Pictures available in ABSE
The top trace (yellow) is taken between C4 and R2. The bottom trace (cyan)
is taken at the base of the transistor. There is a switchercad file, but
the simulation will show allot of distortion that really isn't present in
the prototype circuit, because of lots of circuit capactance I suspect. R1
was something I was playing with to try and tame the voltage across L1/C3
being applied to the base.


Hello all,

I was tinkering with this LC oscillator (Colpitts/Clapp) this weekend. I
arrived at the values of C1 and C2 empirically after starting with a crystal
oscillator circuit. The values in the original circuit created a horrid
waveform that looked allot like the simulation. After much tinkering around
and simulating, I come to the conclusion that getting a perfect waveform is
nearly impossible, especially with big swing. It seems that the transistor
likes to take a bite out of the right half of the peak of the wave.

What is the secret to beautiful waveforms? Do I need another LC resonator
on the output to fix it up? I mean, I'm getting a pretty nice wave now, but
there is still some distortion that you can just see at the top of the peaks
on the yellow trace.

How do you control the peak voltages of an LC resonattor without mangling
the waveform? The waveform at the junction of L1/C3 is of course quite
beautiful, how do I get it from there to the output? ;-)

I realize that I will need a buffer stage(s) before I can make any real use
of the signal, but I want the input to the buffer to be as perfect as
possible.

Thanks :-)

Sorry about the multiple posts. I use OE and it was telling me that it
didn't work, obviously it did.




Anthony Fremont wrote:
Pictures available in ABSE
The top trace (yellow) is taken between C4 and R2. The bottom trace
(cyan) is taken at the base of the transistor. There is a
switchercad file, but the simulation will show allot of distortion
that really isn't present in the prototype circuit, because of lots
of circuit capactance I suspect. R1 was something I was playing with
to try and tame the voltage across L1/C3 being applied to the base.


Hello all,

I was tinkering with this LC oscillator (Colpitts/Clapp) this
weekend. I arrived at the values of C1 and C2 empirically after
starting with a crystal oscillator circuit. The values in the
original circuit created a horrid waveform that looked allot like the
simulation. After much tinkering around and simulating, I come to
the conclusion that getting a perfect waveform is nearly impossible,
especially with big swing. It seems that the transistor likes to
take a bite out of the right half of the peak of the wave.
What is the secret to beautiful waveforms? Do I need another LC
resonator on the output to fix it up? I mean, I'm getting a pretty
nice wave now, but there is still some distortion that you can just
see at the top of the peaks on the yellow trace.

How do you control the peak voltages of an LC resonattor without
mangling the waveform? The waveform at the junction of L1/C3 is of
course quite beautiful, how do I get it from there to the output? ;-)

I realize that I will need a buffer stage(s) before I can make any
real use of the signal, but I want the input to the buffer to be as
perfect as possible.

Thanks :-)

Anthony Fremont wrote:

Sorry about the multiple posts. I use OE and it was telling me that it
didn't work, obviously it did.

Let me recommend Netscape 4.8 !

Graham

"Anthony Fremont" wrote in message
...
Pictures available in ABSE
The top trace (yellow) is taken between C4 and R2. The bottom trace
(cyan)
is taken at the base of the transistor. There is a switchercad file, but
the simulation will show allot of distortion that really isn't present in
the prototype circuit, because of lots of circuit capactance I suspect.
R1
was something I was playing with to try and tame the voltage across L1/C3
being applied to the base.


Hello all,

I was tinkering with this LC oscillator (Colpitts/Clapp) this weekend. I
arrived at the values of C1 and C2 empirically after starting with a
crystal
oscillator circuit. The values in the original circuit created a horrid
waveform that looked allot like the simulation. After much tinkering
around
and simulating, I come to the conclusion that getting a perfect waveform
is
nearly impossible, especially with big swing. It seems that the
transistor
likes to take a bite out of the right half of the peak of the wave.

What is the secret to beautiful waveforms? Do I need another LC resonator
on the output to fix it up? I mean, I'm getting a pretty nice wave now,
but
there is still some distortion that you can just see at the top of the
peaks
on the yellow trace.

