Hardware upgrade capability for 6000 series

  • 1
  • Question
  • Updated 5 years ago
  • Answered
A potential flex owner asked a question on the Flex Net this afternoon about physical upgrades to the 6000 series.  It got me thinking, (which is always dangerous.)  Say in 5 years or so that a much faster, more powerful, processor chip becomes  available at a good price.  Is there any provision for a processor chip upgrade, like people can move from an i3 to an i7 in their PC? Just curious if this had ever been addressed....
Photo of Ken - NM9P

Ken - NM9P, Elmer

  • 4083 Posts
  • 1276 Reply Likes

Posted 5 years ago

  • 1
Photo of W5UN_Dave

W5UN_Dave

  • 315 Posts
  • 30 Reply Likes
Ken, progress marches on, whether we do or not
Photo of Brad - N6SPM

Brad - N6SPM

  • 45 Posts
  • 1 Reply Like
I don't want to discourage the question or conversation, but I have some interesting history in this area. I worked in very high-end DSP hardware and software engineering. Design reuse and extensibility for technology obsolescence mitigation was a big topic of every program. It turned out after many years of incorporating upgrade paths etc., that in many cases it turned out to be less expensive to simply start from scratch rather than doing the hand-stands necessary to design, maintain, then incorporate the technology migration paths. The point or lesson learned was, just because you can do something, it doesn't always yield the most cost-effective solution over time (despite intuition and customers that just know better!). It was an eye opener and made us glad we kept a good accounting of our costs as we went along over time, allowing for such data-based analysis. As a final note, you would be amazed at how many times an ASIC or FPGA vendor claimed pin-compatible upgrades, only for us to later find out all kinds of devilish problems they had not accounted for...

Hoping this is not applicable here and only for the readers amusement.
N6SPM_Brad
Photo of Steve - N5AC

Steve - N5AC, VP Engineering / CTO

  • 1046 Posts
  • 1025 Reply Likes
Brad, good comments ... I had just posted my thoughts when I saw yours.
Photo of Ken - NM9P

Ken - NM9P, Elmer

  • 4082 Posts
  • 1275 Reply Likes
Thanks, Brad, and Steve.  I always learn something from these discussions.
Photo of Steve - N5AC

Steve - N5AC, VP Engineering / CTO

  • 1046 Posts
  • 1025 Reply Likes
Official Response
Ken, we looked at making all the processing a plug in module for the FLEX-6000 when it was designed.  There were a few issues such as:  1) there are a large number of high-speed lines that that would have to go across connectors.  This requires the use of connectors that preserve signal integrity at high speed which are generally fairly expensive.  You often need something like double the connections in a connector system than you would on a PCB because of all of the ground routing requirements (on a PCB you just use a ground plane and a power plane).   Then there's all of the extra work to route across the connectors, verify signal integrity, work any issues, etc.  2) With connectors there are always mechanical concerns -- will the connector fail over time, will the contacts corrode, will there be soldering issues in the factory, will the connector dislodge in a drop event, etc. (reliability)  3) You have to comprehend what you might need in the future ... that can be hard.

We weighed all of this and decided it was best not to incorporate plug in modules for all these reasons including cost, time-to-market, etc.  There are thousands of decisions made in a design process and we try to make the decisions that are in both our best interests.

Also, the key reason to do this is to enable the swap to larger/faster computing components as they become available.  For a PC software person, you just drop in the next processor and you are ready with more power.  It's never this simple in the embedded world.  We always have to trade off the effort to incorporate a new part vs. the advantage of the new part.  We put very current parts in the FLEX-6000 radios -- they are all either the latest generation or one generation back currently, over two years after the design.  This is decidedly not the case for many competitive products.

There is extreme flexibility in the hardware and software in this radio.  If someone said -- we need this radio to be a 5MHz IF radio for a microwave station -- so it needs to receive and transmit a 5MHz swath of data, this is completely possible in the FLEX-6000.  It just requires software.  If someone said "we need only one receiver, but it needs diversity and it needs to continually attempt to decode 20 different digital modulation schemes, trying to figure out which scheme it sees," it's just software.  These two are not likely requests from the ham community, but other things may come up.  Things like a new digital mode that requires 20kHz of bandwidth on HF, etc.  We'd like to be able to do these things when they come up.

And yes, we'll build new radios in the future too.  And over time, they'll have more computing power.  But we've engineered SmartSDR to run on different platforms so we are hoping that if there are things we do in the future on some new platform, that those things will also work on the existing platform whenever possible -- but that's a long time away.
Photo of Ken - NM9P

Ken - NM9P, Elmer

  • 4082 Posts
  • 1275 Reply Likes
Thanks, Steve, for a very informative answer, as usual.  It make a lot of sense here, and was what I had suspected might be the case.  But it looks like the 6000 platform should have enough horsepower for several years to come....

Now I just hope I can keep it running for 20 years, as my trusty TS-850 did!  I doubt I will be able to afford a serious upgrade in my retirement years....<grin>