviation from the published specs.

As a driver it is a easy and good performance 6.5" driver with nice mid range and the ability for some bass extension. My boxes had the size limit of been bookshelf so I used 12l boxes with a 2.5" port and managed to tune it to a f3 of approx 60Hz. With a large box and a BB4 alignment you should be able to get as low as 40Hz from the models.

I found the driver fairly easy to design a crossover for - so they are great performers at a reasonable price.

Have a look at the DIY plans available on www.audioreview.com, this design uses these drivers - quite a good speaker design as well.

Cheers
Conrad
 
wow, i posted that long ago, but thanks for the info.

I would like to ask a separate question about audiofreaks comments.

My audio professor made a big deal about the phase response as well as the magnitude response. By "high order active" what do you mean? like the 24dB/octave linkwitz riley? or something else?

I don't feel that this particular circuit would allow you to get even a good phase response unless you had some really expensive drivers with a very large passband and you crossed to the mid fairly high. Are the xover's you speak of just a regular xover with one of the variable phase circuits similar to those posted in a subwoofer thread not too long ago?

I feel like writing a program that will analyze/optimize a database of driver T/S parameters, enclosure sizes, and crossover slopes/Q's for the best magnitude/phase response based on the model i have. I would probably have to learn a language other than pascal or the TI-85 scripting language :p

At this point, I have little else to do though!

jt
 
jteef said:
wow, i posted that long ago, but thanks for the info.

I would like to ask a separate question about audiofreaks comments.

My audio professor made a big deal about the phase response as well as the magnitude response. By "high order active" what do you mean? like the 24dB/octave linkwitz riley? or something else?

By high order i mean 12dB and higher .... by active i mean an electronic x/over that goes between the preamp and multiple amps in a biamp/triamp etc setup.

I don't feel that this particular circuit would allow you to get even a good phase response unless you had some really expensive drivers with a very large passband and you crossed to the mid fairly high. Are the xover's you speak of just a regular xover with one of the variable phase circuits similar to those posted in a subwoofer thread not too long ago?

if you are using one of the electronic phase coherent x/overs with multiple amps then the phase is about as perfect as you can possibly get it if the drivers are in a suitable enclosure. The enclosure has just as much influence on the final result as does the x/over the amp and the drivers used but all things considered, you can get a very good result if you spend the time in r&d before jumping in and building it.

I feel like writing a program that will analyze/optimize a database of driver T/S parameters, enclosure sizes, and crossover slopes/Q's for the best magnitude/phase response based on the model i have. I would probably have to learn a language other than pascal or the TI-85 scripting language :p

trouble is if you optimize only for magnitude/phase response it will most likely be lacking severely in other areas.

At this point, I have little else to do though!

jt
 
start slow look at building a good 6" 2 way.
Scan Speak / Vifa Speak is about the best choice now given that Dynaudio does not retail drives anymore. Focal is another choice but their tweeter have a foam surrround and their Kevlar drivers do exibhit some nasty breakup. Audax Aerogel drivers are good for the midrange but have lousy Xmax and hence limited LF. Then there is Cabbasse, Accuton, etc...but I like to stick to basics for the begining.

Once you build this one can quite easily add a second woofer and imporve sens but using the second woofers midbass output to compensate for baffle difraction.

Then one can build subs and expand the system to a HT system by adding center adn rear speakers.

Remember start slow and simple. Complex stuff is more likely to discourgage you. Second rule is to build - fearlessly all you loose is abit of wood and a few caps and inductors. I use to attend a lot of forums and seminars etc. and fgound thaat everyone had a theory about how to do this or that but few had any practical advice.

Most of all have fun.
 
While it is true that the 12 and 24dB / octave filters introduce no phase shift of their own, the drivers and their enclosures do.

A sealed midrange is always going to have about 90 degrees of phase at its Fc. so in order to get it to add correctly to the woofer, you're going to need to make the Fc as low as possible, and cross over as high as your woofer would allow. Or use an odd order filter and try to get near 90 degrees of separation. I think it would be pretty difficult to get a good magnitude response with the latter, however. I think with a sufficiently high order filter, the affected bandwidth would be so small as to go un-noticed with most music, but I would hate to settle for 2nd place.

