Active DSP: Please post your software and equipment

I use Jriver in my system to convolve my DRC filters. This in no way could be considered a real time process. I use REW as a virtual check with A*B to see if I got what I was aiming for. Often I haven't so I start again. Measuring in real life is a total PITA for most people with families and restricted volumes. Virtual testing can be quite helpful.

Yes, that was one of my concern with active DSP. Based on some of the feedbacks from those who have done it using miniDSP, every time you make any change in the FIR filter, you have to measure again. As you said, it's definitely a PITA.

One nice thing about all analog xover is that I can use XSim offline to do xover fine tuning without having to measure each time I make a change in the xover. XSim can simulate any changes for me.

And based on what I've read so far, the real weakness of an active DSP setup is in the lack of an integrated simulation software.
 
Last edited:
That is not true it is just that most people have no idea how to use the available software.

VituixCAD can simulate and generate all sorts of filters for active and DSP systems, it can even output the impulse files for you. I suggest to download that and go through the comprehensive guides provided. Truly another gift to the diy community.

You can do the same thing manually in REW as I have explained.
 
"real weakness of an active DSP setup is in the lack of an integrated simulation software." Thats not true at all, Everything you do, requires technique...either you have good technique or you don't. Passive is old technology, sincerely, 2020. Even if you did a passive xover, after you've built it, you're not going to take live measurements?
 
Last edited:
Yes, that was one of my concern with active DSP. Based on some of the feedbacks from those who have done it using miniDSP, every time you make any change in the FIR filter, you have to measure again. As you said, it's definitely a PITA.

One nice thing about all analog xover is that I can use XSim offline to do xover fine tuning without having to measure each time I make a change in the xover. XSim can simulate any changes for me.

And based on what I've read so far, the real weakness of an active DSP setup is in the lack of an integrated simulation software.

Cant see any reason for why one of the filters types would require more measurement than the other really... - I dont think so.

//
 
If it is about seeing the changes within measurements in real time by applying a PEQ etc, REW can show that to you that in real time, just as well as X-sim does it.
If it is about a FIR file convolution, yes, that takes another couple of simple steps within REW after which you can still see the results.

But every mayor change, be it one made in X-sim or REW, should be confirmed with real time measurements to know if it really works the way you want it to work.

I'm guessing you won't be able to see real time changes in your measurements with X-sim, as you'd need to pull out the soldering iron. 🙂

With a measurement setup on standby, one could see changes in real time in measurements after applying a couple of PEQ's depending on the system of choice.

Don't make it harder than it is. This is new for you, so take the time to learn the ropes.
 
The reason in the broadest possible terms is that FIR filters are inherently finite in their extent and therefore, in most applications, an FIR filter is an approximation to a target response. By contrast an IIR filter can be exact (subject to sufficient processing accuracy), for example, if we desire to use an analogue filter as a target response in a crossover design.

Your particular choice of FIR filter, for example, might be more efficiently implemented using IIR blocks for the crossover components and even elements of the EQ, which would leave the phase alignment to a separate FIR filter or realised in the IIR filter zeros - but that distinction is in the end pedantry.

As to the the assertion that IIR filters are always minimum phase, that is not correct. If there are no zeros in the filter - that is, if the FIR part is not used, then the remaining all-pole IIR filter will be minimum phase. But the ability to add zeros in an IIR filter allows non-minimum phase responses too. (Remember also that a non-minimum phase response can be constructed from cascading a minimum phase filter and an all-pass filter, but I will ignore that at your request...)

But a minimum phase response is different from a linear phase response which is necessarily two-sided in time (acausal) and hence precludes the use of recursive IIR filters where the feedback necessitates its causal, one-sided response. A linear phase FIR filter will still be an approximation, however, since it is impossible to find a delay sufficiently long to encompass all of negative time.

With an abundance of digital processing power, there might appear to be little to encourage people to bother to understand matters more fully. But FIR filters are not a subset of IIR filters, and each has their different design compromises that are well worth taking the time to understand if maximum fidelity is the target.

Thanks for the nice reply 🙂

Ok, since no one likes my depiction of IIR as a subset of FIR capability I withdraw it, and use it no more...

I'll simply say my experience is that anything I've been able to do with IIR processing, I've been able to fully replicate with FIR processing, and that FIR processing has also allowed me to make further tuning refinements with its independent handling of phase and frequency.

That's my story and I'm sticking to it 😉
 
Not sure if this thread is still about which active DSP and S/W people are using...

