analog circuit for time-alignment? there's Bessel filters and what else?

[djk]

OK, now we buy five 10mH chokes for 45µS delay at 1Khz. That would require fourty 1mH for 8Khz, and we need about 3mS, so we would only need about 666 of the 1mH chokes.

It works the same way as the opamps.

If you were to optimize the choke and capacitor values you will still need 84 sections for 16Khz bandwidth.

 

[DorinD]

Yes, indeed. Have to study if those mini-inductors of 10mH/45mA/60ohm used in LC filters for power supplies are good for this.

Anyway, seems to be more do-able than using same number of CI's. I just modeled an 90 inductors filter which can delay 900mS till 5kHz, may be used for delaying a mid horn.

 

[m8o]

"Yes, I'll need 6 - 8 opamps in all-pass "

Run the numbers. For a 4Khz tweeter crossover you will need about 36" of delay. ...etc...

Ya, I don't know what the heck I was doing that night I 1st answreed. Working in both LTSpice based on circuits generated from Nuherz's Filter Free on a Delay (all-pass) filter today with either 2nd and 3rd order, I see my error now. The free version only does up to 3rd order. I mis-interpreted how adding sections adds to the total delay (it does) and group delay's freq response (it doesn't, but I thought it did for a reason I won't waste anyone's time with).

Ya, I have to drop this idea.

Wish I could use the Lattice ispPAC81 with something like 700 programmable filters within the IC, in which case this idea would be possible, but it's an obsolete product.... It's 50x - 100x noisier then an ultra low noise opamp, so there we go again. 😉

Just an FYI, I'm figuring I need a bit under 2ms of delay between my particular mid horn and hi-freq horn. I have one of these
http://www.terrasonde.com/products/color.php
so I'll be measuring it exactly once I get around to it. ...the Klipsch K401 horn is l-o-n-g

 

[DorinD]

@m80
Why to drop the ideea so easy; what about BBD's here:

http://www.du.edu/~etuttle/electron/elect39.htm

see also:

http://www.geofex.com/sad1024.htm

you will be more on the analogic side because you'll do only sampling which are memorized as an analog variable (is not "digitized" like in A/D-D/A convertors). Cheers!

 

[phase_accurate]

1.) Most BBDs are made out of unobtainium nowadays.

2.) THD+N of these is/was usually guite bad.

Regards

Charles

 

[DorinD]

MN3006 of Panasonic seems quite good with THD=0.2%, S/N=90dB and delay from 0.32-6.4ms. Problem is where to get such thing.

 

[phase_accurate]

If the bandwidth is also limited at the lower end then one can use less allpass stages to achieve a desired delay.

You will have to use cascaded 2nd-order allpass stages. But beware: The calculation of the coefficients wouldn't be easy!

Regards

Charles

 

[DorinD]

Just find MN3007 at 18.6 euros and MN32007 at 5.69 euros:

http://www.capodimonte.ro/produse.html?keyword=MN3*&where=0

 

[djk]

BBDs were just 'OK' when that was all there was (for surround sound).

 

[m8o]

Quality point seems true, with the SAD-1024 dynamic range @ "S/N > 70db" and distortion @ "less than 1%" ; not what I'd consider too great for a line-stage device.

I'll look to see if there are any advancements from newer silicon devices manufactured in this form however, so thanx for the recommendation. But with everything going digital, I image there's chance they're all obsolete and unavailable products.

edit: wow, that's from the 70's?!
edit2: also, I looked through numerous data sheets on datasheetarchive.com for BBD and Bucket Brigade , and not only is s/n and distortion in the mid-fi category (with THD very clock cycle thus delay dependent), the upper frequency is rolled off at or below 10KHz on all of them. ...I am looking to delay my tweeter after all... 😉

 

[phase_accurate]

IMO multiway horns are not candidates for time aligning in the analog domain. You either use a digital crossover or if you go analog use a crossover that makes the most of the advantages of a horn system (i.e. low IMD, lifelike dynamics ... ) by the use of a steep crossover.
But you would still need phase alignment in the crossover area of course.

Regards

Charles

 

[m8o]

Circuit no 4 here might be the basis for your design:
http://home.comcast.net/~neilrdavis/Plate_amps/
It uses the TAS3004...

Circuit #4 on that link has me re-thinking the 6 LM4780TA chips I have waiting in the wings for me to get started...

Actually, what am I saying. I can use those chips I have now for the lows and maybe mid if I don't need delay, and Circuit #4 for the tweeter and mid if I find I need delay on that too! Wow, I'm like'n that idea a lot! Thanx!
(oh, but I gotta modify the circuit design for transconductance operation ; eh, it can be done p2p on the board I'm sure.)

 

[m8o]

IMO multiway horns are not candidates for time aligning in the analog domain. You either use a digital crossover or if you go analog use a crossover that makes the most of the advantages of a horn system (i.e. low IMD, lifelike dynamics ... ) by the use of a steep crossover.

...I hear you. I'm getting pretty much sold on digital for tweeter now, with analog for the bass/mid if I measure that I don't need delay or find I only need a few usec of delay (Linkwitz himself pretty much mandates its inclusion to make the best of his crossover alignment)

 

[m8o]

I'm pretty sure building a digital eq/delay circuit around one of the chips mentioned [TAS3103 or TAS3004] from the ground-up is beyond me, but buying something complete or in kit form is not! 😀

http://www.groundsound.com/dcn23.html

http://www.groundsound.com/dcnkit.html

http://www.xilica.com/products.html

Discussion:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=89351&highlight=

I guess I gotta "pay ta play" (big time) if I go this route.

 

[DorinD]

Just for fun

I propose a delay circuit for a tweeter which could translate back it's reference plane by 3.5 cm. The 68 ohm resistor models internal resistance of the mini-inductors:

 

[m8o]

lol. ...did you Spice model it? What's the freq response and group delay look like?

 

[DorinD]

There are:

 

[phase_accurate]

I can't read the x-axis properly but it seems to be 100 us ub to 30 kHz, is it ?

This delay can be done with a Bessel ALL-pass using a total of five opamps.

Regards

Charles

 

[Sheldon]

Or something like this?: http://www.thuneau.com/

Yes, it is digital, but makes development easy even if you want to convert some of the functions to analog.

Sheldon

 

[EC8010]

If you were to optimize the choke and capacitor values you will still need 84 sections for 16kHz bandwidth.

That's how it was done at the BBC for matching the stereo delay of land lines that weren't co-routed. I remember using lumps about 120 x 150 x 250 (mm) packed solid with Vinkor inductors and Suflex polystyrene capacitors. I think only two delay equalisers were made that could give 2ms each.