EMRFD Message Archive 2264
Message Date From Subject 2264 2008-10-16 18:13:00 timshoppa Observations of oscillator output waveforms - some with mucho distor Been building oscillators from handbook and EMRFD circuits and
various schematics found in projects. I've got a nice 100 MHz scope
so I can make the oscillator run in the few MHz and I think I can
reliably see the harmonic contents at various places in the
oscillator. It's not a spectrum analyzer but the distortions I see
are usually obvious on the scope screen.
Building LC VFO's, I've been fairly happy with JFET Hartley
circuits. The voltage at the source is a pretty good looking sine
wave with some limiting apparent but not too bad. That shouldn't be
too surprising - it's just a low-impedance tap on the LC tank coil.
LC VFO's with Colpitts circuits don't look as clean as the
Hartley's, but they don't look all that awful.
But playing around with bipolar crystal Colpitts oscillators, wow.
The waveform at the emitter never looks like a sine wave. With some
crystals and feedback capacitor values, it looks like a very
distorted sine wave. With other crystals and feedback capacitor
values, it looks like a series of triangular or sawtooth pulses.
Varying the bias at the base has remarkably little effect... until
the oscillator stops working :-).
Far and away, the best way I've found to get a clean waveform in a
crystal Colpitts oscillator is to sample the current through the
crystal, because there the crystal is acting as a filter. Looking at
the voltage across the "fine-tuning" cap in series with the crystal
seems to be very clean, but even more clever are the diagrams in the
ARRL handbook for putting a low-ohm resistor in series with the
crystal and getting the output through a step-up transformer across
the resistor, or (and this is truly a genius circuit) using a common
base stage with large standing current (small dynamic impedance in
series with the crystal) ala Rohdes, Fig 10.26C in the 2007 handbook.
Is the emitter voltage in crystal Colpitts oscillators just
intrinsically a poor looking wave? Would it look better if we
reduced the gain margin, but then we would have more trouble with
getting the oscillator to start up?
2266 2008-10-16 18:49:56 Wes Hayward Re: Observations of oscillator output waveforms - some with mucho di Hi Tim and group,
I remember the first time I looked at some oscillators after I got my
hands on a high speed oscilloscope. It was indeed a shock because,
as you have observed, the waveforms are truly wild. This is more
the nature of bipolar transistors than it is of the nature of a
Colpitts versus Hartley versus whatever else you might throw into the
game. The same wild things happen with LC bipolar oscillators that
happen with crystal controlled circuits.
You have small signals when a bipolar transistor oscillator is first
turned on. These signals grew from the noise that is always
present. The signals continue to grow. As they do so, you start
to exceed the small signal limits of the device. A bipolar begins
to go really nonlinear as soon as you reach an emitter-base voltage
of about 10 mV peak or thereabouts. Depending on the biasing, the
signals will continue to grow until they are limited. As you go into
limiting, the gain at the fundamental frequency is compressed and
reaches a level that is stable in time. It is just enough to keep
the circuit oscillating. But the extra current that is present, but
not needed to maintain oscillation is shoved off into harmonics.
If you want to look at the mathematics of the process, the Clarke and
Hess text is outstanding. It's an old book, but is still
Generally there is no problem with the oscillator having some ratty
waveforms in it. There may be some harmonic distortion in the
output, depending upon the way that the output is extracted. As you
point out, it is often a good thing to extract it from the tank so
long as you don't load the tank too much in the process. The
transistor is usually operating in a rather deep Class C mode. This
is not a problem so far as performance so long as the limiting is
clean. That is, you want to have current limiting in the bipolar
oscillator such that the transistor never goes into saturation. If
it goes into saturation, the saturation resistance of the transistor
will appear across the tank for part of the cycle and this will lead
to voltage limiting. The low saturation resistance will degrade
tank Q. The other potential problem is if you should drive the
2274 2008-10-17 16:46:37 victorkoren Re: Observations of oscillator output waveforms - some with mucho di To add to Wes response, if you simulated the xtal osc circuit but
with an open loop state, you can see the small signal gain at
The excess open loop gain above 0dB is responsible to the signal
Large open loop excess gain makes the amplifier to distort more to
get unity loop gain at fundamental frequency when oscillating at
close loop state.
