I guess that I must be in the minority - it seems to me that for best AM
fidelity (not selectivity, nor sensitivity), you would use a crystal set
with tuned RF stages, no IF, no heterodyne of any kind. use the tubes for
RF amps if needed, and for audio amplification, and use a tube diode for the
detector.

In article ,
Patrick Turner wrote:

Syl's Old Radioz wrote:

"Patrick Turner" a écrit dans le message

I don't expect anyone to pay 3c for what I say, which could be
seen as OT.

You just met our village idiot it seems...

There is an unspoken rule here..._Ignore_ his posts. Let him talk
to himself.

We don't get into fight with village idiot like you do on
RAT...Keeps rar+p "clean"...;o)

Syl

Well, with all due respects to all gentlemen and possible idiots on
all groups to whom this subject thread is cross posted to, I reserve
the right to decide who I will ignore or not.

I will desperately try not step on anyone's toes as I act in well
intentioned freewill.

I won't budge from the idea that its possible to digitise the signal
from the antenna and simply apply suitable algorithms, and get
digital decoding, without all the phase shift caused by consecutive
tuned circuits.


Chill dude. There is nothing wrong with this idea and the current
technology can do it. The problem is money. It would be expensive to do
this and I would not expect people to pay the price when it would be a
small improvement over the current generation of radios. Heck, I would
not expect people to pay the price for a large improvement.

Digital techniques do not end all distortion and add there own type of
noise by the way.

--
Telamon
Ventura, California

william_b_noble wrote:

I guess that I must be in the minority - it seems to me that for
best AM fidelity (not selectivity, nor sensitivity), you would use a
crystal set with tuned RF stages, no IF, no heterodyne of any kind.
Use the tubes for RF amps if needed, and for audio amplification,
and use a tube diode for the detector.

Actually, this setup intrigues me for local reception, since it
appears to be a quite simple circuit. Are there any schematics of such
a circuit -- any commercially made radio of yesteryear using this
design approach?

Jon

John Byrns wrote:

1. It has been variously stated that the audio bandwidth of AM
broadcasting is either 3.5 kHz, 5 kHz, or 10 kHz. In the US AM broadcast
channels are 20 kHz wide, so audio is effectively limited to a maximum of
10 kHz by law/regulation. It is my impression that most AM stations
transmit audio out to this legal maximum. Of course as HD-radio takes
hold this will change with the analog signal cutting off somewhere around
5 kHz. I know there are at least 2 active broadcast engineers that read
this group, perhaps they could fill us in on what the stations they are
involved with are actually doing as far as audio bandwidth goes?


We've kicked this around the block before - but I guess it won't hurt to
kick it one more time. (I'm only going to address US standards here).

The AM Bandwidth is 10Khz.
However - what has to be taken into consideration are "real world"
filters - and the "stop band" specifications of the NRSC-1. I.E. the
signal must be down 15db (from 100% modulation) AT 10khz(!!!)

Further - it must be:
-30db at 10.5Khz;
-40db at 11Khz; and
-50db at 15Khz.
-50db is .32% modulation (that's point 32 percent - not 32 percent).

Now - depending on how good your processor / final filters are - figure
your "real world" bandwidth from there. Consider "good" filters at 12db
per octave - and "really, really good" filters at 24db per octave (an
octave is 1/2 (going down) or double (going up) a given frequency.

So if you put a 12db per octave filter in front of your transmitter -
the highest unattenuated frequency through that filter will be around
4.5Khz.

You can do much better with a 24db per octave - somewhere around 7.5Khz.
Of course - NRSC-1 is no longer the "newest kid on the block" - the
new one is ITU-R (Recommendation 328-5). To meet that spec. - the
processor filters are set to 6.0Khz.

Amigos and Optimods are set up to meet NRSC-1 power spectrum
requirements "out of the box". The Optimod 9200 (the current top of the
line digital AM processor) is adjustable from 4.5kHz to 9.0kHz in 0.5kHz
steps, plus NRSC - is guaranteed to meet ITU-R (Recommendation 328-5)
and NRSC-1 power spectrum specifications without the need for further
low-pass filtering prior to the transmitter. And as noted -- is
typically set for 6.0kHz for ITU-R.

Amigos or Optimods are probably in 90%+ AM stations in the US (We're
(WMER) is running an Amigo).

Here is the NRSC site for those wishing to get thoroughly tech:

http://www.nrscstandards.org/

And NRSC-1 itself (PDF document - needs acrobat reader)

http://www.nrscstandards.org/nrsc-1.pdf

2. The idea expressed above that a "modern sophicated decompressor
circuit could match the curve of the compressor" seems far fetched to me.

Yeah, I agree: I can't imagine trying to "undo" what either the Optimod
or the Amigo do to the audio; talk about multi-band; mutli-limit;
multi-everything... sheesh.

Here's Orban's Optimod 9200:

http://www.orban.com/orban/products/..._overview.html

You'll find a pop-up menu in the upper right of the page - you can view
the features and specs. from there. Impressive stuff.


best regards...
--
randy guttery

A Tender Tale - a page dedicated to those Ships and Crews
so vital to the United States Silent Service:
http://tendertale.com

In article , Patrick Turner
wrote:

John Byrns wrote:

7. It has been suggested that using a 2 MHz IF frequency would allow
wider bandwidth than the standard 455 kHz IF frequency. I fail to see why
this should be true.

Because for the same Q value, the pass band would be 4 times wider

Where is it written that the same loaded Q must be used for both filters?
If you can change the center frequency, why can't you change the loaded Q?

Within reason, for bandwidths typical of audio
receivers, you should be able to build a filter at 455 kHz that has
effectively the same response as a 2 MHz filter. There is no need to
throw out the 455 kHz IF just to get wide bandwidth.

Its difficult to make a 455kHz typical old IFT produce a nice flat topped
20 kHz wide BW. Its either pointy nosed, undecoupled, or flat topped, critical
coupled,
or over critical or rabbit eared.
I have tried all that.

So you have tried all that and rejected the "pointy nosed", "flat topped",
and "rabbit eared" response curves. I am left to wonder what sort of
response curve you were looking for? Why not settle for a nice "flat
topped" response curve and be done with it?


Regards,

John Byrns


Surf my web pages at, http://users.rcn.com/jbyrns/

In article , wrote:

John Byrns wrote:

1. It has been variously stated that the audio bandwidth of AM
broadcasting is either 3.5 kHz, 5 kHz, or 10 kHz. In the US AM broadcast
channels are 20 kHz wide, so audio is effectively limited to a maximum of
10 kHz by law/regulation. It is my impression that most AM stations
transmit audio out to this legal maximum. Of course as HD-radio takes
hold this will change with the analog signal cutting off somewhere around
5 kHz. I know there are at least 2 active broadcast engineers that read
this group, perhaps they could fill us in on what the stations they are
involved with are actually doing as far as audio bandwidth goes?


We've kicked this around the block before - but I guess it won't hurt to
kick it one more time. (I'm only going to address US standards here).

The AM Bandwidth is 10Khz.
However - what has to be taken into consideration are "real world"
filters - and the "stop band" specifications of the NRSC-1. I.E. the
signal must be down 15db (from 100% modulation) AT 10khz(!!!)

Further - it must be:
-30db at 10.5Khz;
-40db at 11Khz; and
-50db at 15Khz.
-50db is .32% modulation (that's point 32 percent - not 32 percent).

Now - depending on how good your processor / final filters are - figure
your "real world" bandwidth from there. Consider "good" filters at 12db
per octave - and "really, really good" filters at 24db per octave (an
octave is 1/2 (going down) or double (going up) a given frequency.

So if you put a 12db per octave filter in front of your transmitter -
the highest unattenuated frequency through that filter will be around
4.5Khz.

You can do much better with a 24db per octave - somewhere around 7.5Khz.
Of course - NRSC-1 is no longer the "newest kid on the block" - the
new one is ITU-R (Recommendation 328-5). To meet that spec. - the
processor filters are set to 6.0Khz.

