Hi, Paul,

Another option would be to use the "DDS Compiler" block from system generator.  
That can be used to generate sin/cos signals as well.  I'm not sure whether it 
has all the same options as the CASPER sin/cos block, but could be worth 
looking into.

Hope this helps,
Dave

On Sep 19, 2013, at 10:26 AM, Paul Marganian wrote:

> Sorry Andrew, I can't seem to find any of the alternatives you recommended.
> We're using xlinx 14.3, and our mlib_devel's most recent commit was May of 
> this year.
> Thanks again,
> Paul
> 
> On 09/19/2013 11:41 AM, Paul Marganian wrote:
>> Thanks Andrew,
>> I'll look into those other blocks.
>> The problem happens at compile time.
>> Paul
>> 
>> On 09/19/2013 09:45 AM, Andrew Martens wrote:
>>> Hi Paul
>>>> I've recently been having problems with the Casper DSP sincos block.  I'll 
>>>> successfully run a number of simulations, then I start getting these 
>>>> errors at compile time:
>>>> 
>>>> Error in 'test_sin/sincos1': Initialization commands cannot be evaluated
>>>> Error evaluating parameter 'initVector' in 'test_sin/sincos1/rom0'
>>>> Subscript indices must either be real positive integers or logicals.
>>>> 
>>>> The only way to get away from this error seems to be shutting down both 
>>>> the model *and* restarting matlab again.  After a few new simulations 
>>>> however, the problem comes back up, slowing down my workflow considerably. 
>>>>  Any ideas?
>>>> 
>>> Does the problem only happen at compile time or also during simulation? 
>>> 
>>> You can also try the cosin block from the Downconverter subsection. It is 
>>> more sophisticated in BRAM saving and is used in the FFT library for 
>>> coefficient generation (so should work). 
>>> 
>>> Depending on your application you may also look at using the feedback_osc 
>>> block which generates a cos/-sin pair using multipliers instead of using a 
>>> lookup table. It can only generate in-order values though, not in arbitrary 
>>> order as the sincos/cosin blocks do. The feedback_osc is also used in the 
>>> fft family so should work.
>>> 
>>> Regards
>>> Andrew
>> 
> 


Reply via email to