I realized a problem with the AGC design I talked about: The output
impedance will vary a lot with gain. I'm not sure how much of a
problem that is, I guess it depends on what it's driving, but it's
definitely an issue. I'll have to simulate that.
Just need to look into how much of an issue this will be using simulation.
You only need a few different gain steps in a design so may be able to switch in matching networks if problem.
Also may be better to add another buffer stage to isolate/correct for differences.
Another question on that: the input to that MOSFET is obviously going
to be high impedance, but it's coming from a filter that's going to be
low impedance. (high impedance LC filters don't seem to be feasible).
Normally you would put a resistor to ground to terminate the
connection, but then you are introducing noise. Is that an issue?
Since the frequency is fixed, could you put an inductor of the right
value there?
Have to match circuit nodes as well as accept some loss.
Will likly need a resistor to ground to develop signal. Will add some noise but just have to pick best resistor to minimize noise. You can get so called RF resistors to help with this.
Inductors are fine and may be used along with resistors.
Have a look at
web page. Select the tab on the left side that says “Tech Library”. Then under Circuit Design you will see several things on this topic.
Here is little info on input matching.
The problem is the signals are complex in nature.
Also, Bob, you mentioned building boards for testing. How do you do
that? Do you put a zero-ohm resistor between sections and an SMA
connector on each side of it? You probably want some way to enable
power per section. Anything else I need to worry about?
Put coupling capacitors or zero ohm resistors between circuit sections. Need the caps there typically to eliminate any DC voltage levels.
Include a coupling cap tap anyplace you want to test and put a connector on the other side. The add and remove the caps when testing or working.
Put an inductor in series with the power feed to eliminate RF flowing back into power. Also add tank capacitors on power tap points near devices.
As far as building boards, getting the board itself isn't that
expensive, but how much does it cost to populate the board? The
processor is a BGA; I'm not sure I can solder that, and having someone
else handle all the soldering would be nice.
There are several assembly fabricators around that will build boards. American PCB and the local fabricator we used for Pacsat.
The china suppliers will also fab the boards if you use parts they have on hand or can receive on consignment. Most difficult to deal with but they are the cheapest solution.
The US suppliers will have fees for setup and to make it worth their time to take on a project.
I also know a technician that would build things like this probably for under $100 if not too complex. Cindy Stapelton - would have to find her contact info.
If not too many ball on BGA you and solder this with a hot plate. Put flus on the pads with the PCB sitting on the hot plate. Place the chip and line it up. Then plug in the hot plate and heat till reflow and then unplug. In an assembly process they would
print solder paste first but balls probably will have enough for prototypes. If parts are low cost could just build several and test.
Good to have a watch pad with some solder on It you can watch during reflow.
Yes this one may be fine.
Another thing you can do is use a SDR to generate your signals and to test output signals.
You would connect your outputs through a 10 dB or 20 dB (what ever is needed) coupling attenuator. This gets you a 50 ohm termination and lowers the signal to the level acceptable by the front end of the receiver or spectrum analyzer. Have to make sure
not to overload what ever is used.
ADALM-Pluto would work
This is another option sold by Mouser that is a little lower cost.
Using the SDR may take more time because you have to learn to use the device but this may help in other projects.