Designing my first crossover, 2-way, advice on drivers needed...

[nerd of nerds]

Hello!

For a long time now i've been building small rull range speakers, nothing with fostex, just TB and HI-VI.

I really like these but i need some bass and i've never built or designed a crossover before...

Rather than just building someone elses design i thought that i'd do something different...

I want to use this woofer:

http://www.partsexpress.com/pe/showdetl.cfm?&DID=7&Partnumber=295-335

With this tweeter

http://www.partsexpress.com/pe/showdetl.cfm?&DID=7&Partnumber=275-035

I know that isn't going to be the best tweeter in the world but i don't want to go spending a crapload of cash on a tweeter just to kill it accidentally...

I have the boxes designed already, they will be around 30-40 liters (can't remember exactly what i put into winisd at the moment..) and tuned to 30hz

I was thinking of using a 2nd order linkwitz crossover at about 1900hz

i know i will need to attenuate the tweeter down a bit so it will blend better with the woofer, but what else do i need?

Is this doable? or is this a combination of a great cheap woofer with a shitty cheap tweeter? Zaph had great things to say about this woofer in his recent tests...

 

[69stingray]

Nerd-

That is on my "next project" list also. I think the aluminum driver will look great in medium-dark Cherry veneered box.

I actually tried that tweeter in Passive Crossover Designer and could not get it work well (probably my novice skill level).

I am now looking at a 2" full range driver. My hearing cuts out at 14-15 kHz, so this driver is fine.

http://www.partsexpress.com/pe/pshowdetl.cfm?&DID=7&Partnumber=264-806

I think it will match the aluminum cone well also.

Download Passive Crossover Designer (if you have Excel). You can use the Freq. Response and Impedance Files from Parts Express to model a crossover. Now sure how accurate it is, but it is fun to play with. Good luck and tell us (me) how it turns out!

http://www.pvconsultants.com/audio/crossover/pcd.htm

This is how far I got in Passive Crossover Designer:

 

[Zaph]

Nerd of Nerds:
http://www.pesupport.com/cgi-bin/config.pl?read=243649

If you want a cheap tweeter that will work low enough to be used with that woofer, consider the Hi-Vi K1. [shameless plug] I have a pair on ebay that will likely go for super cheap.

69stingray:
Carefull modeling with those PE response curves. They are not accurate at all below 250hz due to gating. As such, the phase for the entire speaker is not accurate, particularly with lower crossover points such as you are using. Not sure if you realize how phase works, but phase at any frequency is determined by the entire frequency response curve with the response closest to the point of calculation being the most important.

To use PE data, you need to generate a predicted bass rolloff and then merge it with the PE data. Also you need a predicted impedence curve for whatever enclosure you are using. Additionally, you need to sum in the predicted baffle step response, regenerate phase and do a hilbert bode transform.

I find the PCD to be a cool piece of software but nearly useless because it relies on other people's data which you can't control and the conditions of measurement are unknown.

 

[nerd of nerds]

Nerd-

That is on my "next project" list also. I think the aluminum driver will look great in medium-dark Cherry veneered box.

I actually tried that tweeter in Passive Crossover Designer and could not get it work well (probably my novice skill level).

I am now looking at a 2" full range driver. My hearing cuts out at 14-15 kHz, so this driver is fine.

http://www.partsexpress.com/pe/pshowdetl.cfm?&DID=7&Partnumber=264-806

I think it will match the aluminum cone well also.

Download Passive Crossover Designer (if you have Excel). You can use the Freq. Response and Impedance Files from Parts Express to model a crossover. Now sure how accurate it is, but it is fun to play with. Good luck and tell us (me) how it turns out!

http://www.pvconsultants.com/audio/crossover/pcd.htm

This is how far I got in Passive Crossover Designer:

from that it looks like you are trying to cross it over at about 700hz...thats WAAAAAAAAY too low, try 1900hz instead...

 

[nerd of nerds]

Nerd of Nerds:
http://www.pesupport.com/cgi-bin/config.pl?read=243649

If you want a cheap tweeter that will work low enough to be used with that woofer, consider the Hi-Vi K1. [shameless plug] I have a pair on ebay that will likely go for super cheap.

69stingray:
Carefull modeling with those PE response curves. They are not accurate at all below 250hz due to gating. As such, the phase for the entire speaker is not accurate, particularly with lower crossover points such as you are using. Not sure if you realize how phase works, but phase at any frequency is determined by the entire frequency response curve with the response closest to the point of calculation being the most important.

To use PE data, you need to generate a predicted bass rolloff and then merge it with the PE data. Also you need a predicted impedence curve for whatever enclosure you are using. Additionally, you need to sum in the predicted baffle step response, regenerate phase and do a hilbert bode transform.

I find the PCD to be a cool piece of software but nearly useless because it relies on other people's data which you can't control and the conditions of measurement are unknown.

k, i'll take a look at that one as well as the vifa.

