Paul Burridge wrote in message
...
Hi all,

I need a simple design for a BPF with a centre frequency of 17.2Mhz.,
corner frequencies one Meg or therabouts either side of that and a
reasonably good shape factor (nothing critical here). I don't mind
having to do a bit of tweaking of values myself if anyone has
something close to that they can post. Passive L/C combinations only,
please. (assume the usual 50R nominal Zs.)
Thanks,

p.
--

The BBC: Licensed at public expense to spread lies.


The simplest filter that'll get you by, is this one. (a couple of tuned
circuits with a bit of top-coupling :-) ...

___ ||C1
-|_R_¦------.----.-------||-------.----.-------, O/p
Source |L2 |C2 || |C2 |L2 |
I/p | | | | .-.
C| --- --- C| | |load R
C| --- --- C| |R|
C| | | C| '-'
| | | | |
=== === === === ===
GND GND GND GND GND

F1=16.2e6 'lower limit MHz
F2=18.2e6 'upper limit MHz
R=50 'Ohms source and load impedance
Pi=3.142

C1 = (F1+F2) / (4 x Pi x F1 x F2 x R)
C2 = F1 /(Pi x F2 x (F2-F1) x R)/2
L2 = 2xR /(4 x Pi) x (1/F1 - 1/F2)

The inductors end up being quite small but a couple of loops of wire cost
nothing. At 25megs the output is down to about 3% , which is not bad.

This particular style is the "3 element" design and from the traditional
"Constant K" filter stable. The pleasantly simple to design "constant K's"
and associated "M derived" filters, can be easily and freely used for most
day-to-day electronics work.

Exact Filter design, or 'synthesis' as they nowadays prefer to call it, can
be a nightmare if you're coming in from scratch. Even using a filter design
programme requires a good knowledge of filter types/performance/spec's
before data can begin to be entered.
If not using a prog' you move onto the more vexing method of having to
design an initial (prototype) low-pass filter, either via precalculated
tables or computing the required co-efficients, then mathematically
transforming the low-pass values into a bandpass (etc) design. For higher
order filters it can be ball-acheing drudgery and even then inductor
resistance has a nasty habit of spoiling that pristine work of art. (not a
problem with DSP filters).

I can't suggest a decent book, as the general purpose books seem over
complicated and destined to be written by maths people for maths people.
(The old GPO line-comm's manuals have useful notes on the 'K and M '
filters).
regards
john

Go to Alt Binaries Schematics -
bg
Paul Burridge wrote in message
...
Hi all,

I need a simple design for a BPF with a centre frequency of 17.2Mhz.,

Go to Alt Binaries Schematics -
bg
Paul Burridge wrote in message
...
Hi all,

I need a simple design for a BPF with a centre frequency of 17.2Mhz.,

John Jardine wrote:
I can't suggest a decent book, as the general purpose books seem over
complicated and destined to be written by maths people for maths people.

I'd suggest Temes and LaPatra, "Circuit Synthesis and Design." Now out of
print, but occasionally you can find a cheap old copy used.

I'd also suggest the original poster does a couple of the exercises (he
might have already), doing what you describe with building a low frequency
prototype filter and then transforming it to a different type and frequency
range, but just for, e.g., a second order filter. There's not much to be
gained by doing the 'drudgery,' as you say, for a higher order filter other
than proficiency in aithrmetic.

---Joel Kolstad

John Jardine wrote:
I can't suggest a decent book, as the general purpose books seem over
complicated and destined to be written by maths people for maths people.

I'd suggest Temes and LaPatra, "Circuit Synthesis and Design." Now out of
print, but occasionally you can find a cheap old copy used.

I'd also suggest the original poster does a couple of the exercises (he
might have already), doing what you describe with building a low frequency
prototype filter and then transforming it to a different type and frequency
range, but just for, e.g., a second order filter. There's not much to be
gained by doing the 'drudgery,' as you say, for a higher order filter other
than proficiency in aithrmetic.

---Joel Kolstad

In article , (Tom
Bruhns) writes:

Download RFSim99 and let it do it for you. Download Filter Design
from AADE's site and let it do it for you. Download NuHertz's demo
version of their filter design software and let it do it for you.
There may even be web aplets that will do it for you. And then you
can play with configurations and shape factors and the inevitable
tradeoffs between complexity and shape factor.