If you have an operating oscillator circuit that you can make
experiments, you can see that the distortion will get smaller when
you change the bias circuit to get smaller current (that reduces
The signal will be very clean when the bias current reduction will
bring the loop gain to a value just a bit higher than one (zero dB),
however any disturbance to the circuit may stop oscillation in this
One practical solution is to use an AGC loop that measures to rf
level output of the oscillator and maintain it to be at a level
somewhat lower than the amplitude you get without the AGC working.
In this case the waveform will be very clean as the AGC controls the
amplifier gain to get loop gain close to one.
By the way I learned at the university (a long time ago) with the
Clarck & Hess book and it is an excellent book, but I'm not sure
whether it is available today.
Victor - 4Z4ME
2275 2008-10-17 16:46:43 timshoppa Re: Observations of oscillator output waveforms - some with mucho di 2276 2008-10-17 19:11:18 Wes Hayward Re: Observations of oscillator output waveforms - some with mucho di Hi Tim and clan,
SPICE simulations of crystal oscillators are much more problematic
than those with LC resonators owing to the extreme high Q of
crystals. The high Q acts like a low pass filter to the extent that
the energy within the crystal cannot change quickly in the same sense
that the DC voltage across a capacitor cannot change instantaneously.
If you have a circuit with extremely low starting gain, you might
manage a simulation that corresponds with measurements. But it can
be quite hard to simulate crystal oscillators in most cases. Much
of this results from the nature of the nonlinear analysis within
SPICE. It's a transient analysis and it starts from a non-
oscillating condition. So you must let the oscillation grow within
Harmonic balance is the other common form of nonlinear circuit
analysis and this is used in a lot of the high end microwave
simulators. I've been able to analyze a crystal oscillator in HB,
but it is quite a bit harder in SPICE. HB does not deal with
transients. Rather, it only examines steady state operation.
Part of the problem here is the difficult related to doing a viable
simulation. There are things that we really can't simulate yet.
But part of the problem with crystal oscillators is real. If you
build one and then key it, you will see that there is a slow buildup
of the oscillation, even if you have none of the usual shaping that
we often use for CW gear. That's the restricted rise time of the
high Q resonator. This is also the thing that often makes simple
crystal controlled transmitters with a keyed crystal oscillator sound
pretty good (minimal key clicks) even when there is no shaping.
Regarding your question about the current limiting with a bipolar
oscillator: Essentially it is a matter of limiting the DC current
so that when all of that current becomes a peak AC value and that is
dumped into the impedance of a parallel tuned circuit, the signal
voltage will still be low enough that the collector voltage never
drops below that of the base, which is the condition for saturation.
(NPN assumed.) I kicked the formal details around a little in Ch 7
of Intro. to RF Design. While all I managed to do there was
to "derive" a rule of thumb, which is not quite science, it formed an
interesting guideline for later computer simulations in SPICE.
It's fun to look at the detailed voltages in SPICE after an
oscillator has achieved steady state operation. Then go in and add
2277 2008-10-17 20:16:55 leon Heller Re: Observations of oscillator output waveforms - some with mucho di I once spent a few hours fiddling with an FET Hartley oscillator to get the
best possible waveform out of it, from the tap rather than across the whole
tank circuit. I managed it, but as Wes has pointed out, I was probably
wasting my time.
Amateur radio call-sign G1HSM
Yaesu FT-817ND transceiver
Suzuki SV1000S motorcycle
2278 2008-10-18 07:44:55 timshoppa Re: Observations of oscillator output waveforms - some with mucho di 2279 2008-10-18 08:10:43 Lasse OT: VNA - any recommendations? I know several in this group have built their own network analyzer, and
I have the urge to do likewise. But I just came across the fact that
TenTec do sell the TAPR design, ready made for a fair price.
So my question to the group is if one design is preferred to any other,
w r t availabilty of parts and the GUI software...
2280 2008-10-18 08:43:31 Dan Rae Re: OT: VNA - any recommendations? Lasse wrote:
> I know several in this group have built their own network analyzer, andLasse,
> I have the urge to do likewise. But I just came across the fact that
> TenTec do sell the TAPR design, ready made for a fair price.