Amigos and Optimods are set up to meet NRSC-1 power spectrum
requirements "out of the box". The Optimod 9200 (the current top of the
line digital AM processor) is adjustable from 4.5kHz to 9.0kHz in 0.5kHz
steps, plus NRSC - is guaranteed to meet ITU-R (Recommendation 328-5)
and NRSC-1 power spectrum specifications without the need for further
low-pass filtering prior to the transmitter. And as noted -- is
typically set for 6.0kHz for ITU-R.

Amigos or Optimods are probably in 90%+ AM stations in the US (We're
(WMER) is running an Amigo).

OK, now we are getting down to brass tacks as my Grandmother used to say,
whatever that may mean. The $64 question is does the Amigo as used at
WMER only do NRSC, or does it have a selection of cutoff frequencies like
the Optimod 9200, and if it does what is the cutoff frequency set to at
WMER?


Regards,

John Byrns


Surf my web pages at, http://users.rcn.com/jbyrns/

Randy,

You get the award for most informative post concerning the "broadcast
standards" in this thread. I was waiting for you to come through. It
takes someone with real broadcast experience to give us the real scoop.
Thanks.

Phil B

"Randy and/or Sherry" wrote in message
...


John Byrns wrote:

1. It has been variously stated that the audio bandwidth of AM
broadcasting is either 3.5 kHz, 5 kHz, or 10 kHz. In the US AM
broadcast
channels are 20 kHz wide, so audio is effectively limited to a
maximum of
10 kHz by law/regulation. It is my impression that most AM stations
transmit audio out to this legal maximum. Of course as HD-radio
takes
hold this will change with the analog signal cutting off somewhere
around
5 kHz. I know there are at least 2 active broadcast engineers that
read
this group, perhaps they could fill us in on what the stations they
are
involved with are actually doing as far as audio bandwidth goes?


We've kicked this around the block before - but I guess it won't hurt
to
kick it one more time. (I'm only going to address US standards here).

The AM Bandwidth is 10Khz.
However - what has to be taken into consideration are "real world"
filters - and the "stop band" specifications of the NRSC-1. I.E. the
signal must be down 15db (from 100% modulation) AT 10khz(!!!)

Further - it must be:
-30db at 10.5Khz;
-40db at 11Khz; and
-50db at 15Khz.
-50db is .32% modulation (that's point 32 percent - not 32 percent).

Now - depending on how good your processor / final filters are -
figure
your "real world" bandwidth from there. Consider "good" filters at
12db
per octave - and "really, really good" filters at 24db per octave (an
octave is 1/2 (going down) or double (going up) a given frequency.

So if you put a 12db per octave filter in front of your transmitter -
the highest unattenuated frequency through that filter will be around
4.5Khz.

You can do much better with a 24db per octave - somewhere around
7.5Khz.
Of course - NRSC-1 is no longer the "newest kid on the block" - the
new one is ITU-R (Recommendation 328-5). To meet that spec. - the
processor filters are set to 6.0Khz.

Amigos and Optimods are set up to meet NRSC-1 power spectrum
requirements "out of the box". The Optimod 9200 (the current top of
the
line digital AM processor) is adjustable from 4.5kHz to 9.0kHz in
0.5kHz
steps, plus NRSC - is guaranteed to meet ITU-R (Recommendation 328-5)
and NRSC-1 power spectrum specifications without the need for further
low-pass filtering prior to the transmitter. And as noted -- is
typically set for 6.0kHz for ITU-R.

Amigos or Optimods are probably in 90%+ AM stations in the US (We're
(WMER) is running an Amigo).

Here is the NRSC site for those wishing to get thoroughly tech:

http://www.nrscstandards.org/

And NRSC-1 itself (PDF document - needs acrobat reader)

http://www.nrscstandards.org/nrsc-1.pdf

2. The idea expressed above that a "modern sophicated decompressor
circuit could match the curve of the compressor" seems far fetched
to me.

Yeah, I agree: I can't imagine trying to "undo" what either the
Optimod
or the Amigo do to the audio; talk about multi-band; mutli-limit;
multi-everything... sheesh.

Here's Orban's Optimod 9200:

http://www.orban.com/orban/products/..._overview.html

You'll find a pop-up menu in the upper right of the page - you can
view
the features and specs. from there. Impressive stuff.


best regards...
--
randy guttery

A Tender Tale - a page dedicated to those Ships and Crews
so vital to the United States Silent Service:
http://tendertale.com

John Byrns wrote:
In article , Patrick Turner
wrote:


John Byrns wrote:


7. It has been suggested that using a 2 MHz IF frequency would allow
wider bandwidth than the standard 455 kHz IF frequency. I fail to see why
this should be true.

Because for the same Q value, the pass band would be 4 times wider


Where is it written that the same loaded Q must be used for both filters?
If you can change the center frequency, why can't you change the loaded Q?

Admittedly it might be a bit more difficult to achieve the same Q at 2
MHz as 455 but we are talking about homebrewing and experimenting here.



Its difficult to make a 455kHz typical old IFT produce a nice flat topped
20 kHz wide BW. Its either pointy nosed, undecoupled, or flat topped, critical
coupled,
or over critical or rabbit eared.
I have tried all that.


So you have tried all that and rejected the "pointy nosed", "flat topped",
and "rabbit eared" response curves. I am left to wonder what sort of
response curve you were looking for? Why not settle for a nice "flat
topped" response curve and be done with it?

Here's a wacky idea that I'll toss out just to see if it flies...

Could one use two garden variety 455kc xfmrs in series, one tuned at
center plus and the other tuned center minus? Impedance matching would
be an issue but maybe such a scheme offers a not so glamourous method of
achieving the wider bandwidth and maintaining the flatness with little ado.

Re wider bandwidth as a whole. On AM sets that I have owned with
excessively wide bandwidth they all tend to sound like crap. I don't
have a wealth of local stations that might be enhanced by the wider
width but on weaker stations the amount of noise and all those AM
"artifacts" seems to go way up making it very unpleasant to listen to.

As a result I think it would make the most sense to use a switchable or
continuously variable bandwidth scheme so as to not be left with an all
or none scenario after so much effort.

John B, you may remember one of my Tandberg receivers that had 4
positions of bandwidth ganged with a switch that somewhat tailored the
audio accordingly. To the ear (or rather to my ear) this seemed very
effective.

-Bill M

Now - depending on how good your processor / final filters are - figure
your "real world" bandwidth from there. Consider "good" filters at 12db
per octave - and "really, really good" filters at 24db per octave (an
octave is 1/2 (going down) or double (going up) a given frequency.


Randy, those figures are not characteristic of modern processors that
use DSP filtering, which is capable of extremely rapid rolloff.

Take a look at http://n2.net/k6sti/speech.jpg . This is a screen shot
of my HP 141T/8553B/8552B spectrum analyzer tuned to a local AM radio
station broadcasting speech. The analysis-filter bandwidth was 300 Hz,
the vertical scale 10 dB/div, and the horizontal scale 5 kHz/div. I
set the storage-screen persistence to maximum and accumulated spectra
for 10-15 seconds. It is easy to see the extremely sharp rolloff at 10
kHz. http://n2.net/k6sti/music.jpg shows a different AM station
broadcasting classical music. The music spectrum is evident, but so is
the brick-wall filtering at 10 kHz. These spectra are typical of what
I observe for AM stations here in Southern California.

If you have a receiver capable of SSB reception, you can easily check
the spectral limits of any AM station. Put the receiver in LSB mode
and tune down frequency from the carrier (or use USB and tune up).
Regardless of program content, it will be obvious where the response
ends. You'll hear the modulation sidebands suddenly vanish. Whenever
I've tried this, my dial has always read more than 9 kHz away from the
carrier.

Brian

Patrick Turner wrote:
Volker Tonn wrote:
Jon Noring schrieb:

In the last couple of years I've posted various inquiries to this and
related newsgroups regarding high-performance, tube-based AM (MW/BCB)
tuners, both "classic" and modern.

Have a look into the "Collins" S-series. These are state-of-the-art
tube sets 'til now. At least it's not the tubes alone but the fabulous
mechanical IF-filters giving outstanding results for a tube set.
Manuals with layout diagrams should be available on the web....