Clearly, a program like i suggested would not be the be-all-end-all design tool. It would merely be a tool.

jt
 
Hi Jt

this is kindof compeletely unrelated to any of your questions, but you keep mentioning your "audio professor"? I was wondering what level of education are you currently undergoing? Is it a university program? And what university? Thanks very much seeing as I have to make these important decisions in my life within the next couple of months, and Audio is definetly one of my interests...

Thanks very much,
Will.
 
BigggWilly,

I graduated last May from Georgia Tech with a degree in Electrical Engineering. They dropped the Audio Engineering "Area of Specialization" when they switched from quarters to semesters in 1998(or 1999, i can't remember). Their new technical interest group is called "Electronic Design and Applications" and information can be found here:
http://www.ece.gatech.edu/research/brochure/electronic.html

This is the basic path I followed. The DSP TIG also does a lot of work with audio, but they do a bit too much programming for my liking.

Dr. Marshall Leach still teaches the Audio Engineering course and there are a couple classes taught by the physics department. The only one I took was "Acoustics" taught by Dr. Eugene Patronis. The mechanical engineering school might also have some classes relating to audio, but I didn't pursue them.

I am guessing that most of the bigger engineering schools have something to offer for those interested in audio.

jt
 
Comment on jt's first posted subject

Hi,

There are several ways to work to your final goal.
You can try and calculate your way around the 830452 with slave 830468 from peerless. In combination with a subwoofer amp which gives you the oppertunity to enchance your bass responce
in the low area plus it enables you to use phase correction which most sub-amps have.
I think building a two-way with resonable bass can be done, but building a two way with -3 db point around 30 Hz is challaging.
Most music do not produce sounds that low. Only digitaly composed music or movie soundtracks can have those low frequenties.

Good luck with your porject

Mazz.
 
Have a look at this unit:

http://www.mass.com.au/masssales/pdf/P21WO-20-08.pdf

I found this in my travels recently. The mid response is very flat, well past 2k which would make it ideal in a 2-way system.

Sensitivity 91dB means minimal tweater attenuation required. Fo of 28Hz

Lacks the nasty mid-band peak of many of the "high-end" units I have looked at.

On paper it looks tempting. Has anyone any experience with this particular driver?
 

ds21

Member
2002-02-11 6:51 am
WI
In fact I just dismantled my two year old P21WO 2-ways last night, had to scavenge some parts for my newest 2-ways. They sound pretty good, first I had them in a 2Ft box crossed at 1800, lots of low-end but muddy, then switched to about a 1Ft box much better. I’ll find some use for them in the future, but for now my new speakers use a Vifa M13SG 5” midwoofer and a D27SG 1” tweeter both are shielded for use in my studio great imaging and an F3 of 55hz.

P.S. I thought that nasty hump in the higher-end drivers were there to try to compensate for baffle-step.
 

SteveG

Account Disabled
2002-01-07 7:20 pm
Newton Falls, Ohio
I have a design for you if you're interested. I have built a set of floorstanding MTM's using two Vifa p17's and a Morel MDT-33, and I'm very happy with them. It might not be high tech enough for you with the drivers you have been describing, but let me know if you're interested. If I remember right, they are getting down to about 42Hz. The midrange is very clean, if a little bit forward. They will really deliver bass, but you usually need a little EQ to get the bass up to where it sounds really full. They image better than many mid-fi designs I have auditioned in stores. The nice thing about these drivers is that they both are pretty flat, and have an easy rolloff to work with, so you can modify the crossover fairly easily. I had the crossovers designed by Madisound. I'm still thinking of optimizing them a little and lowering the crossover point- it's at about 2200 hz right now. I think the midrange would improve with a lower crossover point, but I'm waiting to build my amp first. Just thought I would volunteer the design, if you're interested.
 
Question for DS21:

Hi,

I'm also working on a little monitor using the Vifa M13sg midwoofer and D27sg tweeter. So far i used the 'surround 1' crossoverfilter and enclosure design available from www.audiocomponents.nl
and my own design serial crossover.
I still think i haven't got the best out of these two units with both mentioned filterdesigns. Could you post the crossoverfilterdesign and enclosure volume you are using for the design of your mini-monitors?

thanks in advance and greetings from Holland,

Tube Addict Sander