How do you control the peak voltages of an LC resonattor without mangling
the waveform? The waveform at the junction of L1/C3 is of course quite
beautiful, how do I get it from there to the output? ;-)

I realize that I will need a buffer stage(s) before I can make any real
use
of the signal, but I want the input to the buffer to be as perfect as
possible.

The drive from the amplifier will inevitably be heavily clipped,
this is cuased by the need to have sufficient gain to start oscillation,
the clipping is the efffect of getting rid of the excess gain.

The current waveform is usually quite distorted,
usualy this is perfect to generate frequency multiples with a suitable tank.

There was a low distortion oscillator discussed in here not so long ago.
I think you do better with the inductor in the collector,
at least you need to take the output from the tuned circuit with a
capacitive divider,
but not via the transistor,
wich is easier with the inductor in the collector.

Also a jfet is ideal as it has a much softer clipping.

have you given up with the ne602 and the crystal ?

Colin =^.^=

"colin" wrote in message
...
"Anthony Fremont" wrote in message
...

Also a jfet is ideal as it has a much softer clipping.


I played around with it, changed it to a fet with collector tuned circuit.

It can stil be improved if need be.

ps the 10pf cap can be a trimmer.

Colin =^.^=

colin wrote:
"Anthony Fremont" wrote in message

The drive from the amplifier will inevitably be heavily clipped,
this is cuased by the need to have sufficient gain to start
oscillation, the clipping is the efffect of getting rid of the excess
gain.

Ok, I guess that makes sense. You want plenty of gain, and with plenty of
gain comes clipping.

The current waveform is usually quite distorted,
usualy this is perfect to generate frequency multiples with a
suitable tank.


This would be for overtone oscillators?

There was a low distortion oscillator discussed in here not so long
ago. I think you do better with the inductor in the collector,
at least you need to take the output from the tuned circuit with a
capacitive divider,
but not via the transistor,
wich is easier with the inductor in the collector.

Also a jfet is ideal as it has a much softer clipping.

I found an MPF102 here and doctored up the circuit for it, much nicer
waveform. Much less voltage in the tank though, but it starts fine. This
is fun. ;-) I'm going to wind myself a lower inductance coil and see if I
can get it to run at 10MHz.

have you given up with the ne602 and the crystal ?

Just side tracked for now. ;-) The crystal is kinda restrictive, but I'm
going back to the 602 soon. I'm going to try to make a single conversion
superhet for WWV reception at 10MHz. I have some IF cans of various colors,
so I'm sure I have some for 455kHz. If I can make an LO that will stay
within 1kc of 9.545 at room temperature, I'm going to go with it to beat
against WWV.

Right now my prototype circuit is drifting back and forth over a 1.5kc range
while running at 4MHz. I'm using ordinary ceramic caps to set the
frequency, and everything is on a breadboard. The 2kc drift seems to
coincide with the air conditioning induced temperature changes in the room.
Plenty of room for improvement on that. ;-)

I just wound another coil for 5uH. Things get more picky as the frequency
goes up, but I got it running at 9.5Mhz and it's staying within a few kcs.
Interesting stuff. :-)

"Anthony Fremont" wrote in message
...
colin wrote:
"Anthony Fremont" wrote in message

The drive from the amplifier will inevitably be heavily clipped,
this is cuased by the need to have sufficient gain to start
oscillation, the clipping is the efffect of getting rid of the excess
gain.

Ok, I guess that makes sense. You want plenty of gain, and with plenty of
gain comes clipping.

theres two types of clipping as youve probably figured out, current clipping
where the transistor turns off completly, and voltage clipping where the
transistor saturates,
with a bjt the later cuases problems with storage time, wich is variable
with temprature, hence makes for unstable HF oscillators. the later is
better but is difficult to ensure this happens before the former with out
some form of level feedback.

with jfet however either form of clipping isnt so bad, if you notice the
circuit I put here there is a smal amount of positive gate current at
clipping. but I just slapped those values together.

The current waveform is usually quite distorted,
usualy this is perfect to generate frequency multiples with a
suitable tank.