Since Mr Noring says he has regularly trawled the Net for everyone
else's expertise on AM reception, but got nowhere, because he's
still doin it, why doesn't he gird his loins and put his shoulder to
the task of learning all about AM and radio engineering as spelled
out so clearly in all the old text books, and then damn well build
his own perfect AM radio???

Thanks for sharing your frank coments. They are acknowledged.

The important thing is that the replies to my "trawling" have been
very informative, including yours Patrick, and are not only
benefitting me, but are benefitting many others who are following this
thread in real time.

Whether my trawling is successful or not for my purposes is immaterial
-- if I fail, I fail -- I don't fear failure as some do -- the
discussion is further adding to the information pool for the community
of those interested in some aspect of tube-based AM tuners, and in
that regards I think it has been successful. (With Google archiving
the newsgroups, this information is now being preserved, and is
searchable.)

There are obviously two sides to the engineering of radio receivers:
1) the basic theory and the basic categories of design approaches
(which I am studying -- it helps in that back in 1974 I had the
equivalent of one years' worth of basic electrical engineering courses
at the University of Minnesota, which is now all coming back to me),
and 2) the real-world engineering of receivers/tuners, using real
rather than theoretical components, and the attendant compromises and
work-arounds which inevitably result.

I do agree with Patrick's implying that there is no such thing as a
"perfect radio". I am not seeking the "perfect radio", but a
modern-design tuner "kit" sufficiently meeting the various
requirements I have previously set forth. I believe once a good
design results, that PCB boards can be made, coils can be built by
someone or some company experienced in doing that (I mention coils
since that is the one component difficult to buy right off the shelf
-- thank god no one has to build their own tubes!), and the schematic
with detailed instructions and guidelines sold through diytube (as an
example.)

The target market for the "kit" are those who build their own tube-
based components for their audio system, and want every component to
be a high-performer, approaching audiophile-grade in performance (yes,
AM broadcasts are not "audiophile", but audiophiles want a tuner that
brings out the best in what is there in the signal.) They don't want
to spend their limited time building junk, they don't want to build a
Radio Shack beginners' crystal set. They want very good performance
(which is admittedly a "fuzzy" word), commensurate with their other
components. They just want the tuner kit not to be overly complicated
in design, to work if they follow the instructions and guidelines, and
to meet their (collective) expectations.

And these kit builders are not novices, either, at wielding a
soldering gun, and in chassis and cabinet design -- they are
mechanically- and electronically-inclined, and are now building
audiophile-grade amps and preamps from the many kits now out there. I
also believe that some of the vintage radio collectors, who are
experienced at restoring radios, will also take an interest in the AM
tube tuner kit. (For those who don't know, I'm now restoring a Philco
37-670 console, so I'm not exactly out-of-touch with the radio
collecting world.)

Based on my experience with building audiophile-grade tube amps and
plugging into that community, I think I've laid out pretty well what
they want and expect. Most are not going to become radio design
enthusiasts, they will not live and breathe tuners, building hundreds
of circuits on cake pans in their basement (and I am not disparaging
those who do!) They simply are going to listen to the tuner they
laboriously built from the kit, happy with its performance, and happy
for what they have learned about how radios work "under the hood", in
a general sense. Some will no doubt get the radio bug, and join the
people here, rescuing old radios from the landfill, and restoring
them.

Maybe my focus on TRF-based designs and "channel-based design" have
been diversions. But, from what I've read about real-world TRF designs
(John Byrns messages have been great here), a TRF-based design has
some nice attributes from the audiophile kit perspective, and there
are clever real-world solutions around the selectivity and gain
limitations of the "what's taught in textbooks" regarding basic TRF
design, as John Byrns and others have noted many many times, but which
seems to fall on deaf ears of those who believe that the best
high-performance receiver (however "high-performance" is defined)
*must* be super-het in basic design.

But obviously, the vast majority of commercial designs of
high-performance tube-based radio receivers from the mid 1930's to the
1950's are super-het designs, and many of them are great performers,
so I'll post a parallel message with another call for candidate radios
to inspire the AM tuner kit. After all, if one is to put together an
AM tube tuner kit, it makes a lot of sense to base it on a proven
design from the past -- why reinvent the wheel?

For example, diytube (at http://www.diytube.com/ ) has taken the
venerable Dynaco ST-35 amp design, modernized it some (and to further
improve its audiophile performance), and is now selling the PCB
board with schematics and instructions to diy audiophiles. it is an
excellent performer (I know firsthand -- it is a *very nice* sounding
amp.) diytube is now working on a high-power monoblock tube amp kit
based on the Eico EL34 amp of old -- can't wait until it is released.

Just some thoughts...

Jon Noring

"John Byrns" wrote in message
...
In article ,

2. The idea expressed above that a "modern sophicated decompressor
circuit could match the curve of the compressor" seems far fetched to me.
In the days of yore when audio processing consisted of a single broad band
compressor, and a broad band "peak limiter" one might have contemplated
this, at least as far as the compression part went, but today's audio
processing is much more complex. Processing today involves broad band
AGC, multiband compressors, plus multiband and broadband clippers in place
of the old "peak limiter". It isn't clear to me that this would be easy
to undo, or even possible. I don't know if the multiband aspect creates
problems for reversing the process or not, but how do you undo clipping,
and if there are any feed forward compressors involved it is possible that
the output isn't even a single valued function of the input, making
recovery mathematically impossible.


Sure, the modern decompressor might be farfetched. There's no doubt that
some things can't be recovered. Maybe the best that can be done is to more
or less de-process the AM radio signal. I think audio processing is a
bigger problem than bandwidth, distortion or noise for a real high fidelity
AM tuner. And I think it's a problem which deserves it's fair share of
consideration.

Frank Dresser

Have you considered a test bed for your hi-fi AM radio experiments? Bill's
comments about IF transformers got me thinking that one approach would be to
get an old 70s era stereo receiver with IF transformers and no
crystal/ceramic filters in the IF strip. These things are pretty
inexpensive at the Salvation Army type stores now. You could try stagger
tuning the IF transformers for a wide bandwidth. Ideally, you'd use a sweep
generator, but you could probably do a passable job with a standard signal
generator and some patience. It's not a tube setup, of course, but you'd
get wide bandwidth AM with a reasonably low noise wideband amplifier.

Frank Dresser

John Byrns wrote:

OK, now we are getting down to brass tacks as my Grandmother used to say,
whatever that may mean. The $64 question is does the Amigo as used at
WMER only do NRSC, or does it have a selection of cutoff frequencies like
the Optimod 9200, and if it does what is the cutoff frequency set to at
WMER?

Yes, only NRSC - but you can "tailor" for various "ends" (i.e. talk;
"punch", etc.). - it's the "cheap seats" version - (I.e. CLR) vs. the
optimod (Orban - though now they are the same company).

best regards...
--
randy guttery

A Tender Tale - a page dedicated to those Ships and Crews
so vital to the United States Silent Service:
http://tendertale.com

Phil B wrote:
Randy,

You get the award for most informative post concerning the "broadcast
standards" in this thread. I was waiting for you to come through. It
takes someone with real broadcast experience to give us the real scoop.
Thanks.

(as goofy would say - garsh!)... Hey - when one must keep the station
owner's butt (well - at least his wallet) out of the jaws of pain (in
spite of the FCC's reputation - they can and DO still bite on occasion)
one sorta has to keep up - whether one wants to or not. And in today's
internet published world - finding technical details is (usually) just a
search engine away.

Besides I'm none too proud to ask some of my friends and business
acquaintances (like Olen Booth of Broadcast Engineers) who are eye-ball
deep in this stuff everyday of the world. Guys like him DO know it like
the back of their hand.

best regards...
--
randy guttery

A Tender Tale - a page dedicated to those Ships and Crews
so vital to the United States Silent Service:
http://tendertale.com

Brian wrote:

Randy, those figures are not characteristic of modern processors that
use DSP filtering, which is capable of extremely rapid rolloff.

