Your comments are wanted

[donaldpatten]

I have opened a new web site describing the design, testing and a DIY construction section.

The project started about 4 years ago to design and build a crossover to introduce delay to the low frequencies . Since I have Altec Super Dupllex 605A speakers that I love ?
If you have any input , please pass it on.
Thank you
Don

http://web.mac.com/donaldpatten/iWeb/X-Over/X-Over.html

[waitinonthebus]

Donald, the site is coming along nicely. I have been checking your progress out occasionally. Keep up the good work.

To all passers-by; I have used these crossovers with my 605's and determined that I shall build a set. I have compared them to 2 other crossovers and there is a clear improvement. I have not enjoyed my speakers that much any other time.

If you are a 605 user then this is worth your time. The stock crossover, FWIW, is simply not close to being as effective.

[bfish]

Nice!

[Earl K]

Hi Don,

Really Nice ! / ( some Great Unconventional Thinking going on here. )

- Over the next couple of days I'll try to add into this post ( or maybe additional posts ) some comments & ( constructive criticism ).

- Here's a few comments ( hurriedly written , before I leave on errands ) relating to the web-sites' content & its' structure .

(A) I'd love to see the following pic types a couple of times this size ( I'd like to be able to read all the info in those screen captures . ( I had to download these pics & then blow them up / to see all the pertinant data .




(B) It is extremely useful to the ( more than ) casual browser to understand the design choices that you've made for your custom filter .

Two Examples ;

(i) Why not let people know that C7 & L5 form a "Series Resonant" ( LCR ) circuit that actually boosts HF content ( through resonance ) ?
- For Altec fans ( who've likely never been exposed to this circuit type ) it's extremely useful to know the "hows & whys" about it's implementation .
- This boost is centered around 14,125 hz / ( though I haven't yet figured out its' "Q" .)
- I can see how it's inclusion is fundamental in getting decent UHF response from the 806 driver ( that nosedives after 10K ).
- Sharp-eyed JBL fans should recognize the circuits' , since variations exist within the HiPass portion of the crossovers' within 4430/35 speakers ( as well as the L200B ) .

(ii) In the same vein ; observers may like to know that L6 & C8 form a "series" notch filter centered around 1070 hz.
- Since its' Fc is actually below F3, it would be useful to others if you explained this unconventional approach ( in creating additional slope for the basic single pole HiPass ) .



Bye for now

[

<font color="#FFFFFF" size="1">[ November 09, 2007 03:49 PM: Message edited by: Earl K ]</font>

[donaldpatten]

Earl K :

Thank you for your comments.

With Print Preview I can zoom on my Mac and make a full screen of the pics.

The FFT upper trace ( green )shows the energy , on a frequency basis, in the test pulse (this is not the XO freq resp ). This is the signal into the crossover. The lower ( red ) curve is the noise level for ref.

The lower traces are the O?scope time basis of the test pulse ( green ) into the XO and the LF out of the XO ( red ). This shows the LF time delay.

Right , C7 & L5 give a HF boost 10kHz ?16kHz , L5 = 35? #18AWG 0.27mH & 0.2 Ohms DCR

L6 & C8 are to protect the HF driver. Drivers are not to be driven at or below resonance. ( Do not tell the Woofer Guys ). These drivers peak ( impedance curve ) around 1kHz. L= 88? #28AWG 2.7mH & 5.6 Ohms DCR.

Don

[donaldpatten]

I have updated my website

The primary addition is a crossover electrical freq resp and why the two LC circuits.
Thank you
Don

[Earl K]

"donaldpatten"
If you have any input , please pass it on.

<font size="2" face="Verdana, Helvetica, sans-serif">- Thanks for the voltage drives for the HiPass & Lowpass filters .

from "Earl K"

(B) It is extremely useful to the ( more than ) casual browser to understand the design choices that you've made for your custom filter .

<font size="2" face="Verdana, Helvetica, sans-serif">(i) It would be insightful if you explained your reasoning behind choosing to match up a single pole HiPass filter with the 8 Pole LowPass .

- You really have achieved a very smooth transformation through the crossover area.
Kudos !



-Some here might be squeamish about using such a gentle slope ( 6 db ) vs something steeper which offers greater driver protection .
- That LCR notchfilter does help some in this regard .

(ii) I'd also be interested in seeing some more SIMs ( from your Xover modelling software ) of the non linear group-delay that would result from using other filter types ( in the lowpass ).
- This could help clarify why the Bessel choice is superior ( to other possibilities ) .



<font color="#FFFFFF" size="1">[ November 11, 2007 03:47 PM: Message edited by: Earl K ]</font>

[donaldpatten]

I have added a N1600-C freq resp curve so you can compare

Earl K
I tried a Butterworth two pole LC low pass filter

L = 1.83 mH
C = 5.41 uFd
Rl= 13 ( with Zobel )
Fc= 1.6kHz
Rs= 0

and a Constant Resistance low pass filter

L = 1.91 mH
C = 8.29 uFd
Ro= 12 DCR

The curves are on an old IBM and it refuses to print ,Sorry
They are the same shape with more delay on the CR filter

The Butterworth delay starts out flat with 140 usec delay 40 Hz to 400 Hz
Then a smooth rise to a peak of 170 usec at 1kHz
Falling to 20 usec at 4 kHz and off to nil at 10 kHz

Butterworth feq resp is flat up to 1kHz
-3 dB @ 2k
-11 dB @3k
-24 @ 4k
-47 @ 6k

The delay curve is not very inviting .
Bessel is the only filter that has flat delay and flat freq resp.
As to why 8 poles ? The number of poles is proportional to Fc and Delay.
The more delay , the higher the crossover freq. the more poles


FYI This is a formula I use as a rough check of the LF delay
Td = 10* SQRT(10*Lt*Ct)
Td = time delay in usec
Lt = total inductance in mH
Ct = total capacitance in uFd

To check the schematic delay L1-C4
Plug in L1+L2+L3+L4 = 5.28 and C1+C2+C3+C4 = 21.98
Td = 341usec

Don