DSP Robotics Support

Hello Gang,

Well I have a new problem. I am creating a schematic with many divided square waves in it. I need to see these in my test circuit, but as I add more scopes and square waves the scope displays all the same square wave as the previous and not the divided wave forms.

I have attached an example. Press the switch/button and you will notice the upper display changes. This is not good. I need to see the divides from each flip flop on the scope.

Like I said this is ONLY an example. My true schematic has many, many more divisions and as I add more and more scopes to view them they all change like in the example.

Can anyone fix this. Please, please.
Many thanks in advance!
Later then, BobF.....

In one of your code modules (0 and 1 only) you where writing to an input variable. That was causing the memory glitch. I changed it from:


to:


BTW: I have no idea what you want to achieve with that code, but it doesn't make any sense to me. It's basically the same as: out = tmp & 1;

Hello MyCo,
Many thanks! I should have asked you a long time ago. I use this in several modules. When I did it, it seemed to work so I thought nothing more about it until now. What I am trying to achieve is, any stream that comes in (sine, triangle, square, etc), comes out a square wave that swings from 0 to +1 ONLY and NOT from -1 to +1. The flip flop and several other logic gates I built , would never work without it. Not knowing any programming really screws me sometimes.
If there is still yet a better way, please, please, let me know.

Again thank you so much. This helps me out greatly!

Later then, BobF.....

I guess that is what you want:

Hi MyCo,

So, so easy. Boy do I feel dumb! Again, much thanks.

Take care, BobF.....

MyCo wrote:

Code: Select all

streamin in;
streamout out;

out = (in > 0) & 1;

@MyCo
One Question regarding the programming technique. I see just one condition: "if in greater than zero then set out to 1".
Does DSP code automatically set out to 0 if the condition is not true, or does it just send the current state of out, even if it is undefined? But in that last case, wouldn't out then have the last state (for example a to 1 corrected value)?

The comparison (in > 0) returns either:
Bitmask 0000000000... - for false
Bitmask 1111111111... - for true

The bitmasks also have float values:
Bitmask 0000000000... = 0
Bitmask 1111111111... = NaN (not a number)

When you & (binary and) the bitmasks with any arbitrary number you get:
Bitmask 0000000000... & x = Bitmask 0000000000... (Float = 0)
Bitmask 1111111111... & x = Bitmask of x (Float value = x)

Basically it's just plain and simple binary logic

Ah, finally!
Now I see how it actually works. I was still too much into standard branching, instead of binary logic. Now it makes sense to me. And for human logic this basically means, that with a condition set, out will always be set to a value, either x or the defined one from the condition. With that in mind I start to understand the larger bitmask chains that some people realized.

Say I want to set everything that's greater than 0.5 to 1, else 0.5, but only if it is also greater than -0.5:
in = -1 => out = 0
in = -0.2 => out = 0.5
in = 0.3 => out = 0.5
in = 0.6 => out = 1
etc.

Would this be correct?

Yep, that's correct. I think this would do the same:

Thanks, MyCo!
Another hurdle taken that brings me closer to actually create meaningful dsp code. Couldn't wish for any better teacher regarding programming aspects!

For the other hurdles, like filter design and theory, I'm already learning from the best. Since I worked with him on the SSE-supporting Shared Mem Wavetable Oscillator, Martin is kind of a godfather of dsp for me...