Take a look at http://n2.net/k6sti/speech.jpg . This is a screen shot
of my HP 141T/8553B/8552B spectrum analyzer tuned to a local AM radio
station broadcasting speech. The analysis-filter bandwidth was 300 Hz,
the vertical scale 10 dB/div, and the horizontal scale 5 kHz/div.

Yep and it shows that at around 7.5 it's down 10db - and not symmetrical
at all- I.e. their pushing the positive modulation WAY harder than
negative - which is probably ok for voice - but would sound pretty lame
for music. Their "tilt" would make them "punchy" all right - (make most
listeners punch-drunk - esp. women who (on average) dislike tinny
sounding stuff - and have the ears to hear it.


http://n2.net/k6sti/music.jpg shows a different AM station
broadcasting classical music. The music spectrum is evident, but so is
the brick-wall filtering at 10 kHz. These spectra are typical of what
I observe for AM stations here in Southern California.

So is the -25db roll-off by 7.5khz - this indeed looks like good
conformance to NRSC-1.


If you have a receiver capable of SSB reception,

Yeah - a few - but I also have an Empire NF-105 calibrated RF / Noise
meter that goes from 14Khz to 1000Mhz. It tells me what I need to know.


best regards...
--
randy guttery

A Tender Tale - a page dedicated to those Ships and Crews
so vital to the United States Silent Service:
http://tendertale.com

wrote:
Jon Noring wrote:

Well, being the "OP", I want a high-audio performance, modern
design AM tuner to integrate into my audio system -- and I believe
a lot of tube-o-philes likewise want that -- but not everyone
obviously. There are several reasons why most higher-grade audio
systems use separate components, the reasons of which are obvious
to most everyone. The AM tuner is no different than other audio
components in this regard.

Maybe what you want is the old JW Miller passive AM tuner. No
active devices at all, just a bunch of tuned circuits and a detector
diode.

Yes, possibly. John Byrns has written about the J.W. Miller TRF AM
tuner quite a few times, along with other related ones (and the early
30's Western Electric 10A which seems to have inspired the others.)

See, for example:

http://groups.google.com/groups?selm...&output=gplain

(Is there a schematic of this tuner online? John had it at his site at
one time, but apparently has taken it down.)

Hopefully John will have time to comment on his approach to designing
a workable, improved modernized version of this class of tuners,
suitable for diytube-like kit-building, with working PCB boards. I
know the TRF design is capable of very clean audio, and this is one of
the things which intrigues me about the TRF designs in general, since
the tube-o-philes want very clean sound from their components. Can a
super-het design accomplish the same sound quality? Probably, but I'm
still intrigued with the TRF....

Jon Noring

In article ,
says...
wrote:
Jon Noring wrote:

Well, being the "OP", I want a high-audio performance, modern
design AM tuner to integrate into my audio system -- and I believe
a lot of tube-o-philes likewise want that -- but not everyone
obviously. There are several reasons why most higher-grade audio
systems use separate components, the reasons of which are obvious
to most everyone. The AM tuner is no different than other audio
components in this regard.

Maybe what you want is the old JW Miller passive AM tuner. No
active devices at all, just a bunch of tuned circuits and a detector
diode.

Yes, possibly. John Byrns has written about the J.W. Miller TRF AM
tuner quite a few times
Back in the 70's I built a passive diode detector receiver for the
broadcast station I worked for. The transmitter was only a mile away.
It was flat out to 10 KHz, and we used it for a line-level signal that
was usually used for the control room monitor speakers. We could A-B the
line out going to the transmitter and the received signal. It was good
for understanding how the audio limiting at the transmitter was
affecting different kinds of program material.

--
Regards, Terry King ...In The Woods In Vermont

"The one who dies with the most parts LOSES! What do you need??"

In article , wrote:

John Byrns wrote:

3. TRF receivers have been mentioned, and everyone seems to assume
that a TRF receiver would consist of cascaded single tuned
resonators with RF amplifier stages between. There is no reason why
double tuned circuits, similar to those used in the IF transformers
of a superhetrodyne can't be used in a TRF receiver, with all the
selectivity/bandwidth benefits that brings to the party. For
examples see the Western Electric No. 10A receiver, the J.W. Miller
TRF receiver, the early Altec AM receiver, as well as others.

I did a cursory check on the Internet, but did not yet find any
schematics for the mentioned receivers. Are they online somewhere?
Anyone?

I also found the following article from John posted back in 2000,
where he talks about the double tuned TRFs, such as WE-10A, J.W.
Miller, Collins (which I assume is the same one Volker Tonn
mentioned today), Meissner, and the Weeden (the last of which John
noted to be the best designed of all of them):


http://groups.google.com/groups?selm...&output=gplain

Unfortunately the URLs to the TRF schematics at John's site are not
working.

Try this URL: http://users.rcn.com/jbyrns/BP-TRFs.html

Another bandpass receiver which uses J.W. Miller coils is the Altec
101(B), the schematic for the 101(B) is available on the "Nostalgia Air"
web site.

After reviewing this thread, and considering that the proposed tuner is to
be an audiophile tuner, and given the great affinity of audiophiles for
Western Electric audio equipment it is clear to me that the tuner of
choice would be an updated version of the Western Electric No. 10A
receiver. The tubes used by Western Electric receiver are obsolete and
should be updated to more modern tubes in the revised design. The
detector in the Western Electric design is also inadequate having high
distortion, and due to the square law characteristic it modulates the AGC
voltage at the syllabic rate of the audio signal. Modifications to update
the tubes, resolve the detector and AGC problems were available by the
early 1950's, or earlier.

There is one remaining problem with the W.E. 10A that hasn't yet been
addressed to the best of my knowledge, and that is the aperiodic antenna
input circuit, that is likely to be a source of RF 3rd order IM problems
which will wipe out some weaker signals if you live near a number of
50,000 Watt flame throwers, or even 5,000 Watt stations. This problem
could be eliminated by deleting the first RF amplifier stage, which is
probably serious overkill with modern tubes anyhow, and making the first
bandpass filter a tuned antenna circuit. This is the reason that I prefer
the Weeden and Altec designs, they both have tuned antenna circuits to
minimize RF IM problems.

If one wanted to duplicate an existing circuit the Altec is probably the
best choice as it has a tuned antenna circuit and uses modern 7 pin
miniature tubes. One addition that would be necessary for any of these
tuners today would be the addition of a "NRSC" deemphassis network to
compensate for the preemphassis that is used in AM broadcasting today.


Regards,

John Byrns


Surf my web pages at, http://users.rcn.com/jbyrns/

Jon Noring wrote:

[New Yahoo Group started: "AM Tube Tuners". See end of this message
for more info.]

In the last couple of years I've posted various inquiries to this and
related newsgroups regarding high-performance, tube-based AM (MW/BCB)
tuners, both "classic" and modern.

I'm very interested in building such a tuner to match with
audiophile-grade tube amplifiers and pre-amplifiers now being built by
hobbyists (as well as those sold by commercial vendors.) There are
quite a few nice kits now being marketed for audiophile quality tube
amps/pre-amps, such as those made by diytube (http://www.diytube.com/
-- there are many others like diytube.) So why not similar kits (or
workable designs) for a tube-based AM tuner?

My Hi-Fidelity AM Tuner is a Eico HFT94. I have two of them.
Built both around 1960. Not many AM transmitters left broadcasting
a worthwhile signal in North America. But they do well on what is
left in this area, away from interference.

If anyone is living in an apartment, don't bother trying for good AM.
Even the manufacturers of SS receivers do a poor job with their AM
section,
recognizing the many problems external & out of control of the tuner.

Cheers, John Stewart

Yep and it shows that at around 7.5 it's down 10db - and not symmetrical


Take a look at the screen shot again. Don't the extremely steep
spectral walls--the sudden drop of tens of dB at 10 kHz--suggest
something to you? The inner spectral detail is interesting, especially
since the lopsided spectrum did not occur later in the day on
different program material (see http://n2.net/k6sti/later.jpg), but
it's the spectral boundaries that tell the story.


