[Elecraft] Is the K3 jumping onto "the SDR Bandwagon"? No....

[wayne burdick]

Dave,

I fully understand your concerns here. This is an issue we had planned 
to address anyway.

The K3 has "software-defined" features in the sense that the DSP is 
very easily upgradeable (one-click PC application), and powerful, and 
can handle future operating modes or signal processing tasks. It also 
has a huge amount of flash memory for filter parameter storage. Thus we 
may be able to do very interesting things in the future such as 
auto-tuning of SSB signals, syllabic squelch, DSP noise blanking with 
signal restoration, etc.

We hope to incorporate ideas from K3 owners over time, given this 
flexibility. But all features that are available at time of shipment 
will be very stable, and we won't be depending on the programming 
community to "fix" anything. We'll retain close control over the source 
code for both the DSP and the main microcontroller to ensure high 
quality.

Regarding the microcontroller:

The K3's MCU is the largest and fastest in Microchip's line, and I can 
assure you it is not "way too slow". It runs at 4.5 times the speed of 
the K2's MCU. If a radio with this MCU running at this speed were to 
appear "slow" to the operator, it would be the programmer's fault, not 
the MCU's. In the case of the K3, we can simultaneously handle 38.4 kB 
RS232 I/O; internal CW keying at 50 WPM; rapid update of the LCD; full 
decode and display of PSK31/CW/RTTY text; and a continuous, rich 
interaction between the MCU and DSP. There is no horsepower limitation 
here!

The MCU is supplemented by a very large flash memory on the front 
panel, providing a huge amount of help text and table storage, and 
leaving plenty of ROM for all foreseen upgrades.

I realize that our competition is using PC cores in their radios. These 
have several major disadvantages compared to our Microchip MCU:

   - embedded PCs have much higher current drain (our entire radio can 
draw as little as 0.8 amps,
     vs. 2 to 3 amps for other high-performance rigs)

   - embedded PCs may have very slow boot-up times (the K3 turns on in 
less than 1 second; I believe
     at least one of our competitor's rigs takes 11 seconds because it 
has to
     boot a Microsoft operating system!)

   - embedded PCs and their support ICs generate much more RFI (our 
entire MCU subsystems draws
     something like 20-30 mA, dramatically reducing both conducted and 
radiated emissions, and thus
     simplifying shielding requirements)

   - embedded PCs have much greater cost and complexity (our choice of 
MCU is one reason we can
     offer an extremely competitive radio starting at well under $2000)

As you may know, writing firmware for a microcontroller is vastly 
different from writing code for a PC. I prefer the former, and I've 
been doing it since I was 18. Lyle's been writing DSP code since there 
were such things  ;)

Regarding the K2:

We've learned a lot from developing and supporting the K2. Everything 
we've learned is reflected in the K3, where we've had the luxury of a 
larger enclosure, I.F. DSP, larger MCU, more front panel space, more 
I/O, higher pricetag, etc. And the K3 was designed from the ground up 
to use surface-mount parts everywhere possible to minimize 
manufacturing costs.

The K2 is an ongoing product that we'll be updating as time permits. As 
you may know, we're not a large company, so our engineering resources 
will be stretched at times.

Please send me directly any suggestions for K2 improvements. I maintain 
this list, prioritizing task items based on the needs of current and 
future K2 owners.

Thanks for sharing your views on these topics!

73,
Wayne
N6KR



Dave-Boat Guy wrote:

> Although the K3 seems to take into
> account a lot of the comments we've made over the
> years, it looks like it is jumping on the SDR
> bandwagon.  If any of you have followed the Software
> Defined Radio Bandwagon, as it applies to "amateur
> radio" developments, it simply means that the radio
> manufacturers are more willing to put out a
> "half-baked" radio than to commit to thoroughly
> testing a radio.  Flex Radio and Ten Tec have tried to
> take the approach telling us that we won't need to
> replace hardware for a long time since the software
> can be updated to take into account new features.  All
> I've seen is that the manufacturers who pursue this
> route take advantage of reprogrammability to fix their
> own problems at a later date.  By the time the
> problems are solved, technology has improved and a new
> piece of hardware becomes available and support for
> the older equipment dwindles.  Both of those companies
> have obsoleted their initial releases in 3 or 4 years.
>
> Perhaps the modular approach taken by Elecraft in the
> K3 is the way to go.  I for one am curious if they are
> still using MicroChip PIC's, which are way too slow,
> or if they have chosen a real processor....



---

http://www.elecraft.com