TOPIC:

I'm looking at a SHD unit to potentially replace my preamp and digital crossover. Can anyone tell me if the following settings can be done in the unit (as well as running Dirac)
My speakers are fully active two-way (bass units plus panels)

Stereo bass unit crossover:
LP Frequency: 172Hz (linkwitz riley 48 dB/oct)
HP Frequency: 23 Hz (butterworth 48 dB/Oct)
Bass EQ: Low Shelf. Frequency 30Hz. Gain 10dB, slope 8.4

Stereo treble panels crossover
HP Frequency: 172Hz (linkwitz riley 48 dB/oct)
LP Frequency: 20 kHz (butterworth 48 dB/Oct)
Treble EQ: Low shelf. Frequency 357 Hz. Gain +18dB. Slope 6
Delay: 0.29ms

thanks in advance
 

I think there is a limit to the amount of gain you can add. I think the limit is 10 or 12db. 

Looking at the manual I think Dirac is limited to 12dB but PEQ can go up to +16For the PEAK and SUB_EQ filter types, this is the gain in dB at the center frequency. For the HIGH_SHELF and LOW_SHELF filter types, this is the gain in dB reached at high or low frequencies respectively. A filter has no effect if its gain is set to 0 dB. Gain can be adjusted in increments of 0.1 dB up to +/- 16 dB.

Which is closer but unfortunately not enough

Can I do it via biquads?
And if so, is it possible to use the "standard" crossover functionality for everything else and just biquad for the Low Shelf? Or would everything need to be done via biquads?
 

You might run into clipping if you add gain and Dirac adds more?

My crossover requires that boost to negate dipole cancellation and have the speakers run flat - so I'm ok with that. The question is whether I can 'bypass' the 16dB boost limit if I use biquads 

As for Dirac, I assume it's possible to limit the largest boost it will apply (is that correct ?- sorry for newb questions here it's been years since I used Dirac)

Dirac just does its thing, no control over how much boost it applies directly although you can do things like set your reference line along the bottom of the troughs so that Dirac only ever cuts frequencies and never boosts them.

As for 18dB shelves. Perhaps you could split it into a boost below the frequency and a cut above the frequency. It would takes some thought but 18dB divided by two is well within the 16dB range of the SHD EQs. A 16dB boost below and a 2dB cut above would be closest logically to what you want and equivalent in practice.

The other way of thinking about Dirac is that you don't need to do any of the amplituide balancing you do with traditional crossovers. Just do the required frequency selectivity and allow Dirac to balance the amplitude, which is basically what its for. What Dirac can't do is control your drive units separately so the crossover reduces to just being used for splitting frequencies between drive units. I hope that makes sense, it often appears insane to folk used to using traditional crossvers for frequency selectivity and amplitude balancing. The only real downside to the "let Dirac sort it out" approach is that the SHD with Dirac turned off won't sound great because the amplitude balancing isn't being done by Dirac. In practice does anybody ever run with Dirac off? I certainly can't, it sounds unlistenable without it.

entripy wrote:

Dirac just does its thing, no control over how much boost it applies directly although you can do things like set your reference line along the bottom of the troughs so that Dirac only ever cuts frequencies and never boosts them.

As for 18dB shelves. Perhaps you could split it into a boost below the frequency and a cut above the frequency. It would takes some thought but 18dB divided by two is well within the 16dB range of the SHD EQs. A 16dB boost below and a 2dB cut above would be closest logically to what you want and equivalent in practice.



 

I'm a bit confused here. Do you mean that I could apply the following:
- a flat reduction in gain of -2dB from 357 Hz and up
- a low-shelf increase in gain of +16dB from 357 and below?

entripy wrote:

The other way of thinking about Dirac is that you don't need to do any of the amplituide balancing you do with traditional crossovers. Just do the required frequency selectivity and allow Dirac to balance the amplitude, which is basically what its for. What Dirac can't do is control your drive units separately so the crossover reduces to just being used for splitting frequencies between drive units. I hope that makes sense, it often appears insane to folk used to using traditional crossvers for frequency selectivity and amplitude balancing. The only real downside to the "let Dirac sort it out" approach is that the SHD with Dirac turned off won't sound great because the amplitude balancing isn't being done by Dirac. In practice does anybody ever run with Dirac off? I certainly can't, it sounds unlistenable without it.

