Introduction to designing crossovers without measurement

1. I thought that the Zobel placed before the 8 ohm driver would be seen by the amp, crossover and BSC. Does it change with placement in the circuit? ....

You did mention earlier (#598) that ...."the amp will see the woofer below the crossover frequency"....So, if I cross over around 6000-7000Hz, will the amp be seeing the sum of the impedance of the driver based on the impedance graph at this point and the resistor of the BSC? Will the amp not see the Zobel placed before the driver?

2. Though you have answered this earlier in #598, just wanted to clarify regarding the L-Pad on for the tweeter w.r.t. the Zobel. Do I calculate L-Pad based on the sum of the impedance of the woofer at 6000-7000Hz and the BSC resistor (which is now presenting itself as 20 ohm after 5000Hz) plus the difference in sensitivity OR just the difference in sensitivity if I have a Zobel on the woofer.

Allen... Thanks once again for taking the time to answer my queries. I am sure I am asking questions which would be on the minds of a lot of people wanting to jump the DIY bandwagon but find it difficult to get answers to such questions with an online search.
 
diyAudio Moderator
Joined 2008
Paid Member
1. I thought that the Zobel placed before the 8 ohm driver would be seen by the amp, crossover and BSC. Does it change with placement in the circuit? ....
It will be seen by all, but it does change with placement. The point is that Zobel compensation is used to present a simple load, which means appearing resistive, that is not changing with frequency. There needs to be a reason for doing this. If the reason is to make the load simpler for a filter then it needs to be done directly where the filter sees it. Sometimes on both sides.. you could use Zobel compensation more than once, eg. before the driver, before the filter and after the amp.

However, the likelyhood of the BSC and filter interfering is low. The need to compensate for the amplifier is also low, so this apparently leaves your desire to have 8 ohms as the only reason for wanting this Zobel, I'm not sure this alone is a good reason.

If you put, say a 10 ohm resistor across the amp, it will bring the total impedance closer to 8 ohms, but would not achieve much. If you put the resistor after the amp and after the filter but before the BSC and before the driver it would begin to make some sense. Then the needed 7kHz filter components will be different as discussed a few posts back. But in either case there will be a larger power usage by the compensation resistor than the driver, and this will be dissipated as heat.
L-Pad on for the tweeter w.r.t. the Zobel. Do I calculate L-Pad based on the sum of the impedance of the woofer at 6000-7000Hz and the BSC resistor (which is now presenting itself as 20 ohm after 5000Hz) plus the difference in sensitivity OR just the difference in sensitivity if I have a Zobel on the woofer.
Only use the impedance of the tweeter and only adjust for the difference in sensitivity. All that you need to know about the woofer is that its response has been made what you need it to be, and also what its sensitivity is.
 
Last edited:
1. The Zobel network I have seen, mainly employ a cap and resistor across the speaker. The values I was getting were more like 12.3 uF cap and 9.4 ohm resistor for 7.5 ohm Re and 1.08mh Le.

From your post though, it seems that I would be wasting a lot of amp power by going with a plain resistor across the speaker.
 
diyAudio Moderator
Joined 2008
Paid Member
My point wasn't to use the resistor without the capacitor. An RC compensation works above a certain frequency. At these higher frequencies the resistor works as if the capacitor is not there and uses this power.

If you put such compensation on the woofer, the BSC will change its behaviour and want adjustment.
 
From post #600 .... "If you go first order, you will be putting a capacitor in series with the RL circuit. It can go on either side of the RL, it makes no difference because each part has the same current running through it."

Is the 1st order LPF part a typo. Shouldn't it read inductor in series instead of capacitor if its a LPF?
 
Allen..... I have decided on the following plan for a Sprite type boombox build:

TPA3116D2 AMP+.......LPF 0.3mH Inductor (-3db @ 4250Hz).....BSC (Sprite type 0.9mH Inductor parallel to 16 ohm Resistor).......+Woofer (ND90-8 with 81.5db sensitivity)

TPA3116D2 AMP+.......HPF 2.2uF Capacitor (-3db @ 9025Hz)......L-Pad 12db (6 ohm Resistor in series and 2.7 ohm Resistor in parallel shunt to Tweeter).....+Tweeter (AMT-8 with 89db sensitivity)

I have gone for a 1st order Solen Split type crossover. The idea was to crossover the Woofer around the point the BSC Inductor begins to lose its effectiveness and around the point where the Woofer frequency starts rising.

