I designed a physical modelling synth [mimba] in Pd, it's mainly Karplus-Strong. Documentation, sound samples and downloads are on www.hansroels.be/mim.html

It's an early version (0.90): there are still some minor bugs and it's not really cleaned up yet... It's also rather complicated... but I've created some nice presets (guitar, organ, percussion-like sounds)! I hope you enjoy my work.

@playinmyblues I have not edited any of the mimba files. I've used many other Pd patches on my Windows system without any problems, so I don't think it's anything on my system that is causing the problem. I'll try it on another Windows system when I get a chance.

I found the code that has the divide by zero error (in blue).

@jamcultur it is good to see you found your source of the error. Does that tell you why? And I am curious how you found it.

@jamcultur I am wondering if there is a quick way to search through the files using a text based search. I know you can open Pd files as text files and there is a syntax to them. I think you can edit them as well and produce working Pd patches.

So here are the results from a search on the MacOS cli:
mimba-092b % grep "sin" *
mba.pd:#X obj 1201 543 r $1-mmbsintkeybp;
mba.pd:#X obj 319 354 r $1-mnbsinidelall;
mba.pd:#X obj 347 454 r $1-$2-mnbsinidelalldd;
mba.pd:#X obj 348 493 r $1-$2-mnbsinidelallml;
mba.pd:#X obj 1704 4618 r $1-mmbsinidelsensi;
******************* MY ADDED TEXT TO HIGHLIGHT THE LINE*****************************
mba.pd:#X obj 595 437 expr sin(w$0)sinh(ln(2)/2$f1w$0/sin(w$0));
MY ADDED TEXT TO HIGHLIGHT THE LINE**********
mba.pd:#N canvas 0 83 1920 997 sineIm 0;
mba.pd:#X restore 160 538 pd sineIm;
mba.pd:#X obj 323 536 sin;
mba.pd:#X text 505 456 alpha = sin(w0)/(2Q);
mba.pd:, peak gain = Q) b0 = sin(w0)/2 = Qalpha b1 = 0 b2 = -sin(w0)/2 =
mba.pd:#X text 530 678 b0 = sin(w0)/2 = Qalpha;
mba.pd:#X text 530 759 b2 = -sin(w0)/2 = -Qalpha;
mba.pd:#X obj 125 383 sin;
mimba.pd:#X msg 59 488 ; $1-wtable-ramp3 sinesum 515 1 -1 0.5 -0.5 0.33 -0.33
mimba.pd:#X msg 51 296 ; $1-wtable-ramp2 sinesum 515 1 0 -0.111111 0 0.04
mimba.pd:#X msg 52 88 ; $1-wtable-ramp1 sinesum 515 1 0.5 0.33 0.25 0.2 0.16
mimba.pd:#X obj 147 824 list prepend sinesum 1024;
mimba.pd:#X obj 147 824 list prepend sinesum 1024;
mimba.pd:#X msg 875 315 ; $1-wtable-z1b sinesum 515;
mimba.pd:#X obj 147 791 list prepend sinesum 1024;
mimba.pd:#X text 509 510 rising saw;
mimba.pd:#X text 35 274 soft sine 2;
mimba.pd:#X text 20 201 soft sine 1;
mimba.pd:#X msg 521 916 ; 1-addiwaves write sine-noisy1.txt;
mimba.pd:#X msg 522 853 ; $1-addiwaves read sine-noisy1.txt;
mimba.pd:#X text 165 339 (3) soft sine 3;
mimba.pd:#X text 1334 643 soft sine 1 with octave louder;
mimba.pd:#X obj 171 780 list prepend sinesum 1024;
mimba.pd:#X obj 147 824 list prepend sinesum 1024;
mimba.pd:#X obj 147 791 list prepend sinesum 1024;
mimba.pd:#X obj 679 864 list prepend sinesum 1024;
mimba.pd:#X obj 147 824 list prepend sinesum 1024;
mimba.pd:#X text 193 158 0 soft sine 4;
mimba.pd:#X obj 147 824 list prepend sinesum 1024;
mimba.pd:#X obj 42 374 s $1-mmbsintkeybp;
mimba.pd:en filders te 'clearen' in tien voices tegelijkertijd. Oplossing: tussen
mimba.pd:elke stem bv. 100ms steken , zie oplossing aan de rechterkant van
mimba.pd:#X msg 1041 471 ; 1-addiwaves write sine-noisy1.txt;
mimba.pd:per wavetable)(de langste sinesum messages -zoals 37- hebben nu 36
mimba.pd:#X text 36 318 alle lage tonen maximum 36 harmonieken in sinesum ,
mimba.pd:doen met global previous pitch (en toggle voor aanpassing van vorige
mimba.pd:#X msg 59 488 ; $1-wtable-ramp3 sinesum 515 1 -1 0.5 -0.5 0.33 -0.33
mimba.pd:#X msg 51 296 ; $1-wtable-ramp2 sinesum 515 1 0 -0.111111 0 0.04
mimba.pd:#X msg 52 88 ; $1-wtable-ramp1 sinesum 515 1 0.5 0.33 0.25 0.2 0.16
mimba.pd:#X obj 147 824 list prepend sinesum 1024;
mimba.pd:#X obj 147 824 list prepend sinesum 1024;
mimba.pd:#X obj 147 791 list prepend sinesum 1024;
mimba.pd:#X text 509 510 rising saw;
mimba.pd:#X text 177 311 soft sine 2;
mimba.pd:#X text 162 238 soft sine 1;
mimba.pd:#X msg 521 916 ; 1-addiwaves write sine-noisy1.txt;
mimba.pd:#X msg 522 853 ; $1-addiwaves read sine-noisy1.txt;
mimba.pd:#X text 227 363 (3) soft sine 3;
mimba.pd:#X obj 171 780 list prepend sinesum 1024;
mimba.pd:#X obj 147 824 list prepend sinesum 1024;
mimba.pd:#X obj 147 791 list prepend sinesum 1024;
mimba.pd:#X obj 679 864 list prepend sinesum 1024;
mimba.pd:#X obj 100 1319 r $1-mnbsinidelall;
mimba.pd:#X obj 624 1290 r $1-mmbsinidelsensi;
mimba.pd:#X obj 11 591 hsl 128 15 0 508 0 0 $1-mnbsinidelall $1-mnbrinidelall
mimba.pd:#X obj 829 17 bng 15 250 50 0 $1-mmbsinfopifqin empty i 17 7 0 10
mimba.pd:(read) variations ; 31 = read once with small sine (read) variations
mimba.pd:; 32 = read once with bigger (read) sine variations ; 33 = normal
mimba.pd:#X obj 306 22 r $1-mmbsinfovelom;
mimba.pd:#X obj 555 28 r $1-mmbsinfopifqin;
mimba.pd:#X obj 29 231 r $1-mmbsinfodetails;
mimba.pd:#X obj 416 137 bng 15 250 50 0 $1-mmbsinfovelom empty i 17 7 0 10
mimba.pd:#X obj 907 94 hsl 121 10 -1 1 0 0 $1-mmbsinidelsensi $1-mmbrinidelsensi
mimba.pd:the pitch detuning caused by bypassing/crossfading APF delay., f 66
mimba.pd:#X obj 22 622 r $1-mmbsinfoafterfiltrin;
mimba.pd:#X obj 1016 539 bng 15 250 50 0 $1-mmbsinfodetails empty xtra_details
mimba.pd:#X obj 1650 601 bng 15 250 50 0 $1-mmbsinfoafterfiltrin empty i 17
mimba.pd:; mmbsimplvl ; mmbsxfadeapf ; mmbsxnoilop ; mnbsinidelall ; mmbsxnoistaval
mimba.pd:'drastische oplossing: default naar alle voicespecific send-receives
grep: presets: Is a directory

