On 05/21/2018 01:59 PM, Dan CaJacob
wrote:
Interesting. Not quite what I was looking for.
Basically just doing a design-tradeoff study for grins.
Fixed sample rate and filter/decimate in a "sized-for-economy" FPGA,
or variable sample-rate, and a chain of tuneable analog filters in
front of the
ADC.
A chain of 3 or 4 2nd-order active-lowpass filters using video
op-amps would probably work OK. Use digipots to set the R values,
and use a fixed
value for C, or perhaps selectable C as well (probably only two or
three values).
Or, a small number (4?) of selectable L-C-R low-pass filters, and
fix the sample rates to a small number.
I used to
work at Maxim, but my dealings with those RF guys dealt more
with coffee and the quality thereof.
I looked at the Maxim chip. The filter is relatively steep but
not
complex. Looks like two or three poles maximum. I don't know
specifically about that chip, but that group had SiGe
technology, so
I'm leaning towards variable transconductance to do the
tuning.
You quickly learn just how bad google is at doing a search for
this
technology. The buzz phrase you need is continuous time, so I
suggest
"variable continuous time baseband filter". Sadly OTA gets
links to
"over the air." (And yet, they claim artificial intelligence
will take
over the world.)
The thing to note in any highly integrated analog chip is that
you
don't see the sausage being made. Once you have a system on a
chip, the
metric is the system performance, not the performance per se
of any
individual block. So those filters may not be as great as you
think.
Note the Maxim part shows a tempco on the corner frequency,
which could
imply variable transconductance. One you have bipolar elements
in the
process, variable transconductance is just a matter of tail
current.
(SCF performance was relatively temperature independent.)
Getting back to seeing the sausage, take the old analog modem
market.
Initially the SCF tech was used to make the official Bell 212
filters.
Once the modem was fully integrated, the on-board filters were
simplified for a number of really good engineering reasons,
not just
cost. One was the harmonic distortion of the band split
filters. The
harmonics of the in-band signal were far greater than the
out-band
signal that you were trying to reduce.
Getting back to these baseband filters, if you use a ladder
design, the
filter is relatively immune to component error, well as
opposed
(contrasted) to a chain of biquads. If you go full
differential, the
variable transconductance amp are reasonably linear. But
probably they
limit the number of poles for the same harmonic distortion
problem.
Filters should subtract, not add.
Most of the transconductance based filter designs probably are
derivatives of integrated video filters. Plenty of papers
online for
those designs.
If your goal is to roll together your own filter, TI has app
notes on
how to make analog tracking filters.
On Sun, 20 May 2018 21:19:47
-0400 "Marcus D. Leech" <address@hidden>
wrote:
> On 05/20/2018 09:13 PM, address@hidden
wrote:
> >
> > Can you be more specific about the corner
frequency?
> Corner frequencies step-tunable from perhaps 20Mhz down
to perhaps
> 2MHz.
>
> Many chips, like the R820T2, the MAX2112, and the
higher-integration
> devices like the AD9361 and LMS7002M have
> programmable analog low-pass corner frequencies, for
bandwidth
> tailoring of the complex baseband.
>
> Just not sure how its done. The chips have very low
external parts
> count, so whatever it is, it's got to be done
> internally...
>
>
> >
> > Cell phones use chips that have switchable banks of
capacitors for
> > antenna tuning. st.com
IIRC is a source.
> >
> > I used to design switched capacitor filter chips in
the 80/90s. The
> > technology was killed by oversampled converters and
DSP. The SCF
> > players went into continuous time video filters
using
> > transconductance amps and such.
> >
> >
> > _______________________________________________
> > Discuss-gnuradio mailing list
> > address@hidden
> > https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
>
>
> _______________________________________________
> Discuss-gnuradio mailing list
> address@hidden
> https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
_______________________________________________
Discuss-gnuradio mailing list
address@hidden
https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
--
Very Respectfully,
Dan CaJacob
_______________________________________________
Discuss-gnuradio mailing list
address@hidden
https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
|