FIR for hypex fusion plate!

[Reibekaese]

I have no idea. Ha! Does anyone on here know?

The regular filters in the fusion amps are biquads, so these are regular minimum phase filters.
As per the manual for HFD, the FIR-filters have to be created in third party software. This means they can be whatever you want, linear phase, minimum phase or any kind of excess phase. You need to define your desired filter transfer function and transform it to an (appropriately windowed) impulse response in a suitable format and with a sample rate of 93.75 kHz.

 

[tblb]

in rephase, it is possible to make both, I tried both, but not sure what filter type is best.
I bought a umik, I will make mesurements of my drivers.
Is there some tutorial to make good drivers measurement (with no anechoic chamber) ?

 

[Reibekaese]

in rephase, it is possible to make both, I tried both, but not sure what filter type is best.

The appeal of FIR-filters is that you can easily create linear phase filters, or in general just manipulate amplitude and phase independently, while for minimum phase filters both are linked. Loudspeaker drivers are minimum phase devices, so if you equalize their amplitude response with minimum phase filters, you will automatically linearize their phase response, so that would be the recommended way to go. By the way, I would implement the minimum phase part of the filtering with the biquad (IIR) filters in HFD.
For the crossover it's different, here the minimum phase filters will introduce phase shift, while linear phase filters avoid that. However, it's not easy to decide which is better. Minimum phase will introduce phase shifts and thus a 'bad' looking impulse response. BUT while FIR-filters can be better from a signal theory standpoint and produce a better looking IR, they don't necessarily sound better. At least there's lots of discussion going on about that on this forum and in general.
Problem with FIR is, that they introduce pre ringing which is not masked as much by the human hearing compared to the ringing of causal IIR filters. If that becomes audible is mainly a matter of your crossover slopes but can be influenced by many things.

 

[tblb]

Thanks , I will try some combination but so far , iir crossover filter sound better to my ear .
I have to make mesurements to find what is the best cross over point and slope for bass mid and treble .
First time I go active , and it’s not as simple ….

 

[capslock]

So when did they start manufacturing FAs with the 5.x firmware? I suppose there was a hardware change, so older versions cannot be updated?

 

[nickb]

Hi there! I'm new to the Hypex Fusion amp and new to posting (but not reading) on ASR.

I am attempting to implement FIR crossover filters for the CSS audio 1TDX using Rephase and HFD (driver measurements were taken with REW). Why? Well I bought the fusion amps for my next project (aiming to do a project with a Purifi woofer and tweeter when that comes out). In the meantime I'm curious to compare the excellent passive crossover to an active version and to try FIR filtering and assess objective subjective sound quality differences.

A few questions I'm hoping someone more knowledgeable might be able to help with:
- Within Rephase, is it correct that I can only import one measurement at a time (e.g. either my woofer or tweeter measurement)?
- Can a 4th order filter be created by stacking 2 2nd order filters here like in HFD? E.g. low pass for woofer / high pass for tweeter
- If I need to do a gain correct to match the tweeter (higher sensitivity) to the woofer, should I do that here or just in HFD with channel gain?
- Set the impulse settings to 1500 taps and 93750hz according to the HFD manual, is there anything else I should do before generating the output filter file?

 

[ThatSoundsGood]

Hi there! I'm new to the Hypex Fusion amp and new to posting (but not reading) on ASR.

I am attempting to implement FIR crossover filters for the CSS audio 1TDX using Rephase and HFD (driver measurements were taken with REW). Why? Well I bought the fusion amps for my next project (aiming to do a project with a Purifi woofer and tweeter when that comes out). In the meantime I'm curious to compare the excellent passive crossover to an active version and to try FIR filtering and assess objective subjective sound quality differences.

Sounds fun! I have a passive build with the Purifi woofer and a different passive build with the CSS LDX25 tweeter so I'm interested in your results given my familiarity with these drivers.

A few questions I'm hoping someone more knowledgeable might be able to help with:
- Within Rephase, is it correct that I can only import one measurement at a time (e.g. either my woofer or tweeter measurement)?

Yes

- Can a 4th order filter be created by stacking 2 2nd order filters here like in HFD? E.g. low pass for woofer / high pass for tweeter

If I understand what you're saying, the answer is yes. The way to make a 4th order active filter in the hypex (non-FIR) is to stack two 2nd order filters. The Hypex manual explains this. If you're talking about making a 4th order FIR filter then you will do that in REPhase.

