EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: Terabyte2007 on February 19, 2015, 08:50:03 pm
-
Has anyone mocked up anything in the GHz range successfully on a breadboard? If so, what were your experiences. I have done several hundred MHz with no real issues but I may need to mock up a 3 GHz circuit and I don't want to spend the time or money to make the proper PCB yet. I know there will be capacitance issues and more but it may save me some time if anyone has done this before. I just need proof of concept, doesn't need to be perfect.
-
The best way to do ultra-high frequency stuff I believe is to solder it dead-bug style or to use single sided copper PCB and cut tracks out with a knife or use it as a base for dead-bug style.
-
It really depends on what you're trying to do.
I had a batch of boards made with a 50 ohm line with ground either side and plenty of vias.
It's easy enough to cut bits out of the line to mount series components and there is good ground nearby. That's fine for a ladder circuit, anything else gets a bit more involved.
-
Agreed, you'll never manage to build a circuit on a breadboard which isn't seriously compromised at GHz frequencies. A few pF here or nH there and... well, you can do the arithmetic just as easily as I can. Suffice to say, a bit of wire stuffed into a hole in a breadboard is not going to constitute a well matched 50R transmission line.
-
Study up on Stripline and Manhattan Construction. 3 Ghz on a BB is insane, assume there is a few pF between cells, then calculate the reactance at 3 Ghz. You'll quickly see why this is a bad idea.
My ham radio hobby is dedicated to microwave bands. We get away with thin, double sided FR4 to 4 Ghz or so. This requires extreme care when cutting the striplines. You also have to bond the ground planes as much as possible, perhaps every 4-5 mm with thru board jumpers.
The good news is a 50 ohm trace on FR4 or the modern equivalents is about 0.1" or 2.5 mm wide. Which means prototypes can be made with copper tape, a Dremel tool, or X-acto Knife.
Majaz Vidmar's pages show you some decent pics of what a 2.5 Ghz stripline would look like on FR4.
Scroll down to the picture of the interdigital oscillator, you'll get a idea of how things are done.
Keep in mind there is a ground plane behind that surface in the picture:
http://lea.hamradio.si/~s53mv/spectana/vco.html (http://lea.hamradio.si/~s53mv/spectana/vco.html)
If you buy a piece of controlled impedance Teflon pcb such as Rogers Duroid etc... The traces get a lot smaller.
If you take your product to manufacturing, you'll need to work with the board house to ensure the proper dielectric constant on the board material.
FR4 and its modern clones have a Dielectric constant of 3.9 to 4.6, and a nominal loss tangent of 0.02. They get lossy fast at Ghz frequencies. When I get new stock in, I cut a square, measure its capacitance and thickness, and calculate the rough Dielectric constant. I say rough because it will be considerably different at Ghz frequencies compared to the 1 or 10 KHz where I measure it.
But at least its a low cost option for some one who is new to RF design.
Steve
Steve
-
Great info! Thanks, I really appreciate it.
-
Just search for "RF prototyping methods" and usually you get some, not a lot of useful docs on subject. Here I have attached good doc that I have posted here earlier.
High frequency up to 4GHz can be done on FR4 if all tracks are kept very short and you make sure that critical areas are free of solder mask. I have good experience with FR4 for up to 2.6 GHz. When I say short I mean 1/10 wavelength which is in case of 3GHz will be 9.9 mm. Why FR-4 - you can order your boards cheaply. Keep the bottom layer completely solder mask free and bolt the board onto aluminium block.
For higher frequency (6-24GHz) you need to find low-loss substrate like Rogers Duroid. Luckily in most cases your RF section is single sided and frequently you can avoid via stitching. If you have no HF substrate available then you can try to do something that is even better in terms of loss - keep your critical HF sections suspended in air, connect sections by small pieces of low loss coax. Dry air is best "substrate".
I tried using adhesive copper foil on top of PTFE substrate or Alumina - works very well too.
-
I have used pre made boards eg
http://www.minikits.com.au/electronic-kits/multiplier-kits/1152MHz-Mlt-1, (http://www.minikits.com.au/electronic-kits/multiplier-kits/1152MHz-Mlt-1,)
I modified them a bit with reasonable success up to 5GHz or so.
-
...If you have no HF substrate available then you can try to do something that is even better in terms of loss - keep your critical HF sections suspended in air, connect sections by small pieces of low loss coax.
...good luck with that! A rather dramatically oversimplified recommendation if I may say Alexei.
-
To the OP, it might help if you explain what you are trying to do, there is a world of difference between designing an LNA to an HPA and everything in between, to something digital perhaps?
-
Up at 3GHz the best tool is a decent EM simulator on a PC as this can simulate your circuit layout and can reduce risk (and disappointment) before you make a PCB.
Many years ago I bought a T-Tech PCB milling machine for prototyping RF boards but as has already been suggested, you can also get reasonably good results using a single sided PCB and then stick your microstrip traces and copper shapes on there using sticky copper tape cut with a scalpel. Then drill holes to fit ground pins in the relevant places.
This can work OK up at 3GHz but it depends on what level of performance you are expecting and also what type of circuit you are looking to build...
-
...If you have no HF substrate available then you can try to do something that is even better in terms of loss - keep your critical HF sections suspended in air, connect sections by small pieces of low loss coax.
...good luck with that! A rather dramatically oversimplified recommendation if I may say Alexei.
It is not as uncommon as you think. Take a Log Detector like AD8313 for example - HF section includes impedance matching network made out from 2 capacitors and one resistor and rest of pins are DC or low frequency. Soldering together a contraption from 2 caps and resistor attached to input pin above a hole drilled in pcb is not that hard.
Cut pieces of copper/brass sheet, solder surface mount and through hole components together with copper strips on flat surface. Drill a hole or make a cutout in your board and then suspend whole construction above the cutout leaving it suspended in air. Unless you are going to drop your board from 3rd floor it will work fine. Obviously you cannot make things like distributed filters this way.
Here is nice example of board from commercial radar detector - it combined 24GHz section on low loss substrate attached to lower frequency section on FR-4 without solder mask and then regular FR-4 with solder mask for rest.
-
Fastest thing I ever tried was an avalanche pulse generator; but these produce high frequencies due to physics rather than construction, so it was not surprising to still observe nanosecond edges.
Tim
-
To the OP, it might help if you explain what you are trying to do, there is a world of difference between designing an LNA to an HPA and everything in between, to something digital perhaps?
Right. If you are doing something digital, then you could get up to 10GHz using NRZ signalling with TX pre-emphasis and RX equalization. These are the tecniques used to get USB3.1 or PCIe3 to get up to 10GHz on FR-4.