Nevertheless I'll post my 2cents worth anyway, especially as it alludes to the IIR/FIR Discussion:

Since over 10 years I'm using a DEQX PDC-2.6p (there are much more modern reincarnations of this old item available!) to my continued and repeated satisfaction (yes, I have modified the power supply, the OpAmps and the Capacitors in the mean-time).

DEQXs come with a proprietrary software and measurement set, but all that works well right out of the box. The DEQX approach is

1. to address speaker linearity above a (room dependent) selectable Frequency with quasi-reflection-free measurent for FIR phase/level linearization Filters (using tunable smoothing or filtering aggressveness) and linear phase X-overs.
2. to address room resonance dampening with semi-automatic and/or manual IIR PEQ filters
3. IIR filters are also available to tune tonal balance to personal preference while playing music.

A PC is only needed to do the measurements and calculate the filters which are then uploaded into the DSP unit which is then a remote controlled dig. preamp and EQ with X-over.

Inputs to the DEQX unit are analog and digital, Outputs can be non-symmetric, balanced, coax digital or AES3-digital.

By the way: I do use REW to verify results and to also get ideas for manual equalization 😉

So far I don't see a reason for changing to more modern equipment 🙄😉😀
So I do still recommend DEQX to serious DSP users despite the significant investment (there's a small used items market, though).

Greetings and stay healthy!
Winfried
 
I use Jriver in my system to convolve my DRC filters. This in no way could be considered a real time process. I use REW as a virtual check with A*B to see if I got what I was aiming for. Often I haven't so I start again. Measuring in real life is a total PITA for most people with families and restricted volumes. Virtual testing can be quite helpful.

Here's how a real time process works for me (and my family 🙂)
And why I'm a fan of dual channel....

Dual channel typically uses pink noise which is much less annoying than sine sweeps. (Pink and white noise are often sounds available in sleep machines.)
And since dual channel allows time averaging, volume can be very low and still give accurate results. Like really low.
If that's still an annoyance, dual channel is source signal independent, allowing music instead of noise as the test signal. If the music isn't dense enough across the spectrum, just add a little inaudible pink into the mix.
(at least the dual channel I use, SmaartLive is source independent...I would think they all have to be, but I don't know that for sure)

If tuning with an IIR processor, I turn on the dual channel and adjust response real time, dialing in the number of available EQ's.
After each driver is done, I add xovers and move timings in real time to get adjoining phase traces to overlay.

If tuning with FIR, after the starting raw measurement has been corrected with a FIR file (from whatever software is being used...rephase alone, REW EQ/rephase, DRC, FirDesigner, Acourate, etc),
I turn on dual channel to see how well the FIR file driver corrections worked. Then I use the processor's additional IIR filters in real time to further refine the response.
It's the same process as I use with an IIR only processor, and it shortcuts iterative rebuilds of the FIR file.
When the additional IIRs can't fix the driver response in real time ....i think it means the problems are not minimum phase and not fixable with EQ.
Once I find a set of IIR EQs that improve the FIR file, I just redo the FIR file by adding them in.
All that said, often the starting FIR file is spot on, and there is no need for the additional IIR tuning stage. The only thing needed is to set timings and levels, which is incredibly easy real time.

Another cool things about real time dual channel is the ability to move the mic around while watching response. So you can see in real time changing floor reflections etc, as you move the mic. Or make moving mic averages if you care. It even has phase tracking (at least with Smaart), that essentially removes loopback time continually, so you can see accurate phase from a moving mic.
I've tuned bass reflex ports where the port could slide in length, by watching response change as the port length changed.
On the spinorama, it's helpful to get a picture on axis response, without having to make a whole bunch of measurements. Let's me move xover up or down and watch real time, to get in the right ballpark, before attempting final dial in.
Yep, i love dual channel 🙂

It should be noted that dual channel cannot make THD measurements like single channel REW can...other than looking at one frequency at at time.
That's when I ditch dual channel and go to REW....but that's always after all tuning has been finished... to see what distortion is...

Anyway, just what works for me...plenty of other valid methods out there....
 
Just a sidenote - REW can do RTA with pink noise too, with adjustable parameters. But that is not good for xo work.

However when one does multichannel xo etc, eq of each driver at least one octave beyond xo point is very important and beneficial!

And please do only one thing at the time!
 
What does this mean?
 

Attachments

  • rew loopback timing ref-vert.jpg
    rew loopback timing ref-vert.jpg
    504.2 KB · Views: 163
What does this mean?

That's either semi dual channel or full dual channel (depending on implementation). USB mic.s don't have this capability.

I knew it had that for discreet "sweeps", I just didn't think it had it for real-time measurements (..fft real-time is tough on the computer, and you can even see that problem mentioned in the video above).