- If I need to do a gain correct to match the tweeter (higher sensitivity) to the woofer, should I do that here or just in HFD with channel gain?

I would adjust the gain in the Hypex amplifier settings.

- Set the impulse settings to 1500 taps and 93750hz according to the HFD manual, is there anything else I should do before generating the output filter file?

I think that's it from my recollection. Do some tests to try it and make sure it imports into the Hypex. You can visually see the filter once it's imported.

View attachment 406220

 

[nickb]

Thanks a ton! I implemented the FIR low- and high pass filters for the woofer and tweeter respectively, and adjusted the phase a bit more with the filter linearization (vented box adjustment for the woofer) and paragraphic phase EQ to smooth out some dips and peaks.

After this I measured the impulse response and attempted to align them, delaying the tweeter 0.35ms in HFD based on what I see in REW (Impulse response view). This caused a 360 degree shift in phase around the crossover frequency. Is there a way to avoid this? What is the best way to measure time arrival differences between the woofer and tweeter? My tweeter is in a waveguide, so it sits farther back (visually it looks a bit back from the woofer, but it's hard to say by how much).

 

[Mister.Mole]

For delay measuring you need to measure the driver's separately without any XO filters.

Make sure to not kill your tweeter maybe so set the lower frequency accordingly.

You cannot simply overlay the impulse peaks.

Good explanation here:

 

[fluid]

phase a bit more with the filter linearization (vented box adjustment for the woofer)

Listen for yourself and make your own mind up but I do not like the sound of phase linearizing the system high pass from the woofers rolloff whether sealed or vented. It sounds "wrong" for want of a better way to describe it. Having the phase resemble the minimum phase for the frequency response is what I prefer in listening. Removing excess phase. It is normal for phase to increase as frequency rolls off, changing this relationship to have phase be flat while frequency rolls off, does not seem or sound like a beneficial correction to me.

 

[Abaxial]

I just programmed my Hypex FA253 with FIR filters and I'm absolutely blown away. I used Rephase. The filters are all 48db or 36db an octave in a 3-way box and the phase is nearly perfect. While the software and process are a bit cumbersome, the effectiveness is top notch.

Bit of a bump, but considering a Hypex build myself. How much overall latency did you end up adding to the system?

 

[Busdriver02]

8ms for per channel FIR
24ms delay for output FIR

 

[Keith_W]

Listen for yourself and make your own mind up but I do not like the sound of phase linearizing the system high pass from the woofers rolloff whether sealed or vented. It sounds "wrong" for want of a better way to describe it. Having the phase resemble the minimum phase for the frequency response is what I prefer in listening. Removing excess phase. It is normal for phase to increase as frequency rolls off, changing this relationship to have phase be flat while frequency rolls off, does not seem or sound like a beneficial correction to me.

It is normal in loudspeakers for phase to change as frequency rolls off. But it is not normal in daily hearing.

 

[ThatSoundsGood]

Bit of a bump, but considering a Hypex build myself. How much overall latency did you end up adding to the system?

I never measured it. Why does it matter to you? Are you using them with a television? Are you playing ive instruments through them?

 

[SDC]

I never measured it. Why does it matter to you? Are you using them with a television? Are you playing ive instruments through them?

I don't know his use case but many studio monitors do have switch to turn fir filters off for minimal delay.

24ms is not that bad for what it can do, but still a lot when they try to trim every ms possible.

 

[SDC]

I don't know his use case but many studio monitors do have switch to turn fir filters off for minimal delay.

24ms is not that bad for what it can do, but still a lot when they try to trim every ms possible.

Also Fusion Plate amp is used for many custom studio monitors for it's simplicity and performance.

So knowing the delay and having option to switch between latencies is good feature I guess?

 

[Mister.Mole]

There is no way to have FIR filtering without a delay.

 

[Keith_W]

There is no way to have FIR filtering without a delay.

You can use minimum-phase FIR filtering.

 

[Mister.Mole]

Less delay but still delay.

You can also work with lower sampling rate in the bass region to minimize the amount of taps needed. (Unfortunately not with HFD).

 

[Abaxial]

I never measured it. Why does it matter to you? Are you using them with a television? Are you playing ive instruments through them?

Guitar, yeah. I know you can switch it on and off, but was wondering if it could be made low enough to not make toggling back and forth necessary. Also considering Neumann (appear to be the winners on low-latency linear phase) or Genelec, but I could maybe get more bass + save some money going DIY