# I want to build an LO

Discussion in 'Electronic Design' started by M. Hamed, Jun 8, 2013.

1. ### M. HamedGuest

Hello the most knowledgeable electronics group on earth!

I am trying to build a few LC oscillators for experimenting with mixers. I do not want to use Crystals, Frequency Synthesis, DDS, etc. Let's say 2 MHz and 100 MHz. I don't know where to start!

I have a few RF books and the treatment of oscillator can be one of the following:

1- Too much theory on the analysis of oscillator circuits, phase noise, IMD, etc. No circuits to build. Just demo circuits for illustration, sometimes only the small signal model.
2- Circuits to build but designed at very specific frequencies with very specific components. Very little on how to generalize things into different frequencies. (The ARRL handbook 2010 and Experimental Methods fall under this).
3- Books that give me a design procedure with S parameters but I can't find any S parameter files for the devices I'm interested in using.
4- Cookbooks that have the same problem as (2).

Why is it so hard to build an oscillator at an arbitrary frequency?

2. ### Andrew HolmeGuest

"M. Hamed" wrote in message
Idea 1: find a DIP meter schematic. They typically cover a fairly wide
range using plug-in coils i.e. you just need to change the inductor. And
it's often a single FET circuit.

Idea 2: You could cover quite a wide range using an unbuffered CMOS inverter
(4069UB) with: cap from input to ground; cap from output to ground; inductor
from input to output; maybe a resistor in series with the output. You
should get a sinusoid at the inverter input. It'll need buffering to drive

In both cases you should be able to cover quite a wide frequency range with
suitable choices of L and C. You can lash it up on a breadboard.

But the upper frequency limit for the above circuits will be well below 100
MHz. You will need a different circuit for VHF.

3. ### Guest

With current components, it is no big deal to make a free running
oscillator in the 10-100 GHz range.

However, in order to be usable in communication systems, some
frequency accuracy is needed.

At MF at 1 MHz a frequency accuracy of 1 % would correspond to 10 kHz
(2x5 kHz audio passband). The same accuracy at 10 GHz TV satellite
band would correspond to 2-4 channels.

The real question is that it is no problem making a scanner receiver
handling the 0-5 GHz frequency range, however making a good receiver
or that frequency range is nearly impossible.

4. ### tmGuest

Just try to build an amplifier and most times you will get an oscillator.

tm

5. ### John DevereuxGuest

<http://ee.devereux.me.uk/uhf-vfo.png>

Air-core coil wound on a pencil, can't remember the fet but nothing
special.

No ebay in those days so I just had a DFM. I should hook one up again
and see how good/awful it is now I have lots of test kit.

6. ### Phil AllisonGuest

** A free running oscillator is as accurate as YOU set it.

The real issue is *stability over time* or drift and that is nearly all due
to temperature variations.

** With most AM receivers, one tunes the LO to the desired station by ear or
the uses of an optical indicator. The allowed error is in the hundreds of
Hz. Over time the LO may drift, then one retunes it.

** Communications on VHF, UHF and above pretty much depends on the use of
crystal locked oscillators at both ends.

Luckily they were invented long ago.

** Why is that the "real" question ?

Seems like a smelly red fish to me.

.... Phil

7. ### Tauno VoipioGuest

Maybe I repeat myself: It seems that you need a good book on
basic radio design, with the emphasis on practice.

Please get yourself an ARRL handbook <http://www.arrl.org/>.
It contains plenty of information for your project and

8. ### Guest

Operating in that frequency range was a challenge 50 years ago with
germanium transistors. The parameter spread was very large, so making
a reliable oscillator was hard. To sustain oscillation, the _power_
gain of the active element(s) must be larger than the losses in the
resonator and other passive components. This gets hard, when the
typical fT for an RF germanium transistor was 40-200 MHz.

These days transistors with much higher fT are available. Even some
cheap MMICs have considerable gain up to 5-10 GHz. If the gain is not
enough, cascading an other MMIC does not cost much. Of course, at
frequencies at UHF and above, lumped LC resonators can not be used,
stripline and microstrip constructions are often used.

9. ### Guest

If you want to handle the 2-100 MHz range in a single sweep in a
single oscillator, that is not going to happen.

Of course, you could make a 2-100 MHz sweep with a VFO running between
122-220 MHz and mix it down with a 120 MHz overtone crystal, but the
frequency stability will be quite bad, due to the VFO in the VHF
range.

