I am doing a passive to active conversion, 3-way using a Hypex Fusion amp with DSP and while I am doing the box building I am also looking ahead into how I will set it up.
One thing I am not sure about is how to I apply a gentle slope to the overall response, e.g. Harman curve. For example, if I wanted to apply a 0.5dB (or another tilt) per octave roll off from 200Hz upwards to 20kHs.
Crossover points will be, say, 300Hz and 3kHz.
I need to use the overall EQ function (not per channel), right?.
But is there a way, in Hypex Filter Design, to apply a slope of a certain tilt to everything above a certain frequency?
One thing I am not sure about is how to I apply a gentle slope to the overall response, e.g. Harman curve. For example, if I wanted to apply a 0.5dB (or another tilt) per octave roll off from 200Hz upwards to 20kHs.
Crossover points will be, say, 300Hz and 3kHz.
I need to use the overall EQ function (not per channel), right?.
But is there a way, in Hypex Filter Design, to apply a slope of a certain tilt to everything above a certain frequency?
Last edited:
Of course. You can achieve any response so why not create it artificially and use it as a target..
It would make sense as a global (overall) EQ since you don't want it to interfere with the cross itself. While you can combine them, there are many things you might want to compensate for so it helps to keep them all separate where possible so you can reverse or adjust them individually as needed.
It would make sense as a global (overall) EQ since you don't want it to interfere with the cross itself. While you can combine them, there are many things you might want to compensate for so it helps to keep them all separate where possible so you can reverse or adjust them individually as needed.
Are you asking how to apply the slope itself? I might leave that one for someone who has the same processor as you. However..
In practice I don't often find that I need a filter the same shape as I want the response to be. It's more a matter of meeting the difference between targets and the curves you're trying to fit to them.
In practice I don't often find that I need a filter the same shape as I want the response to be. It's more a matter of meeting the difference between targets and the curves you're trying to fit to them.
Yes, a very basic question: If I want a 5dB slope overall from 200Hz to 20kHz, how do I apply that in Hypex Filter Design?
Is that a single filter e.g. low shelf or is it multiple filters?
In this video, it is done in REW and EQ APO and applies full room correction, not just the tilt. I am after just the tilt and using HFD.
Is that a single filter e.g. low shelf or is it multiple filters?
In this video, it is done in REW and EQ APO and applies full room correction, not just the tilt. I am after just the tilt and using HFD.
Last edited:
First you place this filter direct after the input before split to the different ways. This is so that you dont have to divide the EQ into three pieces.
It's achievable with one singel PEQ use a high-shelf filter.
The values for this PEQ would be:
F: 1000 Hz
Q: 0,38
G: the total level change from bottom to top.
Better to use a high shelf as the gain change will take place higher up where the energy is less and risk for clipping is smaller.
This is a great way to introduce a "tone control" - I think you will be surprised how small change is really detectable when the changes is so broadband.
This is how it looks in CamillaDSP:
A 5dB slope overall from 200Hz to 20kHz would be something like:
F: 2000 Hz
Q: 0,38
G: -5 dB
You have to translate that to Hypex "speak" 🙂
Good luck!
//
It's achievable with one singel PEQ use a high-shelf filter.
The values for this PEQ would be:
F: 1000 Hz
Q: 0,38
G: the total level change from bottom to top.
Better to use a high shelf as the gain change will take place higher up where the energy is less and risk for clipping is smaller.
This is a great way to introduce a "tone control" - I think you will be surprised how small change is really detectable when the changes is so broadband.
This is how it looks in CamillaDSP:
A 5dB slope overall from 200Hz to 20kHz would be something like:
F: 2000 Hz
Q: 0,38
G: -5 dB
You have to translate that to Hypex "speak" 🙂
Good luck!
//
Last edited:
ah, ok, so this is as simple as applying a shelf in the overall EQ section, with a low Q and a suitable gain. Thank you!
