Development process for RF

[xylon89del]

Hi guys,


With your experience in RF design, what is the best method that you can think of to make the development of RF design faster?

FYI , the RF circuit I want to build consists of ADC, mixer, bandpass filter, oscillator, PLL and etc.

1) Is there any development board available?
2) Using ADS simulate all the circuits block then straightaway fabricate a complete PCB ?
3) Or design the circuits part by part , and fabricate the circuits part by part, in the end only combine?
3) Eventhough final product requires SMD, but when prototyping, it is better to use throughhole resistor/cap first or straightaway use SMD resistors/caps?
4) For home-based development, how do you all ensure that you have
all the resistance/capacitance/inductance values that you need when prototyping?
When need the value, then only buy?
Or buy a very wide range of values, and can straightaway use when needed?

 

[klystron]

Please give us more information on your project requirements.

 

[xylon89del]

Please give us more information on your project requirements.

The title of the project is called PC-based spectrum analyzer.
A Low budget project.
I have built up the software on the PC, done the USB communication between PIC and PC.

Now is trying to build up the analog frontend to downconvert the frequency from 250MHz to 1MHz, where mixer , oscillator, PLL & bandpass filter comes into play.

Hope this info is enough.

 

[biff44]

I am not sure 1 MHz qualifies as RF development anymore! That is pretty low frequency.

But, assuming you want to branch out on your next project to higher, even microwave, frequencies, surface mount is the way to go. With standard very low cost surface mount resistors, capacitors, and inductors you can do almost anything up to 2 GHz and beyond.

I would go to ebay or digikey and get component "kits". They will have 20 or 100 pieces of each standard value from x to y. I would use 0805 size chips--they are not too huge, but are big enough to see with your eyes or a large magnifying glass....easy to handle for a newbie.

In addision to the kits, I would also stock a bunch of commonly used other parts, like 0.1 uF ceramic chip caps, 4.7 uF ceramic or tantulum caps, npn and pnp transistors (like mmbt2222a), standard op amps (like op-27), diodes, 5/12/-5V regulators, etc.

I would use surface mount prototyping PCBs, which have a solid ground plane on one side, and lost of little square pads on the other side.

I would chop a small board out of the larger board material, and use "edge-mount" SMA connectors for RF inputs/outputs.

I would drill a small hole thru the board, and run a short piece of fine bus wire thru the hole to the backside ground for each pin or component that needs grounding.

As mentioned you can get a lot of this stuff off of ebay for very reasonable price.

The important thing you are doing buy all of the above is to make prototyping circuits a very easy and quick thing for you. In today's electronics, it is very easy to simulate the performance. And the simulation tools are so good, that engineers tend to believe them--so they end up way out on a limb with a whole string of cascaded parts that are unproven but only simulated. Then you turn on the switch....and none of it works and you are stuck trying to figure out why. If you had breadboarded and tested the more difficult blocks first, you would have noticed the quirks in performance, fixed them, and then ended up with a more robust system.

---------- Post added at 12:15 ---------- Previous post was at 12:02 ----------

here is a board, smt-200, that has a ground plane on back and smt pads on top:
http://www.boardworxsystem.com/index_files/smt200.htm

they stock them at Jameco

 

[xylon89del]

I am not sure 1 MHz qualifies as RF development anymore! That is pretty low frequency.

But, assuming you want to branch out on your next project to higher, even microwave, frequencies, surface mount is the way to go. With standard very low cost surface mount resistors, capacitors, and inductors you can do almost anything up to 2 GHz and beyond.

I would go to ebay or digikey and get component "kits". They will have 20 or 100 pieces of each standard value from x to y. I would use 0805 size chips--they are not too huge, but are big enough to see with your eyes or a large magnifying glass....easy to handle for a newbie.

In addision to the kits, I would also stock a bunch of commonly used other parts, like 0.1 uF ceramic chip caps, 4.7 uF ceramic or tantulum caps, npn and pnp transistors (like mmbt2222a), standard op amps (like op-27), diodes, 5/12/-5V regulators, etc.

I would use surface mount prototyping PCBs, which have a solid ground plane on one side, and lost of little square pads on the other side.

I would chop a small board out of the larger board material, and use "edge-mount" SMA connectors for RF inputs/outputs.

I would drill a small hole thru the board, and run a short piece of fine bus wire thru the hole to the backside ground for each pin or component that needs grounding.

As mentioned you can get a lot of this stuff off of ebay for very reasonable price.

The important thing you are doing buy all of the above is to make prototyping circuits a very easy and quick thing for you. In today's electronics, it is very easy to simulate the performance. And the simulation tools are so good, that engineers tend to believe them--so they end up way out on a limb with a whole string of cascaded parts that are unproven but only simulated. Then you turn on the switch....and none of it works and you are stuck trying to figure out why. If you had breadboarded and tested the more difficult blocks first, you would have noticed the quirks in performance, fixed them, and then ended up with a more robust system.

---------- Post added at 12:15 ---------- Previous post was at 12:02 ----------

here is a board, smt-200, that has a ground plane on back and smt pads on top:
http://www.boardworxsystem.com/index_files/smt200.htm

they stock them at Jameco

Hi, your post is really helpful. Thanks.

Besides that, how you usually test a SMT chip/IC?

Do you use breakout board? or straightaway solder it onto the whole circuit?
For my case, I lack of toolkits for desoldering SMT chips.

 

[biff44]

it depends on your application. If you are talking about digital circuits, a breakout board can be helpful if you are switching less than 100 mhz rate. Above that rate you will start to get a lot of ringing from the extra parasitic inductances.

For a modern op amp with anything bigger than a few MHz of bandwidth, I would be nervous about moutning it on a breakout board since the op amp can become unstable.

For microwave circuits with lots of high frequency gain, instead of a breakout board, I would be tempted to build the circuit on a small (1" x 1").
If I have SMA connectors on the small board, it is easily attached to a bigger system board. Even without the SMA connectors, I will sometimes solder on wires as if there were connectors--at the input side of the board two very short (<.1") wires to joint the 2 ground planes, and a short center conductor wire. Then at the output the same thing--3 short wires. Will probably work up to a few GHz.

Just be careful with "breakout boards" for filters. Something like a SAW filter will have worse out of band isolation if it is not mounted on the main board directly.

 

[Terminator3]

Good thread! I have some questions too. Frequency range 5..10GHz, maybe 24.
1) what kind of Schottky diodes i can buy? I want cheap ones, but in usual SMD black plastic package. Soldering without optics/microscope, but still works good at 10GHz for example.
2) where to buy DROs? What kind of glue i can use?
3) What is that gray or white paint substance used above microstrips in prototype boards? Is it for tuning dielectric coefficient, make wavelenght smaller, tune phase? Usually looks ugly.
Thanks!

 

[xylon89del]

Even without the SMA connectors, I will sometimes solder on wires as if there were connectors--at the input side of the board two very short (<.1") wires to joint the 2 ground planes, and a short center conductor wire. Then at the output the same thing--3 short wires. Will probably work up to a few GHz.

Want to confirm my understanding.

To simulate the SMA connectors with wires, you use one signal wires and two ground wires.
Can I twist them together to make them really close to each other?


I am confused with your 2 ground planes. What do you mean by that? If I am using microstrip , I will only have 1 ground plane unless you mean stripline in your case.