So is the -25db roll-off by 7.5khz - this indeed looks like good
conformance to NRSC-1.


That spectrum was recorded during classical music with little
high-frequency content, hence the sloping spectrum. What's interesting
is where it suddenly vanishes at 10 kHz.

For preemphasis shape, take a look at http://n2.net/k6sti/noise.jpg .
I recorded this later on the same station during a quiet piano
passage. The spectrum beyond 3 kHz is background noise. It shows the
expected preemphasis shape right up to 10 kHz where the spectrum
suddenly vanishes.

I did discover two local signals whose spectrum did not extend to 10
kHz. Both rolled off very rapidly at 8 kHz. When I realized that both
transmitters were in Mexico, I thought I was on to something. I found
a third strong Mexican signal, but its spectrum extended right out to
10 kHz like the US stations.

Brian

John Byrns wrote:

In article , Patrick Turner
wrote:

John Byrns wrote:

7. It has been suggested that using a 2 MHz IF frequency would allow
wider bandwidth than the standard 455 kHz IF frequency. I fail to see why
this should be true.

Because for the same Q value, the pass band would be 4 times wider

Where is it written that the same loaded Q must be used for both filters?
If you can change the center frequency, why can't you change the loaded Q?

The lower the Q, the more IFTs required for a given amount of pass band and
attenuationout of band.



Within reason, for bandwidths typical of audio
receivers, you should be able to build a filter at 455 kHz that has
effectively the same response as a 2 MHz filter. There is no need to
throw out the 455 kHz IF just to get wide bandwidth.

Its difficult to make a 455kHz typical old IFT produce a nice flat topped
20 kHz wide BW. Its either pointy nosed, undecoupled, or flat topped, critical
coupled,
or over critical or rabbit eared.
I have tried all that.

So you have tried all that and rejected the "pointy nosed", "flat topped",
and "rabbit eared" response curves. I am left to wonder what sort of
response curve you were looking for? Why not settle for a nice "flat
topped" response curve and be done with it?

I didn't say I had rejected the flat topped critical coupled IF response.
That's what I use when my radio has the IFT coil distance adjust control set to
minimum BW, but when the coils are slid towards each other after tuning, the BW
becomes wider
without going rabbit eared, because the sum of the responses of the two IFTs is
still
flat topped. I found the mecanical slide method to be more predictable than any
other, and the IFTs don't
drift off Fo, and the rabbit ear shape in IFT1 is symetrical each side of Fo.

Commercially made communication receivers I have seen use the same method to alter
the coupling and hence BW of their three IFTs. There are no long wires going to a
complex switch.

Patrick Turner.



Regards,

John Byrns

Surf my web pages at, http://users.rcn.com/jbyrns/

Jon Noring wrote:

Patrick Turner wrote:
Volker Tonn wrote:
Jon Noring schrieb:

In the last couple of years I've posted various inquiries to this and
related newsgroups regarding high-performance, tube-based AM (MW/BCB)
tuners, both "classic" and modern.

Have a look into the "Collins" S-series. These are state-of-the-art
tube sets 'til now. At least it's not the tubes alone but the fabulous
mechanical IF-filters giving outstanding results for a tube set.
Manuals with layout diagrams should be available on the web....

Since Mr Noring says he has regularly trawled the Net for everyone
else's expertise on AM reception, but got nowhere, because he's
still doin it, why doesn't he gird his loins and put his shoulder to
the task of learning all about AM and radio engineering as spelled
out so clearly in all the old text books, and then damn well build
his own perfect AM radio???

Thanks for sharing your frank coments. They are acknowledged.

The important thing is that the replies to my "trawling" have been
very informative, including yours Patrick, and are not only
benefitting me, but are benefitting many others who are following this
thread in real time.

Whether my trawling is successful or not for my purposes is immaterial
-- if I fail, I fail -- I don't fear failure as some do -- the
discussion is further adding to the information pool for the community
of those interested in some aspect of tube-based AM tuners, and in
that regards I think it has been successful. (With Google archiving
the newsgroups, this information is now being preserved, and is
searchable.)

I don't mind sharing whay I know, but you will be the one left to
decide what works or doesn't work, so you shoukld get away from the PC,
and ito the workshop to try out ideas mentioned in all the responses to your
query.
I for one haven't time for the R&D, but if I was more passionate about
good AM than I already am, it'd be to the workshop I would go,
armed with ideas, and solder on towards lower thd and more BW.



There are obviously two sides to the engineering of radio receivers:
1) the basic theory and the basic categories of design approaches
(which I am studying -- it helps in that back in 1974 I had the
equivalent of one years' worth of basic electrical engineering courses
at the University of Minnesota, which is now all coming back to me),
and 2) the real-world engineering of receivers/tuners, using real
rather than theoretical components, and the attendant compromises and
work-arounds which inevitably result.

I do agree with Patrick's implying that there is no such thing as a
"perfect radio". I am not seeking the "perfect radio", but a
modern-design tuner "kit" sufficiently meeting the various
requirements I have previously set forth. I believe once a good
design results, that PCB boards can be made, coils can be built by
someone or some company experienced in doing that (I mention coils
since that is the one component difficult to buy right off the shelf
-- thank god no one has to build their own tubes!), and the schematic
with detailed instructions and guidelines sold through diytube (as an
example.)

There already have been several excellent SS kit designs for decent AM
released by Oz makers
in past years, but its 35 years since any tube based kits were available.

Nobody seems to think it'd be commercially viable to present yet another
AM kit, because 90% of folks listen to FM.

So you are on your own wanting to make a kit design that could be sold,
and I wish you well with the prototyping of coils and circuits.




The target market for the "kit" are those who build their own tube-
based components for their audio system, and want every component to
be a high-performer, approaching audiophile-grade in performance (yes,
AM broadcasts are not "audiophile", but audiophiles want a tuner that
brings out the best in what is there in the signal.) They don't want
to spend their limited time building junk, they don't want to build a
Radio Shack beginners' crystal set. They want very good performance
(which is admittedly a "fuzzy" word), commensurate with their other
components. They just want the tuner kit not to be overly complicated
in design, to work if they follow the instructions and guidelines, and
to meet their (collective) expectations.

And these kit builders are not novices, either, at wielding a
soldering gun, and in chassis and cabinet design -- they are
mechanically- and electronically-inclined, and are now building
audiophile-grade amps and preamps from the many kits now out there. I
also believe that some of the vintage radio collectors, who are
experienced at restoring radios, will also take an interest in the AM
tube tuner kit. (For those who don't know, I'm now restoring a Philco
37-670 console, so I'm not exactly out-of-touch with the radio
collecting world.)

Based on my experience with building audiophile-grade tube amps and
plugging into that community, I think I've laid out pretty well what
they want and expect. Most are not going to become radio design
enthusiasts, they will not live and breathe tuners, building hundreds
of circuits on cake pans in their basement (and I am not disparaging
those who do!) They simply are going to listen to the tuner they
laboriously built from the kit, happy with its performance, and happy
for what they have learned about how radios work "under the hood", in
a general sense. Some will no doubt get the radio bug, and join the
people here, rescuing old radios from the landfill, and restoring
them.

Maybe my focus on TRF-based designs and "channel-based design" have
been diversions. But, from what I've read about real-world TRF designs
(John Byrns messages have been great here), a TRF-based design has
some nice attributes from the audiophile kit perspective, and there
are clever real-world solutions around the selectivity and gain
limitations of the "what's taught in textbooks" regarding basic TRF
design, as John Byrns and others have noted many many times, but which
seems to fall on deaf ears of those who believe that the best
high-performance receiver (however "high-performance" is defined)
*must* be super-het in basic design.

But obviously, the vast majority of commercial designs of
high-performance tube-based radio receivers from the mid 1930's to the
1950's are super-het designs, and many of them are great performers,
so I'll post a parallel message with another call for candidate radios
to inspire the AM tuner kit. After all, if one is to put together an
AM tube tuner kit, it makes a lot of sense to base it on a proven
design from the past -- why reinvent the wheel?

