Triangle wave???

[sad1024]

Is there anyway to generate a triangle wave, based on the analog model of a comparator followed by an integrator? You could also get a square wave if you sampled the comparator rather than the integrator.

[frank]

Why try to model the analog method in a digital chip?

For triangle wave see this thread http://www.spinsemi.com/forum/viewtopic ... t=triangle

[sad1024]

Not as an lfo. As a large range function generator.

[sad1024]

equ x reg0

sof 0,0
rdax x, 1
skp gez, 1
sof 0, -1
skp neg, 1
sof 0, 0.999
sof 1, 0
mulx pot0
rdax x,1
wrax x,1
wrax dacl,0


Why doesn't this work?

[frank]

Please comment code, makes it much easier to understand your intent at each step and we don't have to try to reverse engineer your thoughts.

[sad1024]

equ x reg0 ;setup registry

sof 0,0 ;clear acc
rdax x, 1 ;read x
skp gez, 1 ;if negative let acc =-1
sof 0, -1
skp neg, 1 ;if positive let acc =0.999
sof 0, 0.999
mulx pot0 ;attenuate acc
rdax x,1 ;integrate acc
wrax x,1 ;write x
wrax dacl,0 ;write to output

[frank]

What are you trying to do here? You don't read any inputs nor initialize the "x" register so if it is a negative value it stays negative and if it is positive it stays positive. There is no input to integrate.

[sad1024]

Sorry, my error. This oscillates but is unstable.

equ x reg0
equ y reg1
equ z reg2 ;setup registry


sof 0,0 ;clear acc
skp run,2
sof 0,-0.5 ; initialize oscillator
wrax x, 0
rdax z, -1 ; feedback from second integrator
mulx pot0 ; attenuate acc
rdax x,1 ; integrate
wrax x,1
wrax dacl,1 ; write to output
mulx pot0 ; attenuate
rdax y,1 ; integrate
wrax y,1
skp neg,2 ; turn into square wave
sof 0, 0.5
wrax z,0
skp gez,2
sof 0,-0.5
wrax z,0 ; write square wave feedback

[frank]

Only had time to look at it quickly but you probably want to also attenuate the integrator feedback values (i.e. rdax x,0.9).

If this is to do a triangle wave I would just use a counter, count up to a fixed value then count down to 0 then back up to a fixed value, etc.