This would be for overtone oscillators?

well not crystal overtones exactly, but frequency multipliyng oscilators,
say to generate over 100Mhz -
crystals tend to be difficult to get much higher than this even with 5th
overtone.

There was a low distortion oscillator discussed in here not so long
ago. I think you do better with the inductor in the collector,
at least you need to take the output from the tuned circuit with a
capacitive divider,
but not via the transistor,
wich is easier with the inductor in the collector.

Also a jfet is ideal as it has a much softer clipping.

I found an MPF102 here and doctored up the circuit for it, much nicer
waveform. Much less voltage in the tank though, but it starts fine. This
is fun. ;-) I'm going to wind myself a lower inductance coil and see if
I can get it to run at 10MHz.

you can still easily get nearly 2x full supplly swing accros the tank with a
jfet.
the resistor can also be replaced by an inductor getting you to exactly 2x
supply.

have you given up with the ne602 and the crystal ?

Just side tracked for now. ;-) The crystal is kinda restrictive, but I'm
going back to the 602 soon. I'm going to try to make a single conversion
superhet for WWV reception at 10MHz. I have some IF cans of various
colors, so I'm sure I have some for 455kHz. If I can make an LO that will
stay within 1kc of 9.545 at room temperature, I'm going to go with it to
beat against WWV.

For interest try putting the crystal across the inductor when it is tuned at
the right frequency and see what happens, then see what difference it makes
at different frequencies. Ive played about with a bag of10mhz crystals to
find 2 that were precisly a some khz apart, eventually I made one lower by
degrading by heating it with a soldering for some time, its to easy to
seriously degrade it though.

unfortunatly most crystals are at a peak of frequency at room temperature,
above and below this temp they start to fall in freq,
untill you reach the higher flex in the tempcooef curve where the freq
starts to rise again.
this can be quite hot and the tempcoeef is then rather steep.

another trick is to look for crystals that will give you a correct frequency
at overtone, 3,5,7th
or use a frequency multiplier x2,x4 etc I spent ages doing this once.
you can even go up in freq then divide by 1-256 with ttl etc.
ofc a pll is the the other option.

Right now my prototype circuit is drifting back and forth over a 1.5kc
range while running at 4MHz. I'm using ordinary ceramic caps to set the
frequency, and everything is on a breadboard. The 2kc drift seems to
coincide with the air conditioning induced temperature changes in the
room. Plenty of room for improvement on that. ;-)

The transistor capacitance changes with temperature, not to mention cheap
ceramic caps too,
god knows what the temp coef of breadboard cap is lol.
not to mention anything other than air cored inductors.
with a low capacitance transistor you would get better stability,
with a lower inductance the transistor cap has less effect too.
you can do a whole lot better than that with just the right parts.

I have a nice temp compensated oscillator wich I got with some other stuff
but not a usefull frequency.
unfortunatly a different crystal didnt match the temp comensation,
but you can use capacitors wich have the opposite temp cooef.
they also make trimmers like that too, although you can just put an ordinary
trimmer in series with it.


I just wound another coil for 5uH. Things get more picky as the frequency
goes up, but I got it running at 9.5Mhz and it's staying within a few kcs.
Interesting stuff. :-)

last colpits I built was a wide range 2:1 1-2ghz VCO with a PHEMT (fet)
this had an extremely low capacitance, but is probably a bit over the top
for 10mhz.
I had 2 wich I tried to keep 455khz apart, very difficult, they liked to
sync up if they got closer than 1Mhz.
(were a few threads about it.)

Colin =^.^=

Anthony...checkout this osc in LTSpice..mbsg
"Anthony Fremont" wrote in message
...
Pictures available in ABSE
The top trace (yellow) is taken between C4 and R2. The bottom trace
(cyan)
is taken at the base of the transistor. There is a switchercad file, but
the simulation will show allot of distortion that really isn't present in
the prototype circuit, because of lots of circuit capactance I suspect.
R1
was something I was playing with to try and tame the voltage across L1/C3
being applied to the base.


Hello all,

I was tinkering with this LC oscillator (Colpitts/Clapp) this weekend. I
arrived at the values of C1 and C2 empirically after starting with a
crystal
oscillator circuit. The values in the original circuit created a horrid
waveform that looked allot like the simulation. After much tinkering
around
and simulating, I come to the conclusion that getting a perfect waveform
is
nearly impossible, especially with big swing. It seems that the
transistor
likes to take a bite out of the right half of the peak of the wave.