This is interesting. By amplitude balancing, do you mean the overall gain of once crossover relative to the other (ie shifting all the bass (or treble) frequencies up or down by the same amount?)
If so, my bass amps and treble amps are the same brand so their gain is equal. I only change the relative gain by a small amount sometimes if I feel like a bit more or less bass. In any event, I'd be happy for Dirac to sort this out automatically

I've never used any SHD EQ so I'm not sure exactly what is available. I would guess perhaps a high shelf that aligns with the low shelf to create an overall 18dB gain change split across the two shelves. It works in theory, it should work in practice but it could be interesting to set up the frequency of the high shelf so that it joins up neatly with the low shelf.

By amplitude balancing I mean anything you would normally do to balance the levels at different frequencies. Such as your low shelf to compensate for the dipole effect. So yes, Dirac is able to compensate for the dipole effect, probably more accurately than you would ever be bothered to measure manually or calculate theoretically. You said you do it to achieve a flat response, but that's Dirac's only job really, to flatten the response.

By distinguishing between amplitude balancing and frequency selectivity I'm trying to be be explicit about the two different aspects that traditional crossovers normally achieve. With Dirac only the frequency selectivity needs to be achieved with the crossovers. e.g crossover low to high at 357Hz or whatever. All you need to do is spread the frequencies between the drive units so that they (just about) join up seamlessly. Yes, even if you crossover really badly with a notch or peak at the crossover frequency Dirac will correct the error (within its limitations).

Then in Dirac you choose the reference you require and Dirac does whatever it needs to to achieve that reference. Compared to the traditional methods its is far quicker and normally enormously better than was typically achieved manually. Some purists still like to set their non-Dirac sound up as flat as possible and give Dirac less work to do, but you no longer have to do that. Its always worth trying this simple approach to begin with.

In one of my systems I don't even high pass the mains, they share frequency range with the sub. It sounds boomy without Dirac but by the time Dirac has sorted it out the mains amps use less power (or go louder).

entripy wrote:

The other way of thinking about Dirac is that you don't need to do any of the amplituide balancing you do with traditional crossovers. Just do the required frequency selectivity and allow Dirac to balance the amplitude, which is basically what its for. What Dirac can't do is control your drive units separately so the crossover reduces to just being used for splitting frequencies between drive units. I hope that makes sense, it often appears insane to folk used to using traditional crossvers for frequency selectivity and amplitude balancing. The only real downside to the "let Dirac sort it out" approach is that the SHD with Dirac turned off won't sound great because the amplitude balancing isn't being done by Dirac. In practice does anybody ever run with Dirac off? I certainly can't, it sounds unlistenable without it.

I'm sorry but this is a really bad way to do things.

Dirac cannot control the relative output of two drivers at any point in the crossover region - it just scales the sum of the two. It is massively sub-optimal to not get the individual driver levels a least roughly matched prior to running Dirac, since otherwise you can end up with one driver operating too far outside of it's optimal operating range. You want equal contribution from each driver at the crossover frequency and it is impossible for Dirac to achieve this.

@Sir Sanders Zingmore the maximum boost Dirac applies at any frequency is 10 dB. As @Entripy correctly said you can't directly limit this by just entering the maximum you like, but you can influence it by where you set the target curve. Not ideal but a control of sorts.

I'm not sure what the maximum boosts you can apply in the plugin are, but it's worth noting that you can apply two shelf filters of the same type to achieve higher gains. I'll try to check for you when I get chance but I don't have my SHD set up right now due to redecorating. I do think the concern above about digital clipping may be an issue though, unless you apply overall attenuation to compensate.

Well I was assuming that the drivers would be approxmately balanced just for sanity. I'm certainly not suggesting you have drivers giving out 12dB levels differences and don't bother to do that basic level match. But that's pretty much the only thing you need to do. Beyond that Dirac can address all the other issues you may want to mess around with and would have to with traditional crossover systems. Its hard to imagine anybody would start with a system where the drivers weren't close to being matched in level, but yes, to the extent that I implied you could do that, you shouldn't.

Good idea about cascading two shelves.

