Help with active crossover settings

[suttondesign]

Hi All. I've built a pair of Heissmann DXT-Mon-182, and I built the analog crossovers. They measure very well. But I also want to use my MiniDSP Flex-8 to try a DSP crossover.

Here's the problem. On Heissmann's site, he specifies the crossover as 1950hz LR 4th order. But he sent me the attached parameters for DSP, and I can't make heads or tails of the 1st and 2d order crossover notations. On the Flex-8, I can only input one high- or low-pass per driver. How do I use Heissmann's 1st and 2d order figures?

FYI, I have already input a LR4th at 1950hz, and I used the DSP parameters on the attached. For the drivers being used, the 1950hz figure falls right where it should per the mfr. spec sheets, and higher or lower tends to increase distortion for one or the other driver. My initial settings got a very good measured result except that the overall tweeter level measured a bit high when I did a near field measurement (and it sounded a bit hot, too).

 

[NTK]

Something seems wrong here. Series connecting a 1st order LPF (or HPF) with a 2nd order LPF (or HPF) is not going to get you a 4th order LR LPF (or HPF). You get a 4th order LR filter by series connecting two identical 2nd order Butterworth filters. You may want to double check with Heissmann.

As to how to implement the filters in the diagram with miniDSP, here are my thoughts:

• miniDSP allows for 1 cross-over filter, and I would implement the 1st order filters (6 dB/octave slope) there.
• Implement the 2nd order filter for the cross-over in the PEQ section.
• miniDSP doesn't seem to offer LPF or HPF as selections in the basic mode for PEQ, so you'll have to create them in the advanced mode.
• Use miniDSP's filter coefficient spreadsheet to calculate the biquad coefficients for the 2nd order LPF (or HPF) filter and load the biquad coefficients
• Proceed to enter the remaining PEQ's

[Edit] The 1st order HPF cross-over filter for the tweeter doesn't look right. Fc is 7293 Hz! I think you need to double check with Heissmann.

 

[suttondesign]

I know, the Heissmann figures are bizarre. My own 1950hz 4th order LR with a broad reduction of 3db for the tweeter and no other PEQ yields the following from the listening position. Ignore below 100hz, where I had an incorrect crossover setting for my subs plus a 59hz room mode.

 

[olieb]

I know, the Heissmann figures are bizarre. My own 1950hz 4th order LR with a broad reduction of 3db for the tweeter and no other PEQ yields the following from the listening position. Ignore below 100hz, where I had an incorrect crossover setting for my subs plus a 59hz room mode.

I am not so sure that the Heissmann settings are bizarre.
Your diagram has a 300dB spread! I cannot see anything useful here, around 1kHz there is unevenness >10dB. You might want to try 50dB spread.
There are two ways to implement a XO. Either equalise for flat first and then apply a textbook LR4, or adjust the EQ to an acoustic 4th order filter curve. Heissmann seemingly went the second route, which is not so straightforward to implement in the MiniDSP, but @NTK explained how it can be done.
Did you use an IR window to have a quasi-anechoic response? As the crossover is high enough it should be possible to get rid of most of the room influence by placing the speaker away from boundaries and using a 4...5ms window.
Then it should become clear whether the Heissmann figures do work or not.

 

[Sokel]

Yoy seem to have a whooping -40dB (!) slope from 60Hz to 20kHz.
That can't be right.

 

[suttondesign]

I am going to go back to the passive crossover network and then apply auto EQ to the whole frequency range of the speakers until I can get better at REW. I know the physical crossover works well for now.

Thank you for your comments here.

 

[suttondesign]

Heissmann confirms the 7293hz 1st order filter owing to the tweeter waveguide’s effects.

 

[NTK]

Heissmann confirms the 7293hz 1st order filter owing to the tweeter waveguide’s effects.

Not convinced. Here are my simulations of the electrical part. Frequency is normalized to 1 kHz, i.e. for the horizontal axis, 1 = 1 kHz.

 

[suttondesign]

Okay, then I will take the advice to EQ in the rew/flex scheme then do a crossover. it will be a fun exercise to compare that with measurements of the analog crossover.

 

[bmc0]

Not convinced.

Why not? A first-order highpass at some high frequency is not uncommon for waveguide-loaded tweeters. You can't expect the electrical responses alone to sum correctly; the drivers (especially the tweeter) will alter the phase considerably through the crossover region.

Here are the electrical filter responses including the peaking EQs (assuming the definition of Q is the same):

 

[NTK]

Obviously the cross-over filters diagram is taken from some loudspeaker design software. It should be trivial for Heissmann to provide an FR plot.

 

[bmc0]

Looks like VituixCAD. It'd certainly be useful if they were to provide at least the individual filtered driver responses.

 

[suttondesign]

Why not? A first-order highpass at some high frequency is not uncommon for waveguide-loaded tweeters. You can't expect the electrical responses alone to sum correctly; the drivers (especially the tweeter) will alter the phase considerably through the crossover region.

Here are the electrical filter responses including the peaking EQs (assuming the definition of Q is the same):
View attachment 479702

this is what heissmann said.

 

[suttondesign]

here was the individual driver responses with the analog crossover.

 

[suttondesign]

Here's the Heissman using the analog crossover I built, and without any subs, showing pre-correction and post-correction with AutoEQ applied in the MiniDSP control console. Measurement was from the listening position.

 

[suttondesign]

Oh, and they sound great, but they have a pretty narrow directivity, much like studio monitors.

 

[Digital_Thor]

Hi All. I've built a pair of Heissmann DXT-Mon-182, and I built the analog crossovers. They measure very well. But I also want to use my MiniDSP Flex-8 to try a DSP crossover.

Here's the problem. On Heissmann's site, he specifies the crossover as 1950hz LR 4th order. But he sent me the attached parameters for DSP, and I can't make heads or tails of the 1st and 2d order crossover notations. On the Flex-8, I can only input one high- or low-pass per driver. How do I use Heissmann's 1st and 2d order figures?

FYI, I have already input a LR4th at 1950hz, and I used the DSP parameters on the attached. For the drivers being used, the 1950hz figure falls right where it should per the mfr. spec sheets, and higher or lower tends to increase distortion for one or the other driver. My initial settings got a very good measured result except that the overall tweeter level measured a bit high when I did a near field measurement (and it sounded a bit hot, too).

I know it might be trivial and well known by most. But sometimes I think people forget that the electrical and acoustical filters are not necessarily the same - meaning that the electrical filter + driver response(in cabinet) = acoustical output - only when everything is linear to begin with.
Example: 12dB filter + 12dB sloping driver in cabinet = 24dB acoustical slope.
My experience, have taught me that you can use almost any setting in a DSP, as long as you respect this rule of thumb.
You could for example have two filters in series in your DSP to shape a driver response to fit better with a neighbouring driver.
I think it helps if you look at a DSP as a 'curve shaper' , as long as you remember that not everything in a DSP, is possible in real life, because your initial limitations always depends on the driver/cabinet design to begin with.

 

[suttondesign]

great info

 

[Ktacos]

Oh, and they sound great, but they have a pretty narrow directivity, much like studio monitors.

Is it? The polars on that thing show a pretty damn wide speaker. I guess people have different opinions on what wide is, but the 182 looks to be past 60 degrees on one side for the majority of the speaker.

 

[suttondesign]

agreed. i am just used to the linkwitz dipole, so going back to a forward firing speaker seems narrow to me!