The guy is struggling to get a crossover developed using his Hypex gear. Forget about "room modes", that's not what he needs to focus on at the moment. Instead try to help him get the basics working, first.
Interesting that you used the word tame here, rather than fix (and I would have called you if you did). It takes time to establish whether the response can be tweaked like this, and hence scheduled for permanent inclusion in a crossover. Response variations, such as room modes are merely incidental occurrences. Sometimes you need to rework at some higher level to get the wanted result.
Hi
Thanks for the interest from everybody. Much appreciated. I take on your suggestions.
The images I sent were nearfield Woofer no filters, nearfiled Mid no filters and farfield all drivers with xovers (to protect Tweeter)
What I was looking for, just to get me to understand Hypex Filter Designer, is what filters to use in Zones 1,2,3,4,5 and 6. I hope that would get me going before checking the PHASE
Thanks for the interest from everybody. Much appreciated. I take on your suggestions.
The images I sent were nearfield Woofer no filters, nearfiled Mid no filters and farfield all drivers with xovers (to protect Tweeter)
What I was looking for, just to get me to understand Hypex Filter Designer, is what filters to use in Zones 1,2,3,4,5 and 6. I hope that would get me going before checking the PHASE
Ok, i try to give some help to response flattening for the 500hz and 2kHz LR4 filters:
Zone 1: forget about it now
Zone 2: extend the flattening to 1kHz (1 octave up from the crossover point) with a /Highself1, 500Hz, 6dB gain/, and tame that skyscraper peak around 1.3kHz with /BoostCut, 1300Hz, Q10, -15dB gain/ apply to the woofer channel.
Zone 3: extend the flattening range from 250Hz to 4000Hz (1 octave up and down) and tame the peak at 4kHz.
To 250Hz /Lowshelf2, 250Hz, Q0.7, 15dB gain/.
To 4kHz /Highself2, 4kHz, Q0.7, 15dB gain/
Taming peak /BoostCut, 4.2Khz, Q9, -6dB gain/ and /BoostCut 4.8kHz, Q10, -6dB gain/
Apply to the midrange channel.
Zone 4-5: requires the tweeter's gated response without filter, just don't do it too loud.
These parameters is just approximate values, always needs to experiment with other values and measure the outcome. Nothing is set in stone but you get a picture what's going if you try these filters what i wrote.
Frequency response measurement is not required at high volume.
Ahh and always start with 0dB gain for the channels.
I don't want to explain where to use what filters, because the filter functions is already good described in the HFD manual at page 14.
Zone 1: forget about it now
Zone 2: extend the flattening to 1kHz (1 octave up from the crossover point) with a /Highself1, 500Hz, 6dB gain/, and tame that skyscraper peak around 1.3kHz with /BoostCut, 1300Hz, Q10, -15dB gain/ apply to the woofer channel.
Zone 3: extend the flattening range from 250Hz to 4000Hz (1 octave up and down) and tame the peak at 4kHz.
To 250Hz /Lowshelf2, 250Hz, Q0.7, 15dB gain/.
To 4kHz /Highself2, 4kHz, Q0.7, 15dB gain/
Taming peak /BoostCut, 4.2Khz, Q9, -6dB gain/ and /BoostCut 4.8kHz, Q10, -6dB gain/
Apply to the midrange channel.
Zone 4-5: requires the tweeter's gated response without filter, just don't do it too loud.
These parameters is just approximate values, always needs to experiment with other values and measure the outcome. Nothing is set in stone but you get a picture what's going if you try these filters what i wrote.
Frequency response measurement is not required at high volume.
Ahh and always start with 0dB gain for the channels.
I don't want to explain where to use what filters, because the filter functions is already good described in the HFD manual at page 14.
Last edited:
Alternatively you can combine the midrange driver's own slopes as a part of the crossover filter.
It looks close to a Q=0.71 2nd order slopes at both sides, you can confirm this with a target response overlay in ARTA.
You already know that you get a LR4 response if you combine 2xQ=0.71 2nd order filter. So you just need to find the -3dB points of the midrange driver, which is around 400Hz and 2.8kHz according to your measurement.
So if you apply a single /highpass2, 400Hz, Q=0.71/ and a single /lowpass2, 2800Hz, Q=0.71/ to the midrange you get a 400Hz and 2.8kHz LR4 response.
This way you don't need to flatten the midrange response. But if the original driver slopes are very different than Q=0.71 the combining to true LR4 response may not works. Anyway it is recommended to filter the peak around 4 kHz.
It looks close to a Q=0.71 2nd order slopes at both sides, you can confirm this with a target response overlay in ARTA.
You already know that you get a LR4 response if you combine 2xQ=0.71 2nd order filter. So you just need to find the -3dB points of the midrange driver, which is around 400Hz and 2.8kHz according to your measurement.
So if you apply a single /highpass2, 400Hz, Q=0.71/ and a single /lowpass2, 2800Hz, Q=0.71/ to the midrange you get a 400Hz and 2.8kHz LR4 response.
This way you don't need to flatten the midrange response. But if the original driver slopes are very different than Q=0.71 the combining to true LR4 response may not works. Anyway it is recommended to filter the peak around 4 kHz.
Last edited:
What? I wrote an example filter to tame the 4-5kHz peak in post #65.
What about thinking? Explain it please!
I meant that the resonance peaks around break-up could be easily equalized and not that defeated completely. What's wrong with that?
