From: radio_rookie on Thurs, Sep 7 2006 1:17 pm

I want to know the importance of intermediate frequency in any
receivers. IF was used in Superhet transceivers. My question is why
doesn't anyone use zero IF now a days. What is the problem of brining
the RF signal directly to baseband? Does the IF stage conditions the
incoming signal? What are the advantages of the IF stage?

Just confused. Can anyone throw some light on this?

This can be a HUGE subject, but, since this is "homebrew"
we can 'distill' it to a few things: :-)

1. Ever-present random NOISE in the front end. Can't
escape it. Since the amount of noise voltage
reaching the demodulator can be reduced by the
square-root of relative bandwidth, IF bandpass
filtering can cut down that random noise, yield
a constant selectivity regardless of RF input.

2. Direct conversion to baseband is subject to dynamic
range limitations v. the amount of RF input power
and RF input selectivity. i.e., a very strong signal
well out of the desired RF input range might mess
with the sampler causing intermodulation distortion.

3. Lowest RF input level (which determines the
"sensitivity" specification) requires a very low-
noise sampler to equate to a full superhet with an
IF chain. Samplers are not noise-free. Samplers
must compete on the tenths of microvolts (or less)
noise with conventional active mixers of now to
meet high-sensitivity specifications of today.

4 Software (as in an SDR architecture) is NOT simple
to implement, even in a very fast processor. While
it is easy to change demodulation modes, one needs to
understand the math behind the demodulation process.
If you have the TIME and the smarts, go for it; if
not, it may be months before your project works and
then it may not work very well.

5. Not all RF input signals are AM or derivatives of
that (on-off keying, SSB on HF). For FM or
combination AM-PM as in the "modem" fashion, it
might be much easier to implement via a separate IF
plus separate demodulator per mode.

6. In the beginning (1918 and Ed Armstrong in Paris
right after WW1), vacuum tubes were NOT what one
could call the best, noise-free, or even with much
gain. The superhet form allowed the same selectivity
(via the IF bandpass) at any desired RF input
frequency; that did not exist before the superhet.
Since that was a quantum-level improvement at the
time, it had a mystique about it that caused nearly
all designers to follow the IF chain idea with its
diode or tube "detector" (really a rectifier-mixer).
The math of modulation had been published in 1915
(John R. Carson of AT&T) but had yet to spread. It
was not intuitive to the non-mathematical and so
few designers got "into" possible new ways to mix
and demodulate. With better tubes that came after,
the IF and 2nd IF and even 3rd IF as discretes was
easier to design and make. That lasted until
roughly 1980 or about 6 decades, all superhets
having IF chains in a familiar arrangement. It was
"comfortable." More importantly, it worked.

7. If you want selectable bandpass filtering at all
frequencies, the IF with its input bandpass
filters at most any bandwidth you want is the
easiest to design-in and build. That way you
lop off the signals on either side as close to
the antenna as you can get.

8. Heterodyning (mixing) down to one frequency, the
IF, makes it easier to work and debug with a
semi-direct-conversion system. Especially so if
the desired RF inputs have many bands.

9. On the other hand, if portability, light weight,
and low power drain is a requirement (as in military
field receivers), plus all sorts of demodulation
modes, the SDR or Software Designed Radio is the
thing to do, using samplers, A-D conversion and
demodulation in a processor subsystem. Note: You
combine the front end of a conventional IF with
the processor sub-system replacing the IF back end
and 'detectors' to get the best of both.

There isn't any one simple answer. It is all a trade-off
between what is desired and what you can design and make
and how much you have to build plus your budget. Its all
wonderfully complex to decide and I love it. :-)

radio_rookie wrote:
Hello,
I want to know the importance of intermediate frequency in any
receivers. IF was used in Superhet transceivers. My question is why
doesn't anyone use zero IF now a days. What is the problem of brining
the RF signal directly to baseband?

Thanks.

Andy writes:
This is the eighth response to your question. All of the others
are accurate. HOWEVER, there is one technical problem with
a homodyne that outweighs the previous resonses...

