amdx wrote:

Hi All,
Please see pictures and diagram at ABSE.

I duplicated a circuit for someone because they got a bad
output waveform.
Well I did too.
Capacitor = air variable set at 365pf
inductor = 41uh aircore
With freq set at 1.3Mhz I have the voltages as follows;
Collector - 11.8vdc
Base - 3.1vac riding on top of 3.2vdc
Emitter - 2.5vac riding on top of 3.9vdc (Don't be critical, it's not ac
if it sets on dc)

As I shift up in freq at about 2.7Mhz the voltages jump.

Capacitor = air variable set at 40pf
inductor = 41uh aircore
Freq = 3.7Mhz
Collector - 11.8vdc
Base - 10vac riding on top of 1.8vdc
emitter - 1.1vac riding on top of 9vdc

The base waveform in both senerios is much more sinewave (ish) than the
output.

Any ideas how to get a decent waveform? The next step will be a buffer,
but...

Note; original schematic used a crystal, I substituted the LC. Could this
be the
cause of the distortion?

Thanks, Mike


This circuit doesn't look very good from any standpoint other than
(possibly) being cheap. It's questionable for a crystal oscillator in
anything but el-cheapo consumer electronics, and just bad for a coil &
cap oscillator. Most of the points have been mentioned already; I'll
just reiterate them with my own spin:

The base biasing is _wrong_. Biasing a bipolar transistor that way will
make it extremely sensitive to device and temperature variations. You
can somewhat get away with it in an oscillator because you're working
with a class C device, so the emitter voltage will rise with rising
oscillation strength, making sort of a poor-man's AGC (with emphasis on
the 'poor').

Bias should be with a resistive divider on the base, adjusted so that
the oscillator starts up nicely yet keeps the emitter voltage from
rising to within 1 volt or so of the collector voltage.

The ratio of C1 and C2 is also extreme. I'm surprised that you get
oscillation at all. The rule of thumb for a Colpitts is to choose
feedback caps with reactances at frequency of around 150 ohms. You can
improve the waveform (and load the circuit less) by retaining the 1nF
"C2", increasing the capacitive reactance to the emitter with a smaller
cap in series to the transistor emitter, and retaining the take-off
point at the hot end of C2.

I wouldn't build an oscillator like this without following it with at
least one buffer stage designed to minimize the effect of following
circuits on the oscillator, and following _that_ with an amplifier
stage. You're not building tube equipment, transistors don't cost that
much -- go ahead and use a few!

If you don't get a book dedicated to RF circuit design (like Hayward's
"Experimental Methods") then at least get a copy of the ARRL handbook
and follow the guidelines there. You can go wrong with them, but not
too much.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Posting from Google? See http://cfaj.freeshell.org/google/

"Applied Control Theory for Embedded Systems" came out in April.
See details at http://www.wescottdesign.com/actfes/actfes.html

"Tim Wescott" wrote in message
...
amdx wrote:

Hi All,
Please see pictures and diagram at ABSE.

I duplicated a circuit for someone because they got a bad
output waveform.
Well I did too.
Capacitor = air variable set at 365pf
inductor = 41uh aircore
With freq set at 1.3Mhz I have the voltages as follows;
Collector - 11.8vdc
Base - 3.1vac riding on top of 3.2vdc
Emitter - 2.5vac riding on top of 3.9vdc (Don't be critical, it's not
ac
if it sets on dc)

As I shift up in freq at about 2.7Mhz the voltages jump.

Capacitor = air variable set at 40pf
inductor = 41uh aircore
Freq = 3.7Mhz
Collector - 11.8vdc
Base - 10vac riding on top of 1.8vdc
emitter - 1.1vac riding on top of 9vdc

The base waveform in both senerios is much more sinewave (ish) than the
output.

Any ideas how to get a decent waveform? The next step will be a buffer,
but...

Note; original schematic used a crystal, I substituted the LC. Could
this
be the
cause of the distortion?

Thanks, Mike


This circuit doesn't look very good from any standpoint other than
(possibly) being cheap. It's questionable for a crystal oscillator in
anything but el-cheapo consumer electronics, and just bad for a coil & cap
oscillator. Most of the points have been mentioned already; I'll just
reiterate them with my own spin:

The base biasing is _wrong_. Biasing a bipolar transistor that way will
make it extremely sensitive to device and temperature variations. You can
somewhat get away with it in an oscillator because you're working with a
class C device, so the emitter voltage will rise with rising oscillation
strength, making sort of a poor-man's AGC (with emphasis on the 'poor').

Bias should be with a resistive divider on the base, adjusted so that the
oscillator starts up nicely yet keeps the emitter voltage from rising to
within 1 volt or so of the collector voltage.

The ratio of C1 and C2 is also extreme. I'm surprised that you get
oscillation at all. The rule of thumb for a Colpitts is to choose
feedback caps with reactances at frequency of around 150 ohms. You can
improve the waveform (and load the circuit less) by retaining the 1nF
"C2", increasing the capacitive reactance to the emitter with a smaller
cap in series to the transistor emitter, and retaining the take-off point
at the hot end of C2.

I wouldn't build an oscillator like this without following it with at
least one buffer stage designed to minimize the effect of following
circuits on the oscillator, and following _that_ with an amplifier stage.
You're not building tube equipment, transistors don't cost that much -- go
ahead and use a few!

If you don't get a book dedicated to RF circuit design (like Hayward's
"Experimental Methods") then at least get a copy of the ARRL handbook and
follow the guidelines there. You can go wrong with them, but not too
much.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Thanks Tim,

The base biasing is _wrong_.
I tried Jim's voltage divider biasing arrangement, it didn't help the
waveform, but I understand
the temperature sensitivity of the original.

The ratio of C1 and C2 is also extreme.
This circuit and all values was taken from Joseph Carr's "Secrets of RF
Design"
I did replace the crystal with the LC. He calls this a 1 to 20 Mhz crystal
osc.

You can improve the waveform (and load the circuit less) by retaining the
1nF "C2", increasing the capacitive reactance to the emitter with a smaller
cap in series to the transistor emitter, and retaining the take-off point
at the hot end of C2.
Sorry, I don't know if I understand what you suggest; Do you mean to say
connect
a small cap from the connection point of C1 and C2 and the connection point
of the
emitter and R2?

I have seen several Colpitts osc. that have an inductor in series with the
emitter resistor,
would that be useful? If so what value for 3 to 6 Mhz.

On another note; I noticed the base waveform is pretty good. Has anyone
seen a design that
lightly couples a signal from the base to maybe a FET and buffers it. Just a
thought!

Thanks,
Mike

amdx wrote:
On another note; I noticed the base waveform is pretty good. Has anyone
seen a design that
lightly couples a signal from the base to maybe a FET and buffers it. Just a
thought!

The collector waveform will always be much richer in harmonics than the
base waveform. If you want to multiply up this is Good. If you see some
random circuit in a handbook keep in mind that they may very well have
intended it to be used for multiplying up 2x, 3x, 4x, etc. so the bias
etc. may have been chosen just to make the waveform at the collector
look un-sine-ish!

There are numerous oscillators that take the output signal from a
tapped-down tank if they want a minimum of harmonic content. Tapping
down helps minimize the effect of loading on the tank Q, but taking the
signal from there will have at least some effect on oscillator
stability. Another place that might like look slightly nicer from a
waveform view (and may be lower impedance) is the emitter. And there
exists oscillators that put a smallish resistor in series with the tank
- the resistor value is chosen small to not degrade the tank Q very
much, and the small voltage across the resistor looks very sine-wavish.

Tim.