> So my question to the group is if one design is preferred to any other,
> w r t availabilty of parts and the GUI software...
> /Lasse SM5GLC
will lead you to the best one that you can build, and it is probably
better than any you can buy ready made without spending 100 times the
In my humble opinion.
2281 2008-10-18 12:04:07 Wes Hayward Re: OT: VNA - any recommendations? Hi Dan, Lasse, and group,
I would agree with Dan. I now have one of the N2PK machines running
and it is wonderful. The dynamic range is outstanding.
The N2PK web site is great with all sorts of great data and software.
Also, there is a Yahoo Group for the VNA that is loaded with good
I am running my VNA from an old laptop computer that runs with Win95
and am using the software that Paul has written. I was not able to
get the VNA to run with my XP machine. It will run with a machine
running Win98. Evidently, you can run the VNA OK with XP if you use
DOS-Box. Also, I understand that there is some software out of
Europe that will run run the VNA with XP. Once I got things
working well with my old laptop, I stopped fighting those issues.
A local fellow who some of you may know built a version of the TAPR
machine. His was a beta build. The whole design was something of
A paper was presented at Dayton, FDIM-07, where some other machines
were discussed. This may be available if you search on the web.
A VNA is a wonderful tool and I'm enjoying having one.
2282 2008-10-18 13:15:53 Homecall Re: OT: VNA - any recommendations? Hi Wes
I also built up a VNA about 5 years ago and did quite abit of Beta testing for Greg Ordy whom developed the Exeter Software which is a truly amazing piece of Software which I advise you download to use with the VNA as its also very easy to use and works with all my Windows XP machines without any problems at all,, BUT you will need a small driver to load only the once,, and if I remember its on Pauls website. The VNA I have used in anger to tune up my bandpass filters in my Software Defined Radio transceiver,, I used to use a Polyscope for sweeping them but it was so heavy and bulky,, the VNA is a delight.
----- Original Message -----
2283 2008-10-18 16:10:07 Art Re: OT: VNA - any recommendations? Hi Lasse,
I do the kits for the N2PK VNA. I cannot advise on other models
I can say that the kits are in stock and available for prompt
shipment, stateside or internationally. The parts for the N2PK are
shipped at actual cost with a very modest markup (just enough to
cover my overhead and to insure that I don't take a loss on the kits).
I will send you additional information in the form of some pdf
documentation on the kits via direct mail.
>I know several in this group have built their own network analyzer, and
>I have the urge to do likewise. But I just came across the fact that
>TenTec do sell the TAPR design, ready made for a fair price.
>So my question to the group is if one design is preferred to any other,
>w r t availabilty of parts and the GUI software...
2288 2008-10-19 16:44:40 Lasse Re: OT: VNA - any recommendations? Thank You All for the comments both on and off the list.
It seems the N2PK do get most positive response, and I guess it is
almost like cheating to buy a ready made unit with less
So I have joined the VNA-group here on Yahoo and will try to get the
project going, good thing is that DigiKey offering free shipping here in
Europe until nov 18. Just what I needed :)
Network analysers have come a long way since my first tries with the HP
8505... and then the 8753A a marvel that still offers superb performance
by any standard.
Sadly the price is really making it unsuitable for hobby work though.
2308 2008-10-22 18:06:37 davidpnewkirk Re: Observations of oscillator output waveforms - some with mucho di > Even more fun was me trying to build a VFO out of a 6AH6 (fordecades
> the recommended VFO tube in the ARRL handbook, probably for no other[deletions]
> reason that its uncontrollably high gain.)
> As a result, this one active device was not just working at onefrequencies,
> frequency, it was oscillating at three entirely different
> 2kHz, 3.5MHz, and 0.4GHz!!!!I'd like to see your circuit, and learn something of your parts values
and placement and lead lengths. It might be possible for some of us to
see some of those effects coming, so to speak, in your actual
implementation, and possibly avoid such problems by prescription.
2310 2008-10-24 09:42:04 timshoppa Re: Observations of oscillator output waveforms - some with mucho di