HF Power Amplifiers





First Contact

--- My first PSK31 using GNU/Radio Weaver generated SSB signal contact was made on Saturday, March 26 at 8:30AM EST with WB4GRB in Athens TN from Dunbar WV using 30-40 watts to a 40mtr hamstick on an apartment deck. My reported IMD was -29 which is great. The PSK31 audio was created with "linpsk" on the same notebook that gnuradio was running on, just looping back headphone out to mic in with a cable. The tx looks substantially like below except the W7IUV amp in the box with the USRP has been removed - it was causing massive parasitic feedback - the solution was to remove 13db of gain and use larger signals from the USRP. Also had to spend time putting ferrite cores on wires and rf chokes and bypass caps on wires entering the USRP case. Not a very good ground in a 2nd floor apartment other than the power safety ground. A ground wire had to be connected from the notebook and the usrp case otherwise massive usb errors (the usb being the only connection between the notebook and the rf system previously). Once the notebook was also grounded those largely went away.



Here's some photos of my power amp chain and USRP which will hopefully work for CW, PSK31 and SSB phone. So far it works ok into a Heathkit cantenna dummy load.



Overall setup with the USRP on top of the 13.9v 25A power supply on the left, under the Dell notebook. The USRP has an internal amp, then feeds the Minicircuits ZHL-32A, then a Communication Concepts Inc. AN779, then a CCI AN762, an SWR/Power meter then a Heathkit Cantenna dummy load under the bench.



Closeup of the AN762 140W power amp - this went together easily and worked right out of assembly as advertised. It is followed by a CCI 20meter band low pass filter (a requirement to meet FCC regs.) It gives us around 13-17db gain.



This is the CCI AN779L (I *wish* they could still make the AN779H as it had a now obsolete module that could work with .1mw input.) I had a bit of trouble with idle current - the kit recommended some bigger resistors and lower supply voltage than the Motorola application note, but I had to lower the supply even further to 10.5vdc to keep it from going into thermal-runaway and burning fingers. 4 3A series rectifiers works fine for lowering the supply voltage. With that and the fan it runs nice and cool and gives us 28-32db gain, enough to get full output from the following stage. Driving this is the Minicircuits ZHL-32A, an amazing but expensive module good for 27db gain from .05-130Mhz. It's 24v but works fine on 13.9 as long as we don't ask much from it (max output 790mW/+29dbm, much more than we need).



This is the USRP and the 1st amp, a W7IUV circuit that gives us 13db gain. There is also wiring for a morse code hand key to one of the TX daughterboard I/O pins.

There are a *lot* of problems with this type of setup - such as having to be very careful not to overdrive anything, which is easy to do, such as accidentally typing '1000' where you meant to put '100' in some GNU/Radio script! But it provides a clean signal as far as I can tell with the dummy load. Sending PSK31 from another notebook with an audio cable, a ssb modulator script and picking it up on a nearby receiver w/ antenna disconnected, IMD is as good as -31 and ocassionally -32db with 20W on 7.072Mhz and 10W on 14.072Mhz