Brett wrote:

> Thank you everyone for your responses.  I hope to start working on
> them today.
> 
> Perhaps I should I have described our motivation for wanted to go from
> digital to analog.  What we have is an RF front end that has a less
> than flat response, mostly in group delay.  My boss konws that I had,
> in another project, taken the response of a distorted IF from a
> network analyzer and created a digital filter that had the inverse
> response.  So with that information my boss wondered if I could take
> that experience a step further and create an analog filter with the
> inverse response of the RF front endo that they could place that
> filter on the RF board.  It seems to be that there must be a better
> way to fix the problems of an RF chain, but I'm just the DSP guy.
> 
> Thanks again
> Brett

How about a frontal attack on the problem?

Guillemin "Synthesis of Passive Networks"; John Wiley
    "      "The Mathematics of Circuit Analysis"; MIT Press
Weinberg  "Network Analysis and Synthesis"; McGraw-Hill

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

Thank you everyone for your responses.  I hope to start working on
them today.

Perhaps I should I have described our motivation for wanted to go from
digital to analog.  What we have is an RF front end that has a less
than flat response, mostly in group delay.  My boss konws that I had,
in another project, taken the response of a distorted IF from a
network analyzer and created a digital filter that had the inverse
response.  So with that information my boss wondered if I could take
that experience a step further and create an analog filter with the
inverse response of the RF front endo that they could place that
filter on the RF board.  It seems to be that there must be a better
way to fix the problems of an RF chain, but I'm just the DSP guy.

Thanks again
Brett

In article rEu1c.69764$aH3.2126163@bgtnsc04-news.ops.worldnet.att.net,
flipdog at john.lai@worldnet.att.net wrote on 03/03/2004 19:21:

>> 
>> 1.  BLT:  invert s = 2/T * (1-z^-1)/(1+z^-1)
>> 
>> 2.  IIT:  convert to parallel and replace each h[n] = a^-n  to
>> h(t) = a^(t/T)  where t = n*T
>> 
>> r b-j
>> 
> Hello,
> I understand that methods like BLT, IT etc assist us to design digital
> filter from analog design.  Why would people interested in the other way
> round?

beats me.  maybe their using some optimized IIR design program and they
wanna design some op-amp filter.  i dunno.

>  What design application would need this approach?

well, there will continue to be the need for analog filters in some
contexts.  whatever comes before the A/D or after the D/A or in some
manufactured products, it might just be less expensive to delete the DSP
step and do it all in the continuous-time domain.  again, i dunno.

r b-j

"Michael Numminen" <michael.numminen@comhem.se> wrote in message
news:sss1c.51137$mU6.207815@newsb.telia.net...
> Hi Brett!
>
> You can use a Analogue Laguerre Network to do this. It is important to
> notice that this filter is an IIR filter with FIR filter structure where
> each digital delay elements is replaced by 1:st order analogue all-pass
> filters. Se T. Oliviera "On Determination of the Optimal Pole Position of
> Laguerra Filters", for more information about that.

Perhaps this addresses the "next step" after R B-J's response.  Anyway, I
was going to say that after you do the mapping then you need to figure out
how to change the rational function into a real filter of R's, L's, C's, op
amps, etc.  I'm sure there are ways but that's a necessary step to get to an
analog filter design from a digital filter design.

Fred

>
> 1.  BLT:  invert s = 2/T * (1-z^-1)/(1+z^-1)
>
> 2.  IIT:  convert to parallel and replace each h[n] = a^-n  to
> h(t) = a^(t/T)  where t = n*T
>
> r b-j
>
Hello,
I understand that methods like BLT, IT etc assist us to design digital
filter from analog design.  Why would people interested in the other way
round?  What design application would need this approach?
Regards,
FD

Hi Brett!

You can use a Analogue Laguerre Network to do this. It is important to
notice that this filter is an IIR filter with FIR filter structure where
each digital delay elements is replaced by 1:st order analogue all-pass
filters. Se T. Oliviera "On Determination of the Optimal Pole Position of
Laguerra Filters", for more information about that.

All the best, Michael Numminen, M.Sc.E.E.

"Brett" <bemsley@yahoo.com> skrev i meddelandet
news:65422a2a.0403030825.41cc6192@posting.google.com...
> Has anyone heard of a method to convert a digital FIR filter back into
> an analog filter?  Can you perform a bilinear or impulse invariant
> transform backwards?
>
> Thanks
> Brett Emsley

In article 65422a2a.0403030825.41cc6192@posting.google.com, Brett at
bemsley@yahoo.com wrote on 03/03/2004 11:25:

> Has anyone heard of a method to convert a digital FIR filter back into
> an analog filter?  Can you perform a bilinear or impulse invariant
> transform backwards?

the quick answer is yes to both.

> Thanks

1.  BLT:  invert s = 2/T * (1-z^-1)/(1+z^-1)

2.  IIT:  convert to parallel and replace each h[n] = a^-n  to
h(t) = a^(t/T)  where t = n*T

r b-j

Has anyone heard of a method to convert a digital FIR filter back into
an analog filter?  Can you perform a bilinear or impulse invariant
transform backwards?

Thanks
Brett Emsley