For example, diytube (at http://www.diytube.com/ ) has taken the
venerable Dynaco ST-35 amp design, modernized it some (and to further
improve its audiophile performance), and is now selling the PCB
board with schematics and instructions to diy audiophiles. it is an
excellent performer (I know firsthand -- it is a *very nice* sounding
amp.) diytube is now working on a high-power monoblock tube amp kit
based on the Eico EL34 amp of old -- can't wait until it is released.

Just some thoughts...

Jon Noring

We await with ardent expectations of the fruit of your your efforts in your
workshop
with a saleable prototype tube based AM radio tuner kit.


Patrick Turner.

Since the superheterodyne patents either started being licensed at
reasonable rates or ran out, few receivers of any other type have been
built. With very good reason.

PCB construction makes more sense for IF strips than for pure
baseband hardware when tubes are employed, but I don't know that
there's a big advantage to doing it with tubes unless you just like to
work with tubes. Doing it with FETs might make more sense. Still, very
good RF boxes were built before the PCB days.

I think you should get some coil components, which are still
available, and either a noise generator and a spectrum analyzer (or
one with a track gen...) or better yet a network analyzer , which will
show both transmission and reflected paths, and just decide what kind
of "haystack" you want, and cobble to suit. RF software exists so that
you can play with precise parms of I and C, but you will be happier
with the cut and try given stray inductance and capacitance and other
variables at 10.7 MHz (or whatever IF you wind up with.) People once
did it without these tools but then it took prodigal amounts of time
and they had techs and test operators who worked cheap.

When you get done, you will have the best fidelity of current AM
broadcast signals available. However, considering as how they've been
Orbanned into submission for "dial punch", and considering commercial
broadcast at least in the US sucks **** through a straw right now
content-wise, it may be a wholly Pyrrhic victory.

Brian wrote:

Take a look at the screen shot again. Don't the extremely steep
spectral walls--the sudden drop of tens of dB at 10 kHz--suggest
something to you?

Yeah - typical processing -- looks like what I would expect with very
good (NRSC) processing.

Let me try and explain it this way: Notice that the signal at 7.5khz is
down roughly 6db from the lower frequency (2-3khz) material. You have to
remember, then -- that makes that 7.5khz material only roughly 1/2 the
volume of the lower frequency (2-3khz) material. The difference here may
not look like much (as it's in db) but it's only HALF as loud. And as
you've noted - it drops off (as best this equipment can do) like a brick
wall past 9.5khz.

The inner spectral detail is interesting, especially
since the lopsided spectrum did not occur later in the day on
different program material

As noted - they probably were really punching talk material. Remember -
you can go way over 100% positive modulation - all commercial
transmitters are designed to do so. You just can't go past 100% negative
(how do you get less than 0 carrier? you don't - you get clipping /
splatter). In fact - most processors won't try to push past -95% -
leaving a "just in case" margin for OOPS!). Sometimes such un-symetrical
modulation causes some strange effects in the sidebands.

The better model optimods can be pre-programmed with many presets - and
chosen as desired. It's all "demographics" now - knowing and appealing
to the (perceived) audience - so changing "the sound" during a broadcast
day is now as routine as sponsors tailoring their ads to suit drive time
commuters, then soccer moms, etc.

(see http://n2.net/k6sti/later.jpg), but
it's the spectral boundaries that tell the story.

Well - let's see -- (relative to 2.5-3khz) they're -6db at 7.5Khz;
rolling on off to about -10db by 10khz - then about another 20db by
12khz. What's your point- looks like NRSC through a pretty good
processor - roll-off pretty much what we've been talking about...

I do note the "spike" in the positive sideband at about 8khz or so - I
feel that is probably a noise spike of some sort - especially since
it's so prominent out of the surrounding "curve"; and also it has NO
correspondence in the negative sideband at all - though it's hard to
tell from a snapshot.

That spectrum was recorded during classical music with little
high-frequency content, hence the sloping spectrum.

Yeah- but even so NRSC pre-emphasis is up to 10db by 10khz - which is
still getting rolled as it approaches the "brick wall".


What's interesting is where it suddenly vanishes at 10 kHz.

Look closer - it's not absolutely a "brick wall". There is some gentler
roll-off before 9khz - then it gets serious. Again - that's what current
state of the art looks like -- consistent with the NRSC spec - and what
I've said about filters / processing.

Again - you don't seem to have a good grasp of what you're looking at in
relation to what you hear. There is no way LA stations (or any other
AM band broadcasts in the U.S.) are going to be "flat" out to 10khz (not
these
days - some (clear channels) used to). It's been noted that the better
optimods (like the 9200) can go way out filter settings (9.5khz) -
Optimods so set would be very useful for ShortWave - and places where
the FCC (or equiv. authority) aren't so strict. I wouldn't recommend
them being so set in LA.

Back to your "music" example - What are they (the station's sidebands)
AT 8-9khz compared to the average at 2-3Khz? That roll-off is the effect
of the filters/processors doing their job in forming (as best they can)
the "brick wall" you do indeed see above 9.5khz.

The picture of "noise" tells us nothing - other than 10khz is down 30db
from carrier. Were it relative to something - that would tell us much.
As it is - it just shows an unknown spectrum of ???? Put a reference in
there (or a known weighted/gated noise source such as specified by NRSC
- like a BruelKjar or equiv.) THEN you can make some solid deductions.

If the material in the 2-3Khz ranges runs at / around -15-20db then
10khz is DOWN roughly 10-15db below that - which is ONLY
(approximately) 1/4 as loud. But since we don't know what the source
material is doing - it's meaningless.

I know dealing in dbs is confusing - but if you're going to deal with
Audio/RF/Spectrum - you HAVE to learn dbs and be familiar with how they
relate to what you hear.


best regards...
--
randy guttery

A Tender Tale - a page dedicated to those Ships and Crews
so vital to the United States Silent Service:
http://tendertale.com

Randy and/or Sherry wrote:
I know dealing in dbs is confusing - but if you're going to deal with
Audio/RF/Spectrum - you HAVE to learn dbs and be familiar with how they
relate to what you hear.

Jeffie hands you another bucket of pearls to cast before the swine.

Jeff

--
"They that can give up essential liberty to obtain a little temporary
safety deserve neither liberty nor safety." Benjamin Franklin
"A life lived in fear is a life half lived."
Tara Morice as Fran, from the movie "Strictly Ballroom"

The picture of "noise" tells us nothing - other than 10khz is down 30db
from carrier. Were it relative to something - that would tell us much.
As it is - it just shows an unknown spectrum of ???? Put a reference in
there (or a known weighted/gated noise source such as specified by NRSC
- like a BruelKjar or equiv.) THEN you can make some solid deductions.


That spectrum was taken during a quiet piano passage with background
noise. The piano, played softly, had little treble, so the spectrum
above about 3 kHz is the product of the program noise spectrum, and
the spectral response of the station, which includes playback
electronics, processor, transmitter, and antenna. The dominant
spectral feature of the station's frequency response is the processor
preemphasis. If the noise spectrum is flat, what you see in the screen
shot is the preemphasis curve. Its absolute level reflects the level
of the background noise, which isn't relevant. But the shape is. The
curve shown is typical of the spectral response you'd expect to see
for a preemphasized AM transmitter. The key point is that it stops
suddenly at 10 kHz, not somewhere below. All of the spectra I've shown
do the same. (The spectra of the two Mexican signals stop at 8 kHz.)

Here's a final screen shot: http://n2.net/k6sti/am1210.jpg . This is
nearby station at 1210 kHz that was broadcasting a live announcer from
a local studio at the time I recorded the spectrum. The carrier is at
the left edge of the screen, the center of the screen is 1220 kHz, and
the horizontal scale is 2 kHz/div. This image shows the upper sideband
in some detail.

If you were designing a high-performance AM receiver, what IF passband
would you use to fully recover the modulation from this signal?

Brian

In article , Patrick Turner
wrote:

John Byrns wrote:

In article , Patrick Turner
wrote:

John Byrns wrote:

7. It has been suggested that using a 2 MHz IF frequency would allow
wider bandwidth than the standard 455 kHz IF frequency. I fail to
see why
this should be true.