Or, you can ask me for a free package of LCie4 that I first wrote
back in '93 and have revised only slightly since then. DOS level,
no fancy screens or prettyness in graphics...but LCie4 allows
individual component value variations, actual unloaded Q values,
and Sensitivity (output response or input impedance) to any
specified tolerance of all-Cs or all-Ls. Works under Win98 and
WinXP.

Does 3 to 9 section Lowpass, Highpass, Bandpass, and Bandstop
and 2 to 9 section "resonator" Bandpass filters with choice of "high"
or "low" impedance coupling by C or L. Butterworth, Tchebyshev
(or Chebyshev, heh heh), or one kind of Elliptic, choice of minimum
L or minimum C configurations on odd sections (or 'orders' as some
insist on calling them).

Output Response and Input Impedance are separate analyses, not
part of the calculation of filter component Synthesis. Q of all Cs or
all Ls is frequency-independent; i.e., same specified value at all
frequencies of analysis. That's a compromise to simplify analysis;
purists want a SPICE-like thing, no doubt, where they can model
everything...:-) Chances are those purists never built a real one to
experience the joy of transmutation of paper into hardware.

A "parts listing" command yields a half-S version of a "schematic"
good enough to show the values of separate components, on-screen
or on paper. LCie4 is fast and thorough, lets one play with values
and affect on response until one gets tired of doing so. :-)

Less than 250K in total package size, sent as an e-mail attachment
in response to polite e-mail inquiry. It's free...do not ask for money
back if dissatisfied...

Len Anderson
retired (from regular hours) electronic engineer person

In article , (Tom
Bruhns) writes:

Download RFSim99 and let it do it for you. Download Filter Design
from AADE's site and let it do it for you. Download NuHertz's demo
version of their filter design software and let it do it for you.
There may even be web aplets that will do it for you. And then you
can play with configurations and shape factors and the inevitable
tradeoffs between complexity and shape factor.

Or, you can ask me for a free package of LCie4 that I first wrote
back in '93 and have revised only slightly since then. DOS level,
no fancy screens or prettyness in graphics...but LCie4 allows
individual component value variations, actual unloaded Q values,
and Sensitivity (output response or input impedance) to any
specified tolerance of all-Cs or all-Ls. Works under Win98 and
WinXP.

Does 3 to 9 section Lowpass, Highpass, Bandpass, and Bandstop
and 2 to 9 section "resonator" Bandpass filters with choice of "high"
or "low" impedance coupling by C or L. Butterworth, Tchebyshev
(or Chebyshev, heh heh), or one kind of Elliptic, choice of minimum
L or minimum C configurations on odd sections (or 'orders' as some
insist on calling them).

Output Response and Input Impedance are separate analyses, not
part of the calculation of filter component Synthesis. Q of all Cs or
all Ls is frequency-independent; i.e., same specified value at all
frequencies of analysis. That's a compromise to simplify analysis;
purists want a SPICE-like thing, no doubt, where they can model
everything...:-) Chances are those purists never built a real one to
experience the joy of transmutation of paper into hardware.

A "parts listing" command yields a half-S version of a "schematic"
good enough to show the values of separate components, on-screen
or on paper. LCie4 is fast and thorough, lets one play with values
and affect on response until one gets tired of doing so. :-)

Less than 250K in total package size, sent as an e-mail attachment
in response to polite e-mail inquiry. It's free...do not ask for money
back if dissatisfied...

Len Anderson
retired (from regular hours) electronic engineer person

On Fri, 05 Mar 2004 16:18:32 +0000, Paul Burridge wrote:

On Fri, 05 Mar 2004 11:28:51 GMT, Active8
wrote:

He already has a book with a fair amount of tables and all the
explainations on how to go about the design. I'm wondering WTF
here. Hey paul. wtf?

Fair point. I just wondered if anyone had a design similar to
requirements *already* lying around they might post it, that's all.
Save me a bit of scrawling and button-pushing. I didn't wish to put
anyone out.
Thanks to all who responded, BTW. Some very useful steers.

Yes. I had to check my disorganized drive to make sure I'd been to
all those links. BTDT. I like having tools like that around to
compare what they say with what I figgered out on paper.
--
Best Regards,
Mike

On Fri, 05 Mar 2004 16:18:32 +0000, Paul Burridge wrote:

On Fri, 05 Mar 2004 11:28:51 GMT, Active8
wrote:

He already has a book with a fair amount of tables and all the
explainations on how to go about the design. I'm wondering WTF
here. Hey paul. wtf?