I am going to refine the search method to include the relationship/equation from the error message to see what happens.

The mimba synth is heavy on the computer cpu because a large part of the sound production happens in a subpatch with blocksize 1, so sample per sample . On my older (linux) laptop the audio production wasn't reliable and started 'cracking' with 3 or 4 voices. On a more recent (linux) laptop it works well with 8 to 10 voices. So I guess that the crackling sound is related to your laptop performance (and the fact that my synth is so cpu-heavy)...
The error message 'expr divide by zero detected' doesn't relate to the performance of the patch. I also get this message, but my patch and synth work fine. This zero division only happens with the first 10 notes that you play, because mimba has 10-voice polyphony... I guess the zero division is related to a wrong or missing initial value somewhere in the patch.

@playinmyblues said:

@jamcultur it is good to see you found your source of the error. Does that tell you why? And I am curious how you found it.

Seeing the code with the error doesn't tell me why the error occurred. I found the code using the findstr command on Windows (similar to grep on Linux). I searched for the code referenced in the error message that was issued to the Pd console. Do you not get the divide by zero messages on your Pd console?

@hansr Pretty sure a wrong order of operation...... as [expr sin....etc] looks for the value of w$0 but it is not set by [unpack] until after [expr ($f3...etc] operates.
So a one time only divide by zero as the patch is first run..
Swapping the first and second outlets of [unpack f f f] should fix it...
David.

@jamcultur I have been getting the error messages but I never had a problem with the sound and did not look at the Pd console for a problem. Looking at your screenshot from a few posts ago, I thought I might be able to find the cause of the error by using grep. I have yet to be able to use the whole expression. It is probably just some syntax in the command that I need to look up.

I see whale-av presents what might be a solution. That could be interesting.

Figuring out Pd is something I really want to do so I am trying to learn a lot as I go along. It all seems to working out this time but it is taking longer than I would like. Oh well.

How to keep track of the developement?