The L.O, has to be EXACTLY the same frequency as the incoming
signal to provide proper demodulation......

With a superhet, depending on the modulation, you can be
many Khz in error ( for AM) or a few dozen off ( for voice SSB).
With homodyne, you may have to process a psuedo-baseband signal which
has frequency components you don't want..... that is a bitch....

In order to achieve the LO to be EXACTLY the same
frequency (phase doesn't matter since you can use I and Q),
it is necessary to achieve frequency lock. That requires a
much higher S/N than a simple superhet with a detector...
So you lose sensitivity. In some systems , you can lock
to a remote carrier, but you aren't really talking about those
methods, I don't. think....

So, the major technical problem is not SIMPLICITY or
BANDWIDTH or NUMBER OF STAGES...,.. it is
how to obtain an LO of the correct frequency.... All the
other problems are simple compared to this.... for most systems...

Andy W4OAH

PS I welcome dissent and would like very much to learn
if anything I have said is in error......

AndyS wrote:

Andy writes:
This is the eighth response to your question. All of the others
are accurate. HOWEVER, there is one technical problem with
a homodyne that outweighs the previous resonses...

The L.O, has to be EXACTLY the same frequency as the incoming
signal to provide proper demodulation......

With a superhet, depending on the modulation, you can be
many Khz in error ( for AM) or a few dozen off ( for voice SSB).
With homodyne, you may have to process a psuedo-baseband signal which
has frequency components you don't want..... that is a bitch....

In order to achieve the LO to be EXACTLY the same
frequency (phase doesn't matter since you can use I and Q),
it is necessary to achieve frequency lock. That requires a
much higher S/N than a simple superhet with a detector...
So you lose sensitivity. In some systems , you can lock
to a remote carrier, but you aren't really talking about those
methods, I don't. think....

So, the major technical problem is not SIMPLICITY or
BANDWIDTH or NUMBER OF STAGES...,.. it is
how to obtain an LO of the correct frequency.... All the
other problems are simple compared to this.... for most systems...

Andy W4OAH

PS I welcome dissent and would like very much to learn
if anything I have said is in error......

I believe that what you said is true, but only for AM. What other kind
of modulation requires better frequency accuracy from a direct
conversion receiver LO than from a superhet LO? I believe that only a
small fraction of today's amateurs are interested in AM reception, but
of course it's the bread and butter of the SWL and BCL.

I've build direct conversion receivers for many years. Their simplicity
is particularly evident when you compare a transceiver having a direct
conversion receiver with one having a superhet receiver -- to be honest,
be sure to include all the extra filtering necessary with the superhet.
The single biggest disadvantage to DC receivers, in my opinion, is the
difficulty of making a good AGC, particularly in conjunction with narrow
audio bandwidth. And they do have their own set of potential problems,
such as unintended AM demodulation, the effects of LO leakage and
radiation, and the difficulties in making a clean, stable, high gain
audio amplifier. But all can be overcome once one understands the causes
of the problems. All but perhaps the last one will be present in a
digital version, too, so a casually designed and/or built one is likely
to be a poor performer.

Roy Lewallen, W7EL

Roy Lewallen wrote:

I believe that what you said is true, but only for AM. What other kind
of modulation requires better frequency accuracy from a direct
conversion receiver LO than from a superhet LO? I believe that only a
small fraction of today's amateurs are interested in AM reception, but
of course it's the bread and butter of the SWL and BCL.

Roy Lewallen, W7EL

Andy responds

I don't disagree with anything you've said. However , i've
found it a lot easier to make the final frequency adjustments
in a superhet since it can be done at a lower frequency.
Trying to sync up at 915 is more challenging than at 455 khz,
obviously. And for systems like FSK and SSB, you gotta
do it somewhere..... And with FM, since the beat formed
by the carrier depends on the modulation index of the
received signal, getting rid of it can be squirrelly.....

