discrete RF prototyping

[Da_Stier]

Hi all,

I find myself to build a lot of discrete RF prototypes lately.
This thread is meant to show, how "far you can go" with normal prototype boards and careful planning of your circuit.
It also is a kind of shameless self commercial for my prototype boards:
https://highvoltageforum.net/index.php?topic=1560.0



The first prototype I want to show is a PIN diode attenuator.
PIN diodes are special in a way that makes them pretty useless as a diode, however pretty usefull as a current dependent resistance.
This makes it easy to implement a RF attenuator with them.

This is the circuit, it is a very basic implementation of a PIN diode attenuator.


This is the circuit built around a piece of coplanar waveguide prototype board.
The resistors are 0805 packages, the four PIN diodes are still quite a bit smaller, more of a 0402 size.


The prototype works pretty well up to 3GHz.
The attenuation is pretty constant for each control voltage (except in the lower range of course, which is normal).
This tells me, that the losses of the construction technique is relatively small.







The next prototype I want to show is a small buffer amplifier.
It is used as a universal buffer / gain stage.
It is made from a discrete NPN transistor (BFP420) with a 25GHz transition frequency.

This is the (very basic) circuit.


This is the finished prototype, once again based on a pice of coplanar waveguide.


By carefully choosing the location, even throughhole parts can be used up to several GHz.
However you have to be sure about the effects of parasitic effects.


This is the S21 characteristic of the little buffer.
For the transistion frequency of the transistor used, it is not all that great.
For such a simple build however, it works pretty well.


If a peak at some specific frequency is needed, some more effort has to be put in input / output matching.


The last prototype I want to add for now is a simple series tuned Colpitts - or Clapp - VCO.
If you are interested in VCO topologies and construction, this website is what you want:
https://www.qsl.net/va3iul/High_Frequency_VCO_Design_and_Schematics/High_Frequency_VCO_Design_and_Schematics.htm
The circuit I built is a slightly modified version of the Clapp osciallator from the website.

This is the circuit (and the boad I guess)


The circuit is built on a small pice of my SMD protoboard.
The small pads are quite a good fit for 0805, 0603 and 0402 components.
You have to be careful about adding parasitic capacitance however, since there is a GND plane on the bottom.
The GND connections are made by drilling 1mm holes and soldering in small pieces of silver plated copper wire.



Another common technique is to use off the shelf items for electronic / RF prototypes.
The most common example is propably the pipe cap filter.
It uses some copper plumming caps as cavity resonators for filters / oscillators, etc.
This is an example of such a filter, it uses a 28mm pipe cap.
It is tunable from around 2GHz to around 8GHz. 
The insertion loss is quite good at around 1dB.
Also the Bandwith of such filters is pretty narrow with a Q factor of well over 200.
I don't know about much better filters that can be built so easily.













I hope this motivates some of you to tinker more in the RF frequencies.
You can get some rather promising results from relatively simple and cheap materials.
For example by combining a few of the shown blocks, it would be rather easy to build a simple FMCW radar.
Another idea might be a small VCO driven transmitter.
Just stay at very low power and don't interfere with anything.


Greetings,
Michael

[Da_Stier]

Hi all,

I don't want to doublepost, however I think I got another interesting example made today.
This time I prototyped a driver stage for an upcoming LDMOS amplifier project.
I want the driver to be able to deliver around 33dBm or 2W at 2400MHz.
To do so I cascaded a MiniCircuits PHA-102+ and a MiniCircuits PHA-202+.
The PHA-202+ is to deliver the necessary power, the PHA-102+ is meant to add 14 more dBs of gain.
This combination should deliver around 25dB of gain with a 1dB compression point of around 30dBm.
Not quite what I want but I had all the components around so I wanted to give this a try.

The interesting prototyping technique on this is the QFN package with exposed pad on the PHA-202+.
I mounted the IC together with some coplanar waveguide onto a piece of SMD protoboard.


I used a strip of 0.8mm thick brass sheetmetal as both a GND connection as well as a heatsink.


It is soldered to the back GND plane as well as the GND of the waveguide.


This is the finished circuit with the deadbug PHA-202+ and the PHA-102+ mounted directly into the waveguide.


The S21 response of the unit looks pretty good and is pretty flat across a wide bandwith.
It also matches the expected gain of the circuit.


At 1GHz, the maximum output power with reasonably good third order interception is around 33dBm or 2W which is promising.
The third order products are around -60dBc at that power.
Unfortunately at 2.4GHz, where I want to use the amplifier, the 1dB compression point is at 25dBm.
The maximum output achievable is around 26dBm.
At first I thought that my construction adds loss or is bad at this frequency, however after reading the datasheet again - and more closely this time -, it matches perfectly.



So after all, the parts work very well and are pretty straight forward to use.
However they are not quite what I need for my project, so the prototyping might have saved me a PCB revision.
Also I'm pretty happy to see, that these techniques still work at "medium" power of multiple Watts output.




Greetings,
Michael

[johnf]

Very nice Michael
I've always used double sided copper clad pcb and scoured pads with a sharp craft knife and used dead bug style but only up to 1GHz using most of the pcb for ground

[Da_Stier]

Thanks John. 

I've always used double sided copper clad pcb and scoured pads with a sharp craft knife

I built a lot of stuff like this, in the dead bug / Manhattan style.
However at higher frequency this gets kind of tricky due to parasitic inductance and capacitance of the leads.
Actually pretty similar problems to Bridge design, now that I think about it.
But the biggest problem are the modern packages, since most parts only come in tiny SMD packages.


Greetings,
Michael

[Da_Stier]

Hi all,

I have a few more prototypes and test pieces, that I wanted to share.
I think some of the pieces are rather interesting.

The first thing is a 2.4GHz cavity filter.
I followed the article from Matjaž Vidmar S53MV.
http://lea.hamradio.si/~s53mv/cavity/cavity.html
His measurements were spot on and easy to follow.
I built the version with a center frequency of 2360MHz and a bandwith of 140MHz.
(This covers the complete 13cm HAM band)
I didn't have the right connectors available so I used some single hole ones and made little adapter plates.

This is what I came up with:









I was surprised, how much you can tune it, both for the bandwith, as well as the center frequency.
I am pretty happy with the S21 response, however I might come back and do some more tuning.



Another thing I recently did, was to build my own RF adapters.
I have a literal sorting box full of all kind of RF adapters.



However I still come across some combination I don't have from time to time.
The ones I needed today were SMA male to QMA female.
I could only find very expensive ones for over 200$ a piece.
So I decided to build some.

I started with an edge launch QMA connector, that I salveged from some old cellular boards.



I machined some adapter parts and came up with the following:



I also needed a QMA female to QMA female adapter.
I did the same thing to turn down some edge launch connectors.





These are the final adapters:



They won't be as good as the 200$ ones but they work fine for what I do.
I measured them on my VNA up to 3GHz and everything looked perfect there.


I hope you liked the look at some different prototypes and found an idea or two to keep in mind. 




Greetings,
Michael