You can definitely boost more than 16 dB using a single biquad. I think the bigger question is how were your filter parameters developed and do they match miniDSP convention?

For example on your shelf filters is the frequency specified the center frequency of the shelf or something else? I assume that slope is the shelf slope in dB/octave? For either calculating biquads or using the miniDSP standard filters you will need to specify the filter parameters in center frequency and Q.

Michael

Hi @mdsimon2,

The filters/crossovers are designed by the speaker manufacturer (Sanders electrostatics). They come programmed into the Venu360 crossover that ships with the speakers

From what I can tell, the frequency specified is the frequency at which half the gain is achieved and I think the slope is as you specify

I'm trying to add a screenshot of the filter but I don't seem to be able to add images?

imgur.com/a/kKQPB51

Managed to insert an imgur link but I can't for the life of me work out how to post an actual image


 

If you have the ability your best bet would probably be to measure the Venu360 and make sure the SHD response matches. That being said I dug up some references I previously found relating 2nd order shelf filter slope in dB/octave to gain and Q. 

www.musicdsp.org/en/latest/Filters/197-r...dio-eq-cookbook.html
webaudio.github.io/Audio-EQ-Cookbook/audio-eq-cookbook.html

I also tried implementing your desired filters using biquads in the SHD and I was able to measure the shelf mid point slope and confirm it matches what you think you want.

Here is 357 Hz shelf example.

f0 = 357 Hz
s = 6 dB/octave
dBgain = 6 dB
Fs = 96000 Hz

A = 10^(dBgain / 40) = 10^(6 / 40) = 2.818
w0 = 2 * pi * f0 / Fs = 2 * pi * 357 / 96000 = 0.0234
S = (s * log(10) * sin(w0) * (A^2 +1)) / (40* log(2) * w0 * abs(A^2 -1))
S = (6 * log(10) *sin(0.0234) * (2.818^2 + 1)) / (40 * log(2) * 0.0234 * abs(2.818^2 - 1)) = 0.642
Q = 1 / sqrt((A + 1 / A) * (1 / S - 1) + 2)
Q = 1 / sqrt((2.818 + 1 / 2.818) * (1 / 0.642 - 1) + 1) = 0.515

Resulting biquad in miniDSP format is:

biquad1,
b0=1.02457523555732388232,
b1=-1.97248045370098168405,
b2=0.94942346728247417609,
a1=1.97314401014927520706,
a2=-0.97333514639150442438,

drive.google.com/file/d/1UIZJQxkCPRWQFIR...0Oq/view?usp=sharing

I also did the same procedure for the 30 Hz shelf filter, the resulting Q is 0.844. Biquad is:

biquad1,
b0=1.00067945908548061595,
b1=-1.99825244667383183916,
b2=0.99757983743992728609,
a1=1.99825478854298910214,
a2=-0.99825695465625063907,

drive.google.com/file/d/1EL_NZYCS34h_EIB...hkT/view?usp=sharing

Michael

See below for total system response from the SHD at -12 dB volume for your specified filters. As mentioned by others previously you have about 12 dB of boost in this configuration and need to worry about digital clipping. Probably best if you use the SHD as volume control as opposed to downstream volume control as it will give you the most flexibility. To confirm, there are no other overall level adjustments between the bass and treble channels?

drive.google.com/file/d/1fiXJ2AZT_VyBC0Z...53P/view?usp=sharing

Michael

mdsimon2 wrote:

See below for total system response from the SHD at -12 dB volume for your specified filters. As mentioned by others previously you have about 12 dB of boost in this configuration and need to worry about digital clipping. Probably best if you use the SHD as volume control as opposed to downstream volume control as it will give you the most flexibility. To confirm, there are no other overall level adjustments between the bass and treble channels?

drive.google.com/file/d/1fiXJ2AZT_VyBC0Z...53P/view?usp=sharing

Michael

 

My plan is definitely to use the SHD as volume control. One of the benefits as I see it is that I can replace both my preamp and my Venu360 crossover with a single box

Regarding other overall adjustments, in my current setup I sometimes boost or lower the bass crossover by 1 or 2 dB depending on my mood/what Im listening to.To be clear that's an equal shift of the entire crossover region up or down