Hardman crossover filter with Hypex

[Geertidow]

Dear forum members,

Out of curiousity I'd like to try a Hardman crossover.
I find it difficult to find a way to build that filter using Hypex Filter Design (for my FA-123 plate amps).
See: http://linea-research.co.uk/wp-cont...Tech Docs/CrossoverFilters White Paper -C.pdf for more information.

Does anyone know how to design this in the filter design software?

Best regards,

Geert

 

[AllenB]

I would lay out the wanted response in a crossover simulator and find an alternative.

 

[Geertidow]

It just seems like an interesting crossover, but I don't think I can recreate that wanted crossover.

 

[4real]

Looks like an elliptical filter. Check Micromodeler DSP - Interactive, Web Based Digital Filter Design Software, it might help you create the correct coefficients.

 

[Douglas Blake]

Is this schematic any help?

Hardman crossover filter | Consort3's Blog

 

[AllenB]

Can you take those transfer functions, like in the link above, and trace them but scale the frequency at the same time?

 

[Boden]

In case the Hypex can load a .frd file to be used as a target, you could simulate the Hardman filter in Vituixcad and take the .frd filter curves (aka filter transfer functions) from Vituixcad and import those in the Hypex Wizzard. But then again, I do not know whether the Hypex allows a .frd import.


Keep in mind the Hardman filter does not provide any baffle step. Many moons ago I simulated the Hardman filter in Calsod, but could not get it properly optimized for a 6,5"midwoofer/1" tweeter combo. Furthermore, the different acoustic centres at x/o may make matters more complicated.

 

[Geertidow]

I might be in a little further than I can comprehend, but I'll see what I can do. I may contact Hypex and ask them.

 

[Geertidow]

Hypex does not know if it's possible .

 

[AllenB]

You just need a similar looking curve. You might make one from other common filter types. You would want to engage an overlay to use as a visual target, then try filter combinations.

a. Does Hypex designer allow an *.frd file as an overlay?
b. Does Hypex designer show phase or delay (as this was the original claim of Hardman)
c. Can you make or find a plot of a Hardman crossover to trace, to use the response as a target in Hypex designer, that also shows phase or delay.

If Hypex designer doesn't do it, other crossover simulators will. People here could also assist with that.

 

[AllenB]

I thought I'd share something I made for another thread. It shows the idea of fitting one filter to another. It shows that the other filter type can be fairly simple and doesn't always need to be the same type. It also shows prioritising of the higher level output.

No claims that this is helpful and it is a starting point not a finished example.

 

[Geertidow]

Hey Allen,

a. I've asked Hypex technical support if this is possible.
b. Hypex does show phase in the phase graph, but I don't think that really corresponds to actual phases.
c. I'm not sure, I've only found the curve from the document posted above, which doesn't show phase or delay.

I'll await the Hypex answer.

 

[AllenB]

a. This was so you could find a filter by comparison. My example above uses a different crossover simulator but the deal is the same, because once you find the filter you want it should be easy to set it up over in hypex designer.

b. You are probably right, I would guess that it only shows phase for the filter and not the driver as well. Phase like this is still interesting while you are trying to find a style of filter that fits with your goals.

For example, the above plot.. The article says group delay was preferred in the Hardman crossover. I've shown two plots below (group delay is shown black). The one with the red phase trace is the normal looking filter from above and the one with the blue phase trace is the hardman type.

If you assume that the below 4k region is the more important, the group delay is more relaxed for the hardman type. By the way, don't be put off because I didn't make the dip more sharp like the example. Changing it to do this is not hard.

Some like to take out the axial peak around driver breakup, and would line that up with the filter dip if the resulting crossover frequency fits in with the tweeter.

 

[Douglas Blake]

You just need a similar looking curve. You might make one from other common filter types. You would want to engage an overlay to use as a visual target, then try filter combinations.

To do that in DSP you would use a fairly broad notch filter followed by a high pass with attenuator 1/2 octave higher.

 

[mark100]

Hi Gertidow, gang,
if you do find that the hypex software can generate bi-quads from a measurent like being suggested/hoped, I have the linea processor you linked and would be willing to make some Hardman transfers at a few chosen frequencies...if that helps start to roll the ball

 

[Boden]

Hello Mark100


Is the Linea software capable of creating filter coeffs for any given filter transfer function?
From a quick glance I gathered the free version can only work up to 4th order. For complex, non textbook, transfer functions, that include e.g. baffle stap and several notches, that is not enough.

How are you coping with all that?

 

[mark100]

Hi Boden,

The Linea software, System Engineer, only loads named 'menu' filters into the processor, without any type of export that i'm aware of.
4th and 8th are the Hardman menu options.

So really, all i can offer is transfer functions taken at chosen xover frequencies, if the Hypex software can use them. To help jump start.
I guess other EQs could be mutiplied in with the Hardman xover if desired, before giving it all to bi-quad generation ???.

(Must admit, the Hypex's bi-quad route is what's kept me away from their amp packages that otherwise look very nice....)

 

[soundbloke]

What is the difference between a Hardman alignment and Thiele's earlier notched alignments? Am I missing something?

 

[Geertidow]

Hi Mark, thanks for your offer.

I am currently thinking on producing 4th or 8th order Hardman filters at 500 and 2800 Hz.
Possible for 8th order even 500 and 2500 Hz.

The biquad allows for custom q's etc. but I don't think it takes functions.
This project may be a bit too difficult for me (yet..)

 

[mark100]

Glad to Geertidow, if you think you can use a particular transfer, just let me know.
fwiw, i don't think i could do what you're trying either

@ soundbloke, hi, i sure dunno..everybody is always trying to find a new xover technique. That's the hobby i guess, and it is fun, digging in.
I bet there is a lot of satisfaction in truly understanding all the IIR xovers.
And perhaps doubly fun if you can build them passive.
Me, i never need anything past IIR BW's and LR's, and only then to go with complementary linear phase FIR, when i need to keep latency down.
But i just view xovers as a means to an end, not something that's fun like making different speaker builds.

We each have our fun, huh?
I wanna hear mine without too much thinking/complication. lol

My hat is off to folks who like to dig deep into IIR crossovers ! They are freakin complicated imo