For the L-Pad I have calculated the 7.5db difference in sensitivity and around 4.5db of BSC sensitivity loss of the Woofer for a total of 12db loss. Will adjust according to sound. Might have to compensate for the BSC Resistor??

I hope I am on the right track with this. Thank you for all the help.
 
So, I got hold of the old Paul Carmody Sprite speaker I had built last year and given away to my brother. It was based on a TD7498 bluetooth amp with the Dayton ND90-8 drivers. The Sprite contour filter (BSC) was a 0.9mH coil in parallel to a 16 ohm resister. The plan was to make it into a 2-way speaker with the addition of an AMT-8 tweeter. I thought I'd give it try with the old box before I built a completely new one for me. The tweeter would be placed outside with all the other added components.

I added the 0.3mH LPF before the BSC on the ND90-8 and a 2.2uF cap on the tweeter. This is a 1st order Solen Split crossover at around 6200 Hz. It gives a 2LR sort of -6db dip at the crossover point. The only difference is that the slopes are like the Butterworth 1st order slope. I padded down the tweeter with a 12db L-Pad (6 ohm resistor in series with 2.7 ohm resistor in parallel).

The sound has gone from good boombox type sound to really great. The bass is more defined, the mids are now more clean and the treble has got so much of space and clarity. I'm listening to them with a big grin and can imagine that they would make superb separates.

Now to build my own box for them and put all the guts in. My suggestion to all those who started out with the Sprite build, try this out. Its a huge improvement over the Sprite sound, though it adds to the cost as you've got to buy tweeters, extra LPF coils, HPF caps and the L-Pad resistors. But, its worth it.

Thank you Allen for your tremendous thread and inputs. Its been a good learning experience.
 
Last edited:
hello and thank you for the excellent write up... I will be building my first ever speakers and stuck at the 2nd step, lol (i.e. flattening the tweeters impedance).

If you don't mind I would like to post my questions as I go so that it might be helpful to others, too.

The home theater system will be 5.2.2 atmos:
- denon s730h receiver
- subwoofer amp (still looking)
- L/R/C: visaton BG20 full range (8 ohm) + visaton DTW72 tweeter (8 ohm) + visaton br15.34 port hole in an approx. 41-liter vented box
- surrounds and ceiling speakers: 4x peerless TC9FD18 (8 ohm) in a 1.5-liter sealed box
- subwoofers: 2x infinity 1260w's in 140-liter boxes

I started modeling in boxsim and attaching the bg20 and dtw72 simulations. BG20 has an enclosure of 84x28x17 cm (height/width/depth) and dtw72 has 3x28x17 cm (couldn't merge them yet in one enclosure - my first trials).

Now, I have started to get numbers from your equations and so far I got:

flattening the woofer's impedance:
BG20:
Resistor: Rdc = 6.20 ohm >> 6.20 x 1,25 = 7,75 ohm - ?? watts (how to determine the wattage / what is the importance?)
Capacitor: Le = 0.8 mH >> 0.8 (7.752) = 13.32 uF

flattening the tweeter's impedance:
DTW72:
Resistor: ?? (in your example the graph shows a peak at 550hz at 12 ohms but DTW72 doesn't have any peaks like that and at 8 ohms the freq. is at 180hz so how am I going to determine the resistor value and the crossover will be at 180hz? I thought I can use the tweeter especially after 5kHz)

Sorry for the newbie questions :confused:

Best
 

Attachments

  • BG20_01.JPG
    BG20_01.JPG
    354.4 KB · Views: 476
  • DT72_8ohm_01.JPG
    DT72_8ohm_01.JPG
    348.8 KB · Views: 472
Last edited:
diyAudio Moderator
Joined 2008
Paid Member
The power rating of resistors should be high enough that they are not damaged by heat. It is not simple to predict how much you need. Low efficiency drivers, and woofers, and when used at high power, need to be larger than high efficiency drivers, and tweeters, and when used at low power. The resistors I use typically vary between 1/4W and 20W. It is OK to use a resistor with a higher power rating than you need. It can be useful to have a stock of 5W resistors of different values to use. When you use them on a woofer, it is best to play music through them then feel them. If you cannot keep your finger on them then you need to be using a higher rated part.

