DSP Tools

[halunke86]

Hello,

I have a little understanding problem with the numbers format of systemgenerator:

I want to implement a CIC-FIR combination, using the given compilers. My input data comes from an ADC (1MPSP) and is an unsigned. CIC compiler needs signed data, so I used a concat block (with additional zero as first figure) and reinterpret it as signed (alternatively i used a convert block : 13 bits, signed).

For filtering and later signal processing,  I wanna use full precision /  an expanded number format (to reduce rounding errors).

Now to my problem: After signal processing, I need again a 12bit unsigned format for DAC, but just using a convert block doesn't lead to a useful solution.

I found out that expanding the numbers format (by CIC / FIR) leads to bigger numbers (control output gateways to simulink), so I understand why results at DAC aren't correct.

Is there a solution or do I have to stay in a 12-bit-system (or better: 13-bit)

Thx for help

Benny

PS: I attached

[ywu]

It's not clear what exactly problem you're getting. If the problem is that the bit width of CIC-FIR output is too big, you can use a slice block to get upper 12 or 13 bits from the output and then "reinterp" it to the desired format.

---

halunke86 wrote:

Hello,

I have a little understanding problem with the numbers format of systemgenerator:

I want to implement a CIC-FIR combination, using the given compilers. My input data comes from an ADC (1MPSP) and is an unsigned. CIC compiler needs signed data, so I used a concat block (with additional zero as first figure) and reinterpret it as signed (alternatively i used a convert block : 13 bits, signed).

For filtering and later signal processing,  I wanna use full precision /  an expanded number format (to reduce rounding errors).

Now to my problem: After signal processing, I need again a 12bit unsigned format for DAC, but just using a convert block doesn't lead to a useful solution.

I found out that expanding the numbers format (by CIC / FIR) leads to bigger numbers (control output gateways to simulink), so I understand why results at DAC aren't correct.

Is there a solution or do I have to stay in a 12-bit-system (or better: 13-bit)

Thx for help

Benny

PS: I attached

---