Crystal oscillator oscillates at more than one frequencies simultaneously!

Hi,
This idea is based on a very stable and robust oscillator I have previously developed, that is able to oscillate at 1MHz to > 30MHz by just changing the crystal. It works on 1M as well as 50R (with reduced level of course).

My new idea is based on that one, but now, I have used two or three crystals, each of different band.
I have tested the oscillator and it outputs more than one frequencies at once without problems!
The harmonics are unmeasurable (<50db), like in the original circuit with the single crystal.

My question is: what uses may such a circuit find?

Some ideas:

1. By using different frequencies crystals (not too close enough to lock each other) and because the oscillator components are the same for both crystals, one can achieve a very stable oscillator without an oven, by mixing these two frequencies. If the crystals have quite close characteristics their up-converted or down-converted frequency should be extremely stable, since they drift at the same amount prior to mixing.

2. The thing can be easily ovenized for multiple frequencies (oven oscillators operate at one frequency), since there is no switching of crystals (difficult to do inside an oven). This, provided that the crystals inside the oven have the same temperature knee points.

3. In contrast to harmonic multipliers, where the drift is multiplied in higher harmonics, this circuit will have the same drift. Also any higher frequency can be generated depended on the crystal you plug in, not just multiples.

What other uses could such a circuit find?
It would be interesting to know your ideas.

WWV broadcasts beeps spaced one second apart, on several shortwave frequencies (2.5 MHz, 5, 10, etc). Maybe your circuit could create something similar? Not sure if you'd have trouble with overtones in case you use frequencies which are integer multiples.

Your diagram shows three crystals in parallel; they are tightly coupled. You should expect to see the sum and difference frequencies, in addition to harmonics. Most high frequency crystals are designed to be run at 2nd or 3rd overtones and for that you need to keep them lightly loaded.

No, that is not the case. I see only the fundamental of each crystal, all of them appear at the output of the oscillator together. Maybe it is because this oscillator produces very clean sinewaves (all harmonics below -50dbc). However, as I expected, I need to set up the oscillator gain so that greater gain crystals do not take up so much gain, because then some lower gain crystals could not produce output. But when I set the gain to be less, the lower gain crystals may not oscillate. So it is a balance between gain and output power. I have succeeded up to 3 crystals. The oscillator output is clean (no harmonics) and all 3 carriers appear at the output at the same time.



That is another use, to broadcast at more than one frequencies at once using just one transmitter and a multi-band antenna, but the amplifiers must be able to handle the simultaneous different signals without distorting.

My fft has not got enough resolution to see if two very closely spaced crystals will inter-lock.

No that wont change anything.
Say you have two crystals one at 10MHz and one at 11MHz. They have a stability of 1ppm so they drift to 10.000 010MHz and 11.000 011MHz.
Mix them to their sum and you get 21.000 021 MHz with is the same stability - 1ppm.
Mix them to their difference and you get 1.000 001MHz and that is 1ppm all over again.

In reality they would not drift equal amounts, so you might end up with much worse or better numbers, it would be random just like the randomness of the the drift of a single crystal.

If you could measure the general drift versus temperature you could compensate for the drift to some degree.
This is what is being done in TCXO (Temperature compensated crystal oscillators)
http://www.abracon.com/Oscillators/A...MHz-512242.pdf

Hi,

I can′t see HOW one could mix two independent clock signals as sum or difference.

I assume when you somehow join both signals you will always get 21.000 012 MHz as well as 1.000 001 MHz

Klaus

Yes this is the case with mixing (multiplying). You get the sum, the difference and some other products, depended on the mixer type and characteristics. But his response was related to the frequency stability of these two mentioned signals, generated by the same oscillator.

I believed that because one oscillator is used with multiple resonators, the difference or the sum of the oscillator output signals would be always the same, as both would drift by the same amount.
However, his reply shows that this is not the case.

I've seen this post, and Barkhausen told me: let's give a try and replicate the phenomenon.

I am afraid that somewhere is a misunderstanding.
Whatever crystals I used (various frequencies, various packages, various manufacturers) never ever the oscillator oscillate on two SIMULTANEOUS frequencies.
I used various JFETs (J309, J310, BF245, BF256, MPF102), with various bias currents, but never got two simultaneous oscillation frequencies.