Last edited:
What's wrong with that? Breakup is not just resonance peaks. You wouldn't see this on a two dimensional plot like given above. You'd have to know this separately.
Consider this link. 2/3 down the article there is a CSD. It doesn't even tell the whole story, but might help.
So far we've only mentioned the peak that you can see on the plot.
Consider this link. 2/3 down the article there is a CSD. It doesn't even tell the whole story, but might help.
So far we've only mentioned the peak that you can see on the plot.
Yeah, i know well that a break up is not just a frequency peak. But the timing related problems (e.g. ringing) is out of reach to solve.
If the peak is equalized and the crossover point is not in the close proximity then a LR4 crossover is usually enough. But if you want to use 96dB/octave or whatever brickwall sloped crossover, then use it, it's your choice.
I think you're being ridiculous
Why?
Some ppl are prefer a steep filter like that. Not me, but some.
Anyway, to be constructive for this thread, what are your advises here?
I appreciate all your help. I will study what you say and try this weekend. Takes some time to set up living room for measurements.
I cannot agree with you that the filter funstions are well described in HFD manual P14. If they were, I wouldn't be asking.
I have just rec'd a reply from Hypex tech dept. where I asked when to use Shelf rather than Asym Shelf and how to use Allpass filters. Latter I gather to correct Phase?? The reply from Hypex. "Sorry but I do not have that information" Well who has?
Zone 1 left out because it's room related. You need an un-gated measurement at the listening spot to do there something.
Zone 4-5 left out because we do not know yet what causes the peaks and dips here. Do identify you need to isolate the tweeter response from the whole as i wrote.
Tweeter: I am not going to risk a 400€ tweeter without a crossover. What may be low for you, may be high for the tweeter, so I'm just not going to risk it.
The Farfield measurement with all drivers and tweeter and crossover has a dip at approx 3.5KHz (Zone 4) No idea yet what causes it. Manufacturer's graph is flat in this area.
The Slope down from 10K or 12K down to 20KHz (Zone 5) approx 5dB follows manufacturer's slope down of 5dB. So I think this can safely be flattened
Woofer: 10dB slope down from 60Hz to 20Hz (Zone 1) is similar to manufacturer's graph from 40Hz to 20Hz also 10dB
This I think can also be flattened. The measurement was nearfield so hopefully room modes didn't come into it. I have many listening areas so think nearfield is appropriate for the woofer.
The Farfield measurement with all drivers and tweeter and crossover has a dip at approx 3.5KHz (Zone 4) No idea yet what causes it. Manufacturer's graph is flat in this area.
The Slope down from 10K or 12K down to 20KHz (Zone 5) approx 5dB follows manufacturer's slope down of 5dB. So I think this can safely be flattened
Woofer: 10dB slope down from 60Hz to 20Hz (Zone 1) is similar to manufacturer's graph from 40Hz to 20Hz also 10dB
This I think can also be flattened. The measurement was nearfield so hopefully room modes didn't come into it. I have many listening areas so think nearfield is appropriate for the woofer.
If you don't want to risk your tweeter that's ok. Then you need a gated measurement from the tweeter only with their filter in place to check how it follows the LR4 response. Do an LR4 target response overlay in ARTA.
I think the 3.5kHz dip may crossover related. That's wat we want to fix.
The woofer's nearfield is not what you hear at your listening spot because of the massive room and boundary gain you got from the combination of wall mount and your room's limited dimensions. Your room is not an anechoic chamber.
The gain from the room can be 9dB to the low-end. So please do a non-gated farfield measurements at the listening spot. You may not need any DSP boost to the woofer.
I think the 3.5kHz dip may crossover related. That's wat we want to fix.
The woofer's nearfield is not what you hear at your listening spot because of the massive room and boundary gain you got from the combination of wall mount and your room's limited dimensions. Your room is not an anechoic chamber.
The gain from the room can be 9dB to the low-end. So please do a non-gated farfield measurements at the listening spot. You may not need any DSP boost to the woofer.
Ok, if we can believe the factory measurements then the tweeter's frequency response is almost completely flat in the desired band, starting to roll-off from 2kHz in endless wall. That means if you apply a LR4 crossover filter, you get a LR4 response down to 2kHz then the slope start to steepens which is may not too bad if your desired crossover point is above 2kHz but depends on other factors.
The spike at 7kHz can be reduced with a high Q boostcut filter with -3dB gain.
The 3.5kHz dip may comes from the too steep response of the midrange driver, because you applied a LR4 filter to an already 2nd order sloped response.
The zone 5 may need a slight boost with a highself2 20000Hz Q=0.7 6dB gain, but i don't think it matters too much.
The spike at 7kHz can be reduced with a high Q boostcut filter with -3dB gain.
The 3.5kHz dip may comes from the too steep response of the midrange driver, because you applied a LR4 filter to an already 2nd order sloped response.
The zone 5 may need a slight boost with a highself2 20000Hz Q=0.7 6dB gain, but i don't think it matters too much.
Last edited:
Zone 5 just to understand how it's done
Why Highshelf2?
Why at 20000Hz (the) highest freq not the lower 10-12k?
Why Q = 0.7?
5dB-6dB seems OK
Zone 1 I would like to flatten because the measured downward slope is similar to manufacturer's and I can hear lack of bass without needing to take measurements
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.