upconverting a kHz/MHz signal to GHz

I want to know how in practice we can take a kHz/MHz signal to a GHz band?

Suppose we have an FSK modulated signal at around 300 kHz (or 20 MHz). I want to take it to 3.5 GHz, so about 3499700 kHz higher (or 3480 MHz higher).

• Should I build an oscillator with a frequency of 3499700 kHz (or 3480 MHz) to use it as a mixer?
• Is it possible in practice to have a precise 3499700 kHz oscillator?
• Is there any IC useful or should it be by PLL?
• How?
• Is there any other and better solution?
• Hey! I'd focus on fewer questions :) makes it easier to answer all of them, and the last three ones aren't exactly well-asked. Anyway, I went ahead and proofread and corrected your question; there's only one capitalization of the kilohertz unit, it's kHz, not KHz, nor Khz ;) just a pet peeve of mine, but really, the decimal prefixes should be something that you don't mix up with units like Kelvin (K). Oct 22, 2018 at 8:59

I want to know how in practice we can take a kHz/MHz signal to a GHz band?

You guessed right: a mixer.

Should I build an oscillator with a frequency of 3499700 kHz (or 3480 MHz) to use it as a mixer?

Assuming these frequencies are the difference between your source and your target frequencies, yes.

But you don't use an oscillator as a mixer, you use its output to feed it to a mixer.

Is it possible in practice to have a precise 3499700 kHz oscillator?

Define "precise"!

Precise is a relative term. If you build a house, your standard of length needs to be accurate to say 0.2 % (so that if your house is 10m long, its length error will be no more than 2cm). If you build an aircraft engine, you might want to be better.

With oscillators it's usually way, way more precise than that, even on cheap standards.

Technologically, RF devices typically use synthesizers to generate local oscillator (LO) frequencies from a reference oscillator, which might be running, for example, at 10 MHz. A synthesizer typically contains all the parts necessary to derive that frequency – a control loop for a voltage-controlled oscillator, that oscillator, some sort of frequency multiplier, filtering to actually get the multiple you want, control and compensation circuitry…

The precision and stability (two different things!) of the resulting oscillator usually mainly depends on the precision and stability of the 10 MHz reference, and to a smaller extent (at least in the frequency regions you mention) on the noise and precision of the control loops.

Is there any IC useful or should it be by PLL?

This question makes no sense. All the ICs that might make sense here contain at least one form or another of a PLL.

How?

Good!

Is there any other and better solution?

To what?

You seem to want a mixer + filters + a synthesizer. Mixers and filters can be bought as modules for moderate prices (e.g. from minicircuits), synthesizers often either take the shape of signal generators in laboratory settings, or as synthesizer ICs + support circuitry in actual devices.

• @ Marcus: Thanks for your answers. By "precise", I meant a resolution in the order of kHz for an oscillator of "3.5 GHz", because the original signal is at around 300 kHz. By "How?" I meant how is it possible to get a signal of 3499700 kHz by PLL? Because if I use for example a reference signal of 100 kHz, I should apply a multiplication of 34997 to it, which looks a little unusual. I also wanted to see if there is any other solution to transfer the mentioned signals to 3.5 GHz ones? Oct 22, 2018 at 12:15
• so, as I tried to illustrate, you need to measure the frequency error in relation to the frequence. "order of kHz @ 3.5 GHz" means $\frac{1\cdot10^3}{3.5\cdot 10^9}\approx 3 \cdot 10^{-7}=300\,\text{ppb}$. Oct 22, 2018 at 13:41
• That means that you'll need a reference with 300 parts per billion error. That would be a very good GPS disciplined oscillator or an atomic clock. Oct 22, 2018 at 13:47
• Regarding "make 348.... from 100 kHz": I've answered exactly that in my answer; please reread for further information. Oct 22, 2018 at 13:48
• because 300 ppb is so hard to achieve, every real-world communication receiver has means to detect the frequency of the transmitter and to adjust itself so to be able to receive. Oct 22, 2018 at 13:49