FPGA Central

Dear All
Could you sugest how to implement digital integrator working above 2 kHz.
In fact I need to obtain quadrature signal y[n] from given x[n].This mean
that it has to be some sort of IIR filter with perfect -90 deg phas
response for frequencies above 2 kHz.
Which structure should I take?
What fs and what coefficients must be?
What precision (digits after point)will be requared?
Thank you

On Mar 25, 6:42 am, "andrey_kiev" <sachk...@yahoo.com> wrote:
> Dear All
> Could you sugest how to implement digital integrator working above 2 kHz.
> In fact I need to obtain quadrature signal y[n] from given x[n].This means
> that it has to be some sort of IIR filter with perfect -90 deg phase
> response for frequencies above 2 kHz.
> Which structure should I take?
> What fs and what coefficients must be?
> What precision (digits after point)will be requared?
> Thank you

You leave many questions unanswered. Ignoring them for now, I'll
propose the following method for developing quadrature components from
a real bandpass signal: Create a prototype FIR LPF and mix the
coefficients with sine and cosine of center of your passband. You now
have a pair of bandpass filters in quadrature that you can apply your
input signal to.

John

Thanks John
I already implemeneted LPF with H(z)=0.024*(1+z)/(z-0.982) and fsampl=5
kHz.So this is a combination FIR+IIR,and it seems it works as integrato
above 2 kHz (at least in simulations)



>On Mar 25, 6:42 am, "andrey_kiev" <sachk...@yahoo.com> wrote:
>> Dear All
>> Could you sugest how to implement digital integrator working above
kHz.
>> In fact I need to obtain quadrature signal y[n] from given x[n].Thi
means
>> that it has to be some sort of IIR filter with perfect -90 deg phase
>> response for frequencies above 2 kHz.
>> Which structure should I take?
>> What fs and what coefficients must be?
>> What precision (digits after point)will be requared?
>> Thank you
>
>You leave many questions unanswered. Ignoring them for now, I'll
>propose the following method for developing quadrature components from
>a real bandpass signal: Create a prototype FIR LPF and mix the
>coefficients with sine and cosine of center of your passband. You now
>have a pair of bandpass filters in quadrature that you can apply your
>input signal to.
>
>John
>

>On Mar 25, 6:42 am, "andrey_kiev" <sachk...@yahoo.com> wrote:
>> Dear All
>> Could you sugest how to implement digital integrator working above
kHz.
>> In fact I need to obtain quadrature signal y[n] from given x[n].Thi
means
>> that it has to be some sort of IIR filter with perfect -90 deg phase
>> response for frequencies above 2 kHz.
>> Which structure should I take?
>> What fs and what coefficients must be?
>> What precision (digits after point)will be requared?
>> Thank you
>
>You leave many questions unanswered. Ignoring them for now, I'll
>propose the following method for developing quadrature components from
>a real bandpass signal: Create a prototype FIR LPF and mix the
>coefficients with sine and cosine of center of your passband. You now
>have a pair of bandpass filters in quadrature that you can apply your
>input signal to.
>
>John
>
hii
I'm new to this but I think that Quadrature signal exact -90 degree phas
shift can also be obtained by using applying hilbert tranform to you
input signal and as far as my knowlwdge goes a hilbert transform II
filter can be designed in MATLAB by using "hilbiir" function if you wan
to get the coefficients
if u are directly running on matlab platform you can also use hilbert(you
input signal vector) directly to get the complex vector whose real part i
your input signal and the imaginary part is the quadrature part of you
signal...

I'm still new to this please tell me if i were wrong some where...

Regards

Bhargava

andrey_kiev wrote:

> Dear All
> Could you sugest how to implement digital integrator working above 2 kHz.

Take one each adder. Then use.

> In fact I need to obtain quadrature signal y[n] from given x[n].

And this relates how to first request?

> This means
> that it has to be some sort of IIR filter with perfect -90 deg phase
> response for frequencies above 2 kHz.
> Which structure should I take?
> What fs and what coefficients must be?
> What precision (digits after point)will be requared?
> Thank you

Poorly formed question.

andrey_kiev wrote:
> Dear All
> Could you sugest how to implement digital integrator working above 2 kHz.
> In fact I need to obtain quadrature signal y[n] from given x[n].

