Impedance matching / LPF question

[neddie]

Hi to all.
RF is not really my strong point , so I could use some advice here.
I have a simple Collpits oscillator running at 433.92Mhz. It has a
strong output signal at 433Mhz , but unfortunately also at 868Mhz.
In fact the signal at 868Mhz is also very strong. I would like to
attenuate this harmonic , and the others
that I can't see(My Spectrum analyser only goes to 1Gig), which are
definitely there.
I want to put in a 5th order low pass filter. I have Chris Bowick's RF
design book , which shows some
easy ways to design a filter for this. The problem is I don't know how
to calculate the output impedance
of the oscillator , which is obviously needed in the calcs. I'm
assuming the load , which is a 1/4 length piece
of wire to be about 75 ohms , give or take.
The data sheet I have for the BRF93a only shows S11 at 400 and
500MHz , so I have to interpret the
impedance at 433Mhz , but it gives ma a starting point. This is also
at Vce of 8V and 25ma , which is not exactly where I'm sitting , but I
think is close enough.
I get a s11 of 0.211 , angle -137deg.
Plotting this on a Smith chart shows an impedance of 35.3-J10.5 ohms.
I assume this to be roughly the output impedance of the transistor.
I'm not even sure if I'm on the right track at all :0(
Assuming this is correct , Do I use this as the source impedance in my
calculations , or do I then have
to take into account the other components around the transistor.(I
assume so)
It would then be (attached is a schematic in LTSpice)
(35.3-j10.5 + 47) // (L1 // (C1+c2)) . Hope this makes sense. It's the
output impedance of the transistor
in series with the emitter resistor. All this in parallel with the
parallel combination of the inductor and
the 2 caps in series , phew.....
What I need to know is am I on the correct path at all , or am I
missing the boat all together :0(
Any pointers would be great.
Cheers
Rob


Version 4
SHEET 1 880 680
WIRE -16 16 -160 16
WIRE 96 16 -16 16
WIRE 176 16 96 16
WIRE 176 32 176 16
WIRE -16 80 -16 16
WIRE 96 96 96 80
WIRE -160 144 -160 16
WIRE 176 144 176 112
WIRE 288 144 176 144
WIRE 432 144 288 144
WIRE 448 144 432 144
WIRE 560 144 528 144
WIRE 576 144 560 144
WIRE 672 144 656 144
WIRE 704 144 672 144
WIRE 784 144 768 144
WIRE 176 160 176 144
WIRE 288 176 288 144
WIRE -16 208 -16 160
WIRE 112 208 -16 208
WIRE 784 224 784 144
WIRE 432 240 432 144
WIRE 560 240 560 144
WIRE 672 240 672 144
WIRE 176 272 176 256
WIRE 288 272 288 240
WIRE 288 272 176 272
WIRE -16 288 -16 208
WIRE 176 288 176 272
WIRE 288 288 288 272
WIRE -160 400 -160 224
WIRE -16 400 -16 336
WIRE -16 400 -160 400
WIRE 176 400 176 368
WIRE 176 400 -16 400
WIRE 288 400 288 352
WIRE 288 400 176 400
WIRE 432 400 432 304
WIRE 432 400 288 400
WIRE 560 400 560 304
WIRE 560 400 432 400
WIRE 672 400 672 304
WIRE 672 400 560 400
WIRE 784 400 784 304
WIRE 784 400 672 400
WIRE 176 416 176 400
FLAG 176 416 0
FLAG 96 96 0
SYMBOL res 160 272 R0
SYMATTR InstName R1
SYMATTR Value 47
SYMBOL ind 160 16 R0
SYMATTR InstName L1
SYMATTR Value 68n
SYMBOL cap 272 176 R0
SYMATTR InstName C1
SYMATTR Value 1.5p
SYMBOL cap 272 288 R0
SYMATTR InstName C2
SYMATTR Value 3.3p
SYMBOL voltage -160 128 R0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
SYMATTR InstName V1
SYMATTR Value PULSE(0 12 0 10n)
SYMBOL res -32 64 R0
SYMATTR InstName R2
SYMATTR Value 10k
SYMBOL cap 80 16 R0
SYMATTR InstName C5
SYMATTR Value 100n
SYMBOL cap 544 240 R0
SYMATTR InstName C6
SYMBOL cap 656 240 R0
SYMATTR InstName C7
SYMBOL ind 544 128 R90
WINDOW 0 5 56 VBottom 0
WINDOW 3 32 56 VTop 0
SYMATTR InstName L2
SYMBOL ind 672 128 R90
WINDOW 0 5 56 VBottom 0
WINDOW 3 32 56 VTop 0
SYMATTR InstName L3
SYMBOL res 768 208 R0
SYMATTR InstName R3
SYMATTR Value 75
SYMBOL cap 768 128 R90
WINDOW 0 0 32 VBottom 0
WINDOW 3 32 32 VTop 0
SYMATTR InstName C8
SYMATTR Value 1n
SYMBOL cap 416 240 R0
SYMATTR InstName C3
SYMBOL npn 112 160 R0
SYMATTR InstName Q1
SYMATTR Value BFR93a
TEXT 0 -16 Left 0 !.tran 0 100u 0
TEXT -88 360 Left 0 ;433.92Mhz SAW
RECTANGLE Normal 0 336 -32 288