What is the secret to beautiful waveforms? Do I need another LC resonator
on the output to fix it up? I mean, I'm getting a pretty nice wave now,
but
there is still some distortion that you can just see at the top of the
peaks
on the yellow trace.

How do you control the peak voltages of an LC resonattor without mangling
the waveform? The waveform at the junction of L1/C3 is of course quite
beautiful, how do I get it from there to the output? ;-)

I realize that I will need a buffer stage(s) before I can make any real
use
of the signal, but I want the input to the buffer to be as perfect as
possible.

Thanks :-)

begin 666 colposcillator3.asc
Attachment saved: C:\0DNLOAD\colposcillator3.asc

end

Check the tank Q. You have the emitter resistor in parallel with the
main tank capacitance. At 3MHz, Xc is 377 ohms. With an emitter
resistor of 1k, the tank Q is just over 2. This is not adequate for
low phase noise.

The series resistance of 1 ohm for the inductor is a bit optimistic.
This would give a Q of 377 for the inductor. Values near this can be
achieved at 3MHz, but it takes special care. A more probable value
is 37, which means increasing the series resistance to 10 ohms.

The capacitance ratio of 5n / 140pf = 35 is quite large. With an
inductor Q of 37, the oscillator may fail to start at higher values
of emitter resistance. So you are constrained to operate at emitter
resistances that give a low tank Q, thus poor phase noise.

Regards,

Mike Monett

"Mike Monett" wrote in message
...
begin 666 colposcillator3.asc
Attachment saved: C:\0DNLOAD\colposcillator3.asc

end

Check the tank Q. You have the emitter resistor in parallel with the
main tank capacitance. At 3MHz, Xc is 377 ohms. With an emitter
resistor of 1k, the tank Q is just over 2. This is not adequate for
low phase noise.

The series resistance of 1 ohm for the inductor is a bit optimistic.
This would give a Q of 377 for the inductor. Values near this can be
achieved at 3MHz, but it takes special care. A more probable value
is 37, which means increasing the series resistance to 10 ohms.

The capacitance ratio of 5n / 140pf = 35 is quite large. With an
inductor Q of 37, the oscillator may fail to start at higher values
of emitter resistance. So you are constrained to operate at emitter
resistances that give a low tank Q, thus poor phase noise.

Regards,

Mike Monett

you could replace the emitter resistor with an inductor,
say 100uh.
I played around with the jfet ltspice circuit I posted,
it showed less than -40db harmonic,
virtually lost in the digitising noise.

Colin =^.^=

colin wrote:
"Anthony Fremont" wrote in message
...
colin wrote:
"Anthony Fremont" wrote in message

The drive from the amplifier will inevitably be heavily clipped,
this is cuased by the need to have sufficient gain to start
oscillation, the clipping is the efffect of getting rid of the
excess gain.

Ok, I guess that makes sense. You want plenty of gain, and with
plenty of gain comes clipping.

theres two types of clipping as youve probably figured out, current
clipping where the transistor turns off completly, and voltage
clipping where the transistor saturates,
with a bjt the later cuases problems with storage time, wich is
variable with temprature, hence makes for unstable HF oscillators.
the later is better but is difficult to ensure this happens before
the former with out some form of level feedback.

with jfet however either form of clipping isnt so bad, if you notice
the circuit I put here there is a smal amount of positive gate
current at clipping. but I just slapped those values together.

Ok, I think I'll stick with JFET based oscillators from now on. I like the
waveform, much better anyway. ;-)

The current waveform is usually quite distorted,
usualy this is perfect to generate frequency multiples with a
suitable tank.

This would be for overtone oscillators?

well not crystal overtones exactly, but frequency multipliyng
oscilators, say to generate over 100Mhz -
crystals tend to be difficult to get much higher than this even with
5th overtone.

I have some crystals for 2M HAM stuff in an old Kenwood TS-700A. I assume
these really don't oscillate at ~150MHz, but are in the 30MHz range but cut
to produce lots of harmonics.