If you want 2-100 MHz from a single oscillator, that is a frequency
range of 1:50. If only one component is tunable, typically the
capacitor, this requires a 2500:1 capacitance range. This is clearly
impractical due to stray capacitances. Clearly both the capacitance
and inductance must be tuned, but getting 50:1 or similar ratios for
both the inductance and capacitance is hard.

With only a tunable capacitor (capacitance diode) trying to reach
larger frequency ranges larger than 1:3 is asking for troubles, as
this already required a 9:1 capacitance range, including stray
capacitances.

In the old days, when tubes and transistors were expensive, elaborate
mechanical switching was used to switch in various inductances and
capacitances in series/parallel. These days, when the active component
costs practically nothing, it makes much more sense to build a
complete oscillator for each subband and only connect power to the
oscillator to be used at a particular time.

You might for instance build an oscillator for 2-6 MHz with a big
varactor intended for MW (AM) receivers.

The next oscillator could cover 5-15 MHz, next 12-36 MHz and finally
33-100 MHz, all these with low capacitance varactors, several in
parallel at the lower bands. There are plenty of designs for these HF
and VHF frequencies, which should be easily scalable to those
frequency ranges.

One other alternative would be to build a 50-100 MHz VFO followed by
five divide-by-two frequency dividers followed by fixed low pass
filters at each tap. Since the square wave output from the divider
contains a lot of 3rd harmonic, to simplify the low pass filters, the
VFO tuning range should be less than 1:3, in this case 1:2.
I do not like the recent ARRL handbooks, the older ones have better
technical contents. Try to locate a 10-20 years old ARRL handbook from
a library or eBay etc.

10. ### Phil AllisonGuest

** The OP did not ask for anything like that and you are being

FUCKING tedious WANKER !!!!

by repeatedly inventing *straw men* for your self aggrandisement.

Piss OFF !!!!!!!!

..... Phil

If someone is not already credited on this, you should be.

Many times the result is a very poor, inefficient one at that, as
compared to dedicated (intended to be such at design time) oscillator
circuits.

12. ### tmGuest

LOL. Yes, it was meant to be tongue in cheek. Another one is "the better you
make the amplifier, the more it will oscillate".

tm

13. ### Fred AbseGuest

As long as it doesn't give you the Clapp.

14. ### Robert BaerGuest

Oh, it is VERY easy; just build an amplifier!

15. ### Robert BaerGuest

What happens when you cross a Colpitts oscillator with an Armstrong
super-regen?
You get an ArmPitts oscillator.

17. ### John SGuest

I must be doing something wrong. I simulated it in LTSpice and I get
5MHz oscillation with a 10MHz tank. And it is a long way from a sine wave.

Version 4
SHEET 1 880 680
WIRE -160 -256 -288 -256
WIRE 16 -256 -160 -256
WIRE 128 -256 16 -256
WIRE 400 -256 128 -256
WIRE -288 -208 -288 -256
WIRE 16 -160 16 -256
WIRE 128 -160 128 -256
WIRE -288 -112 -288 -128
WIRE 16 -16 16 -96
WIRE 128 -16 128 -80
WIRE 128 -16 16 -16
WIRE 400 -16 400 -256
WIRE 128 32 128 -16
WIRE 336 32 128 32
WIRE 128 192 128 32
WIRE -160 240 -160 -256
WIRE 64 240 -160 240
WIRE 128 336 128 288
WIRE 400 336 400 80
WIRE 400 336 128 336
WIRE 128 384 128 336
WIRE 128 480 128 464
FLAG 128 480 0
FLAG -288 -112 0
SYMBOL npn 64 192 R0
SYMATTR InstName Q1
SYMATTR Value 2N2369
SYMBOL npn 336 -16 R0
SYMATTR InstName Q2
SYMATTR Value 2N2369
SYMBOL res 112 368 R0
SYMATTR InstName R1
SYMATTR Value 4.3k
SYMBOL voltage -288 -224 R0
WINDOW 123 0 0 Left 2
WINDOW 39 0 0 Left 2
SYMATTR InstName V1
SYMATTR Value 5
SYMBOL ind 112 -176 R0
SYMATTR InstName L1
SYMATTR Value 16u
SYMATTR SpiceLine Rser=20
SYMBOL cap 0 -160 R0
SYMATTR InstName C1
SYMATTR Value 15.8p
TEXT -322 504 Left 2 !.tran 0 10u 5u

18. ### John SGuest

Oh. I thought he wanted the Xc and Xl to be 1000 ohms each at 10MHz,
hence the Zo of 1000 ohms. What do you think he meant about Zo?

19. ### John SGuest

He has me kill filed.