diyAudio (http://www.diyaudio.com/forums/index.php)
-   Solid State (http://www.diyaudio.com/forums/solid-state/)
-   -   problem with active allpass filter for time delay (http://www.diyaudio.com/forums/solid-state/227640-active-allpass-filter-time-delay.html)

 davidel94 12th January 2013 07:08 PM

problem with active allpass filter for time delay

6 Attachment(s)
hi I have a problem with an allpass filter, which function is to delay by 140us the high frequency bandwidth of an active crossover, made up with second order Linkwitz-Riley filters. I found on the internet a texas instrument note and I designed the filters using those circuits with the way described at page 46 and the coefficient they give at page 63. one of the pics shows the response of the difference of the outputs of the two filters and I obtain a flat response, as it should be, while another shows the sum, each stage is inverting thus the output of the allpass has already opposite phase than that of the treble output, of the output of the allpass with the output of the bass and I get a bad response. I don't understand where is my mistake or if LTspice is making something wrong.

 davidel94 12th January 2013 07:14 PM

I forgot to say that I choose a frequency of 10kHz which give a coefficient of 1.4 which is a 5th order filter, using the Tgr0 of 1.506 gives a frequency of 10757Hz that I used in the calculations.

 CharlieLaub 12th January 2013 07:32 PM

I think that the first order stage you have used is reversing the phase. You can try using the other first order all pass configuration. See this link for more info:
Linkwitz Lab: Active Filters - Delay Compensation

Just swap the positions of C5 and R13. That will likely fix the problem.

-Charlie

 davidel94 13th January 2013 07:30 AM

No it doesn't solve the problem because I get two dips in the response of about -3dB each around the crossover frequency which is 2.5kHz

 DF96 13th January 2013 01:36 PM

I don't understand what you are asking. An allpass works by adding lots of phase shift in a certain frequency region, but keeping the amplitude the same. Above and below that region it will either invert or not, depending on the design. For example, the simplest allpass will swap from inverting to not inverting as the frequency is raised (or the opposite).

If you sum/difference phase shifted signal with unshifted signal then you are likely to get sharp dips with one of them because of cancellation.

 CharlieLaub 13th January 2013 04:21 PM

Quote:
 Originally Posted by DF96 (http://www.diyaudio.com/forums/solid-state/227640-active-allpass-filter-time-delay-post3323792.html#post3323792) I don't understand what you are asking. An allpass works by adding lots of phase shift in a certain frequency region, but keeping the amplitude the same. Above and below that region it will either invert or not, depending on the design. For example, the simplest allpass will swap from inverting to not inverting as the frequency is raised (or the opposite). If you sum/difference phase shifted signal with unshifted signal then you are likely to get sharp dips with one of them because of cancellation.
I'm seconding this explanation. I had assumed that davidel94 understood this. Are you just doing a sim of the circuit? You should make a plot of the relative phase angle between the outputs with and without the delay, and the behavior of the amplitude will become more clear to you.

Why do you intend to use the delay and why choose 140us?

-Charlie

 jcx 13th January 2013 04:41 PM

changing the phase of just one path in the XO region will change the |sum|

you should move the phase EQ to the input of both sections of the XO circuit

 davidel94 15th January 2013 04:30 PM

the time delay is needed in order to make treble sound wave come to the ear at the same time of bass air wave, because these two signals come from two different loudspeaker which are not on the same horizontal axis. so accordingly to what you said the problem is in the phase.

 Davey 15th January 2013 04:47 PM

David,

You're analyzing the electrical summation of the circuit. Once you start making corrections for acoustic differences (physical offsets in this case), you should be analyzing the acoustic summation/results.

Cheers,

Dave.

 jcx 15th January 2013 04:48 PM

have you moved the phase EQ to the input before split to both sections of the XO yet?

All times are GMT. The time now is 11:59 PM.

Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.