PA design without models
Part 1 of my question:
I am trying to design a PA using SPB-2026z which is now RFPA2026. The datasheet is here:
http://www.dielco-store.com/media/da...s/spb2026z.pdf
The PA is suitable for 700-2700MHz, and I want to use it for 1 GHz. However, I don't find much model information on the product aside from the application circuits at 1.8 GHz, 1.9GHz and 2.1GHz. I have simulated the load network for the 1.8GHz application circuit, and I can do a rough translation of the impedance down to 1 GHz, but the performance is not really great. So my question is, what could I do to optimize the performance? I do not have access to load-pull equipment if you're wondering.
Any thoughts are welcome. If there's any way to model the transistor inside, even better.
Part 2 of my question:
I found this application note for the above IC:
http://www.rfmd.com/sites/default/fi...nSPACENote.pdf
and I simulated the impedance networks, and tried the same impedances into my design. However, the gain was poor (for reasons I won't go into, my measurements were not what I would call very accurate, but they were informative). Looking at the application circuit, I note a 0.6nH inductor in shunt to the input path. In the PCB layout, they are using a 0 ohm resistor for this small inductor. This is extremely puzzling to me as it seems too small (0.6nH is about 3 ohms at 900MHz). Given that the IC is pre-matched to 5-ohm, the input matching network here seems to be way off.
Any comments on this application circuit? Am I missing something?
thanks,
Aaron
since impedance only provided for 1.8GHz, 1.9GHz and 2.1GHz, i suggest you ask the impedance at 1GHz because i believe there is no way to predict impedance at 1GHz based on 1.8GHz,1.9GHz and 2.1GHz.
i dont think its wrong to shunt inducto in the input matching path. its only wrong to shunt a capacitor!
I used SPB-2026Z at 1.3 GHz, and works fine, getting slightly more than 2W output power. I tuned manually the input/output match.
Let the printed TL as they are for 1.8GHz application, and tune C5, C6 for maximum PA gain, and L1, C2, C7, C8 for maximum PA output power.
Why don't you measure the s-parameters of that device on a simple test PCB ? ( There will simply be a supply and 50 Ohm TLs at input and output ) Then you can match the input/output regarding to your frequency.
After that you tweak a bit to get the right specs.
Thanks for all of the replies.
I think its gonna come down to getting s-parameter data somehow, plus a heavy dosage of manual tuning.
I've asked, but all I got was s-parameters for the 1.8GHz application circuit...
what's wrong with a shunt capacitor? Anyway my point is that the inductor seems small at 900MHz to be in shunt with the signal path. Also, the designers IMO are pretty brave to put a 0-ohm resistor in shunt with the signal path. I have analyzed the 1.8GHz application circuit, and the real part of the input is 5-ohm, which is in line with what the datasheet said. For the 900MHz application note, however, the input impedance isn't anywhere close to 5-ohm.
Did you simply re-solder each component until you got what you wanted? Seems tedious. I'm asking because eventually I need to try other frequencies and other PAs, and I don't want to have to go through such a long procedure every time.
I will probably do that eventually, but I don't have a VNA handy so there's some effort involved in borrowing equipment etc.