TOPIC:

I just got a miniDSP SHD and I have the following noobie question.

In thinking without knowing all the details a thought occurred to me that DIRAC achieves linear phase crossover. Is that correct.?

So does DIRAC make the effort of configuring a crossover with the rephase tool in order to achieve a FIR crossover redundant and unnecessary?

In short, does DIRAC achieve automatically what I could achieve manually using the Rephase FIR Tool?

In other words, would it be faster and easier to set up IIR high and low pass filters and let DIRAC do the work that the rephase tool achieves for FIR filters?

Or have I not understood what DIRAC does? I have understood that it is much more than just EQ.

I am creating an active two way system crossing from a woofer to a tweeter, rather than a sub-satellite system.

Thank you for your insights.

You can't actually do FIR crossovers with the SHD, so you use IIR and rely on Dirac Live to unwind the phase. I'm not sure I've checked the actual phase but you can see it in the impulse response. The crossover is still limited to the slopes/shapes that you get with IIR, whereas if you were designing FIR filters with rephase you have more control over that..

Thank you for your answer.

So you seem to be suggesting my thought was partly correct: that DIRAC achieves linear phase crossover (automatically and with less effort)

So my process will be to measure and set up IIR crossovers and then let dirac do the rest?

Yes

BTW and OT does your handle refer to tubes?

I’ve had excellent results using both


2x4hds as a dac and fir crossover and
Ddrc22d upstream.


Although Dirac will un-twist the phase in the sum of the responses from all the drivers in a multi-way, as John noted. I’ve really been liking a linear phase crossovers with Dirac better as the entirety of all responses are linear phase to begin with.

Than Dirac can focus on the room and not the speaker and do what it does best.

IMHO, it works better this way, however Dirac by itself is still very good, so don’t think for a second its not good without a linear phase crossover. The sum is what you hear anyways, so the end result is mostly the same, I’ve found with a rephase crossover there is virtually no driver interaction issues in any of the crossover stop bands which could be a result if you listen off axis at all or with uneven path lengths or if the way your crossovers behave with how Dirac moves the phase (maybe too steep a slope or high angles) Your crossovers May sound good and the change Dirac makes to the sum may not be how the two drivers or more want to play together with the correction applied. With linear phase it’s a non issue.

Hope that helps, if you have the means , both is very good.

Andrew

I currently use REW and rePhase to generate filters for use in Roon (running on a Mac Mini).

I am thinking of buying an SHD.

The current SHD product brief says that up to 4 FIR filters can be stored.

If so there has presumably been an update since the original post saying that FIR was not available - is that correct?

The product brief doesn't mention the maximum number of taps available for FIR filters - is this information available?

miket wrote:

I currently use REW and rePhase to generate filters for use in Roon (running on a Mac Mini).

I am thinking of buying an SHD.

The current SHD product brief says that up to 4 FIR filters can be stored.

If so there has presumably been an update since the original post saying that FIR was not available - is that correct?

The product brief doesn't mention the maximum number of taps available for FIR filters - is this information available?

I can't see a way to add FIR filters using the latest SHD plugin. I also have a 2x4 HD so know how to do this but the equivalent capability on the SHD is reserved exclusively for Dirac Live filters as far as I'm aware.

Edit: just checking the Product Brief I believe the statement is technically correct, it's just that the only way to enter FIR filters is via Dirac Live.

Thanks. Useful to know.

Interesting that current product brief says "FIR Filter Storage - Up to 4 filter configuration files stored on the unit".

Looking at the manual there's a reference to importing PEQ (IIR) filters exported from REW.

I've put in a support request to try to get clarification.

miket wrote:

Thanks. Useful to know.

Interesting that current product brief says "FIR Filter Storage - Up to 4 filter configuration files stored on the unit".

Looking at the manual there's a reference to importing PEQ (IIR) filters exported from REW.

I've put in a support request to try to get clarification.

Yes, you can add up to 10 biquad filters per output using the PEQ option, and these can be as produced by for example REW. I've done exactly this in the past. This can be in addition to Dirac's filters, and similarly crossovers, delays, and gain adjustments can all be made as well. I make use of these features to integrate my subwoofer prior to running Dirac.

I'm certain you can't manually enter FIR filters, and if you could then you wouldn't be able to use Dirac as well.

Meshplate wrote:

Thank you for your answer.

So you seem to be suggesting my thought was partly correct: that DIRAC achieves linear phase crossover (automatically and with less effort)

So my process will be to measure and set up IIR crossovers and then let dirac do the rest?

So Dirac makes the left and right impulses match as much as possible within its algorithm and power and mostly the response between left and right and the time domain characteristics. It makes a decision that will not loose sound quality at what comes at it , and believe me in a car (for me) it’s a mess when it comes to crossover slope shift and Dirac

I run 3 2x4hds with rephase crossovers linear phase for a 4way
And then have a upstream ddrc22 for Dirac that seems to be the cats meow for correction

I’ve run Dirac without the linear crossovers and its still good it will make the phase the same for both channels , but it’s not linear phase always, it’s the same transfer function for left and right that targets the Dirac impulse, so yeah some phase smoothing will be used but nit directly looking at “linear phase” as much as a clean impulse


To achieve linear phase in crossovers use an fir at each driver and make the crossover, then run Dirac with another box upstream

Best of luck
Andrew