dirac delta function

knight23 · Jan 31, 2007, 09:27 PM

i need help on how to make a dirac delta function to check the impulse response of some filters i made.

d[n-m] = 1 if n=m and 0 otherwise

any idea on how to make one on dane? thanks!

Max M. · Feb 1, 2007, 05:06 AM

something like:

Code:

input  n, m
output d

macs d, n, m, -1 
skip ccr, ccr, 8, 1    
macs d, 0x1, 0, 0 ; or macs d, 1 ... if you need a fractional 1 as answer
end

[color=gray](although i can't really see why you need it in dane - isn't it easier to just record an impulse response and analyse it in whatever software you want?)[/color]

knight23 · Feb 2, 2007, 03:05 PM

i'm doing some LMS filters and i just wanted to get the filter coefficients.

Code:

control m; treat m as either 0 or 1
output d
static z=0
temp y, x,n

macs x, 0, 1, m
macs y, x, z, 1; when m = 1, 1st iteration y = 1, succeeding iterations y = 2
macs n, 1, y, -1 
skip ccr, ccr, 0x100, 1; n != 0    
macs d, 0x1, 0, 0
macs z, x, 0, 0
end

thanks max m.! i didn't know how to use the skip ccr thing. whats the difference between a fractional 1 and an integer 1?

Russ · Feb 2, 2007, 06:04 PM

Quote:

Originally Posted by knight23

whats the difference between a fractional 1 and an integer 1?

I am guessing that what you really want to know is what the different is between using '1' and '0x1' in your code. This is related to how Dane translates numbers entered in decimal, for different instructions.

For instructions expecting the parameter to be a fractional value:

'1' is translated (by Dane) to 0x7fffffff (fractional 1), and is treated by the DSP as 0.9999999995343387126922607421875 (2's Compliment Value of 2147483647 (0x7fffffff)))

0x1 is not translated, and is treated by the DSP as 0.0000000004656612873077392578125 (2's Compliment Value of 1 (0x1)).

For instructions expecting the parameter to be an integer:

1 is translated (by Dane) to 0x1, and is treated by the DSP as 1 (integer 1).

0x1 is not translated, and is treated by the DSP as 1 (integer 1).

i.e.
Integer parameters are treated as 1's Compliment numbers.
Fractional parameters are treated as 2's Compliment numbers.

If you want to see how Dane translates the decimal numbers used in your code, load the plugin in the DSP and then choose 'Edit Dump' from the context menu.

As far as output goes, 0x1 is the lowest possible (positive) value, and fractional one is the largest possible (positive) value. Also, it should be noted that signals entering the DSP are divided by 4, and signals exiting the DSP are multipied by 4.

Russ · Feb 2, 2007, 07:41 PM

BTW: I am not completely sure what you are trying to do with your code, but I do not think it is going to do what you want.

'output d' is a static register, so once it changes to 1 in your code, it is going to remain 1 (it has no initial value). Going by your intial question, I am guessing that is not what you want.

'fractional 1' + 'fractional 1' does not equal 2 (fractional 1 is the max value, so the result saturates to fractional 1). When implementing a counter, you should enter the values in hex.

If I understand your intent correctly, you probably want something more like the following:

Code:

control m=0; treat m as either 0 or 1    
output d
static z=0
temp y, x, n
temp tmp_m;  adjusted m

interp tmp_m, 0x0, m, 0x2; scale m range to 0x0..0x1 (0% to 49% : tmp_m=0, 50% to 100% : tmp_m=0x1)
macs d, 0, 0, 0; initialize d to 0
macints x, 0, tmp_m, 1; 
macints y, x, z, 1; when m = 1, 1st iteration y = 1, succeeding iterations y = 2
macs n, 0x1, y, -1 
skip ccr, ccr, 0x100, 1; n != 0    
macs d, 1, 0, 0  
macs z, x, 0, 0

knight23 · Feb 2, 2007, 10:15 PM

just to be consistent...

Code:

macints y, x, z, 1; when m = 1, 1st iteration y = 1, succeeding iterations y = 2
macints n, 0x1, y, -1; n = 0 if y = 1 
skip ccr, ccr, 0x100, 1; n != 0    
macints d, 1, 0, 0  
macints z, x, 0, 0

thanks for all of your help!

jbobchin · Jun 13, 2007, 08:17 AM

This is a good plugin to quickly test out the various kx reverbs. I wanted to see if the CLEAX4Reverb was usable as a multichannel reverb but I don't like the "pulsing" decay echoes. Be careful not to blow out your speakers though. Sounds like plugging in your guitar to a live amp! It's safer to turn the volume low and watch the levels dance on ProFX:Peak6.