If your tweeter impedance is smooth enough to work with then you don't need the parallel resistor.
 
The power rating of resistors should be high enough that they are not damaged by heat. It is not simple to predict how much you need. Low efficiency drivers, and woofers, and when used at high power, need to be larger than high efficiency drivers, and tweeters, and when used at low power. The resistors I use typically vary between 1/4W and 20W. It is OK to use a resistor with a higher power rating than you need. It can be useful to have a stock of 5W resistors of different values to use. When you use them on a woofer, it is best to play music through them then feel them. If you cannot keep your finger on them then you need to be using a higher rated part.

If your tweeter impedance is smooth enough to work with then you don't need the parallel resistor.

Thank you sir! I will continue the calculations. Best.
 
While attaching foam near the edge of the baffle can reduce the ripples in the response (particularly on-axis) caused by baffle edge diffraction, the basic baffle step shift in response of 6dB will always remain regardless of the type of damping material used. All you're doing is getting a little bit closer to the theoretical smooth 6dB shelf response of a sphere, with less overshoot than you might otherwise get with a rectangular baffle.

I'm sure you know that, but some people reading could possibly mistake your comments to suggest that somehow the basic 6dB shift was being reduced, reducing the need for baffle step correction, which isn't the case.

With typical edge radii of front baffles I'd also suggest that foam is only effective above about 2Khz or so, and only at frequencies where the active driver has wide enough (180 degree) dispersion to illuminate the baffle in the first place.

For a midrange driver you would probably get more reduction of baffle diffraction ripples (especially below 2Khz) by simply offsetting the driver a bit, and for a tweeter (which is where the foam would be most effective) you can more or less eliminate baffle diffraction by using a tweeter incorporating a waveguide - on many such designs little or no signal is radiated at right angles along the plane of the baffle so there is nothing to diffract from the cabinet edges at those frequencies.

Whether there is any worthwhile effect depends entirely on the directivity of the drivers - small drivers with very wide dispersion may see a significant improvement, larger and/or directivity controlled drivers won't see any useful improvement in my experience.

Certainly worth trying though.

Hi Simon, what would you class as a “typical baffle radii”?
 
OR
as long as one is happy with the sound his speakers produce...
If his speakers are imperfect, after initial honeymoon, he is bound to try and experiment (with guidelines given here) and try to improve upon the sound. There are always sim tools available to get closer to the reality of perfect crossover. I say "closer" ;)

(these things are applicable to those who dont have measurement set up)
BTW I am no. 615
 
Hi Guys,
In most 2-way speakers consists 1.woofer 2.tweeter.
So from which driver human (male % female singers) voice comes?.
In the thread inexpensive ultra OB project one is large woofer GRS and one peerless vifa 3.5 full range driver. so in this scenario is it the same ?.
Thank you.
 
diyAudio Member
Joined 2007
Allens answer is an arguement in favour of 4-Way speakers with big subwoofers. Crossed as low as practical.
I read a while ago that Frank Sinatras voice had a fundamental frequency of 88Hz and was considered a Baritone which makes me wonder what Ivan Rebroffs/Barry Whites fundamental frequency were. I always thought the Sinatra was a tenor
 
Allens answer is an arguement in favour of 4-Way speakers with big subwoofers. Crossed as low as practical.
I read a while ago that Frank Sinatras voice had a fundamental frequency of 88Hz and was considered a Baritone which makes me wonder what Ivan Rebroffs/Barry Whites fundamental frequency were. I always thought the Sinatra was a tenor

Hmm. This explains why I haven't been satisfied until now with vocals.
Now I'm using three 15" 98db/octave woofers per channel, from 50 to 1000 Hz, a planar mid and a planar tweeter above that, all dipole, excepting the dual 15" subwoofer.
First time I've had effortless clean vocals at all listening levels.