Shift a 100 MHz sine wave in frequency

Thread Starter

Tar9etPractice

Joined Apr 2, 2019
3
Hi all!

I'm working on a project where I want to shift a sine wave in frequency. The general idea is that frequency is being used as an input to a system. Changing the frequency is how I provide feedback. The frequency of the source sine wave is in the range of 100 MHz, and I want to dynamically change its frequency by up to +/- 0.01%.

Is there a circuit that can readily do this?
 

danadak

Joined Mar 10, 2018
4,057
What is the source ?

If its analog LC tank you could use a varicap to modulate it.


You could do it with a DDS chip by using a micro to measure freq, and calculate
it to +/- 10 Khz. More involved for sure.


You could do it using a PLL with non integer divide in the fdbk loop.

http://www.ti.com/lit/an/swra029/swra029.pdf


Or do it with a mixer and filter.


Regards, Dana.
 

Thread Starter

Tar9etPractice

Joined Apr 2, 2019
3
use FM (frequency modulation)
I don't think I can use FM in this case. For this application, it's important for the signal to be close to a sine wave. Basically, I want as inputs two sources V0*sin(2pi f1 t) and V0*sin(2 pi f2 t) and an output that's as close to possible to V0*sin(2 pi (f1 + f2) t).
 

panic mode

Joined Oct 10, 2011
1,829
create oscillator for 100MHz with varicap. apply voltage to varicap to set oscillator frequency. this is the simplest FM setup and output remains sine wave.

you can do more sophisticated methods using DDS or PLL

mixing signals works of course but you get sidebands and when they are close (in your case they are EXTREMELY close), you will never be able to filter one out. output will be mix of sinewaves, not a single sine.
 

Thread Starter

Tar9etPractice

Joined Apr 2, 2019
3
create oscillator for 100MHz with varicap. apply voltage to varicap to set oscillator frequency. this is the simplest FM setup and output remains sine wave.

you can do more sophisticated methods using DDS or PLL

mixing signals works of course but you get sidebands and when they are close (in your case they are EXTREMELY close), you will never be able to filter one out. output will be mix of sinewaves, not a single sine.
This looks like a good option for me, thank you!
 

DickCappels

Joined Aug 21, 2008
6,301
upload_2019-9-6_23-50-2.png

You could replace the three 2N4401 transistors on the right with a single varicap.

Notice the abundance of 2N4401, 1 k resistors and 330 Ohm resistors? That's because I have a lot of each. 2N4401's weren't intended to be used as varactors, but they work just fine. Don't be afraid to experiment with the parts you have on hand.

Build this on a ground plane board. Layout is important. If you place the parts relative to one-another in correspondence with their placement on the schematic, it should work pretty well. For making one or two of something, I don't like to bother with a printed circuit boards. I just cut off a 2.5 x4 cm piece of copper clad board and soldered grounded parts onto the copper cladding fiberglass, and the ungrounded parts directly onto the leads of the grounded parts. This "dead bug" construction is a very speedy way to put high frequency circuits together and it works pretty well at these frequencies. One thing to be careful of is to make sure all the components are packed tightly so as to minimize parasitic resulting from long leads. In the ones I built, the components are leaded, though in one of them I used chip capacitors for the bypass capacitors, though I couldn't see any difference in performance.

It was necessary to shield the circuit so that it would not be affected by other circuits nearby, particularly additional RF stages. The loose coupling for the RF, provided by the 330 build-out resistor on the emitter follower further serves to isolate the oscillator from its load. These result in greater stability. The shield was made by forming a five-sided box (one side missing) from tooling copper (a thin copper sheet) and soldering it onto the groundplane of the copper clad board. Once, for something similar, I made a shield box by soldering pieces of copper clad board together. Nearly anything will do as long as its reasonably low resistance material (brass shim stock should work if you have a piece laying around).

The inductor is made by winding 8 turns of #24 insulated solid copper wire on a 5 mm screwdriver. I used a conductor from a piece of category 5 quad twisted pair, left over from wiring the house with Ethernet and this seems to work well and to be easy to handle. Stripping the vinyl insulation is much easier than scarping or burning off enamel.

This was easily tuned up and down the bottom of the FM band using a 10 turn panel mounted potentiometer. The upper limit to the frequency modulation is set by the 39 k input resistor and the shunt capacitance (I would guess around 20 pf max), so that shouldn't be a concern until the modulation frequency reaches many MHz.

Raise the frequency to 100 MHz by taking a turn or two off the coil. You can also shift the center frequency by compressing or stretching the coil.

upload_2019-9-6_23-49-8.png
 

SLK001

Joined Nov 29, 2011
1,543
I don't think I can use FM in this case. For this application, it's important for the signal to be close to a sine wave. Basically, I want as inputs two sources V0*sin(2pi f1 t) and V0*sin(2 pi f2 t) and an output that's as close to possible to V0*sin(2 pi (f1 + f2) t).
Basically, you don't understand what FM is. FM is the ONLY analog way to do what you want (and probably the simplest way). The purity of your output signal will depend entirely on your oscillator circuit design.

You could use a 100 MHz crystal oscillator and pull it with a varactor, but I think that ± 10k Hz would be difficult. A more realistic xtal pull at 100 MHz would be ± 1k Hz. This is due to the need to use a fairly high overtone grind for the crystal.
 

crutschow

Joined Mar 14, 2008
24,709
You could feed the signal into a mixer as danadak suggested,where the output is the sum and difference frequencies of the two input frequencies.
 

DickCappels

Joined Aug 21, 2008
6,301
From my limited experience in the area you should be able to get almost 5 kHz pull on a 100 MHz crystal.

Does the frequency shift have to be analog? It might be done with a phase locked loop with a digital divider in the feedback or a Direct Digital Synthesizer like the Analog Devices AD9910.
 

SLK001

Joined Nov 29, 2011
1,543
You could feed the signal into a mixer as danadak suggested,where the output is the sum and difference frequencies of the two input frequencies.
The problem with this, is that BOTH the +10 kHz and the -10 kHz are present at the output (in addition to the 10 kHz and the 100 MHz signals), so there is no way to discriminate the desired signal from the undesired.
 
Top