Corrective circuit phase shift?

[prickears]

I'm hoping someone can take the timeout to help me to understand this.

With regards to passive XO's, Is it a simple case of........................

A) In series with a given driver, C = + 90deg, L = - 90deg,
then in parallel, C = - 90deg, L = + 90deg?

B) Resistors do not shift phase?

Then, corrective circuits, shift phase accordingly? (ie, LCR = 0deg, zobel = - 90deg)???

Thanks, Mick.

[LineArray]

All-pass filter - Wikipedia, the free encyclopedia


I think you may be looking for something like an allpass filter.

What do you like to compensate ?

Kind Regards

[speaker dave]

Originally Posted by prickears

I'm hoping someone can take the timeout to help me to understand this.

With regards to passive XO's, Is it a simple case of........................

A) In series with a given driver, C = + 90deg, L = - 90deg,
then in parallel, C = - 90deg, L = + 90deg?

B) Resistors do not shift phase?

Then, corrective circuits, shift phase accordingly? (ie, LCR = 0deg, zobel = - 90deg)???

Thanks, Mick.

C and L will give an ultimate phase shift of +-90 degrees, but only when terminated into a resistor and at frequencies well into their stop band. That is to say they aren't universal phase shift devices that give 90 degree shift to anything they are connected to.

For example if you place a capacitor between an amplifier and a resistor, it will create a first order rolloff cutting off towards low frequencies. The corner frequency depends on the capacitor value and resistor value. Where their reactances are equal there will be a cutoff corner and 3dB of loss. You would also measure 45 degrees of phase shift there (half the ultimate value). At lower frequencies the phase shift would tend towards a maximum of 90 degrees.

Voltage across and current through would always have the 90 degree shift property but that doesn't help you.

Hope that makes sense.

David

[prickears]

Thanks David,

Yes, that does make sense and helps alot. This is a suprisingly difficult subject to find any real info on, so I apprieciate you taking the time to help.

Thanks also LineArray,

Not looking to compensate anything really at the moment, however, I did just make up a BSC circuit for the mid driver on my active three way set up. It made a huge improvement, Although I believe this to be at least partly attributed to phase shift.
Also knowledge for future reference, I plan on buying books on such subjects, just as soon as I'm back on my feet and back at work again.

Thanx guys, Mick.

[wintermute]

Hi Mick, I have always found phase issues the hardest thing to get my head around! It wasn't until I saw a diagram of some sine waves which showed how the phase worked that It finally sunk in (and that was just understanding what it meant by a 90 deg phase shift!)

Are you still having problems with the tendon you severed, or have you had another mishap?

Tony.

[prickears]

G'Day Tony,

I 'think', I'm on the right track, as to understanding this phase thingy, I reckon i've seen the wave diagram you mentioned, (two waves criss/crossing 90 deg out of phase?). Just trying to work out how each branch of the crossover throws these waves back and forth, in phase + phase, and - phasing back into phase, and so forth, (WTF!?).

Wrist is fine,(Thanks for your concern), I stretched my medial and posterior cruciate ligaments, to within an inch of there lives, so they are badly torn ATM, (one hard ball get too many, Footy). Surgeon, says I should be able to walk properly again in about 6-8 weeks, as for shearing.........hhmmm. Time will tell, I guess.

Thanks Tony,

Kind regards, Mick.

[wintermute]

ooo that sounds very nasty!!! Hope you are back on your feet (literally) soon!

One thing I have been having some problems coming to grips with re phase is whether it is the electroics or the actual acoustic slope of the crossover that determine the phase shift.

I found that in simulation (and the implemented crossover seemed to agree) that the phase was fine if the aoustic slope was fine) regardless of how I got that slope.

The current sim has a 4th order (lclc) on the woofer and a 2nd order CL on the tweeter but the accoustic slope is 4db for both, and it gives a very deep null when one of the drivers is reversed indicating that the phase must be right... So I think it is more than just the electrical interaction of the components in the crossover, I think that the slope affects it too... just anecdotal, don't take my word for it though

Tony.

[prickears]

Thanx Tony,
Does this tweeter have a natural 2nd order roll off, to cause this, or is there a corrective filter of some sort that is making it exibit a second order slope, then the xo itself doubling it? If that makes sense.
It seems likely to me that your simulator is also predicting phase shift, based on component values and therefore, predicted slopes.
The way I understand it, a chebychev filter has a different phase shift at xo to a butterworth of the same order, and therefore will exhibit a different slope shape, (Q?), whilst still arriving at 12 db/octave, (electricaly at least).
Does your sim allow you to test LCR filters and predict the phase shift (if any), or a zobel which I would think would give a +90deg shift, But I think this would depend on how far removed it is from fX?
Which sim is it BTW?

(sorry I'm not real good at getting stuff from my head into typed form, I only hope you can see WTF i'm on about here!)

Mick.

[prickears]

Filter & Crossover Types for Loudspeakers — Reviews and News from Audioholics

OK I understand this much (I think), at ^this^ point we have an electrical crossover. We know its phase characteristics, how it behaves b4 it arrives at its given roll off rate (Q), so in an ideal world, we could have a perfect XO.

Would it be fair to say that a 2nd order chebychev actually gets louder before rolling off and arriving 12 db down @ 1 oct, so boosts, then has a sharper roll off than L/R?

So then to arrive at a flat accoustical response one must shape these different filter shapes to the accoustical characteristics of the given driver, THEN and only then, if in correct phase, we have true matching accoustical slopes, and hopefully a XO that doesn't draw attention to itself?!

'IF' thats how it all works, what effect does a corrective filter have on all this? (WRT Phase only!!).

Does it simply effect the whole show as discribed in my first post????.....

Mick.

Ps. can't wait to get some testing gear!!!!

[wintermute]

Hi Mick,

My current crossover has a 2nd order Bessel slope centred at 3Khz. The woofer has a 4K notch filter on it (ie a cap a coil and a resistor in parallel) and the tweeter simply has a cap on it. The tweeter needs to be inverted (as per normal 2nd order) or I get a deep null.

I don't actually know what the phase shift of the notch filter is (I'd guess it is comlex??), maybe I was just lucky that this combo worked out. Not sure. But I put it down to more to do with the accoustic slopes than the electrical characteristics, since for the 4th order network sim, I had 4th order electrical on the tweeter and 2nd on the woofer (as opposed to what I think is nothing on the woofer 1st order on the tweeter) for the currently implemented crossover.

I'm using speakerworkshop for the sim, but I have paid less attention to the phase than I probably should have Perhaps it is less of a problem since the drivers are time aligned.

Tony.