Hei Jukka!
Good summary...I'll add a few comments and I'm sure
Tom will too also.
At 11:06 AM 5/12/03 +0300, jukka.klemola at nokia.com wrote:
>The core of this is all receivers today have a wide
>(> 10 kHz bandwidth) first intermediate frequency filter.
>And that largely ruins the receiver no matter
>how good the second or third intermediate frequency
>stages are.
Absolutely correct. The fundamental problem is
rejection of unwanted distortion products at the first IF
so that they do not appear in subsequent IF passbands.
Just to review how we got here in the first place, the
following features created the problems we have today.
1. General coverage receive capability dictated that
the first IF be at a very high frequency in order to minimize
image problems (e.g. MP family is at 70 MHz).
2. The ability to receive FM mode required a minimum BW
of ~15 kHz, thus this became the standard first IF BW for
the current generation of general coverage radios.
3. Besides #2 above, it is more difficult to economically
manufacture narrowband crystal filters at high frequencies
and probably impossible to do so with DSP. For example,
a 1 kHz crystal filter at 70 MHz is more difficult to build
than the same bandwidth at 9 MHz. For DSP, it is practically
impossible because the ADC needs to run at twice the
highest frequency to be sampled (2 X 70 = 140 MHz) which
is probably done only for military applications today.
Thus a different design is called for which allows
a lower frequency first IF by making the receiver restricted
to ham-bands only. I say "different" because this is NOT
new. This is the way all our receivers worked before the
days of general coverage (Drake R4C for example). At a
lower first IF, it is quite practical to provide multiple
crystal filters for the different modes to be received so you
are not just limited to the widest bandwidth mode (e.g. FM).
>The current second and third intermediate frequency
>filters can have better ability to filter than the
>current DSPs we see in Amateur radio receivers for the
>second or third IF.
>
>The pricing estimate for a good-enough DSP unit has
>maybe fallen to a half of Tom's year 2001 estimate.
>And in two-three years is again a half of the current
>price.
>
>
>This is what we now wait for:
>When comes the time we can have a DSP or two DSPs in
>a box without 500Hz-3kHz analog filters limiting the
>performance of the receiver or two receivers and the
>transmitter in the same box.
>And taking into account the box must be below 4000 USD
>for the consumer equivalent to year 2003 USD pricing.
I believe you may be missing a fundamental point here.
The ADC/DSP/DAC technology used in current DSP radios is
being driven by the digital cellular market (GSM, CDMA, etc).
The extremely high volume of this market is what makes these
components available at very reasonable prices for IF's in the
kHz range. I do not see a similar market at the moment which
will drive DSP volumes at the higher frequencies necessary
for a first IF to allow general coverage receive. I believe this is
what Tom meant by his statement, but maybe he will clarify:
> DSP filtered radios, without a narrow IF filter upstream of the DSP,
> simply won't do that. The last time I looked A/D conversion with
> enough speed and bits would cost a thousand bucks.
I would add that a 24-bit ADC with a sampling rates of >100 MHz
will probably not be available at a reasonable cost anytime soon!
We will more likely see ADC's at 18 MHz (2 X 9 MHz) but again
I am not sure what application will drive the cost down like the
cellular market has done for the kHz ranges.
>We are waiting for the first company to offer us a
>decent upgrade how to filter the first intermediate
>frequency stage better than the curreny factory version
>radios that all use wide, poor shape factor filters.
The simple way around this problem is to find inexpensive
narrowband crystal filters in the 45-70 MHz IF range which would
plug into the first IF of current generation radios. Of course this
is not that simple since this would also defeat the general
coverage features of those radios. Perhaps Inrad could provide
these filters as an interim solution.
>We already have been:
>Waiting to see factory version radios using good
>filters at the first IF.
>Now with Orion we see it means having lower frequency
>for the first IF.
>
>
>My addition to this:
>We must remember why the first IF was set well above
>30MHz and hope the manufacturers take measures to
>overcome those effects already in the factory versions
>of the radios.
It will probably require different designs similar to
what Ten Tec has done to do it right. As Doug Smith stated:
>The winning combination therefore includes a synthesizer
having exceptional phase-noise characteristics, a strong
analog signal path, an ADC with large dynamic range and
a DSP that preserves signal integrity
The only thing really missing in current radios is "a strong
analog signal path" which Orion has overcome with its
hybrid approach. Until we see economical DSP's in the
45-70 MHz range, Ten Tec's approach is likely to be the only
practical solution. BTW, if anyone wants a simple comparison
of the two types of radios being discussed, look at the block
diagrams for the Main and Sub receivers of the Orion here:
http://www.doug-smith.net/orion.htm
Orion's Sub receiver is quite similar to current generation radios
like the IC-PRO2. It has general coverage and uses a 15 kHz
BW roofing filter at the 45 MHz first IF with DSP at the 14 kHz
third IF. The 24-bit ADC/DAC and 32-bit DSP are also similar
to those in other DSP radios.
>And like VR2BrettGraham pointed out, the radios must
>be such we can operate with a linear (= some RF) in
>the schack.
Absolutely true! And as you said so well before, let
the competition begin and may the best radio win. After all,
we should all be strong believers in the benefits of competition!
73, Bill W4ZV