Phase problems combining LP and HP xo in Speaker Workshop

[soongsc]

I don't think I understood these instructions. Can you help?

I understood the following:

• select one design point: I presume it's the mic placement point.
• get a minimum phase for the tweeter: I presume this means I measure the tweeter's FR and then unwrap the phase using "Delay".
• Use the same marker location to measure the woofer: this is how I measure tweeter and woofer nowadays anyway... one has to keep the latency, mic position and marker positions constant, in order to get coherent readings.
• See how the woofer would look like: What do you mean by this? What do I see, and what am I supposed to be looking for?

And that last sentence about difference between the acoustic centres, I couldnt' understand at all. How do I know the location of the acoustic centre of a driver?

1. Yes.

2. Yes again. When the phase is unwrapped, there are some small adjustments necessary. When this is done. The difference between the reference impulse time and the marker time is the time of flight, that with the speed of sound, you can calculate the acoustic ceneter distance from the mic. The difference between that and the distance between the mic and baffle is where the acoustic center is relative to the baffle.

3. Correct.

4. This is the part where the trade-off is. If you've done the first three, post what you get and we can figure it out. But you want to keep this chart first. Then you can move the marker around like you did with the tweeter until you've calculated the acoustic center location. Since the line between the mic and the woofer will not be perpendicular to the baffle, you need to consider this when calculating the acoustic center location.

 

[tcpip]

I wrote to Roman Bednarek for help, and he's been trying to help me very patiently, clearing all my confusion and giving me tips. He tried to design an xo using my SPL measurements, and he first removed equal amounts of delay from both drivers (my measurement of 4.532msec seemed okay to him) and then put them together with an Fc of 2.8kHz, sort of close to Jbateman's Fc. He thinks my idea of 2K for this tweeter is too low (quite understandable). Like Jbateman, he too has got a very sharply phase-aligned Fc.

His xo is:

An externally hosted image should be here but it was not working when we last tested it.

His combined SPL curve with drivers in phase is:

An externally hosted image should be here but it was not working when we last tested it.

and with drivers in opposite phase is:

An externally hosted image should be here but it was not working when we last tested it.

What do you think? I think this is hard to beat, and I'll just live with the higher Fc... for all I know, it may give me better performance than 2K, given this tweeter's characteristics.

The essential differences I can see between this one and Jbateman's design are:

• This one has a slightly higher Fc, and
• It's been able to use this higher Fc to reduce the 3-5KHz mild hump in the SPL which Jbateman had been unhappy about in his own design.

All in all, it's an amazing learning experience for me.

 

[jbateman]

Build it and see how you like the sound!

The one I offered earlier was just to show a typical circuit...I didn't spend much time trying to perfect it at all.

 

[tcpip]

Build it and see how you like the sound!

The one I offered earlier was just to show a typical circuit...I didn't spend much time trying to perfect it at all.

Yes, but it had bloody good phase coherency.

You know, your design taught me a lot, because I'm now beginning to see that you may not always be able to get both flat SPL and good phase coherency at any arbitrary Fc, even though the individual drivers' curves look okay. I'm beginning to finally understand the tradeoffs that xo designers do.

I'm going to build Roman's xo design and I'll see how it performs. And if I can, I'll build one of yours and one of mine too, to see what really happens to the sound with a shift in Fc or a change in the magnitude at the 2-5kHz range... all these are just paper facts for me till I actually slog over the building and listening.

Thanks a lot for your help.

 

[lndm]

Ahh, the fun part

I agree, a simmed design will usually sound good, but it is only a starting point.

Using common sense, and knowledge learned by simming, I will play with component values for an extended time. Every now and then when I think something has come together, I will sim what I have to get a new perspective.

 

[tcpip]

Now for BSC

Can you guys help me with BSC? I want a sort of gentle (i.e. 3dB) shelf for BSC for my drivers. Now that I have the data for the xo, and I know the impedance curve for it, how do I add the BSC circuit?

I read various articles about BSC, and also downloaded and tried out the program called The Edge. It seems pretty good, but it only gives me a 6dB step, like most other documents and articles. How do I get a shallower shelf?

My front baffles are about four feet tall and about 9.5" (23.75cm) wide.

 

[tcpip]

Using common sense, and knowledge learned by simming, I will play with component values for an extended time. Every now and then when I think something has come together, I will sim what I have to get a new perspective.

Isn't it difficult/painful playing around with component values if you have to wind your own coils each time? How do you play around with coil values? R and C are easy.