Hi, am I reasoning correctly from your posts that you are not using any crossover filter simulator to help with passive / active conversion?
House curve with DSP would be set after the whole thing is ready and built and in place and measured in room, because for example distance form speakers to walls affect balance. In room measured sound is not on-axis response but more like power response of the speaker, which naturally has a tilt with flat on-axis response depending on the size and shape of the speaker (that sets the DI). It might already have suitable slope in room.
House curve with DSP would be set after the whole thing is ready and built and in place and measured in room, because for example distance form speakers to walls affect balance. In room measured sound is not on-axis response but more like power response of the speaker, which naturally has a tilt with flat on-axis response depending on the size and shape of the speaker (that sets the DI). It might already have suitable slope in room.
yes, I intend to potentially apply a house curve as part of the final steps of setting up the speakers in room
I use global eq with minidsp but only for lowest room modes.
With HFD I must make mode/tonality eq-adjustment to each way's eq because global is not working.
The FA board must be connected to the computer to make EQ box green, but still I can't find a way to make any settings... Fw 1.46
Harman Curve is result of the loudspeaker's on-axis and off-axis responses in the end user's room, measured around listening spot with long gating and smoothing, then averaging. In most "normal" sized rooms it happens naturally or even in exaggerated form... Listening to music with two speakers for a few days is more important than "curve" fitting based on meaasurements. Measurements will help to find appropriate eq settings to get best sound from basically flat and smooth loudspeakers.
With HFD I must make mode/tonality eq-adjustment to each way's eq because global is not working.
The FA board must be connected to the computer to make EQ box green, but still I can't find a way to make any settings... Fw 1.46
Harman Curve is result of the loudspeaker's on-axis and off-axis responses in the end user's room, measured around listening spot with long gating and smoothing, then averaging. In most "normal" sized rooms it happens naturally or even in exaggerated form... Listening to music with two speakers for a few days is more important than "curve" fitting based on meaasurements. Measurements will help to find appropriate eq settings to get best sound from basically flat and smooth loudspeakers.
Last edited:
... which is a lot. I think the harman curve is at listening position and the straight FR you want to correct , which I assume is straight/flat, must be taken close to the speaker (1m) / near field. Don't forget to think about used gating for the input to EQ in the two different locations... its a pitfall...
So you EQ to flat from near filed measurement. The, if you take this EQ and move the mic back to LP listening position - you will see HF drop - so remember that the HK curve is at LP so even if the HK curve dips 5 dB in the top, it doesn't necessarily mean that you should apply a 5 dB drop to the HF... Measuring is knowing 😉 - with the reservation that measuring and interpretation is hard...
//
In terms of where you are at, don't worry about this until the speakers drivers are all working correctly first. When it's time to adjust for a house curve it will be easy with the right software to help you get the right biquad values. You really can ignore this problem until late in your process.
Do your filter design and EQ each driver first. Make sure to solve individual driver EQ problems and driver integration issues before applying global EQ. That is, if you have a hump or valley where two drivers meet, solve that by fixing your high and low pass filter settings. Don't just set filter parameters with your eyes closed and expect global EQ to fix your issues well. Once you've done that it's easy to apply a tilt.
Then there's several types of software to help you. I use OmniMic and it's super easy to use optimizer, but VituixCAD also has similar tools, including a tilt tool (no optimizer needed) which you can adjust and see the results, and I'm sure REW has a similar thing.
Generally the process is to take the FR at the listening location and import that into Omni/Vituix/REW. In Omnimic they have an optimizer. In Vituix there's a tilt filter. Adjust until the output looks right to you. Then export the biquad settings to Hypex.
In VituixCAD you can set the correct DSP type in Options, and then when you configure the level tool export the biquad coefficients.
I should point out that generally speaking Vituix is a pretty good tool for setting everything. That is, import your raw driver FR into Vituix and you can then configure all your filter settings from there for your quasi anechoic design.
Last edited: