23cm Amplifier - PCB Artwork

After plugging the various parameters into a micro strip calculator I had the line length and width needed for matching. The draft of the PCB looks like this :

If you have this device and would like the artwork in PDF format just ask.

73's

23cm Amplifier Development using RFSim99 - Part 2

Following on from Part 1, here is how I plan to replace the matching netowrks with microstrip.

Step 1 : Input network

1. Remove the LC components on the input network.
2. Press Simulate and read off input impedance of 4.96R+j3.93R
   
   
3. We know that one full revolution of the Smith Chart is 23cm/2. By inspection we can see that we need a transmission line around 62 degrees (135 degrees/2)
   
   
    
4. Insert a 4cm transmission line (62degrees electrical length at 1296MHz) and press Simulate
   
5. Not quite long enough. Trial and error guesswork gives us 4.4cm. The alternative to this inspection approach is to use a program like Iowa Hills Smith Chart which I prefer. But I'll stay on-topic.
6.  Now we need to move along the arc towards the origin with a series capacitor. I started with 6pF, since it was there on the output network. There was little movement so I halved it each iteration. I found 0.82pF was the closest standard value. You can buy these!
   
   
   

Step 2 : Output network.

1.  Repeat along the lines in Step 1. It is a nice coincidence that the output transmission line is the same length as the input line. Since we hit the arc running towards the origin a littel closer to the origin than the input network did, we can use less reactacne. I guessed 1pf. Turns out 1.2pF is better.

Results

We still have stability and good matching. Gain is  unchanged at 18dB. But now we have two standard value capacitors and some PCB drafting to do.

I'll post about the artwork next.

23cm Amplifier Development using RFSim99 - Part 1

I have an assortment of MMIC's recovered from a cellular base station. Since an engineer previously selected these as suitable for a device where linearity matters I thought they would be good contenders for UHF and microwave projects. The SXB4089, a 500mW output device, looked interesting for a 23cm transmitter so I decided to document my approach. This post concerns the steps leading up to  generating PCB artwork for a test board.

Step 1 : Set up RFSim99

1. Load the S2P file into a unmatched schematic and press Simulate
2. If not auto-scaled set the lower frequency to 50MHz via the Graph Limit Setup command
3. Repeat for the upper frequency of 6.05GHz
4.  Start with 500 points and reduce until we get something close to 1296Mhz. 497 is just right

Already we note that matching will be needed on input and output where the return loss is too high (S11=-0.84dB, S22= -2.46dB)

Step 2 : Check for Instability

1. Switch to the Smith chart. S11 and S22 plots show the unstatble region is outside the smith chart boundary
2. With S11 sweep the frequency by dragging the slider. Very small possibility of instability at 630MHz noted.
3. Sweep frequency for S22 display, note large range of unstable loads at 473MHz.

Step 3 : Fix instability

1. I try a range of resistors in series with Port 2. Nothing appeals.
2. Repeat for Port 1. I quickly establish that a minimum of 1.5 ohms brings stability on input and output at all frequencies.  
   
3. Let's be a little conservative and use 2.2 ohms going forward.

Step 4 : Matching

1. Drag the slider so the frequency is 1296MHz.
2. Press Auto Match
3.  Be amazed at how clever Stuart Hyde is. I'd like to meet him to say thank-you in person.
4. A conjugate match simultaneously brings input and output to 50 ohms and updates the schematic
5. Sweeping with the slider still shows no instability

Results:

A stable amplifier with a gain of 18dB matched on input and output to 50 ohms.

You could change the values to he nearest standard value, and press Simulate to see what happens to gain, matching and stability. However, in Part 2 I will cover how I plan to replace the matching networks with micro strip.

NanoVNA - Test fixtures for measuring SMD's Summary

Well, I'm really impressed. I never expected such a low cost instrument to perform so well. Granted, as frequency increases you get some odd results, but nothing that detracts from this piece of test gear. Perhaps I'll buy a second unit and dedicate it to SMD measurements. Then I could recall calibrations like 0-30MHz, 0-100MHz, 0-300MHz and 0-900MHz with ease.

I can also recommend the test fixture I made. Simply a SMA connector onto a PCB. The track and pads allow for calibration ie

    Short - bridge the track with a piece of copper tape, perhaps like that shown in the picture which
                was there to short out some imperfections,

    Open - with nothing in place

    Load - a 50 ohm resistor bridging the gap. Could be two 100 ohm resistors soldered together but
                you can buy or recover 50R chip resistors. 

Once calibrated same your settings and if you need to recover a part or measure somehting it becomes very easy. It never appeared to be introducing errors which is a credit to the people behind the NanoVNA.

A few more pics just to whet your appetite to replicate this:

The 22nH Inductor

0201 12nH Inductor

Measured with ease

1 47nH

2 56nH

3 6.8pF

4 200 ohms

5 1.5pF

6 0 ohms