As I understand it a MEMS oscillator will output a square wave and a crystal oscillator will output a kind of triangle wave. The question is when you mix RF with an oscillator signal to produce the IQ baseband frequency, should you first smooth the local oscillator frequency, like should you do a square to sine conversation before mixing?
\$\begingroup\$
\$\endgroup\$
1
-
\$\begingroup\$ Your typical crystal oscillator will output a square wave as well, at least the sort used in digital systems (commonly a Pierce oscillator). Things get a bit weird if you push the limits of frequency, but unless you're working in the gigahertz it'll be pretty square. But of course you can also design a crystal oscillator to output a sine wave as well, if you want--it's a matter of how you drive the system. Same for MEMS. \$\endgroup\$– HearthCommented Oct 21, 2022 at 19:45
Add a comment
|
2 Answers
\$\begingroup\$
\$\endgroup\$
3
The mixer will mix every RF frequency that it receives with every LO frequency that it receives.
A square or triangle wave is a base wave plus harmonics at odd-number multiples of that. E.g. a 1Hz square wave is a 1Hz sine wave plus a 3Hz sine wave plus a 5Hz sine wave etc.
So, check if any of the combinations are problematic.
If so (and it sounds like they will be, because you have no filtering) - it's extremely normal to have filters before and after mixers to suppress unwanted frequency combinations. So figure out which frequencies you need to keep and which frequencies you need to suppress to get the output you want, and add corresponding filters, on either input or the output of the mixer, as needed.
If you are using a diode-ring mixer then you can't do much to improve the LO quality - you will have to live with it being a square wave - and so there is not much point filtering the LO.
By the way, you tagged the question with "sdr". The need for filters to suppress unwanted frequencies from mixing is one of the reasons that many radio circuits have quite limited frequency ranges, and one of the challenges when developing an SDR. For example, the HackRF One has two different selectable filters on the RF side.
-
\$\begingroup\$ So, if I select 5mhz as my target IF and place a 5 MHz bandpass filter in it then all the harmonics will get filtered out and I would get a decent approximation of a sine wave? \$\endgroup\$ Commented Oct 23, 2022 at 18:26
-
\$\begingroup\$ @UdayV.S.sharda if none of the other RF+/-LO combinations result in a 5MHz. let's say you are shifting 50MHz down to 5MHz with a 45MHz LO; Because your LO would contain 45MHz, 90MHz, 135MHz, etc, you'd be shifting 50MHz down to 5MHz but also 95MHz, 140MHz, etc. And 40MHz, 85MHz, 130MHz, etc. So you would want to have a 50MHz bandpass filter on the RF side as well as a 5MHz filter on the IF side. Lots of filtering! If you want different RF frequencies, the RF filter needs to be adjustable. It doesn't need to be as precise as the IF filter. \$\endgroup\$ Commented Oct 24, 2022 at 15:12
-
\$\begingroup\$ @UdayV.S.sharda btw: I haven't worked on these systems myself but I've read about them. \$\endgroup\$ Commented Oct 24, 2022 at 15:13
\$\begingroup\$
\$\endgroup\$
What i found out is a crstyal can be used to generate all kind oscilation dpending on how it is driven. If the amp driving crystal has a low gain then you will get a sine wave but if increase the gain then the wave will start clipping the edges of the voltage side(say after gain your sine voltage peak to peak is 3v but your amp has a swing range of 2v then the voltage at the times it supposed to be above 2v the amp will output a flat 2) and give you a clipped sine wave if increase the gain a lot you will basically get a square wave.