 

[[email protected]]

Hi to all.
RF is not really my strong point , so I could use some advice here.
 I have a simple Collpits oscillator running at 433.92Mhz. It has a
strong output signal at 433Mhz , but unfortunately also at 868Mhz.
In fact the signal at 868Mhz is also very strong. I would like to
attenuate this harmonic , and the others
that I can't see(My Spectrum analyser only goes to 1Gig), which are
definitely there.
I want to put in a 5th order low pass filter. I have Chris Bowick's RF
design book , which shows some
easy ways to design a filter for this. The problem is I don't know how
to calculate the output impedance
of the oscillator , which is obviously needed in the calcs. I'm
assuming the load , which is a 1/4 length piece
of wire to be about 75 ohms , give or take.
The data sheet I have for the BRF93a only shows S11 at 400 and
500MHz , so I have to interpret the
impedance at 433Mhz , but it gives ma a starting point. This is also
at Vce of 8V and 25ma , which is not exactly where I'm sitting , but I
think is close enough.
I get a s11 of 0.211 , angle  -137deg.
Plotting this on a Smith chart shows an impedance of 35.3-J10.5 ohms.
I assume this to be roughly the output impedance of the transistor.
I'm not even sure if I'm on the right track at all :0(
Assuming this is correct , Do I use this as the source impedance in my
calculations , or do I then have
to take into account the other components around the transistor.(I
assume so)
It would then be (attached is a schematic in LTSpice)
(35.3-j10.5 + 47) // (L1 // (C1+c2)) . Hope this makes sense. It's the
output impedance of the transistor
in series with the emitter resistor. All this in parallel with the
parallel combination of the inductor and
the 2 caps in series , phew.....
What I need to know is am I on the correct path at all , or am I
missing the boat all together :0(
Any pointers would be great.
Cheers
Rob

Version 4
SHEET 1 880 680
WIRE -16 16 -160 16
WIRE 96 16 -16 16
WIRE 176 16 96 16
WIRE 176 32 176 16
WIRE -16 80 -16 16
WIRE 96 96 96 80
WIRE -160 144 -160 16
WIRE 176 144 176 112
WIRE 288 144 176 144
WIRE 432 144 288 144
WIRE 448 144 432 144
WIRE 560 144 528 144
WIRE 576 144 560 144
WIRE 672 144 656 144
WIRE 704 144 672 144
WIRE 784 144 768 144
WIRE 176 160 176 144
WIRE 288 176 288 144
WIRE -16 208 -16 160
WIRE 112 208 -16 208
WIRE 784 224 784 144
WIRE 432 240 432 144
WIRE 560 240 560 144
WIRE 672 240 672 144
WIRE 176 272 176 256
WIRE 288 272 288 240
WIRE 288 272 176 272
WIRE -16 288 -16 208
WIRE 176 288 176 272
WIRE 288 288 288 272
WIRE -160 400 -160 224
WIRE -16 400 -16 336
WIRE -16 400 -160 400
WIRE 176 400 176 368
WIRE 176 400 -16 400
WIRE 288 400 288 352
WIRE 288 400 176 400
WIRE 432 400 432 304
WIRE 432 400 288 400
WIRE 560 400 560 304
WIRE 560 400 432 400
WIRE 672 400 672 304
WIRE 672 400 560 400
WIRE 784 400 784 304
WIRE 784 400 672 400
WIRE 176 416 176 400
FLAG 176 416 0
FLAG 96 96 0
SYMBOL res 160 272 R0
SYMATTR InstName R1
SYMATTR Value 47
SYMBOL ind 160 16 R0
SYMATTR InstName L1
SYMATTR Value 68n
SYMBOL cap 272 176 R0
SYMATTR InstName C1
SYMATTR Value 1.5p
SYMBOL cap 272 288 R0
SYMATTR InstName C2
SYMATTR Value 3.3p
SYMBOL voltage -160 128 R0
WINDOW 123 0 0 Left 0
WINDOW 39 0 0 Left 0
SYMATTR InstName V1
SYMATTR Value PULSE(0 12 0 10n)
SYMBOL res -32 64 R0
SYMATTR InstName R2
SYMATTR Value 10k
SYMBOL cap 80 16 R0
SYMATTR InstName C5
SYMATTR Value 100n
SYMBOL cap 544 240 R0
SYMATTR InstName C6
SYMBOL cap 656 240 R0
SYMATTR InstName C7
SYMBOL ind 544 128 R90
WINDOW 0 5 56 VBottom 0
WINDOW 3 32 56 VTop 0
SYMATTR InstName L2
SYMBOL ind 672 128 R90
WINDOW 0 5 56 VBottom 0
WINDOW 3 32 56 VTop 0
SYMATTR InstName L3
SYMBOL res 768 208 R0
SYMATTR InstName R3
SYMATTR Value 75
SYMBOL cap 768 128 R90
WINDOW 0 0 32 VBottom 0
WINDOW 3 32 32 VTop 0
SYMATTR InstName C8
SYMATTR Value 1n
SYMBOL cap 416 240 R0
SYMATTR InstName C3
SYMBOL npn 112 160 R0
SYMATTR InstName Q1
SYMATTR Value BFR93a
TEXT 0 -16 Left 0 !.tran 0 100u 0
TEXT -88 360 Left 0 ;433.92Mhz SAW
RECTANGLE Normal 0 336 -32 288