An integrator will give you a 90 degree phase shift, but it'll have an
extremely strong frequency/amplitude characteristic. If you really want
a quadrature version of your signal than an integrator isn't the right
way to go.

> This means
> that it has to be some sort of IIR filter with perfect -90 deg phase
> response for frequencies above 2 kHz.

You can get a 90 degree phase shift with an IIR filter over some band of
frequencies, but (to my knowledge at least) this is most often done in
this day and age with FIR filters.

> Which structure should I take?
> What fs and what coefficients must be?
> What precision (digits after point)will be requared?
> Thank you

What are you really trying to do? How are you getting x?

You have a signal x and you want it's quadrature y -- OK. What's the
bandwidth of x, and does it go down to DC? What's the band over which
the y must be 'good'? What amplitude and phase variations can y have
from a perfect quadrature of x and still be good? None of your
questions make much sense in the absence of these requirements.

You'll find that your biggest constraint on a FIR filter length is going
to be the ratio between the lowest frequency that you need to maintain
good phase and gain properties and the sampling rate. Similarly, the
complexity of your IIR filter will be constrained by the ratio between
the lowest and highest frequencies where you need to maintain
'goodness'. Both of these complexity requirements will be eased or
tightened as you ease or tighten your requirements for phase and
amplitude accuracy.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Do you need to implement control loops in software?
"Applied Control Theory for Embedded Systems" gives you just what it says.
See details at http://www.wescottdesign.com/actfes/actfes.html

On Mar 25, 11:42 pm, "andrey_kiev" <sachk...@yahoo.com> wrote:
> Dear All
> Could you sugest how to implement digital integrator working above 2 kHz.
> In fact I need to obtain quadrature signal y[n] from given x[n].This means
> that it has to be some sort of IIR filter with perfect -90 deg phase
> response for frequencies above 2 kHz.
> Which structure should I take?
> What fs and what coefficients must be?
> What precision (digits after point)will be requared?
> Thank you

You could do

y(k)=y(k-1)+k*input

where k depends on the integrator gain (bandwidth ie unity gain cross-
over frequency)

Hardy

HardySpicer wrote:
> On Mar 25, 11:42 pm, "andrey_kiev" <sachk...@yahoo.com> wrote:
>> Dear All
>> Could you sugest how to implement digital integrator working above 2 kHz.
>> In fact I need to obtain quadrature signal y[n] from given x[n].This means
>> that it has to be some sort of IIR filter with perfect -90 deg phase
>> response for frequencies above 2 kHz.
>> Which structure should I take?
>> What fs and what coefficients must be?
>> What precision (digits after point)will be requared?
>> Thank you
>
> You could do
>
> y(k)=y(k-1)+k*input
>
> where k depends on the integrator gain (bandwidth ie unity gain cross-
> over frequency)

That's an integrator. It's what he asked for (and nearly what Owlet
already told him) but it's not what he needs.

Jerry
--
Engineering is the art of making what you want from things you can get.
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯ ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Yup, a Hilbert Transform is definitely the way to go for this. If you nee
something of less complexity, you could also get by with a couple o
cascaded all-pass filters to generate a quadrature signal that is vali
within some range of input frequencies.

>HardySpicer wrote:
>> On Mar 25, 11:42 pm, "andrey_kiev" <sachk...@yahoo.com> wrote:
>>> Dear All
>>> Could you sugest how to implement digital integrator working above
kHz.
>>> In fact I need to obtain quadrature signal y[n] from given x[n].Thi
means
>>> that it has to be some sort of IIR filter with perfect -90 deg phase
>>> response for frequencies above 2 kHz.
>>> Which structure should I take?
>>> What fs and what coefficients must be?
>>> What precision (digits after point)will be requared?
>>> Thank you
>>
>> You could do
>>
>> y(k)=y(k-1)+k*input
>>
>> where k depends on the integrator gain (bandwidth ie unity gain cross-
>> over frequency)
>
>That's an integrator. It's what he asked for (and nearly what Owlet
>already told him) but it's not what he needs.
>
>Jerry
>--
>Engineering is the art of making what you want from things you can get.
>ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï ¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿ ½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï ¿½ï¿½ï¿½ï¿½ï¿½
>