I know you aren't teaching speaker design 101, but do you think a 2khz crossover point would work well with 2nd order?

And do i need any baffle step or crap like that? iirc the baffle will be 9" wide and about 24" tall, dayton driver will be centered and tweeter will be abote the woofer, port will be below the woofer.

 

[nerd of nerds]

On 2nd thought, would the dayton reference series rs125 be a good driver for this tweeter? i know i'd need a subwoofer if i went this route, but i would have needed one eventually any way...

 

[morbo]

A couple of points:

1. 700hz is not really low at all for that 2" TB driver, unless you are going 1st order. I've crossed the 3" version (871) as low as 350hz 2nd order, and it didn't strain. I would reconsider using it though, the cheap little widerangers from Tang Band/Aura/Hi Vi are hit and miss, sure there are some proven diamonds in the rough, but the majority are 'the rough'. Stick with the known quantities, drivers that have been measured and proved to perform well - TB 871, Aura NS3, Hivi B3N, etc. Start by looking at Zaph's excellent mini-speaker roundup (http://www.zaphaudio.com/minitest/).

2. The little Hi-vi 3/4" tweeter is not well suited to crossing that low. Notice the Fs of 1400hz, and the operating range of 3500-25khz. This is a supertweeter. Since the woofer you want to use needs to be crossed low, you would be much better off with the tweeter Zaph recommended here, or some other tweeter with a <1000hz fs. Again I would stick to the known quantities in the bottom price range, do lots of searching for the model, if you cant find any feedback/info, move on. In the higher price ranges, you can choose a driver from a name vendor (vifa, seas, Scan, peerless, dayton, usher, morel etc etc) and expect generally good performance, at worst mediocre. The se bargain basement drivers (<$20) have some real duds among them, so you have to select carefully.

3. What else you need besides the drivers and the idea to pad one down - I would reccommend some measurement gear and a copy of speaker workshop or somthing like that. Designing with stock XO calculators and no measurements is a crapshoot, and getting something good will more than likely cost you a lot in prototypes. You can of course use the FRD tools which are quite nice, but as Zaph said, you rely on other people's measurements, your own knowledge of when to use which measurement, and which intermediate adjustments of the data are neccessary.

 

[morbo]

If you want to stick with that tweeter, you probably want a woofer that will roll off smoothly without nasty breakups, since you should be crossing no lower than say 3.5k. Something like the CSS Wr125, or the GR Research m130, soft or paper cone, smooth top end rolloff will make the crossover design a lot easier. Or you could go with a smaller metal 'widerange' like the aura ns3.

 

[nerd of nerds]

yeah, i'm working on all the measurement equipment right now.

i just realized that my mobo can handle it...the last motherboard i owned had shitty onboard sound that couldn't handle input and output at the same time 🙄

So i'm guessing that i need:
A mic

A clamp for the driver, or a sealed box, or an infinite baffle? which of these would work better?

whatelse would i need?

 

[morbo]

Do the measurements with the drivers flush mounted in the baffle exactly as they will be in the finished speaker. The simplest way to get up and running with measurement gear is to get a Behringer UB802 mixer (mic preamp + phantom power) and the ECM8000 measurement microphone. Any decent soudcard should do, I use one of the old Soundblaster PCI 128 cards based on the Ensoniq chipset. You can find some reviews and comparisons of sound cards here: http://www.pcavtech.com/soundcards/summary/index.htm. The list is rather dated, and there are lots of discontinued models there. But - you can probably pick some of them up real cheap (like the PCI card I have). I assume Speaker Workshop has loopback-style calibration routines that should compensate for any sound card irregularities, like JustMLS does.

 

[69stingray]

69stingray:
Carefull modeling with those PE response curves. They are not accurate at all below 250hz due to gating. As such, the phase for the entire speaker is not accurate, particularly with lower crossover points such as you are using. Not sure if you realize how phase works, but phase at any frequency is determined by the entire frequency response curve with the response closest to the point of calculation being the most important.

To use PE data, you need to generate a predicted bass rolloff and then merge it with the PE data. Also you need a predicted impedence curve for whatever enclosure you are using. Additionally, you need to sum in the predicted baffle step response, regenerate phase and do a hilbert bode transform.

I find the PCD to be a cool piece of software but nearly useless because it relies on other people's data which you can't control and the conditions of measurement are unknown.

Thanks Zaph-

I knew that in regards to the bass roll-off. I was using PCD for the crossover stuff for more of an educational experiment. I think it is a good tool to play around with. I sure of made that a little more clear in my reply.

I have nosed around the FRD Consortium a little more and have gotten up the nerve to try to work through their process of using their tools to design the crossover with the combined *.FRD files from the different programs. Wish me luck, I am sure I will have plenty of questions.