Power Response VS Summed Response

ani_101

What should we optimize for - power response or summed response.

While playing in PCD, i came across the following scenario - The power response seems to be higher than the summed response. The summed response is lower than the woofer response - indicate a little bit of cancellation from the tweeter.

Usually I see a dip in the power response around the XO region, but in this case i see the opposite - the Power response is flat compared to the summer response. 

The Phase seems to have gone for a toss... inverting the tweeter doesn't seem to do much to the summed or power response - slightly wonky wobble, but nothing more. Trying to get the woofer and tweeter phase to overlap or even come close seems to destroy anything on the summed or power response - so this is probably going to get thrown away and started from scratch. But i wanted to understand power response and summed response what what should be prioritized over what?

The woofer is forth order parallel with a parallel LR Contour filter and tweeter is third order parallel.

dcibel

What you show in your screenshot will happen when the drivers are not in-phase on your design axis. Here you can see that you have 15 degrees of separation at the crossover frequency, and at 3kHz where you see the bump in power response your drivers are about 25 degrees out. When drivers are not in-phase at your design axis, at some point off-axis where the phase does line up, the response will be increased as a result.

Play around with these buttons to simulate off-axis response and all should become clear. Make sure you've entered the driver piston diameters and offset geometry correctly.

As for optimizing for power response or summed response - yes to both. Acoustic design is all about balancing compromises. A lot of what makes up power response is unavoidable by the driver geometry, but the crossover order does have some effect, and for a dome tweeter with a flat face plate, what you usually see is a bump in power response above the crossover frequency. Consequentially, you might find that making a small dip here in the on-axis response makes for a better overall presentation.

Ed_Perkins

That plateau in the tweeter response below the xover point probably contributes too.  IIRC the power response of LR filters have a dip in power response around the xover while BW filters have flat power response.  But that's only if your final acoustic roll-offs match those xover target curves - which the tweeter doesn't. 

ani_101

So follow up question. Normally the rule of thumb (or maybe an old audiophile's old wife's tale) is to aim for good phase overlap in and around the XO region (which lends to a deep reverse null).

So given that this sim is not that - how viable is this particular sim - usable or shouldn't be considered at all. As I am getting good phase overlap with deep reverse null, but at a cost of a smooth summed / power response. So what is more important / desirable smoother FR with not so good phase integration or slightly bumpy FR but with better phase response?

jhollander

For even order x-o's I might have a great null but lousy phase tracking which could mean there is a lobe off axis.  So I wouldn't ignore phase alignment for on axis response without looking for off axis lobes.

My Limus design had OK phase alignment, and I thought decent off axis performance. But I got comments that it sounded hot.  I discovered a big lobe vertically off axis.

ani_101

with a bit more fidgeting around, this si the outcome.

V3, this is not that close, but the Summer response is relatively flat and the Rev null has a deeper null.

ani_101

This is the next one - more Phase overlap, but a bit shallower null. Also a little dip in the 5k region. Theoretically, which one is more acceptable (without voicing, just on paper based on the wriggly lines?

V4, little dip in 5k and shallower rev null

dcibel

ani_101 said:

So follow up question. Normally the rule of thumb (or maybe an old audiophile's old wife's tale) is to aim for good phase overlap in and around the XO region (which lends to a deep reverse null).

So given that this sim is not that - how viable is this particular sim - usable or shouldn't be considered at all. As I am getting good phase overlap with deep reverse null, but at a cost of a smooth summed / power response. So what is more important / desirable smoother FR with not so good phase integration or slightly bumpy FR but with better phase response?

What you've shown really isn't that bad, in fact the overall power response is quite good. Remember that the "design axis" doesn't need to be a straight line toward the tweeter, for example if you intend for the speaker to me mounted high up on a wall, a design axis angled downward might make sense.

What I would call your "design axis" is the axis where phase does line up at the crossover point. Using this axis for your "on-axis" design point guarantees that the response will only drop off as you move off axis. If phase doesn't line up, at some point off axis you will have a peak.

What I would do if I were you on the original design, is adjust the off-axis points to determine the specific off-axis angle where phase does line up, and looking at that angle and the response at the angles where you expect to be as a listener, decide if that works for you.

Wolf

Neither really has bad phase alignment. If the FR is flatter and a deeper null, pick that one.
Place a cap across your padding resistor of about 1-2uF to lift top octave.

jhollander

Me, I'm building one and measuring the response to see how close the sim matches.  That will tell me if I got the offset and tails right.  It also gives me a chance to move my mic around and look for issues.  

Looks like you might have some impedance issues to work out.

ani_101

jhollander said:

Looks like you might have some impedance issues to work out.

2 ohm stable design? 

haven't played around with impedance much - what's the process for bumping up impedance...

ani_101

jhollander said:

Me, I'm building one and measuring the response to see how close the sim matches.  That will tell me if I got the offset and tails right.  It also gives me a chance to move my mic around and look for issues.  

I am going to start a new thread - for measuring and using the various tools to sim and measure, NF, splicing, Baffle step, tails, etc. Need a solid grounding in the basics.... more to understand how everything interacts...

Billet

I kinda like the first one, the power response looks very smooth and tracks the on axis response well.  My first order systems have a similar phase issue at the crossover point and they sound fine to me.

However, I do agree that there may be some off axis peaks and valleys.  You also try some different Z offsets for the tweeter to see the effect of vertical angles on the response.  The power response is an average. 

If you were to find the vertical angle where the two drivers are in phase at the crossover point, it will probably be louder at that frequency.  Would you actually hear it?  I don't know, maybe?  Good questions.

 

ani_101

Wolf said:

Neither really has bad phase alignment. If the FR is flatter and a deeper null, pick that one.
Place a cap across your padding resistor of about 1-2uF to lift top octave.

@Wolf, Padding Resistor on the Tweeter? BTW there is no padding resistor on the tweeter.

Would that be R10 in PCD, and would the Cap be C8?

ani_101

So this is what i got for a draft XO...

The dip  at 4k is around 3db and 5 db for the power response, but the phase and reverse null is pretty good. The impedance stays around 4 ohms throughout, so that's good - see next post...

I think I will start with this and see how it measures once I have the XO put together.

ani_101

Other graphs for the V6 posted above.... 

Summed Response, Transfer Function and Impedance.
Impedance stays at/above 4ohms.



Tweeter in Reverse:



Woofer Low Pass:



Tweeter High Pass: