Welcome, Guest
Username: Password: Remember me

NOTE: This is a "Community" forum. Please be mindful that community members are here to help as part of a community effort. We therefore appreciate your effort in keeping this forum a happy place!

If you have a specific issue (e.g. hardware, failure) and want help from our support team, please use our tech support portal (Support menu - > Contact Us).
Thanks a lot of your help in making a better community.
  • Page:
  • 1
  • 2

TOPIC:

OpenDRC-DI with Acourate 6 years 4 months ago #19290

  • Draki
  • Draki's Avatar Topic Author
  • Offline
  • Junior Member
  • Junior Member
  • Posts: 23
  • Thank you received: 4
Have DRC-DI and Acourate for several months now, and it is working as expected.

My only quibble is that the correction filter generated with the option "FIR+14IIR Full range" is not working, at least not correctly. It switches off and on, sometimes only one channel is working and the other is muted. As if it is trying too hard and can't make it.

In practice I am using the "FIR+6IIR,keep all crossovers" filter type, with the filter control set to "0" and the smoothing at the middle - the default settings. This,I assume, is less demanding on the processor, and this way all is fine.

But I am curious if this behavior is typical or I have a problem with the DRC-DI? Does anybody else have a similar experience? Maybe the DevTeam could comment?

Thanks.
Draki

Please Log in or Create an account to join the conversation.

Last edit: by Draki.

OpenDRC-DI with Acourate 6 years 4 months ago #19307

  • devteam
  • devteam's Avatar
  • Away
  • Administrator
  • Administrator
  • Posts: 8965
  • Thank you received: 1291
@ Draki,

Our guess (based on what we've seen from few filters by the latest AcourateDRC) is that something isn't right in the filter generation making the DSP just clip internally. In the miniSHARC, when we are above 0dBFS, we just mute..

You'd want to talk to AudioVero about it because that's their issue ultimately since they are the one generating the filters. We've seen few filters with some odd +20dB gain inside.. easy to overload the DSP in that case.

You might sometimes be able to get the signal back (e.g. attenuate at the input the signal) as a trick, but ultimately, fixing the filter generation would be the step #1.
Check the PEQ section and you'll see what we mean.

Devteam
miniDSP, building a DSP community one board at a time.

Please Log in or Create an account to join the conversation.

OpenDRC-DI with Acourate 6 years 4 months ago #19314

  • uli.brueggemann
  • uli.brueggemann's Avatar
  • Offline
  • New Member
  • New Member
  • Posts: 9
  • Thank you received: 7
It can happen under certain conditions that cascaded IIR do not behave numerically stable.
One factor is also the applied numeric resolution.
Right now I can report that inspecting the filters with Acourate is always stable, maybe because of the 64bit floating point calculations. Btw Acourate also orders the sequence of poles and zeros to get a stable behaviour.

Unfortunately it seems that there are some restrictions with OpenDRC. As e.g. the frequency plots only cover parts of the IIR filters and are even limited in y-axis it is difficult to check what happens inside OpenDRC. If some algorithm also applies clipping and muting in between the output signal is a result of unknown factors. So e.g. a 32 bit float number is not limited to max. + 20 dB. And all IIR filters together do not exceed +-20 dB. So why is this a problem?

Right now the only chance is to reduce the number of filters.

Maybe a different approach to get IIR filters promises more stability. I will check if an approximation of a given correction curve by standard biquad peaking filters gives a more stable result.

- Uli
AudioVero
The following user(s) said Thank You: Draki

Please Log in or Create an account to join the conversation.

OpenDRC-DI with Acourate 6 years 4 months ago #19345

  • devteam
  • devteam's Avatar
  • Away
  • Administrator
  • Administrator
  • Posts: 8965
  • Thank you received: 1291
@ Uli,

Thanks for the feedback!

What kind of correction you're trying to apply here with these 20dB (and upward) filters? Besides potential damage to the speaker (that's a whole lot of boost to apply, no? :-), in a DSP 32bit isn't the limitation you're correct. There is an output card at one point of time in the chain that will clip at 0dBFS. Agree? So let's say you input -10dBFS into the Audio unit, then apply 20dB, how do you come out to the IO card that's capped at 0dBFS for 24bit resolution?? :-)

That's what we're talking about here and one of the reason why we don't think that showing a PEQ section with more than +/-20dB makes sense. There are indeed cases (hard to reproduce but we know they are there) where the Sharc (or IO card) "will" mute the Audio instead of outputting crazy noises (e.g. above 0dBFS) and damage the speakers. That's the most likely-hood here. I'd recommend that you apply a "maximum level of boost" in your algorithm to account that a DSP need to deal with internal headroom or for example control a level at the input to include some headroom, that's what other software would do typically.. Anyway, our 2cts just trying to help. :-)

DevTeam
miniDSP, building a DSP community one board at a time.

