Gauging interest for an SDR PA

GNU Radio
1

Given the availability of SDR hardware (USRPs, BladeRF, HackRF, etc)
covering VHF through S/C bands, is there any interest in a wide band
power amp to complement this hardware? GaN seems to be ubiquitous now,
and there are medium power, 48 VDC parts available in low cost SMT
packages. So I think it’s feasible for something like:

50 MHz - 3 GHz bandwidth
Class AB
5 - 30 VDC supply (high efficiency, high frequency boost supply)
10 dBm drive
37 dBm Psat
Robust to full mismatch (open/short)
Logic level enable
ALC with VSWR monitoring (serial, I2C, etc)
Small (cigarette pack or smaller)
Low cost
Open source

2

On Tue, Nov 26, 2013 at 9:37 PM, Louis B. [email protected]
wrote:

Small (cigarette pack or smaller)
Low cost
Open source

This sounds neat to me. Would you plan on having harmonic filter
banks on the input and/or output of the amplifier? How do you think
you would tackle the ALC with the different arbitrary bandwidths of
the SDR signals? Also, curiously, do you have an efficiency in mind
that you want to design for?

Brian

3

Hi,

Open source
It does sound really neat.

However as Brian pointed out, I think on-board filtering would be a
must. I’m not even sure if just harmonic filtering is enough.

The output of all those SDR isn’t necessarily the cleanest from a RF
PoV and amplifying that to 37 dBm without paying attention is
definitely a bad idea. Combine that with the fact a lot of people
buying those newer / cheaper SDR have not much RF experience ( you
know, like connecting TX output to RX input directly ) and that could
be a recipe for disaster with people broadcasting noise all over.

Then of course there is the legality aspect. Although it’s pretty easy
to fit SDR into the “test equipment” category with their low power
output, a PA will definitely not fit. Not much an issue for people
building it themselves, but for sale / distribution it could be
another matter.

Cheers,

Sylvain
4

I was thinking the user could just screw on coaxial low pass filters.
For instance, the VLF-xxxx line from Minicircuits are available from 80
MHz to 6 GHz, at about $22 each. An output filter bank would be doable,
but it would probably be 0.4 dB insertion loss per switch as S band, so
it adds up pretty quick. Peregrine makes some SOI switches which are
essentially DC coupled and have P1dB=40 dBm, so that may be an option,
along with relays.

In band filtering may be too specific, and get costly as frequency
increases.

Ha ha, yes, burning things out would be a problem. Can’t protect from
some user error. Even TX/RX monopoles with one meter separation may
blow that 2nd stage amp on the WBX card. As far as the amplifier
itself, these 48V GaN parts have 160V breakdown, so running them at ~30
VDC rails should make them pretty robust. Allot of GaAs parts breakdown
at only 2.5 times the rail. Conversing with a guy from Nitronex, he
said 37 dBm with a 30 VDC rail should be feasible without any matching
at the output.

Efficiency would be poor, probably at most 20% at P1dB, but that would
be the trade off for that much bandwidth. I figure true SDR (with the
computer) is already inefficient to begin with. Heat could be taken
care of with a processor cooling fan. There could be space left for the
user to add matching to shape up the drain waveforms, with user
adjustable gate bias, just limited enough so that nothing would be
cooked. Maybe instructions of where to solder parts for 2m, 70cm, ham
bands etc.

I have had good use with the RMS detectors from Linear Tech and Analog
Devices. Typically 0.2 dB variation over 5% bandwidth when used in an
analog ALC loop. There may have to be a LUT to flatten the detection
form VHF to S band. They should be pretty insensitive to modulation
bandwidth as long as it is a fraction of the RF bandwidth, and the video
filtering is low enough. Maybe not a true ALC, but I/O available to
feed back to the SDR board. Linear Tech now has a detector with built
in ADC and 3 wire interface. It would be rough power detection as not
much coupler directivity could be had for that bandwidth, really non at
all; just a resistive tap with some equalization to flatten the
response.

I can see how marketing as a radio would be an issue. It would have to
be test equipment, yet, anyone can freely buy a multi-KW, multi-GHz PA.
I’m not sure how the FCC views that stuff. ITAR would be a concern to
with the GaN; could be a show stopper. Anyway, just a thought for a
project. I can key up repeaters now directly from the WBX card, but I
have a 10 dBi antenna 20 feet up. 5 watts would allow desktop
experimentation, again, given nothing is blown up.

5

The latest GaN devices seem impressive, many rated at DC to 4GHz.
But the reality is whether impedance matching circuit can reach 50MHz to
3GHz bandwidth.
The problem is power efficiency and thermal dissipation.
Especially when you combine boost DC supply, you’re not going to have a
heat sink fitting into a cigarette pack size.
Best power efficient of RF device is at saturated power, but often SDR
signals have high peak-to-average power ratio, and must be backed off so
only the power peaks reach saturation.

Seems alot of these challenges are addressed with whats known as “class
E”
or drain modulation / polar modulation.