There was a low distortion oscillator discussed in here not so long
ago. I think you do better with the inductor in the collector,
at least you need to take the output from the tuned circuit with a
capacitive divider,
but not via the transistor,
wich is easier with the inductor in the collector.

Also a jfet is ideal as it has a much softer clipping.

I found an MPF102 here and doctored up the circuit for it, much nicer
waveform. Much less voltage in the tank though, but it starts fine.
This is fun. ;-) I'm going to wind myself a lower inductance coil
and see if I can get it to run at 10MHz.

you can still easily get nearly 2x full supplly swing accros the tank
with a jfet.
the resistor can also be replaced by an inductor getting you to
exactly 2x supply.

Ok, I've been doing allot of tinkering yesterday and have some neat stuff
working here. :-) I picked up some 1mH inductors to use in the oscillator
and buffer. Beautiful results, until the kickback from the 74HC14, but I
can live with it.


have you given up with the ne602 and the crystal ?

Just side tracked for now. ;-) The crystal is kinda restrictive,
but I'm going back to the 602 soon. I'm going to try to make a
single conversion superhet for WWV reception at 10MHz. I have some
IF cans of various colors, so I'm sure I have some for 455kHz. If I
can make an LO that will stay within 1kc of 9.545 at room
temperature, I'm going to go with it to beat against WWV.

For interest try putting the crystal across the inductor when it is
tuned at the right frequency and see what happens, then see what
difference it makes at different frequencies. Ive played about with a
bag of10mhz crystals to find 2 that were precisly a some khz apart,
eventually I made one lower by degrading by heating it with a
soldering for some time, its to easy to seriously degrade it though.

unfortunatly most crystals are at a peak of frequency at room
temperature, above and below this temp they start to fall in freq,
untill you reach the higher flex in the tempcooef curve where the freq
starts to rise again.
this can be quite hot and the tempcoeef is then rather steep.

another trick is to look for crystals that will give you a correct
frequency at overtone, 3,5,7th
or use a frequency multiplier x2,x4 etc I spent ages doing this once.
you can even go up in freq then divide by 1-256 with ttl etc.
ofc a pll is the the other option.

Thanks for the info, I'll have to get back to the crystals, but I'm having
to much fun with coils now. ;-)

Right now my prototype circuit is drifting back and forth over a
1.5kc range while running at 4MHz. I'm using ordinary ceramic caps
to set the frequency, and everything is on a breadboard. The 2kc
drift seems to coincide with the air conditioning induced
temperature changes in the room. Plenty of room for improvement on
that. ;-)

The transistor capacitance changes with temperature, not to mention
cheap ceramic caps too,
god knows what the temp coef of breadboard cap is lol.
not to mention anything other than air cored inductors.
with a low capacitance transistor you would get better stability,
with a lower inductance the transistor cap has less effect too.
you can do a whole lot better than that with just the right parts.

I have a nice temp compensated oscillator wich I got with some other
stuff but not a usefull frequency.
unfortunatly a different crystal didnt match the temp comensation,
but you can use capacitors wich have the opposite temp cooef.
they also make trimmers like that too, although you can just put an
ordinary trimmer in series with it.

I added a PIC chip for stability. ;-)

I just wound another coil for 5uH. Things get more picky as the
frequency goes up, but I got it running at 9.5Mhz and it's staying
within a few kcs. Interesting stuff. :-)

last colpits I built was a wide range 2:1 1-2ghz VCO with a PHEMT
(fet) this had an extremely low capacitance, but is probably a bit
over the top for 10mhz.
I had 2 wich I tried to keep 455khz apart, very difficult, they liked
to sync up if they got closer than 1Mhz.
(were a few threads about it.)

Cool, but way beyond my reach. Yeah mechanical systems are like that. I
collect wind-up watches and many beat at the same rate (18,000 BPH). You
can put a bunch in say a sock drawer (box with hard, flexible bottom to
carry vibrations) and they'll all sync up after a while.

Murray Bloom wrote:
Anthony...checkout this osc in LTSpice..mbsg

Thanks, I'll check it out. :-)