Please Log in or Create an account to join the conversation.

OpenDRC-DI with Acourate 6 years 4 months ago #19358

  • uli.brueggemann
  • uli.brueggemann's Avatar
  • Offline
  • New Member
  • New Member
  • Posts: 9
  • Thank you received: 7
Let me try to explain:
a given signal gets described by 14 IIR filters and 1 FIR filter (6144 taps). The IIR filters are the result of solving an equation system. See the example at www.audiovero.de/freedownload/Amplitude.png (btw. I cannot add an attachment here or preview my answer). The given signal in red color gets replaced by the 14 IIR filters (green color) and the FIR filter (brown color).
You can see that the IIR filters mainly concentrate on the lower frequencies (where the FIR filter is weak due to the few taps only). You can also see that all IIR filters together result in a curve <= 0 dBFS.
Loading the filters into the OpenDRC plugin results in a display of the PEQ filter section www.audiovero.de/freedownload/PEQ.png and of the Xover filter section www.audiovero.de/freedownload/Xover.png. As you can see the Xover display does not show the 8 IIR filters correctly.
Anyway the 6 PEQ and 8 Xover filters together do not exceed 0 dbFS ! But with real applications it may haven that a single IIR filter has such a behavior. This will be compensated by another IIR filter in the cascade of course.
So at the end in combination with floating point computations the filters should behave well. I can only imagine that because of the cascading some errors in calculation add up and thus lead to an overflow.

- Uli

Please Log in or Create an account to join the conversation.

OpenDRC-DI with Acourate 5 years 6 months ago #24163

  • Mitch
  • Mitch's Avatar
  • Offline
  • New Member
  • New Member
  • Posts: 7
  • Thank you received: 4
Hi all, I am having the same issue as described above: very good frequency response correction and better stereo image thanks to acourate calculation, but loud cracks when switching preset, sometimes one side disappears even at low level, and sometimes a loud white noise when switching on the unit.
This happens booth in Analog units and digital units so it is not hardware related.
Uli answered kindly to this that I should try to change the parameter called "excess phase" in the the filter generation macro window (i am using Acourate pro and not Acourate Drc, where this parameter has to be changed inside an xml file ).
It behaves better now that I used excess phase 2/2 instead of the default 2/3 settings, but I stil have sometimes instabillity.

M'y guess si that using only 6IIR is the best Way of having a stable filter, but the basses are not as good, full and Acurate, so it is a Pitty!

Maybe having only one preset loaded instead of 4 improoves stability in my experience, but I am not sure.

I would have loved to tweak concerts sound systems with Acourate and Drc in minimum phase but thèse instabillities make this impossible as you can imagine !
A workaround would be to have one Drc for Fir and a second one for IiR or even the IIR of the foh digital console ...
But where could i get the correct IIR settings to copy in my eq since these are biquads values and not normal parametric values....

Of course of some kind of update coule solve this I would be a happy man ! :woohoo:

Please Log in or Create an account to join the conversation.

Last edit: by Mitch.

OpenDRC-DI with Acourate 5 years 6 months ago #24228

  • devteam
  • devteam's Avatar
  • Away
  • Administrator
  • Administrator
  • Posts: 8965
  • Thank you received: 1291
@ Mitch

Thanks for the follow up.

Feel to send us the XML file so we can have a quick look at it. As we've mentioned before, we have no control over the types of filter generated here. Interestingly enough, we know that based on other implementations, REW (biquad only) or rephase (FIR + IIR), we never run into these issues of instability... So something obviously down to what's being fed as biquad filters I'm guessing. If you go FIR only (i.e. don't copy over the IIR manually), is the issue still here?

Devteam
miniDSP, building a DSP community one board at a time.
The following user(s) said Thank You: Mitch

Please Log in or Create an account to join the conversation.