Hello Rob,

I would recommend you to make a band pass filter (or a low pass filter
with extreme ripple) as this requires less components for the same
attenuation. The antenna itself (for example) a resonating loop) can
be part of the bandpass filter. Don't make the BPF too narrow in
bandwidth to avoid component tolerance issues.

Regarding the antenna. When the physical size of your 433 MHz device
is very small with respect to a quarter wave length, the impedance of
your antenna may deviate significantly from 75 Ohms. The behavior of
a quarter wave depends on the ground. The ground is in your case your
transmitter + supply wiring (if present). When you fed the quarter
wave antenna over a large flat piece of metal, you will see something
between 30 and 50 Ohms.

Best regards,

Wim
PA3DJS
www.tetech.nl

 

[Andrew Holme]

I have a simple Collpits oscillator running at 433.92Mhz. It has a
strong output signal at 433Mhz , but unfortunately also at 868Mhz.
In fact the signal at 868Mhz is also very strong. I would like to
attenuate this harmonic , and the others
that I can't see(My Spectrum analyser only goes to 1Gig), which are
definitely there.
I want to put in a 5th order low pass filter. I have Chris Bowick's RF
design book , which shows some
easy ways to design a filter for this. The problem is I don't know how
to calculate the output impedance
of the oscillator , which is obviously needed in the calcs. I'm
assuming the load , which is a 1/4 length piece
of wire to be about 75 ohms , give or take.

[snip]

It's more a case of what input impedance the filter presents to the
oscillator. If the oscillator works with a load RL, then you could design a
filter with that input impedance.

 

[neddie]

I have a simple Collpits oscillator running at 433.92Mhz. It has a
strong output signal at 433Mhz , but unfortunately also at 868Mhz.
In fact the signal at 868Mhz is also very strong. I would like to
attenuate this harmonic , and the others
that I can't see(My Spectrum analyser only goes to 1Gig), which are
definitely there.
I want to put in a 5th order low pass filter. I have Chris Bowick's RF
design book , which shows some
easy ways to design a filter for this. The problem is I don't know how
to calculate the output impedance
of the oscillator , which is obviously needed in the calcs. I'm
assuming the load , which is a 1/4 length piece
of wire to be about 75 ohms , give or take.

[snip]

It's more a case of what input impedance the filter presents to the
oscillator.  If the oscillator works with a load RL, then you could design a
filter with that input impedance.

Does that mean I should load the oscillator with a resistive load to
find out where I can get max power out and maintain oscillation ,
then use this as my source resistance for the filter calcs?
Cheers
Rob

 

[Andrew Holme]

[snip]

Does that mean I should load the oscillator with a resistive load to
find out where I can get max power out and maintain oscillation ,
then use this as my source resistance for the filter calcs?

Sucking maximum power from an oscillator is not a good idea. It lowers the
working Q of the resonator and increases phase noise. In the extreme, it
may also make startup unreliable. It is common practice to follow the
oscillator with a buffer stage. This improves frequency stability by
isolating it from load impedance variations. In your case, it sounds like
you want to feed the antenna directly from the oscillator. Without a
buffer, you may find hand-capacity effects around the antenna troublesome at
433 MHz. There's a trade-off: the more you load the oscillator, the more
output you will get - up to a point, at the expense of stability and purity.
But yes you could test with a load resistor and then design your filter for
that input impedance.

