After failing to get much gain out of these things at 2304 using my
modelling approach (i.e. come up with new match, hack the board, test,
fail)....  I just monkeyed with one board today. 

I have a 2.4 GHz GaAs 1w stage driving one of the two 19125's... Which
drives the 2nd 19125 (they're just cabled together with pigtails and barrels
so I can test individual stages).

Not sure what gain I'm getting... My setup isn't perfectly calibrated, but I
know I'm getting 40-60 watts out.  Still need more tuning and more drive,
but it's a start.  I was at the 20w max level before moving to the GaAs PA
driver, so at this point I may still be drive limited.  I have a better
driver around I can try to get going soon.

Gotta heatsink the thing too, at 10-12 amps total current on that hunk of
aluminum it gets pretty toasty quickly.

Anyway, just thought I would provide this little update... I'll continue to
work on it as I have time.

-Lance

Yeah, it's really nice to be able to make small coax toasty.  After messing
with it this morning (and getting more drive)... I've seen 80-100 watts.
I'm probably not drive limited, but I think I only have about 3-5 watts max
driving the first 19125 (which should be enough if both stages were
reasonably well matched for gain and power). 

The 100 watts (+/- 1 dB probably) tends to only come on the initial power
up, I expect this is due to heating in the device.   I can get about 60-80
watts regularly now.

The efficiency of my setup is really poor though, and I start to suffer
significant voltage sag when this thing (not optimally tuned) draws 15-20
Amps (probably 3-5 amps are in the drive stage (?).   Ideally I would
monitor the current to both stages separately, but I've not yet setup for
that.

We should be looking at isolators that can handle this kind of power (I know
Dave and others have found some...).   What the heck would people (rovers)
use to provide the 15-20 amps at 26v? (assuming a single stage was the
output, the 1900 MHz data sheet performance {who knows how close we can get
to} only shows 40% efficiency... So 100w output means 250w DC input.

We'll also need heatsinks (this is painfully obvious in my setup... I have
the PA resting on a heatsink with a wet paper towel in-between).

Anyway.. It's still a fair bit of work to go, but it's at least somewhat
encouraging.    Linearity/Stability totally TBD...

-Lance

Here are the AML 2304 MHz amp mods, or at least the first pass.

No real new work has been done on this front, but I've finally documented
them.

I did not carefully measure (lazy and only have one directional coupler) the
gain, but I suspect it is in the 10-13 dB range.  I've seen 60-80 watts out
of a single stage with a 26V supply.

Make sure that you take note of the comments I added to the notebook-scan,
and the correction (output network dimension update).   If you can't see the
comments in yellow, let me know please!  They do not appear to print by
default, you can select "print with comments" but it gets a little ugly.

Here's the .pdf of the notebook page ...and here's a snapshot of one I was driving
with two stages combined...

-Lance

1. Take the uP board off of an amplifier and the upper cover, and expose the bias and power supply board. 
2. Look for the large IR transistor and the silver aluminum electrolytic cap in the lower corner near the power connector. 
3. Now look up about 1 inch above the IR device, for the narrow rectangular ferrite chokes FLT8 and FLT7. 
4. The traces from the top sides of these 2 are the wires that go to the uP controller and enable the amp. As they are without the uP, there is no voltage applied. 
5. Jumper these 2 points together and run them thru about 1K to any convenient point of +5V on the board with a jumper wire. (there is a 5V regulator located near the middle of the pcb) 
6. Now when you apply +26 to the connector, the amp should work.
   

Amplifier Pictures