New Flex radio?

[Terry Fox]

For the last few days, Gerald Youngblood of Flex Radio Systems has been teasing the Flex reflectors about a big announcement at Dayton, and on their website come Friday.  There has been a lot of guessing.  On Tuesday, he sent “Imagine digital starting at the antenna connector...” as the message subject, and “...and a bit of arithmetic. Imagine with us May 18th.” as the message body.

If one follows this, it may mean that Flex is FINALLY introducing a DDC/DUC SDR rig.  I hope they are, as the QSD/QSE technology is getting pretty old as far as being leading edge.  QSD/QSE rigs remind me of old SSB phasing rigs, like my old SB-10 sideband adaptor and some others I have known (and cussed at).  Once crystal filtering became the norm, it was hard to justify a phasing rig anymore, and technically the phasing rigs were worse on the air, as you had to constantly adjust the phasing to reduce the unwanted opposite-sideband and/or carrier.

We have been moving toward the 21st century version of this.  QSD/QSE versus DDC/DUC.  A rig based on a DDC/DUC doesn’t have any of those nasty differential gain or phasing problems that plague analog QSD/QSE rigs.  Yes, it has other issues, such as cost, and dynamic range.  But, have you seen the price of a Flex 5000?  It’s pretty costly for being QSD/QSE based.  If you are paying $2,500 or more for a rig, the $130 cost of a GOOD 16-bit, 130MHz sampling A/D converter kind of gets lost in the noise.  And, that’s at single quantities.

My hope is that Flex has finally produced a good DDC/DUC rig, that uses ethernet connectivity instead of firewire (another peeve of mine), and at an affordable price.  With the N2ADR-design boards selling for under $800, the Hermes coming out at about $900, the Russian ZS1SDR, and other DDC/DUC boards, Flex could be left behind if they don’t update their product line with a DDC/DUC-based rig.  And, that would be too bad, as Flex tends to have a complete rig solution, where the others mentioned are just the central module, with filters, amplifiers, T/R switching and other supporting parts needed.

If they do come out with a DDC/DUC-based SDR rig, and if it doesn’t use Firewire, and if it’s somewhat affordable, count me in!  My Flex 1500 will be part of the WB4JFI sell-a-thon to raise funds to acquire one.  If it has an open hardware/software interface, that would be even better.  Most of my computers use Windows, but I like the ability to connect that new, expensive hardware to my Linux and Mac boxes as well.

However, if this announcement is just another iteration of PowerSDR, I will be VERY disappointed.

Note:  QSD typically means Quadrature Sampling Decoder (or sometimes Detector), QSE means Quadrature Sampling Encoder, DDC means Digital DownConverter, and DUC means Digital UpConverter.  The QSD is also known as a Tayloe Mixer/decoder.  Since QSD/QSE hardware creates I and Q signals in the analog domain, any hardware imbalance in the QSD itself, the QSD output buffer, and (typically) sound card input circuitry up to the A/D converter, will cause an imbalance of the I or Q signal that will throw off the digital signal processing after the A/D converter.  For this reason, QSD/QSE rigs must have the ability to add/subtract offset values to counteract any gain or phase impurities added by the analog hardware.  A QSE operates exactly like a QSD wired backwards.  After the soundcard D/A converter output, there are analog circuitry for the sound cards output buffers, the audio/IF input buffers to the QSE, and the QSE itself.  So, any differences in gain or phase in any of this analog hardware will cause an imbalance in the QSE, resulting in poorer opposite sideband suppression, and also carrier leakage.

Since the DDC-based receiver typically has only one analog path (for HF rigs) before the A/D converter, there is much less possibility of any imbalance.  Once the signal is digitized, the digital downconverter is implemented in the digital domain.  As long as it is well designed, there is very little chance of an imbalance between the I and Q signals.  Yes, there are opportunities for problems inside the DDC, but using longer data words for the samples, and careful inter-stage design to reduce math problems limits the potential for problems.  The biggest problem with DDC-based rigs on HF has been the dynamic range of the A/D converter, followed by the noise figure of the A/D converter.  Even a good 16-bit A/D converter has only a theoretical maximum of 96dB, and in the real-world it has less.  The DDC following the A/D converter can make up for this with decimation and filtering, often referred to as “processing gain”.  Once again, the DUC is essentially a DDC operating in reverse.  All stages up to the D/A converter are implemented in the digital world.  Following the D/A converter, we are back to the real analog world for amplification and filtering, with all the attendant issue there.

So, QSD/QSE requires fiddling with the sample values in the digital world to offset analog problems/imbalances.  DDC/DUC do not require that fiddling, but requires careful design of the key modules, usually implemented inside FPGAs.

At least I hope that’s right!!
Terry, WB4JFI


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://amrad.org/pipermail/tacos/attachments/20120516/ee35e8d3/attachment.html>