Jan 31, 2007, 09:27 PM	#1
knight23 DriverHeaven Newbie Join Date: Sep 2006 Location: cebu, philippines Posts: 9 Rep Power: 0	dirac delta function i need help on how to make a dirac delta function to check the impulse response of some filters i made. d[n-m] = 1 if n=m and 0 otherwise any idea on how to make one on dane? thanks!

Feb 1, 2007, 05:06 AM	#2
Max M. h/h member-shmember Join Date: Dec 2002 Location: Evil Empire Posts: 2,639 Rep Power: 69	something like: Code: input n, m output d macs d, n, m, -1 skip ccr, ccr, 8, 1 macs d, 0x1, 0, 0 ; or macs d, 1 ... if you need a fractional 1 as answer end [color=gray](although i can't really see why you need it in dane - isn't it easier to just record an impulse response and analyse it in whatever software you want?)[/color] Last edited by Max M.; Feb 1, 2007 at 05:12 AM.

Feb 2, 2007, 03:05 PM	Thread Starter #3
knight23 DriverHeaven Newbie Join Date: Sep 2006 Location: cebu, philippines Posts: 9 Rep Power: 0	i'm doing some LMS filters and i just wanted to get the filter coefficients. Code: control m; treat m as either 0 or 1 output d static z=0 temp y, x,n macs x, 0, 1, m macs y, x, z, 1; when m = 1, 1st iteration y = 1, succeeding iterations y = 2 macs n, 1, y, -1 skip ccr, ccr, 0x100, 1; n != 0 macs d, 0x1, 0, 0 macs z, x, 0, 0 end thanks max m.! i didn't know how to use the skip ccr thing. whats the difference between a fractional 1 and an integer 1?

Feb 2, 2007, 07:41 PM	#5
Russ HardwareHeaven Extreme Member Join Date: Jan 2005 Posts: 5,561 Rep Power: 62	BTW: I am not completely sure what you are trying to do with your code, but I do not think it is going to do what you want. 'output d' is a static register, so once it changes to 1 in your code, it is going to remain 1 (it has no initial value). Going by your intial question, I am guessing that is not what you want. 'fractional 1' + 'fractional 1' does not equal 2 (fractional 1 is the max value, so the result saturates to fractional 1). When implementing a counter, you should enter the values in hex. If I understand your intent correctly, you probably want something more like the following: Code: control m=0; treat m as either 0 or 1 output d static z=0 temp y, x, n temp tmp_m; adjusted m interp tmp_m, 0x0, m, 0x2; scale m range to 0x0..0x1 (0% to 49% : tmp_m=0, 50% to 100% : tmp_m=0x1) macs d, 0, 0, 0; initialize d to 0 macints x, 0, tmp_m, 1; macints y, x, z, 1; when m = 1, 1st iteration y = 1, succeeding iterations y = 2 macs n, 0x1, y, -1 skip ccr, ccr, 0x100, 1; n != 0 macs d, 1, 0, 0 macs z, x, 0, 0

Feb 2, 2007, 10:15 PM	Thread Starter #6
knight23 DriverHeaven Newbie Join Date: Sep 2006 Location: cebu, philippines Posts: 9 Rep Power: 0	just to be consistent... Code: macints y, x, z, 1; when m = 1, 1st iteration y = 1, succeeding iterations y = 2 macints n, 0x1, y, -1; n = 0 if y = 1 skip ccr, ccr, 0x100, 1; n != 0 macints d, 1, 0, 0 macints z, x, 0, 0 thanks for all of your help!

Jun 13, 2007, 08:17 AM	#7
jbobchin DriverHeaven Newbie Join Date: Mar 2005 Posts: 8 Rep Power: 0	This is a good plugin to quickly test out the various kx reverbs. I wanted to see if the CLEAX4Reverb was usable as a multichannel reverb but I don't like the "pulsing" decay echoes. Be careful not to blow out your speakers though. Sounds like plugging in your guitar to a live amp! It's safer to turn the volume low and watch the levels dance on ProFX:Peak6.