Always I get only one oscillation frequency, most probably the priority is given by the crystal with higher Q.

Sometimes (when perhaps the Q of the crystals are appropriate) I get a free running oscillation, which doesn't match any crystal frequency.
I used various crystal combinations, for example: 4MHz and 10MHz, 9.6MHz and 10MHz, 14MHz and 10MHz, 5MHz and 14MHz, 11MHz and 12MHz, and many other combinations.

I appreciate your experiments!
I have tested the schematic in post #1 with wide spaced crystal frequencies. I do not know if the J108 is critical, but I have found it to work well at all HF frequencies and the trimmer control, not to be very touchy. The gate-source capacitor is missing, and this is formed by the internal capacitance of the j108. So I suspect this is critical somehow, as j310 has much less capacitance. I have tested the circuit at around 7MHz, 10MHz, 14MHz. The max number of crystals it could accept in my case, was 3. After that, some frequencies refused to oscillate. The trimmer had to be reset for every new crystal inserted in the circuit, so that all crystals could produce output. Of course the output signal level of each crystal was not the same. To compensate that, I put a series variable capacitor with each crystal, which alters it's frequency a bit, but more importantly alters its output level (less coupling). This allowed all output signals to appear at the same level at the output.
I have built the whole thing into breadboard, and tested on 1M and 50R on FFT, although 50R is loading it too much, but it still works.

I found in my box also a J108 and retest the circuit with mentioned crystals, new series capacitors, and different loads.

But again, I NEVER get two SIMULTANEOUS oscillation frequencies.

Barkhausen oscillation criterion mention that a feedback amplifier can sustain a steady oscillation if the loop gain is equal to unity in absolute magnitude, that is, | β A | = 1, AND the phase shift around the loop is zero or an integer multiple of 2π: ∠ β A = 2 π n , n ∈ 0 , 1 , 2 , …

For the same given feedback amplifier, to get at 2 different frequencies a phase shift necessary to obtain 2 different steady oscillation frequencies, is practically impossible.

Then it is weird why it works for me?
I forgot to mention in my previous post that it works, provided that the frequencies are not very close a few KHz (so that they do not interlock) and they are not very far apart (for example test with two crystals that are between 5-12MHz or so. do not test with 3 and 27MHz, it won't work because the trimmer setting is much different in these two frequencies.)


What types of crystals did you use?

I might be able to post a short video when I go back home, to show you the circuit and the FFT results.

Doesn't matter what type of crystals I use. I do have many high quality low ppm crystals that behave the same.

Here is about basic oscillator theory. It's impossible to get the necessary phase shift at two different frequencies and get two simultaneous steady oscillations.

Anybody else who want to try this experiment, it will not take more than 15 minutes to arrive to the same conclusion.

if you actually TRY to make an oscillator with three parallel frequency resonators, you most likely will have it oscillate at only ONE of those frequencies. It is because the gain of your active device might favor one frequency slightly over the others, and as oscillations build up and the active device gets gain compression, only the one In the sweet spot keeps sustained oscillations. The other two will snuff out.

It is not my intention to prove why it works in my case. You are much more experienced than I probably and I cann't disagree in what you say.
I only mention my results of this experiment, which worked for me without problems and tested with different crystals, as long as the criteria mentioned above are satisfied.

It would be interesting to try to find out what is the difference that makes mine working and yours not.



This was the case when I used 4 crystals. It could not "support" all 4 of them. Using 3 crystals was the max I could have and that with careful adjustment of the source trimmer.

Here it is, as promised, with 3 crystals in place, 160m, 40m and 20m.

Are you sure you don't use separate oscillators?
The screenshot picture of the spectrum analyzer have good resolution and contrast, but the picture of the breadboard circuit looks like was taken by the inventor of photography, Nicéphore Niépce.

Hey, I am not fooling you. The circuit picture is like this, because it was taken from outside my glove box I use to solder electronics in. The circuit is inside this polypropylene box, which is not 100% transparent. I also did some digital zoom to bring it closer. If you look closer, you will see only one active design.

I do not know why it did not work for you, It worked in my case. I used cheap crystals from ebay.