On Mon, Oct 19, 2015 at 2:07 PM, Grant Saviers <grants2@pacbell.net> wrote:
> Not having any particular axe to grind, I'll pile on a bit with some
> comments.
>
> 1. The superhet/SDR vs direct sampling radio manufacturer and technology
> competition is and will continue to be very good for the ham community.
>
Agreed.
>
> 2. The early days of CD audio yielded, "I can't stand the sound" for
> which there were sound engineering reasons. Today with 24bit 196KHz
> sampling and playback, I think it is impossible to discern that analog
> signals via LP vinyl are technically superior. However, that won't sway a
> number of folks who "like the sound of analog". Sounding "different" will
> always appeal to some segment Direct Sampling vs superhet/IF DSP vs pure
> superhet. btw I suspect that the inherent amplitude/phase distortions at
> the audio level of different amateur radio filtering techniques affects
> intelligibility more than is commonly recognized.
>
Simply my opinion here, but in my experience booth analog and digital can
yield results that will please even the most discerning audiophile.
Generally, the things that go wrong are design errors that result in
distortion. The designer of a radio often has so many things to
check/verify/test that things can go without a thorough analysis. It is
these mistakes that make it into a final product that cause issues.
Engineers of both the analog and digital variety are equally susceptible to
this malady ;-) BUT, it's a lot easier to fix software after something
has shipped than it is to replace components on the board.
>
> 3. Unfortunately, the ham radio market isn't currently large enough for a
> major Si firm to design for our needs. The requirements are driven by the
> cell base station market and perhaps a bit of military needs. So sample
> rates are going up beyond the current 300 MSPS rather than more bits of
> resolution which would solve the overload concerns. However, as the
> technology of fast + wide A to D's disperses, there may be a boutique firm
> that takes up the challenge as has happened in the audio market where the
> best A/D's are not made by Analog Devices, TI, or Linear Tech.
>
You are correct that cellular is driving the industry. We are ALL
benefitting from this as many of the parts we use are cheap because they
are purchased in the millions by the cellphone industry, making the pricing
competitive.
Oversampling (higher sampling rates) does add bits and dynamic range, but
in and of itself does not protect from overload (you can use attenuation to
shift the dynamic range and reduce overload if you are oversampled enough,
though). Overload is a function of raw input voltage, just like it is with
any other transceiver. The key spec in an ADC for overload is the max
input voltage, not the bits (although some tricks can be played with
impedance transformation).
Every design employs measures to prevent overload. The roofing filter is
designed to protect the superhet, but if you get a strong signal inside the
roofing filter, the superhet will overload. But RMDR is the bigger issue
for virtually ALL modern transceivers regardless of architecture. If you
have a receiver with a BDR of 140dB, but an RMDR of 95dB, you have a 95dB
receiver, not 140dB. The operator doesn't really care about the method of
blocking -- if you can't hear the signals you're trying to hear, it doesn't
matter so much why. Even if you can prevent overload with a high BDR, if
your RMDR is lower, the phase noise from that big carrier will wipe out
weak signals reducing your dynamic range.
> 4. It seems that the overload issue is now confined by consensus to duplex
> operations on the same band since high pass and bandpass filters and stubs
> take care of most BCB/MW/multi multi overload situations. I agree that
> overload otherwise is a minor concern.
>
It is important to check to see if your radio has a BCB/etc filter to
reduce overload possibilities from offending signals. Some do, some don't.
>
> 5. As always there will be a range of good and not as good implementations
> of technologies. Also, the terminology of "SDR" is a mess since it is
> applied to several generations - 1. outboard audio DSP, 2. integrated audio
> DSP, 3. integrated IF DSP, and 4. direct RF sampling. Many posts seem to
> me to confuse generations 3 and 4 and perhaps different capability radios.
>
I couldn't agree with this more. Even inside of a technology,
implementation issues can cause drastic performance differences. For
example, in a direct sampling radio, the ADC output bits are multiplied
inside an FPGA by an NCO and the resulting number is a larger number of
bits than either two multiplicands. If selection of the number of bits
and the scaling are not done correctly, the receiver will perform poorly,
appear to suddenly and dramatically overload, etc. We'd use floating
point, but today it uses too many resources so most designs use fixed point.
>
> 6. Moore's Law continues and more MIPS and FPGA gates will become cheaper
> and better. It seems to me that the direct sampling technology offers a
> number of opportunities for better signal processing than IF SDR's. Maybe
> not, time will tell. FLEX has a teaser with "wide band noise reduction" in
> their latest 6000 release. What is that?
>
The short version is that we look at very wide bandwidths (>20MHz) and
perform the blanking at the higher sampling rate. It is easier to discern
noise from signals due to the differences in rise time and the signal
characteristics and then do a better job surgically removing the noise
while leaving desired signals intact.
>
> 7. One thing unlikely in affordable superhets in the opportunity to decode
> many signals on many bands simultaneously. The contest rule writers
> have/are going to have increasing difficulty with what that capability
> means to contesting. e.g. a 5 band skimmer for SSB? (since my Samsung S5
> has nearly perfect speech recognition in a noisy car, is that too far out?).
>
> 8. The content (syntax and semantics) of contest and DX messages are
> mostly proscribed, so our brains ability to discern (or make a good guess)
> of content in very poor signal to noise and QRM/QRN situations is pretty
> impressive and varies hugely among operators. (rant: why do DXpedition
> callers insist on telling me their state or they are QRP or are running 50w
> to a dipole in the basement?)
>
> 9. This thread has been quite interesting and informative.
>
Agreed.
>
> 73,
> Grant KZ1W
> ex TX5D, TX7G, E51MKW
>
73,
Steve, N5AC
Stephen Hicks, N5AC
VP Engineering
FlexRadio Systems™
4616 W Howard Ln Ste 1-150
Austin, TX 78728
Phone: 512-535-4713 x205
Email: steve@flexradio.com
Web: www.flexradio.com
Click Here for PGP Public Key
<https://sites.google.com/a/flex-radio.com/pgp-public-keys/n5ac>
*Tune In Excitement™*
PowerSDR™ is a trademark of FlexRadio Systems
_________________
Topband Reflector Archives - http://www.contesting.com/_topband
|