10 GHz oscillator and switch
I want to build a pulsed 10Ghz source with kHz switching frequency. I thought to build a 10GHz oscillator and then using a GaAs switch. what do you think?
How I can approach this project first? What I can use to design the PCB and what kind of problems I should face (impedence matching, reflections, insertion loss)?
I do not care so much about the insertion loss since the output of the system will be fairly low power and it is a second order problem.
I would also like to build the 10 GHz oscillator, i see around some nice chip (VCO+PLL) from Hittite that will do the job.
I would like to use coplanar waveguide to have everything on a just a double layer PCB. Any suggestion on how I made it (tips, software, calculations)?
Thank you to all!
If you pulse the oscillator in any way you can expect it to be
unstable at the beginning and end, including some thermal-
time-scale frequency drift. Probably better to run it full time
(or allow time to stabilize before emitting anything). I think
you should use a RF switch after a buffer amp to keep from
kicking the frequency around, if this is a comms system with
any adjacent channel interference requirements.
Before you get all excited about designing and building, you
would be well off to consider how you will debug and test
what you've made. The gear for that, dwarfs the cost (and
probably, effort) of one-offing the radio itself. Have you got
your 20GHz spectrum analyzer and, if things go very wrong,
your 20GHz channel bandwidth 'scope and probably a VNA?
Without these, how will you know anything about what you've
built?
Making a board look like 50 ohms in that frequency range
is a lot of art, and it sounds like you're looking for all of it.
Thank you for the quick response dick_freebird!
What you suggest is what I was planning to do. A continuous oscillator that is switched by a non-reflective GaAs switch.
The instruments are not a problems. I have everything I need: spectrum analyzer, VNA and Scope :).
Do you have any suggestion in the planning, structuring this project?
Up to know I can have an analog generator up to 6GHz, I should receive also a 10 GHz oscillator just for testing the switch, but in any case I should build my own.
Thank you again.
You will want to carefully review the ill effects of loss tangent on substrate and controlled impedances with many ground micro-vias is a must.
The impedance matching is the most important thing I need to take into account. How I can take into account the impedance when doing the PCB?
If you are a beginner, I would suggest to see first how others do it. Buy a cheap satellite Low Noise Block Down Converter, and open it. On the CB you can see the amplifier stages as well as the mixer and DRO.
Many amateurs have modified the LNB to operate as a keyed transmitter. Simply take the DRO output and using a short coax or stripline, connect it to one LNA HEMT stage. Use the gate bias to key its gain.
Learning by such experiment an show you how to use existing structures for your purpose. Then you can try design yours and compare the results.
Hittite MMICs are good and will give you good results. Impedance matching in microstrip is a standard procedure in microwave circuit design. It is explained in books and on this Forum, but in real life you must do experiments to learn how to run a good design.
Hi I was thinking of buying a DRO at 10 GHz and attach via the SMA the eval board of the switch. This, just to have a proof that it can work. Does it sound good?
The only non reflective switch I found for 10 GHz non reflective is from Hittite. Do you know other solutions?
if it were me, i would start off with a 10 GHz DRO running CW, add a 10 dB pad at the output, then add a 10 dB gain amplifier, then add a spst non-reflective 50 dB isolation switch. You should be able to run that at KHz switching rates without too much frequency movement.
BTW, it does not matter if it is phase locked or not, you will see an instantaneous frequency jump when you turn the RF output on and offl.
You could build a non-reflective switch by terminating one port of a SPDT switch to 50 Ohm.
Do you mean a mechannical SPDT? Because I think I will need a Microwave Switch if it is integrated, no?
Well that could be an option but mechanical switches can be expensive.
I was referring to the fact that you don't need a non-reflective switch; you could probably implement one yourself.
Thank you z3phyr!
I would like to know I can proceed on the PCB design... is it necessary a simulator as AWR or MWO? Can I match the impedance just from the calculus and design directly in altium?
Since for the beginning I will start with a pre-made source I need just to design the PCB for the switch.
It's probably OK if you can make sure that you get the impedance right and that you follow good RF PCB design practices, e.g. maintaining continuous ground plane under RF traces, shield RF traces with many vias and copper pours, etc.
It's also a good idea to follow reference designs provided by the vendor. An example can be seen on Page 10 of this document:
https://www.hittite.com/content/docu.../hmc347lp3.pdf
Thanks again z3phyr!
Since I need the source to work in CW and then pulsed by the switch, what would be good to have a PLL +VCO or just a VCO?
Only the VCO will be pretty simple since the output will be the needed frequency with a non negligible power (for my application of some dbm)
Thank you!
I think it really depends on your need. For example, do you need to hit a particular frequency with great accuracy? If so and you want to use a VCO, you'd have to have a very low noise power supply with precision voltage output.
If you are uncertain, I'd go with a VCO first as it is a lot simpler and easier to acquire.
I do not need a really precise frequency and as you suggested I will go for a VCO.
Now, let's imagine I start to draw the PCB, I put a VCO then connect the switch with a straight line of length L and width W.
I choose to design with coplanar waveguide, let's say following this nice tutorial : https://www.hittite.com/content/docu...w_cost_pcb.pdf
afterwards how can I be sure that the impedance between the VCO and the switch is 50?
put that same line structure to another location on your PCB that you can connect two sma connectors to it, then use VNA to measure the line itself (+2 connectors and the launch of those connectors).
I would like to suggest this project: https://hackaday.io/project/1682-Sim...ost-FMCW-radar
This guy is trying to do a FMCW radar. Is this the approach I should follow?
I found several interesting thing:
1) He did some breakout boards for each components. It is very interesting and useful.
2) He uses a VCO.
3) He uses a LNA at the output of the VCO. Is is necessary for my application? I need less than 50 mW on 50 Ohm...
Yes, should be a good one. The project is loosely based on the original MIT OCW course: http://ocw.mit.edu/resources/res-ll-...uary-iap-2011/
I've also taught a similar class at our university: http://ucdart.net/SeniorDesign
You may find some relevant information on the website.
I am going deeper in the projects, I developed a board on FR4 material. I know it is not the best solution but this is what I have and I have to use it.
Thus, I am playing a bit with TXLINE from Microwave office.
I put all the parameters needed, tangent loss, dielectric constant and so on. Then I want to build a coplanar waveguide with the ground. Now what I do not understand is why when I specify 50 ohm of impedance does not matter the length of the line I have to build.
I mean, if I change the parameters such as the width of the trace or the gap the impedance change, but why when I change the length only the electrical length change?
I am a bit lost...
Moreover, attaching to my VCO (adapted) pcb an unmatched network what would happen? There is no spec in the VCO for the VSWR.
Thanks!