can i generate square wave of (1 - 5 GHz) using mosfet astable circuit ?

DickCappels

Joined Aug 21, 2008
6,819
I remember a paper that described cross-coupled MOSFETs being used as a regenerative detector, for 2.5 GHz and maybe 5 GHz, so yes it can be done. The oscillator in the paper was on a single integrated circuit. Not sure how well you could do it with discreet components.
 
hi all,

can i generate square wave of (1 - 5 GHz) using mosfet astable circuit ?


regards
A square wave at 1 to 5 GHz!? Wow, that is going to take some serious engineering. The result won't be even marginally 'square' if you can't get to at least the 7th harmonic without phase errors and even then it will be a very soft square wave and I cannot see how you might verify the result. The best CRO I have ever used was a LeCroy worth $250k and that had a 1GHz bandwidth and definitely could not have shown a square wave at that frequency even if it could be generated. You'd be able to see the harmonics and their phase on a spectrum analyser and they can go to many 10's of GHz but you'd need to visualise the waveform for yourself because I don't think there is a time domain instrument that could show it to you.

A 1GHz signal has a period of 1ns, 500ps per half cycle and 5GHz is just 100ps per half cycle. To keep it 'square' you'd be looking for a rise and fall time in the order of 10ps for 5GHz would you say? Still a bit soft for a square wave. At these frequencies everything is an RF circuit so you'll need to brush up on impedance matching and transmission line theory as a minimum to be in the running for success with this.

Consider this, the speed of an electric current in a wire is about 60% the speed of light (spooky how accurate that rule of thumb is) so about 5.5 nanoseconds / m and a quarter wave length of 10 pico seconds (from the above) is 1.8mm. This puts the 2mm package sizes of some of the parts suggested already into a context that makes sense and also makes those parts seem maybe a little on the big side. For example, if you had a lead connecting the source of a fet to ground that was even just 1mm long then the lead will interact with the harmonics of your signal and conspire to ruin your square wave by modulating the amplitude and shifting the phase of the harmonics.

Unless you are an RF guru and engineering superstar I'd have to say the job is the work of a lifetime. Sorry to be a party pooper and I hate to be the bearer of bad news but I think if you have to ask, and in a forum like this, then this job is too big for you. I also hope that does not offend you, it is simply the reality of what you are asking about (unless you are the guru in which case I unreservedly take it all back and you have my apologies).

Can I ask, what do you need it for?
 
I remember a paper that described cross-coupled MOSFETs being used as a regenerative detector, for 2.5 GHz and maybe 5 GHz, so yes it can be done. The oscillator in the paper was on a single integrated circuit. Not sure how well you could do it with discreet components.
And was that a square wave output oscillator?
What was the output, ie was it LVDS? ECL?
I would have expected a detector to be more of a single frequency thing (sine wave that is).
 

-live wire-

Joined Dec 22, 2017
912
This type of thing requires serious engineering skills. You need to manage parasitic capacities, inductances, and other undesirable effects. At 5GHz, a 100pF capacitance is a .3 ohm path from power to ground. A 1nH inductance is a 30 ohm resistance. There is a skin depth of 1 uM through most conductors.
 
It was a subcircuit in a regenerative receiver. The output was, if I recall correctly, data.

Let's please try to stay on topic.

While at it you. mctjo,might pick up some ideas from the attached series of slides.
The slides you have offered and mentioned in a previous post seem to me to be sine wave oscillators and the mathematical pages seem to be descriptive of sine wave oscillators specifically for communications systems. None of it actually applies to square wave oscillators or so it seems. When you said it could be done, this square wave oscillator at 5GHz, what exactly did you have in mind?
 
This type of thing requires serious engineering skills. You need to manage parasitic capacities, inductances, and other undesirable effects. At 5GHz, a 100pF capacitance is a .3 ohm path from power to ground. A 1nH inductance is a 30 ohm resistance. There is a skin depth of 1 uM through most conductors.
Frightening numbers, and that is only the fundamental frequency! :) If you were to settle for a sloppy square wave and therefore only needed to get to the 7th harmonic then your band of interest extends to 35GHz and you need to ensure no appreciable phase differences over the band 5GHz to 35GHz and even then the square wave, not to put to fine a point on it, is going to look pretty awful. ;) Or it would if there were a CRO anywhere on the planet that faithfully could show it to you.

I hate having to admit technical defeat as much as anybody, but this one I would throw up the white flag for.

I'm hoping the OP can fill us in on the background and maybe there is another way to help them or another way to solve the problem that gave rise to this question.
 
It was a subcircuit in a regenerative receiver. The output was, if I recall correctly, data.

Let's please try to stay on topic.

While at it you. mctjo,might pick up some ideas from the attached series of slides.
Given the question was for a "square wave oscillator" it is you that is off topic, not anybody else.
 

ericgibbs

Joined Jan 29, 2010
11,651
hi mtcjo,
May I ask what is your application that requires a 1 to 5 GHz square wave.?
Also you have not stated the required amplitude of the square wave signal.?
E
 
You can start with DIY pulse generator (mod it to clock with osc) -

http://www.starlino.com/build-a-rea...time-using-an-ultra-fast-sige-comparator.html

https://github.com/podonoghue/Jim_Williams_Pulse_Generator/blob/master/README.md

As far as scope to observe waveform, sampling scopes, like 7000 series Tek with
sampling timebase and vertical amp plugins can take you to 10 Ghz for a few
hundred dollars.

Regards, Dana.
CRO technology has come a long way since 2004! You can get a CRO now up to 10's of GHz and with plug-ins up to 100's of GHz! There are several Tectronix CRO's now offering this. Who knew! This is fantastic.
I couldn't find anywhere that sold this technology for a few hundred dollars though. Where did you see that? I for one, and possible the OP could make use of such an instrument if you could just point us in the right direction that would be great.

BTW, the generator you linked to has a rise time of 50ps but the OP needs a half period of 100ps and a square wave. So the 50ps, while impressive, isn't quite the ticket. It would look more like a sawtooth waveform, not a square wave.
 
Top