Re: [Discuss-gnuradio] Customized Block Giving Incorrect Output

[Michael Dickens]

Hi Tellrell - OK so now you're getting somewhere! It might help to know that "input_items" is a 3-dimensional array: [stream#][vec_Length#][data_vector], where: "stream#" is the input stream number index starting with 0; "vec_Length#" is an index into incoming data organized by the vector length; and "data_vector" is the actual "vec_Length" of data of the type requested. In your block's case, the first index will always be [0]; the 2nd index will vary depending on the number of vectors provided to the block for work by the scheduler ... use "len(input_items[0])" to get the number to iterate over inside general_work(); finally, the 3rd index will always be in "range(vec_Length)", for whatever value of "vec_Length" is provided at instantiation ... this length is guaranteed by the scheduler. I think with this info you should be able to write the Python code to do whatever your block needs to do, whether like the version you started with or more like what Marcus is suggesting. Keep us posted! - MLD

On Tue, Feb 13, 2018, at 4:39 PM, Tellrell White wrote:

Updates:

@ Michael I followed your advice and "vectorized" the complex to mag^2 block creating a variable for vector length equal to 1024, which I set as the vector length of this block. I noticed that this changed the color of the output port of the block.

Next, I "vectorized" the custom ED block as well. One question I do have is is there a way to find out the length of the input vector that is passed to the block just to confirm that there are 1024 values being passed? I checked the length of the input_items used in the work function and it prints 1, which I'm assuming is equal to an array of length 1024 since this is the vector length parameter i set in the ED block and also, because this is the vector length I set for the complex to mag^2 block as well.

As a result of making these changes to the custom block, now, I'm simply taking the array of input_items, normalizing them, and then comparing to a threshold as before. I'm assuming this is all that needs to be done assuming the block is taking in vectors of length 1024.

@Marcus I think your question does warrant some consideration and perhaps is the better approach. Besides this approach being easier, and I'm assuming less of a strain on the cpu are there any other reasons for this approach?

I've attached an updated flow graph used for testing

On Mon, Feb 12, 2018 at 11:36 AM, Müller, Marcus (CEL) <address@hidden> wrote:

Just another thought: Why convert every single FFT output vector from

linear to dB with a logarithm (that's a very complicated function!)

just to then compare it to a threshold, if you could also just convert

the threshold to linear once?

Best regards,

Marcus

On Mon, 2018-02-12 at 10:21 -0500, Michael Dickens wrote:

> In GRC, you open the "complex to ||^2" block settings & set the vector length to whatever you want. I'd advise using a variable that's defined in GRC, and then use it for any blocks that require the vector setting; that way you can change the variable value & all blocks are updated with it. Hope this is useful. - MLD

>

> On Mon, Feb 12, 2018, at 10:17 AM, Tellrell White wrote:

> > Thanks for the response. That's exactly what I'm trying to accomplish.  You mention the "complex to ||^2 can be vectorized. My question is how exactly do you go about doing that?

>

> _______________________________________________

> Discuss-gnuradio mailing list

> address@hidden

> https://lists.gnu.org/mailman/listinfo/discuss-gnuradio

Email had 1 attachment:

• ener_dtec_sim1.grc
  
    28k (application/octet-stream)