Like youself, I've built both types , both as home projects,
and as commercial products, for a long long time. In the
DC versions, if I use I/Q and combine them to form the
audio, the AGC is the same as with superhet SSB. And I generally
use PIN diodes before the front end for the first AGC stage...
But, that's just my own preference.....

Andy W4OAH

I don't disagree with anything you've said. However , i've
found it a lot easier to make the final frequency adjustments
in a superhet since it can be done at a lower frequency.
Trying to sync up at 915 is more challenging than at 455 khz,
obviously. And for systems like FSK and SSB, you gotta
do it somewhere..... And with FM, since the beat formed
by the carrier depends on the modulation index of the
received signal, getting rid of it can be squirrelly.....

You mentioned FM. I have a general question about the Tayloe mixer. Is it
possible to receive NBFM or FSK from VHF with it? I'm a little lost in the
question how to use the for outgoing phases or I/Q to demodulate FM. All I
found on the Net was doing shortwave SSB demodulation.
Are there other analog switches going higher than the mentioned 70MHz for
FST3253 or 74HC4066? Perhaps video switches? What is better: Tayloe 4-phases
or a switched mixer with one output only? In the latter case I would need a
conventional IF filter after the switching mixer?

If I need a DSP at the baseband doing math with the phases or I/Q I would
think of an www.wavefrontsemi.com DSP AL3101/2CG DSP-1K will suffice? Wolud
it suffice? It is a little simple DSP mainly for doing FIR - but FAST and
with 24-bits including audio AD converters ip to 50KHz. The DSP runs with
50MHz up to 1000 instructions long until it repeats the prog. The nice think
is a very low pin-count package and cheap too.

There is maybe a middle way with a device like the www.cypress.com PSoC
family of mixed-mode Microcontroller with programmable analog cells. There
even exists a PSoC app note describing a heterodyne FSK receiver for 130KHz.

Maybe a www.microchip.com dsPIC is better?

Would it better having the IF not at zero but at a usually higher IF, say
25KHz (remember the NFBM!)?

regards -
Henry

Henry Kiefer wrote:

You mentioned FM. I have a general question about the Tayloe mixer. Is it
possible to receive NBFM or FSK from VHF with it? I'm a little lost in the
question how to use the for outgoing phases or I/Q to demodulate FM. All I
found on the Net was doing shortwave SSB demodulation.

Provided you implement the Tayloe mixer with sufficient baseband
bandwidth (probably about 20 KHz for NBFM), you can mix, say, a 2 meter
FM signal to baseband and demodulate it with an audio discriminator
(some sort of audio frequency to voltage conversion scheme). Mix the
signal to be either exclusively in the upper or lower sideband of the
output. So, for example, a NBFM sig at 146.000 MHz mix with the Tayloe
mixer set to 145.990 MHz and select the upper sideband. The upper or
lower sidebands are, as you probably know, selected by phase shifting
and summing circuits following the Tayloe mixer (implemented in
software in SDRs). Note that SDRs like the Flex-Radio SDR 1000 do not
process incoming signals near 0 Hz anyway, but mix the desired signal
centered around about 11 KHz (I think).


Are there other analog switches going higher than the mentioned 70MHz for
FST3253 or 74HC4066? Perhaps video switches? What is better: Tayloe 4-phases
or a switched mixer with one output only? In the latter case I would need a
conventional IF filter after the switching mixer?


The Tayloe mixer is a passive mixer terminated in large capacitors, and
similar performance can be obtained at VHF, UHF and microwaves with two
FET ring mixers driven with quadrature signals and also terminated with
capacitors (that is, no wideband transformer on the mixer outputs, but
capacitors followed by HiZ input differential audio amps). I've done
this with the Peregrine Semiconductor FET mixers.

If I need a DSP at the baseband doing math with the phases or I/Q I would
think of an www.wavefrontsemi.com DSP AL3101/2CG DSP-1K will suffice? Wolud
it suffice? It is a little simple DSP mainly for doing FIR - but FAST and
with 24-bits including audio AD converters ip to 50KHz. The DSP runs with
50MHz up to 1000 instructions long until it repeats the prog. The nice think
is a very low pin-count package and cheap too.