OpenDRC-DI with Acourate 5 years 6 months ago #24249

  • uli.brueggemann
  • uli.brueggemann's Avatar
  • Offline
  • New Member
  • New Member
  • Posts: 9
  • Thank you received: 7
"Simple" biquad filters like calculated by REW seem to be more safe. Acourate is using the solution of a matrix equation and indeed the filters can also behave unstable. To overcome this Acourate re-orders the poles and zeros. But it is still possible, that under given circumstances the IIR filters still behave unstable. So maybe I need to step back and to calculate also the "standard" biquad peaking filters.
Unfortunately the display of the OpenDRC plugin is also not very helpful with the given axis limitations. So it is difficult to find the reason = which condition is exactly causing the non-stability. The OüpenDRC is like a closed box here. With the Acourate environment the IIR filters have never been found to be unstable but with 64 bit accuracy there are less problems anyway.

A quick fix could be the usage of REW biquads in combination with the Acourate FIR.

Uli
The following user(s) said Thank You: Draki, Mitch

Please Log in or Create an account to join the conversation.

OpenDRC-DI with Acourate 5 years 6 months ago #24251

  • Draki
  • Draki's Avatar Topic Author
  • Offline
  • Junior Member
  • Junior Member
  • Posts: 23
  • Thank you received: 4
About 10 month after the OP: IME in practice there is really no need for more than 6IIR + FIR. And in that setup the DRC-DI+Acourate works stable. YMMV.
The following user(s) said Thank You: Mitch

Please Log in or Create an account to join the conversation.

OpenDRC-DI with Acourate 5 years 6 months ago #24304

  • Mitch
  • Mitch's Avatar
  • Offline
  • New Member
  • New Member
  • Posts: 7
  • Thank you received: 4
Hello All !, Thank you for the replays ;-)

@Draki :
Happy to see that with 6IIR the problem is fixed for you.
I will use this setting when it is necessary to have a good stability, for live small events for instance
However, I have listened to the different settings (10IIR vs 6IIR vs 12IIR) and I find that for the studio where I am with a Genelec medium system the sound is much fuller ( basses) and clearer ( mids) when using all the available IIR.

With my colleague we are searching for the best possible sound in order to do a top mixing job.
Acourate with Mini dsp AN has been an incredible improvement after having tried 4 other FIR softwares,...
we are super happy with Acourate.
We use a target curve going down 6dB from 1khz up to 20khz (a straight line).
The measurements have to be well made of course...
We have strong standing waves issues and it solves almost all of it,so we have real basses.
Also the center image is improved, and the highs are much smoother than without the sound processing !

@Uli:
Happy to know that a convolver plugin is stable, however for me it is more complicated in a day to day job because we monitor through an analog console . And we really like the external processor option.
That's the reason why I am hoping to fix the instability with all the available IIR.
If it is possible for you to find a way to identify the IIR problem through a "standard biquad filtering" as you say, that could be the solution, we can't know that before having tried I guess?

For now I am trying two things:
1) only load one or two pre-sets per Open DRC to see if it works better.
In the past we had only one Acourate pre-set and three other pre-sets (REW) loaded and it was stable once started up.
We had anyway some annoying white noise issues on processor start up that needs to unplug and replug the power of OpenDrc.
The white noise issue has been there even before we tried Acourate in fact, ....

2) Have a left processor and a right processor: I load the XML on both machines and then deactivate everything one of the sides on each of them.(FIR + IIR)
The stereo image and the sound is as good as with one stereo machine, and when switching between two pre-sets , there is less pop noise, so maybe the OpenDRC has more processing headroom and a better stability or a better power distribution between the different parts of te machine like that...

@dev team: you are right that the Biquads filters given by Acourate are pretty extreme sometimes, but the sound is really good with them so they are at least well calculated even if maybe not stable.
I send you the XML file attached.
1) is full phase
2) is minimum phase
The instability (one side distorts and goes off) happens after sometime like half an hour, when having sound through the machine.
Probably the problem will be replicated at your place, but I didn't spent several days switching between seversl minidsp's and maybe some machines will behave better than others, difficult to be sure.
I only know that here two different machines with the same 4 filters had the same issue.
If you test on your side you could load all 4 pre-sets with the same XML to help replicate the issue.

I will come back to you if I find that having less pre-sets loaded or a right and a left machine sets the problem...
Attachments:

Please Log in or Create an account to join the conversation.