Because for the same Q value, the pass band would be 4 times wider

Where is it written that the same loaded Q must be used for both filters?
If you can change the center frequency, why can't you change the loaded Q?

The lower the Q, the more IFTs required for a given amount of pass band and
attenuationout of band.

Patrick, you are missing the point, the issue was the merits of a 2.0 MHz
IF frequency vs. a 455 kHz IF frequency with respect to
bandwidth/selectivity, my point was that for the sort of bandwidths we are
talking about for audio, a 455 kHz IF can provide virtually identical
"pass band and attenuation out of band" with exactly the same number of
IFTs as a 2.0 MHz IF frequency. The loaded Qs result from the design
specifications in both cases, and are what they are. There may be
architectural advantages to using one or the other IF frequency in a
radio, but so far only the bandwidth/selectivity has been mentioned and in
that regard an IF of 2.0 MHz offers no significant advantage over a 455
kHz IF for the reception of the full audio bandwidth.

To illustrate this consider the example of the following calculated
response curves for both a 455 kHz IFT and a 2.0 MHz IFT:

One One
455 kHz IFT 2.0 MHz IFT
Q = 15.167 Q = 66.667

Fc-60 kHz -24.30 dB -24.18 dB
Fc-50 kHz -21.22 dB -21.09 dB
Fc-40 kHz -17.56 dB -17.42 dB
Fc-30 kHz -13.22 dB -13.07 dB
Fc-20 kHz -8.72 dB -8.60 dB
Fc-15 kHz -7.09 dB -7.02 dB
Fc-10 kHz -6.27 dB -6.24 dB
Fc-05 kHz -6.04 dB -6.04 dB
Fc kHz -6.02 dB -6.02 dB
Fc+05 kHz -6.03 dB -6.03 dB
Fc+10 kHz -6.19 dB -6.22 dB
Fc+15 kHz -6.86 dB -6.96 dB
Fc+20 kHz -8.34 dB -8.50 dB
Fc+30 kHz -12.75 dB -12.95 dB
Fc+40 kHz -17.15 dB -17.32 dB
Fc+50 kHz -20.88 dB -21.01 dB
Fc+60 kHz -24.01 dB -24.12 dB

The only advantage the 2.0 MHz IFT shows is marginally better symetry of
responce about the ceter frequency, the response of the two IFTs is
virtually identical.

Within reason, for bandwidths typical of audio
receivers, you should be able to build a filter at 455 kHz that has
effectively the same response as a 2 MHz filter. There is no need to
throw out the 455 kHz IF just to get wide bandwidth.

Its difficult to make a 455kHz typical old IFT produce a nice flat topped
20 kHz wide BW. Its either pointy nosed, undecoupled, or flat
topped, critical
coupled,
or over critical or rabbit eared.
I have tried all that.

So you have tried all that and rejected the "pointy nosed", "flat topped",
and "rabbit eared" response curves. I am left to wonder what sort of
response curve you were looking for? Why not settle for a nice "flat
topped" response curve and be done with it?

I didn't say I had rejected the flat topped critical coupled IF response.

Then what did you say? You said you had "tried all that" but now it
appears that you were telling a little fib and hadn't actually tried a 455
kHz IF designed to produce the desired response.


Regards,

John Byrns


Surf my web pages at, http://users.rcn.com/jbyrns/

"Sam Byrams" wrote in message
om...


When you get done, you will have the best fidelity of current AM
broadcast signals available. However, considering as how they've been
Orbanned into submission for "dial punch", and considering commercial
broadcast at least in the US sucks **** through a straw right now
content-wise, it may be a wholly Pyrrhic victory.

I have a wideband AM radio, and the some stations sound better than others.
One of the Polish language stations sounds pretty good in wideband. I'm
sure my Grandmother would have loved hearing good fidelity polkas, but I'd
prefer jazz. A couple of weeks ago they played a Classical recording before
sign-off which sounded great. The sound was well balanced and had lots of
dynamic range.

The gospel station sounds over treble boosted on my radio, but other wise
it's not bad. The oldies station is a bit of a disappointment. They have
great material, but it sounds a bit flat.

There are a couple of stations which don't sound very good.

I made a mistake in my previous post. I got off on a "testbed" tangent. It
would have been better to say "test drive". Wideband AM isn't necessaraly
bad, it can be pretty good. But I wouldn't get into a big project unless I
gave the local stations a good listen in order to judge if it's really
worthwhile.

Frank Dresser

Brian wrote:

The picture of "noise" tells us nothing - other than 10khz is down 30db
from carrier. Were it relative to something - that would tell us much.
As it is - it just shows an unknown spectrum of ???? Put a reference in
there (or a known weighted/gated noise source such as specified by NRSC
- like a BruelKjar or equiv.) THEN you can make some solid deductions.

That spectrum was taken during a quiet piano passage with background
noise. The piano, played softly, had little treble, so the spectrum
above about 3 kHz is the product of the program noise spectrum, and
the spectral response of the station, which includes playback
electronics, processor, transmitter, and antenna. The dominant
spectral feature of the station's frequency response is the processor
preemphasis. If the noise spectrum is flat, what you see in the screen
shot is the preemphasis curve. Its absolute level reflects the level
of the background noise, which isn't relevant. But the shape is. The
curve shown is typical of the spectral response you'd expect to see
for a preemphasized AM transmitter. The key point is that it stops
suddenly at 10 kHz, not somewhere below. All of the spectra I've shown
do the same. (The spectra of the two Mexican signals stop at 8 kHz.)

Here's a final screen shot: http://n2.net/k6sti/am1210.jpg . This is
nearby station at 1210 kHz that was broadcasting a live announcer from
a local studio at the time I recorded the spectrum. The carrier is at
the left edge of the screen, the center of the screen is 1220 kHz, and
the horizontal scale is 2 kHz/div. This image shows the upper sideband
in some detail.

Would not the use of pink noise through a low pass filter
and used as the carrier signal modulation be a better way to see the
frequency contour on an analyser, why noise + piano?



If you were designing a high-performance AM receiver, what IF passband
would you use to fully recover the modulation from this signal?

Brian

There are 5 divisions where there seems to be a signal, so to get
the 10 kHz of AF BW involved so that the 10 kHz response was 1 dB down at
10kHz,
about 30 kHz of IF BW would be required, ie, 15 kHz each side of the IF centre
F

This would be somewhat difficult using normal high Q 455 kHz IFTs.

I think one might have a much better chance if one used 2 MHz IFTs,
perhaps 3 of them, and settled for -3dB at 10kHz each side of 2MHz centre F.
Then an emphasis RC filter could boost the 10 kHz back up a bit.


In 1982 in the Australian magazine Electronics Australia, there was an
elaborate AM tuner kit
offered for sale for aud $250 back then which is about equal to usd $1,400
now.
It had 10 different coils types including 3 well damped 455 kHz IFTs, RF
coils, and 9 Khz whistle filter,
5 j-fets, 6 opamps, one ceramic filter, and 3 signal transistors, a 3 gang
variable tuning cap,
and lots of diodes, and R&C bits, and that doesn't include the +15v PS.

The set had non tuned RF input coil feeding 1st RF LC, then fet RF amp,
2nd RF LC, then two untuned balanced transformers and a two fet PP balanced F
converter
feeding IFT1, a fet IF amp, IF2, a 2nd IF fet amp, then IFT3.
The oscillator has a three winding coil, and 3 bjts.
The AF detector is a CA3016 with shunt FB to linearise the detection.
AVC is via TL071, 741, and UAA180m is used to drive
sigal leds and tuning meter.
Output audio is filtered by two TL072 and
with a passive bridged T filter.

I doubt that any of the coil components would be findable today.

The final AF response was - 3 dB at 10 kHz on the "wide" bw setting.

A tube kit to do the same thing today would cost at least the same shirtload
of money,
probably more.

Imagine trying to build any tubed radio today in small batch numbers, and in
doing so
include for an extra 3 tubes to achieve the low thd and wide BW
of the 1982 EA circuit.
It would make the cost greater than a tube power amp.

