Saturday, May 10, 2014

Back in the Saddle part deux

This is a shot of the panadapter of a Signature Flex, and a copy of CWskimmer.  The bands have been lousy but finally on Friday things opened up a bit.  I followed the grey line across Europe until it  hit Ireland

It was fun hearing another epoch of operators wake up, make some coffee, and stumble into the shack to flip on the radio and see what DX was there to greet them.  For some it was me.  I worked about 30 barefoot countries in about 3 hours until I finally hung it up for the night.  It was great fun!

The panadapter shot above is 40M with contest mania last night.  There was some European thing going on and QSO parties to beat the band.  Note the new waterfall above.  I didn't think it would be, but it has turned out to be quite useful.  More on the new display in a future blog post.

The New Architecture

I was talking to Gerald K5SDR at the Orlando Hamcation a couple months ago, and he demonstrated and explained the new architecture to me.  Flex switched from a radio that was basically a modem with a really powerful PC based communications software program, to a radio that is basically a dedicated big honkin number crunching data engine.  The way they do the number crunching is with a FPGA or Field Programmable Gated Array.  If you look at the upper left of the blog you will see a Xilinx chip.  A Xilinx chip is at the heart of the new radio.  A FPGA is not a computer but it can be under the control of a computer.  A FPGA is basically a bunch of logic gates that can be programmed to do specific tasks.  The programming kind of turns the chip into a custom "radio" IC and the nature of how the IC behaves can be changed on the fly, from outside (or inside) the IC.  What this means is the chip is a cross between a device that has a static program burned into it, like the controller in your microwave (which is a kind of radio transmitter) and PowerSDR on a general CPU which is totally dynamic, but suffers from ADD (Attention Deficit  Disorder) since its attention is always trying to be interrupted by things like painting the screen and wondering whats happening on the USB line.  By dynamically setting up the task at hand, the FPGA can then proceed to process without all the darn interruptions.  The FPGA is programmed in a native language which means it is not dependent on higher level language to tell it what to do.  This all means it is blazing fast at its tasks, and this is why it's a game changer!    

SDR is what I call a near real time endeavor.   It happens in virtual time.  What you want is for the process to be as close to real time as possible and the FPGA is the best tool for that job.  Add to that a big honkin ADC (analogue digital converter) and now your set to do some damage to the RF spectrum.  The Signature radio basically works by presenting raw RF directly to the converting chip much like light is directly applied to the CMOS of a camera.  This is called direct digital conversion, no base bands or mixers are involved.  Flex calls this direct conversion device the SCU or spectral capture unit.  It captures a continious set of snapshots of an entire hunk of spectrum 0 - 72 MHZ wide and sends it into the radio for further processing.  The processing function in the radio SLICES this 72 mhz wide data into 14 mhz slices and puts that 14mhz into a panadapter called a "slice receiver".  The radio depending on the model can display either 4 or 8 of these 14 mhz wide "slice receiver" panadapters on a computer screen.  This means the 6700 can display all of the RF data between about 30 khz and 72 mhz all at once.  In any slice one can drill down the resolution until the graticule is only 400hz wide.  This means a 2.7 khz ssb signal will occupy about 1/4 of the screen.  Amazing resolution!  Since all of this is being done in the radio and not in a general purpose computer it is all done very quickly and very efficiently. 

Here are some shots of computer utilization vs SSDR in various forms of stress

This is a shot of utilization with no programs open with my i3 ivy bridge processor running the chip's internal 4000 HD graphics driving 2 1920 x 1080p HD monitors

This is utilization running one slice of SSDR 1.2.1 with the waterfall and fps wide open.

Here are 2 slices wide open.  Note the i3 is throttled back barely percolating

Here is one slice with one instance of skimmer with one DAX channel active.  Skimmer and DAX use a lot more clicks.  Skimmer is joined to SSDR via the SDR-Bridge program

Here are 2 slices of SSDR with 2 skimmers and 2 DAX channels open using W2RF fine SDR-Bridge program.  As you can see utilization has gone up dramatically but this still works fine as far as functionality.

Here is a shot of my station in the heat of DX battle!  The top slice is the DX and this is locked on his freq.  This slice corresponds to the right Skimmerfall.  The bottom slice is the pounce for the kill transmitter slice.  It corresponds to the left Skimmerfall.  Operation is dead bang simple listen to the DX, look for the 599, click red 599 dot on the left skimmer.  Hit the keyer memory with W9OY.  Hit 5nn and TU and log the sucker!

I have the new 6300 so I am limited to only 2 slices but it is entirely adequate.

So back to Gerald K5SDR.  This is why I admire his courage.  He and his team grew to know this IS the correct direction to take SDR.  He took the risk, his team delivered and my new radio experience is the result!  KUDO!  

More to follow on the 6300.

73  W9OY