There is maybe a middle way with a device like the www.cypress.com PSoC
family of mixed-mode Microcontroller with programmable analog cells. There
even exists a PSoC app note describing a heterodyne FSK receiver for 130KHz.

Maybe a www.microchip.com dsPIC is better?

Only familiar with the dsPIC, though the others sound okay. The dsPIC
would work, though its 12 bit A/D doesn't give a lot of dynamic range.
Plenty for NBFM though, especially since you can run the signal(s)
through a limiter first. For NBFM sixteen bit DSP is sufficient.

Regards,
Glenn

radio_rookie wrote:
Hello,
I want to know the importance of intermediate frequency in any
receivers. IF was used in Superhet transceivers. My question is why
doesn't anyone use zero IF now a days. What is the problem of brining
the RF signal directly to baseband? Does the IF stage conditions the
incoming signal? What are the advantages of the IF stage?

Just confused. Can anyone throw some light on this?

Thanks.

Hey radio rookie,

Thanks for posting this excellent question.
The answers you received have taught me more than a few textbooks have.
Some very good minds have weighed in on your query.

Keep up the good work and welcome to ham radio.

To the respondents, I say "Thanks!" for approaching ideal NG activity!

Makes one proud to be a ham.

John
AB8O

"john" wrote in message
...
radio_rookie wrote:

Some very good minds have weighed in on your query.

I hate to sound "college professor-y", but you really need to be familiar
with the literature.

One of the challenges with anything on the Internet is that anyone can say
anything. That is one of the great advantages, too! The problem is,
especially in a technical forum like this, you will get responses from folks
who don't have a clue but want to sound knowledgable. What is perhaps
surprising, is that sometimes, especially on this NG, some of the truly
giant minds in the field will also weigh in. Unless you are familiar with
who is who, you will have a tough time assessing the responses.

But if you review the books and magazine articles on the subject, and browse
the archives of various other forums, you will see that some of the leading
experts have posted here. Recognize that they might not sound too different
than someone who is simply guessing, and be critical in your analysis of the
responses you have received.

...

xpyttl wrote:
"john" wrote in message
...

radio_rookie wrote:


Some very good minds have weighed in on your query.


I hate to sound "college professor-y", but you really need to be familiar
with the literature.

One of the challenges with anything on the Internet is that anyone can say
anything. That is one of the great advantages, too! The problem is,
especially in a technical forum like this, you will get responses from folks
who don't have a clue but want to sound knowledgable. What is perhaps
surprising, is that sometimes, especially on this NG, some of the truly
giant minds in the field will also weigh in. Unless you are familiar with
who is who, you will have a tough time assessing the responses.

But if you review the books and magazine articles on the subject, and browse
the archives of various other forums, you will see that some of the leading
experts have posted here. Recognize that they might not sound too different
than someone who is simply guessing, and be critical in your analysis of the
responses you have received.

..


Your advice is taken. But, look thru the thread: there was reasonable
consensus and I recognized some "big hitters".

I didn't mean to suggest that one learns best on the NG. Rather, my
intent was to support and give kudos to those who bring the NG up to a
reasonably high standard: no flames, insults, diversions...you know, the
usual stuff.

Here are some of my "favorite" NG remarks
"Get help"
"You have serious issues"
"What's YOUR mensa number?"
"I spent xx years at xx and xx years building xx and have xx degrees, so
I know what I'm talking about"

As one monitors these NG's, one begins to "smell" the diff between
the knowledgeable and the blow-hards. The only problem comes when they
are one and the same. Luckily, this is uncommon.

I appreciate your advice, "Professor"

John
AB8O

"jawod" wrote in message ...

Your advice is taken. But, look thru the thread: there was reasonable
consensus and I recognized some "big hitters".

Yes, this particular thread seemed to gather surprisingly few space cadets

...