Fair point. I just wondered if anyone had a design similar to
requirements *already* lying around they might post it, that's all.
Save me a bit of scrawling and button-pushing. I didn't wish to put
anyone out.
Thanks to all who responded, BTW. Some very useful steers.

Yes. I had to check my disorganized drive to make sure I'd been to
all those links. BTDT. I like having tools like that around to
compare what they say with what I figgered out on paper.
--
Best Regards,
Mike

On Thu, 04 Mar 2004 23:43:21 +0000, Paul Burridge
wrote:

Hi all,

I need a simple design for a BPF with a centre frequency of 17.2Mhz.,
corner frequencies one Meg or therabouts either side of that and a
reasonably good shape factor (nothing critical here). I don't mind
having to do a bit of tweaking of values myself if anyone has
something close to that they can post. Passive L/C combinations only,
please. (assume the usual 50R nominal Zs.)
Thanks,

Since your too lazy to do this, here, have a few designs. You can run
these on LTSpice. What is reasonably good shape factor?
Mark

Butterworth, pi-input, 1st-order low-pass prototype
Vin 1 0 AC 2.0000V
Rin 1 2 50.000
L1 2 0 26.899n
C1 2 0 3.1831n
Rout 2 0 50.000
..AC DEC 396 253.58kHz 1.1667GHz
..PROBE
..END

Butterworth, pi-input, 2nd-order low-pass prototype
Vin 1 0 AC 2.0000V
Rin 1 2 50.000
L1 2 0 38.041n
C1 2 0 2.2508n
L2 2 3 5.6270u
C2 3 4 15.216p
Rout 4 0 50.000
..AC DEC 792 5.7378MegHz 51.560MegHz
..PROBE
..END

Butterworth, loose coupled, 2nd-order low-pass prototype
Vin 1 0 AC 2.0000V
Rin 1 2 50.000
C1 2 0 2.0654nF
L1 2 0 38.046nH
C2 2 3 185.01pF
C3 3 0 2.0654nF
L2 3 0 38.046nH
Rout 3 0 50.000
..AC DEC 792 5.7378MegHz 51.560MegHz
..PROBE
..END

Butterworth, pi-input, 3rd-order low-pass prototype
Vin 1 0 AC 2.0000V
Rin 1 2 50.000
L1 2 0 53.798n
C1 2 0 1.5915n
L2 2 3 7.9577u
C2 3 4 10.760p
L3 4 0 53.798n
C3 4 0 1.5915n
Rout 4 0 50.000
..AC DEC 1188 10.224MegHz 28.937MegHz
..PROBE
..END

Butterworth, loose coupled, 3rd-order low-pass prototype
Vin 1 0 AC 2.0000V
Rin 1 2 50.000
C1 2 0 1.4607nF
L1 2 0 53.798nH
C2 2 3 130.84pF
C3 3 0 1.3299nF
L2 3 0 53.798nH
C4 3 4 130.84pF
C5 4 0 1.4607nF
L3 4 0 53.798nH
Rout 4 0 50.000
..AC DEC 1188 10.224MegHz 28.937MegHz
..PROBE
..END

Butterworth, loose coupled, 8th-order low-pass prototype
Vin 1 0 AC 2.0000V
Rin 1 2 50.000
C1 2 0 511.07pF
L1 2 0 137.94nH
C2 2 3 109.63pF
C3 3 0 457.95pF
L2 3 0 137.94nH
C4 3 4 53.121pF
C5 4 0 527.60pF
L3 4 0 137.94nH
C6 4 5 39.985pF
C7 5 0 543.91pF
L4 5 0 137.94nH
C8 5 6 36.809pF
C9 6 0 543.91pF
L5 6 0 137.94nH
C10 6 7 39.985pF
C11 7 0 527.60pF
L6 7 0 137.94nH
C12 7 8 53.121pF
C13 8 0 457.95pF
L7 8 0 137.94nH
C14 8 9 109.63pF
C15 9 0 511.07pF
L8 9 0 137.94nH
Rout 9 0 50.000
..AC DEC 3170 15.208MegHz 19.453MegHz
..PROBE
..END