Sample based delay lines with feedback

Dec 2, 2008 at 2:38am

Sample based delay lines with feedback

I’m trying to create a simple comb filter, and before I start, I don’t want to use [comb~].

I need to build a delay line with feedback. Delay needs to be as small as one sample. [delay~], as the help file correctly states, does not work in with feedback. Does anyone know of a delay external or other object that will allow this? Moving back and forth between Windows XP and OS X 10.4.

Thanks,
Brennon

#41136
Dec 2, 2008 at 7:57am

#146233
Dec 2, 2008 at 8:10am

On 02 Dec 2008, at 02:39, Brennon Bortz wrote:
>
> I’m trying to create a simple comb filter, and before I start, I
> don’t want to use [comb~].

maybe, before you start, you want to tell us why you don’t want to
use comb~?

#146234
Dec 2, 2008 at 10:19am

Feed delay back into itself by running a second cable from it’s output thru a [*~ 0.9] and back into itself.
adjust number for more or less feedback, beware, 1 means full feedback.

jrp

#146235
Dec 2, 2008 at 3:39pm

You might want to look at Eric Lyon’s stuff. I think he has a sample accurate delay called [vdb~ ].

http://www.sarc.qub.ac.uk/~elyon/LyonSoftware/MaxMSP/

His paper on this subject is at:

http://www.sarc.qub.ac.uk/~elyon/LyonPapers/SampleAccurate-Lyon-ICMC2006.pdf

Not sure it does everything you want but hope it helps out,
Mitch

#146236
Dec 2, 2008 at 7:09pm

#146237
Dec 2, 2008 at 7:31pm

Sorry Brennon, I did not look at your location (Belfast). I look like a bit of a dork, oh well.
Mitch

#146238
Dec 2, 2008 at 8:26pm

> Sorry Brennon, I did not look at your location (Belfast). I look like
> a bit of a dork, oh well.
> Mitch

No worries! I look like the dork for not checking Eric’s objects!

#146239
Dec 3, 2008 at 1:04am

All of the [el.vd*~] objects seem to blow up when fed back. I need an object without an internal feedback mechanism that I can manually feedback into the signal, as one is able to do with [tapin~]/[tapout~] delay lines. Any ideas? Take a look and see what I’m talking about:

– Pasted Max Patch, click to expand. –
#146240
Dec 3, 2008 at 1:11am

Quote: jayrope wrote on Tue, 02 December 2008 10:19
—————————————————-
> Feed delay back into itself by running a second cable from it’s output thru a [*~ 0.9] and back into itself.
> adjust number for more or less feedback, beware, 1 means full feedback.
>
> jrp
—————————————————-

I can’t seem to accomplish it through this method, either…

– Pasted Max Patch, click to expand. –
#146241
Dec 4, 2008 at 9:44pm

#146242
Dec 4, 2008 at 11:00pm

Quote: Brennon Bortz wrote on Wed, 03 December 2008 01:04
—————————————————-
> All of the [el.vd*~] objects seem to blow up when fed back. I need an object without an internal feedback mechanism that I can manually feedback into the signal, as one is able to do with [tapin~]/[tapout~] delay lines. Any ideas? Take a look and see what I’m talking about:
>

vd~ is deprecated, in favor of vdp~. I’ve found that generally these objects do not blow up when using their internal feedback reasonably. However, they do follow the Max convention of specifying time arguments in milliseconds, not seconds as is the case in your example. Also, please note that there is a feedback loop in your patch that would kill a horse :) Although it does not solve your problem, the following patch shows how feedback does (and does not work) in Max. You can still blow it up, but no more than any other similar feedback network.

Cheers, Eric

– Pasted Max Patch, click to expand. –
#146243
Dec 4, 2008 at 11:44pm

Quote: Eric Lyon wrote on Thu, 04 December 2008 23:00
—————————————————-
Although it does not solve your problem, the following patch shows how feedback does (and does not work) in Max. You can still blow it up, but no more than any other similar feedback network.

Thanks, Eric! I hadn’t thought of adding a separate delay line between your objects. You’re right, it doesn’t fix my specific problem, but might be an option in a different scenario.

Ultimately, I took a different approach, and now have a patch that I’m happy with. Now on to the next bit…

Thanks again,
Brennon

#146244

You must be logged in to reply to this topic.