Another way to reduce loading might be to replace the 3.3p cap with two in
series (e.g. 6.8p + 6.8p) and take your output from the middle.

 

[neddie]

[snip]

Does that mean I should load the oscillator with a resistive load to
find out where I can get max power out and maintain oscillation ,
then use this as my source resistance for the filter calcs?

Sucking maximum power from an oscillator is not a good idea.  It lowersthe
working Q of the resonator and increases phase noise.  In the extreme, it
may also make startup unreliable.  It is common practice to follow the
oscillator with a buffer stage.  This improves frequency stability by
isolating it from load impedance variations.   In your case, it sounds like
you want to feed the antenna directly from the oscillator.  Without a
buffer, you may find hand-capacity effects around the antenna troublesomeat
433 MHz.  There's a trade-off: the more you load the oscillator, the more
output you will get - up to a point, at the expense of stability and purity.
But yes you could test with a load resistor and then design your filter for
that input impedance.

Another way to reduce loading might be to replace the 3.3p cap with two in
series (e.g. 6.8p + 6.8p) and take your output from the middle.

Thanks for the help , I'll try what you suggested. I'm not going to
try and get to much power from the transmitter ,
so I won't try and "overload" it to much. Range does not have to be
huge , so maximum power is not that
much of an issue.I'm more interested in getting the principal
correct.I also assume over loading the transmitter
would make the harmonic issue worse!!

As to my original post. If I wanted to match an impedance to the
output of the transmitter , would my
methodology have been correct?

Cheers
Rob

 

[Baron]

Andrew Holme Inscribed thus:

[snip]

Does that mean I should load the oscillator with a resistive load to
find out where I can get max power out and maintain oscillation ,
then use this as my source resistance for the filter calcs?

Sucking maximum power from an oscillator is not a good idea. It
lowers the
working Q of the resonator and increases phase noise. In the extreme,
it
may also make startup unreliable. It is common practice to follow the
oscillator with a buffer stage. This improves frequency stability by
isolating it from load impedance variations. In your case, it sounds
like
you want to feed the antenna directly from the oscillator. Without a
buffer, you may find hand-capacity effects around the antenna
troublesome at
433 MHz. There's a trade-off: the more you load the oscillator, the
more output you will get - up to a point, at the expense of stability
and purity. But yes you could test with a load resistor and then
design your filter for that input impedance.

Another way to reduce loading might be to replace the 3.3p cap with
two in series (e.g. 6.8p + 6.8p) and take your output from the middle.

He will probably find that a tuned buffer amp also helps with a
reduction in the second harmonic !

 

[neddie]

-- snip --

You don't necessarily _want_ a good impedance match to the oscillator.
In fact, if your circuit is the power oscillator that it sounds like it
is, you have to carefully juggle the "good oscillator" characteristics
of your oscillator (frequency stability, phase noise if you care,
reliable and rapid starting, etc.) against the amount of power that you
can take out of it.

So take care about trying to load the oscillator with a "perfectly
matched" load -- you may well be loading it with an "oscillation killer".

If you have the time to delve, a good book on oscillator design may
help.  Here's the 2nd edition of the one I have; I can only claim it to
be "adequate", so if someone chips in with a better suggestion go with it:

http://www.powells.com/partner/30696/biblio/1884932304

Here's the 1st edition, with a floppy by gawd (if it still has it's
disk), at a much more reasonable price:

http://www.powells.com/partner/30696/biblio/0136425135

I think Hayward goes into oscillator design a bit, too, and I trust his
approach -- but the Rhea book is _just_ about oscillators:

http://www.powells.com/partner/30696/biblio/0872594920

--

Tim Wescott
Wescott Design Serviceshttp://www.wescottdesign.com

Do you need to implement control loops in software?
"Applied Control Theory for Embedded Systems" was written for you.
See details athttp://www.wescottdesign.com/actfes/actfes.html

Thanks for the tips.
In fact I think I do have a copy of Wes Haywards book around here
somewhere. Time for a tidy up :0)

I see what you are saying about trying to get to good a match , and in
the process killing the
oscillator.I'm still not sure if my methodology that I described in my
first post is correct.
Any comments?

 

[neddie]

I think the subject has been flogged quite thoroughly.  If I were to add
anything it would be to remember that because of the interaction between
the oscillator and the filter, you can't design the filter in isolation.

So you're kind of left with designing (and understanding) the whole
thing, instead of doing it in bits the way you might like to.

--

Tim Wescott
Wescott Design Serviceshttp://www.wescottdesign.com

Do you need to implement control loops in software?
"Applied Control Theory for Embedded Systems" was written for you.
See details athttp://www.wescottdesign.com/actfes/actfes.html

Thanks to all for the help :0)