DSPRelated.com
Forums

reconstruction filter

Started by maxascent January 10, 2006
Hi

I want to design a filter after a dac for a dds system. It will be a
passive 9th order elliptic filter. Problem is I want to add a circuit to
compensate for the sinx/x of the dac. 

Thanks

Jon
maxascent wrote:
> Hi > > I want to design a filter after a dac for a dds system. It will be a > passive 9th order elliptic filter. Problem is I want to add a circuit to > compensate for the sinx/x of the dac.
It is generally more straightforward to preemphasize the signal while it is still digital, You can compensate for sin(x)/x and also, if need be, the reconstruction filter's affect on the passband. You must have very special requirements to call for a 9th-order filter. Would you care to describe the constraints that make it advisable? Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
"maxascent" <johnbean_uk@hotmail.com> wrote in news:sNydnc7WcdM4k1neRVn-
tg@giganews.com:

> > Hi > > I want to design a filter after a dac for a dds system. It will be a > passive 9th order elliptic filter. Problem is I want to add a circuit to > compensate for the sinx/x of the dac. > > Thanks > > Jon
It's not particularly easy to build a 9th order elliptic filter. Tolerances of parts, tuning etc is very difficult. Can you sample at a higher frequency to simplify your reconstruction filter? -- Al Clark Danville Signal Processing, Inc. -------------------------------------------------------------------- Purveyors of Fine DSP Hardware and other Cool Stuff Available at http://www.danvillesignal.com

maxascent wrote:

> Hi > > I want to design a filter after a dac for a dds system. It will be a > passive 9th order elliptic filter. Problem is I want to add a circuit to > compensate for the sinx/x of the dac.
The 9th order analog elliptic filter is going to be monstrous. I don't understand why do you need it for the DDS. Probably the system could be designed more optimally. Anyway if the filter itself does not scare you, then the analog compensation of sin(x)/x should not be a problem at all. It is just a low shoulder filter. Vladimir Vassilevsky DSP and Mixed Signal Design Consultant http://www.abvolt.com
"maxascent" <johnbean_uk@hotmail.com> wrote in message
news:sNydnc7WcdM4k1neRVn-tg@giganews.com...
> > Hi > > I want to design a filter after a dac for a dds system. It will be a > passive 9th order elliptic filter. Problem is I want to add a circuit to > compensate for the sinx/x of the dac.
Wow, 9th order filter. That must end being pretty big. You can pre-compensate for the sinx/x of the DAC in the digital data before you send it out the DAC. That way, your output data will not have the sinx/x distortion and you can avoid some analog headache in trying to create a circuit.
> > Thanks > > Jon
Bhaskar Thiagarajan wrote:
> "maxascent" <johnbean_uk@hotmail.com> wrote in message > news:sNydnc7WcdM4k1neRVn-tg@giganews.com... > >>Hi >> >>I want to design a filter after a dac for a dds system. It will be a >>passive 9th order elliptic filter. Problem is I want to add a circuit to >>compensate for the sinx/x of the dac. > > > Wow, 9th order filter. That must end being pretty big.
It's about the size. For faithful adherence to the designed response, the components will need to be trimmed to maybe 0.1%, and stay that way with temperature and aging. ... Jerry -- Engineering is the art of making what you want from things you can get. &#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;
Bhaskar Thiagarajan wrote:

 > "maxascent" <johnbean_uk@hotmail.com> wrote in message
 > news:sNydnc7WcdM4k1neRVn-tg@giganews.com...
 >
 >> Hi
 >>
 >> I want to design a filter after a dac for a dds system. It will be a
 >> passive 9th order elliptic filter. Problem is I want to add a circuit to
 >> compensate for the sinx/x of the dac.
 >
 >
 >
 > Wow, 9th order filter. That must end being pretty big.


It's not about the size. For faithful adherence to the designed 
response, the components will need to be trimmed to maybe 0.1%, and stay 
that way with temperature and aging.

   ...

Jerry
-- 
Engineering is the art of making what you want from things you can get.
&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;
I can give you a bit more info regarding the system. I am sampling at 50MHz
for a max frequency of 20MHz and generating a sine wave. I need such a high
order filter to enable me to achieve -70dBc. 

maxascent wrote:

> I can give you a bit more info regarding the system. I am sampling at 50MHz > for a max frequency of 20MHz and generating a sine wave. I need such a high > order filter to enable me to achieve -70dBc.
This does not seem to be a good design. The DDS ICs allowing for 12 bits resolution at the clock rate over 300 MHz are available from ADI. Or, you can lock the analog PLL on your DDS, using the PLL like a narrowband filter. Vladimir Vassilevsky DSP and Mixed Signal Design Consultant http://www.abvolt.com
maxascent wrote:
> I can give you a bit more info regarding the system. I am sampling at 50MHz > for a max frequency of 20MHz and generating a sine wave. I need such a high > order filter to enable me to achieve -70dBc.
Why a Bessel filter? Other designs give sharper cutoff. Have you looked at cascading 2 4th-order Butterworth switched-capacitor chips? Jerry -- Engineering is the art of making what you want from things you can get. &#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;