 

[lndm]

I just buy a bunch of inexpensive inductors and put them in series or parallel as required, then change for better ones when I know what I want. This isn't ideal, but it works for me.

As far as BSC, put the tweeter aside and tune your woofer crossover to bring the upper midrange down to the level of the bass. Then make your tweeter match. At the very least (speaking broadly), you will need to increase the value of series inductor on the woofer, pad the tweeter, and do some more tweaking.

I prefer to incorporate BSC into the crossover rather than using a separate circuit.

Unfortunately, this often means you will lose some sensitivity. You could also experiment with room placement.

 

[tcpip]

I just buy a bunch of inexpensive inductors and put them in series or parallel as required, then change for better ones when I know what I want. This isn't ideal, but it works for me.

Very practical. I guess I can give it a shot if I get some lying-around inductors some day.

As far as BSC, put the tweeter aside and tune your woofer crossover to bring the upper midrange down to the level of the bass. Then make your tweeter match. At the very least (speaking broadly), you will need to increase the value of series inductor on the woofer, pad the tweeter, and do some more tweaking.

I was hoping for a separate circuit block to just add at the input to the composite (i.e. both tweeter and woofer) xo. That way, I'd get both drivers' levels to turn down. This is my first design project, and I feel it might be better if I took the simple and less-than-optimal route out once in a while, just to see what I hear.

 

[lndm]

I was hoping for a separate circuit block to just add at the input to the composite (i.e. both tweeter and woofer) xo.

If you look at the total impedance after adding the crossover, it may appear difficult to do anything with.

You could correct the impedance first, then put an inductor in series with the system, and a resistor in parallel with the inductor. (Then maybe correct the impedance again)

This wouldn't be my preference but it should work.

 

[tcpip]

If you look at the total impedance after adding the crossover, it may appear difficult to do anything with.

You could correct the impedance first, then put an inductor in series with the system, and a resistor in parallel with the inductor. (Then maybe correct the impedance again)

I tried working with the impedance. Created a pseudo-driver with the same impedance curve as that of the complete xo. Created a spurious FRD file with a flat freq response, from 30Hz to 20000Hz. Attached this FRD file as frequency response of the pseudo-driver. Then put this driver into a blank network, and added an L-R combo. It's not getting me anywhere. Let me first correct the peak in the impedance curve, and then see if I can get anywhere with BSC.

It's turning out to be harder than I thought.

 

[tcpip]

I have a first draft of a BSC

Want to take a look at it? It's in this SWD file. I'll also try posting some simulated FR curves with and without BSC, so that you can get some idea without opening the SWD file.

Incidentally, I followed the apparently sub-optimal approach of adding BSC to the entire composite xo, so that I could leave the existing xo circuit untouched. I've first added an impedance compensation circuit to fix the resonant peak of the earlier xo at 1.1KHz, then added an LR block for the actual BSC. I now have, as a fringe benefit, an xo with a lovely flat-looking impedance curve at some 10 Ohms impedance.

Please let me know what you think.

 

[tcpip]

Re: I have a first draft of a BSC

Incidentally, I followed the apparently sub-optimal approach of adding BSC to the entire composite xo, so that I could leave the existing xo circuit untouched. I've first added an impedance compensation circuit to fix the resonant peak of the earlier xo at 1.1KHz, then added an LR block for the actual BSC.

Got a note from Roman about his own xo. He says it already has about 4dB of BSC built-in... that's how he chooses his woofer inductor. I guess I'll then leave out my additional xo and just build and listen to his xo to see what it sounds like.

Quite amazing, how the experts do these things so simply. I wouldn't know how to copy his approach if I tried.

 

[lndm]

I'm sorry I don't use Speaker Workshop, I use xoversim from the FRD. I'd like to point out that the 'sub-optimal' hint I may have implied is only a logical conclusion I have drawn from my opinions, nothing more.

Anyway, you seem happy with your design to this point and that strikes me as a good thing. The flat impedance sounds especially good, I believe in doing this for any decent system.

FWIW, I went poring over some old crossovers I made for my ScanSpeak 7" drivers on 9.5" baffles. I used between a 1.5mH and 3.5mH inductor in series to bring the mids down to the level of the bass. This range was due to my experimenting with sensitivity versus bandwidth. I also used to experiment with room loading, but felt the speaker should either use the room fully or not at all. Of course it's never that simple

Now I use a 1mH inductor as I use a rear loaded horn to bring the bass up to the level of the mids