Last edit: by Mitch.

OpenDRC-DI with Acourate 5 years 6 months ago #24306

  • Mitch
  • Mitch's Avatar
  • Offline
  • New Member
  • New Member
  • Posts: 7
  • Thank you received: 4
@ DevTeam :
I would like to add a question to you:
is there a possibility that the problem is related to the power distribution inside of the Open DRC?
1) when powering up I get sometimes a small white noise while the machine makes it's powering routine and then it disappears
2) sometimes something goes wrong while powering up and the noise is super loud and doesn't stop until I restart the machine
3) When switching between pre-sets one side disappears or a loud crack is made by the analog output, as if the current supplied to the analog output electronics was interrupted by the pre-set changing.
4)I get more white noise issues when turning up all the studio at the same time with a switch than when waiting that everything is on and then turning up the Open DRC...
5) When getting instability with an Acourate filter, one side distorts and then turns off, like if the analog output of the Open Drc had not enough power supplied to work properly over the time.
Could it be that using the largest amount of processing available in the OpenDRC the power is not sufficient because the processor eats more power for it's calculation, ad then the analog output IC starves (growing audio distortion before one side turns of...) ?

The issue could maybe be setted by having a larger current supplied to the OpenDrc with an larger external power supply?
Or maybe my wall power outlet changes a little over time (230-235-240-225 volts) and it makes a difference in the power output of the transformer current...and then it drives the OpenDRC crazy ...
I this case should I use an uninterruptible power supply ?

Thank you for your help dear Dev Team ! B)

Please Log in or Create an account to join the conversation.

Last edit: by Mitch.

OpenDRC-DI with Acourate 5 years 6 months ago #24319

  • devteam
  • devteam's Avatar
  • Away
  • Administrator
  • Administrator
  • Posts: 8965
  • Thank you received: 1291
@ Mitch,

These issues of distortion are quite odd indeed. We're assuming that you're using the latest firmware + latest plugin for loading such filter?
If you're not sure, please load it as step#1. If it's still happening, we can investigate with the 2 files you've provided above.

Devteam
miniDSP, building a DSP community one board at a time.

Please Log in or Create an account to join the conversation.

OpenDRC-DI with Acourate 5 years 6 months ago #24322

  • Mitch
  • Mitch's Avatar
  • Offline
  • New Member
  • New Member
  • Posts: 7
  • Thank you received: 4
Hi Dev team, I can confirme that my plugin /firmware/adobeair are all up to date....
The problem was there before and after update...

I can confirm after using the XML on two different OpenDrc machines that the distortion and then mute of the sound happens only on the right side.
I tried using one right and one left OpenDRC with the same XML with only one side filtering enabled and the same distortion and mute happens...
So it is maybe a power of the analog issue in the analog output (filter consumes too much power) or maybe it is filter gain related.
1)I will plug a more powerful power supply instead of the standard to see the behaviour...
2)I will also generate the same filter with less gain (the faulty one has a 0dB gain, no more, no less volume than without a filter).

Thank you for testing the XML if you can understand what happens inside the guts of the machine... :blink:
Best regards
Mitch


Mitch

Please Log in or Create an account to join the conversation.

Last edit: by Mitch. Reason: more details

OpenDRC-DI with Acourate 5 years 5 months ago #24589

  • Mitch
  • Mitch's Avatar
  • Offline
  • New Member
  • New Member
  • Posts: 7
  • Thank you received: 4
Hello all, I found a way to make my mini dsp stable so I report here:
-I tried changing the power supply, still instability
-I tried having one left and one right mini dsp, still instability
-I changed the filter design in Acourate: More Excessphase (3_3), and preringing (1_1). And... BINGO it works now.

SO : the phase correction settings in Acourate will change the stability of the filter...
Best regards
Mitch
The following user(s) said Thank You: devteam

Please Log in or Create an account to join the conversation.

OpenDRC-DI with Acourate 5 years 5 months ago #24590

  • Draki
  • Draki's Avatar Topic Author
  • Offline
  • Junior Member
  • Junior Member
  • Posts: 23
  • Thank you received: 4
Hi Mitch

What Acourate software are you using: the "standard" or the "DRC"?

Please Log in or Create an account to join the conversation.

  • Page:
  • 1
  • 2
Moderators: devteam