My paper files have around 20 different circuits for AM tuners including
a fairly simple synchrodyne ( or direct conversion ) two tube sets which
use a 6EJ7 for an RF amp, and followed by a 6BE6 synchronous detector.

I tried building one, but audio output was low, stability was difficult,
and and a superhet proved far better.

Then there were several if not many synchrodyne and some homodyne designs in
Wireless World
over the years, but these were all chip based, except the early
D.G.Tucker circuit of 1947.

Mr Noring wants some miraculously simple cheap design solution to drop out of
the sky.
He should pray to the God of Triodes, perhaps He will send a schematic
directly.

But then perhaps He won't, but there is much information on
AM reception out there in the old publications which mainly lay slowly rotting

in university basement achives if they havn't all been chucked out years ago.
I spent quite some time reading all I could and my copied paper files consist
of a couple of hundred sheets.

Why the heck would I ever want to re-invent the wheel with AM?

Better to consider the wisdom of the past before deciding on something novel.

The tubed tuners are in the minority in my files.

The commercialisation of the complex synchronous receiver types
was exceedingly limited, because when such receivers were concieved,
nearly everyone was farewelling AM for serious listening,
and going to FM.
But AM was good for the cricket, football, news, talkback and pop trash.
Rarely if ever was there any Bethoven.
And nearly everyone started using cheap japanese SS portables
in 1965. There was 3kHz of BW, if you were lucky.

I think one can get ceramic filters with 20 kHz of BW, -3 dB,
Murata part number CFU455E2 offers -6 dB at +/- 12.5 kHz.
The attenuation 10 kHz away from the -3 dB point is 90 dB.
These are usually low impedance input devices, maybe 1 kohm, so they need to
be driven with an untuned IF transformer with a low impedance secondary, or
cathode follower.
Don't apply DC to any of the pins on ceramic filters.

Patrick Turner.

John Byrns wrote:

In article , Patrick Turner
wrote:

John Byrns wrote:

In article , Patrick Turner
wrote:

John Byrns wrote:

7. It has been suggested that using a 2 MHz IF frequency would allow
wider bandwidth than the standard 455 kHz IF frequency. I fail to
see why
this should be true.

Because for the same Q value, the pass band would be 4 times wider

Where is it written that the same loaded Q must be used for both filters?
If you can change the center frequency, why can't you change the loaded Q?

The lower the Q, the more IFTs required for a given amount of pass band and
attenuationout of band.

Patrick, you are missing the point, the issue was the merits of a 2.0 MHz
IF frequency vs. a 455 kHz IF frequency with respect to
bandwidth/selectivity, my point was that for the sort of bandwidths we are
talking about for audio, a 455 kHz IF can provide virtually identical
"pass band and attenuation out of band" with exactly the same number of
IFTs as a 2.0 MHz IF frequency. The loaded Qs result from the design
specifications in both cases, and are what they are.

Sure. But the same Q would give wider BW at 2 MHz.
I have not ever done this, so I guess at what the final response could be.



There may be
architectural advantages to using one or the other IF frequency in a
radio, but so far only the bandwidth/selectivity has been mentioned and in
that regard an IF of 2.0 MHz offers no significant advantage over a 455
kHz IF for the reception of the full audio bandwidth.

I supect it might, and one article in Wireless World refered to using
10.7 MHz.



To illustrate this consider the example of the following calculated
response curves for both a 455 kHz IFT and a 2.0 MHz IFT:

One One
455 kHz IFT 2.0 MHz IFT
Q = 15.167 Q = 66.667

Fc-60 kHz -24.30 dB -24.18 dB
Fc-50 kHz -21.22 dB -21.09 dB
Fc-40 kHz -17.56 dB -17.42 dB
Fc-30 kHz -13.22 dB -13.07 dB
Fc-20 kHz -8.72 dB -8.60 dB
Fc-15 kHz -7.09 dB -7.02 dB
Fc-10 kHz -6.27 dB -6.24 dB
Fc-05 kHz -6.04 dB -6.04 dB
Fc kHz -6.02 dB -6.02 dB
Fc+05 kHz -6.03 dB -6.03 dB
Fc+10 kHz -6.19 dB -6.22 dB
Fc+15 kHz -6.86 dB -6.96 dB
Fc+20 kHz -8.34 dB -8.50 dB
Fc+30 kHz -12.75 dB -12.95 dB
Fc+40 kHz -17.15 dB -17.32 dB
Fc+50 kHz -20.88 dB -21.01 dB
Fc+60 kHz -24.01 dB -24.12 dB

The only advantage the 2.0 MHz IFT shows is marginally better symetry of
responce about the ceter frequency, the response of the two IFTs is
virtually identical.

The equality in performance depends on a large Q difference, with
544 kHz Q much lower than 2MHz Q to get the same BW.

The Q of a typical 455 kHz IFT is higher than you have indicated, because
the impedance of the LC circuit at Fo is required to be high to suit
pentode loading, and to get high gain.

If the Q was real low, and hence the Fo impedance, you
would probably need 3 IFTs.

I have never tried 3 very damped IFTs.




Within reason, for bandwidths typical of audio
receivers, you should be able to build a filter at 455 kHz that has
effectively the same response as a 2 MHz filter. There is no need to
throw out the 455 kHz IF just to get wide bandwidth.

Its difficult to make a 455kHz typical old IFT produce a nice flat topped
20 kHz wide BW. Its either pointy nosed, undecoupled, or flat
topped, critical
coupled,
or over critical or rabbit eared.
I have tried all that.

So you have tried all that and rejected the "pointy nosed", "flat topped",
and "rabbit eared" response curves. I am left to wonder what sort of
response curve you were looking for? Why not settle for a nice "flat
topped" response curve and be done with it?

I didn't say I had rejected the flat topped critical coupled IF response.

Then what did you say? You said you had "tried all that" but now it
appears that you were telling a little fib and hadn't actually tried a 455
kHz IF designed to produce the desired response.

What I said was what I said.
You are confused.

Build a radio with 2MHz and measure it, maybe it works better.

Just don't knock the idea before trying it, or condemn the idea
with postulations about what might be.
These things must be tried and measured, to really know.

Patrick Turner.



Regards,

John Byrns

Surf my web pages at, http://users.rcn.com/jbyrns/

Brian - you seem to have totally missed my point. My point was NOT that
"nothing exists" near 10khz - my point IS that in designing a receiver
for maximum fidelity one must consider that between the audio "source"
and the received signal - there are a lot of factors to consider - one
such is that there IS a roll-off as one approaches 10khz.

I never suggested that there was "nothing there" - only that if one were
to try to "reconstruct" the audio as it existed from the original source
-- processing needs to be taken into consideration - including the NRSC
- and any roll-off approaching the "brick wall".

How do you know that the 10khz signal in your last picture wasn't
originally 20db higher than shown here - before being "processed"?

(at this point Jeff adds):

Jeffie hands you another bucket of pearls to cast before the swine.

Oh, give 'em time - they'll learn - you didn't "pop-out" spouting this
stuff either... well- come to think of it --- never mind.

One more time - without knowing exactly what was going into the
transmission chain - your pictures are annecdotal at best. While they
show conformance with NRSC-1 - they on NO WAY tell us anything about
what indignities the audio suffered on it's way through the chain. You
are assuming that something that "looks flat" is. Let me remind you -
if it's flat after being pre-compensated 75us (10db at 10Khz)- SOMETHING
ate some of it!!!!!

To answer your last question (bandwidth to fully recover modulation)

Go look at John Byrns recent post where he shows the curve comparison
between two IFTs. Note that even in this very broad filter - there is
STILL some loss at Fc +/- 5khz. (You're the one that said "FULLY" - i.e.
total - no loss). Even these won't do that. And in fact - no circuit is
that ideal - too many trade offs - so one settles for practical.

To accomplish "usual" standards of fidelity - Johns numbers show that
these particular IFTs would have a "passband" of 40khz. Can you
determine why that is?

The problem (in this case using IFTs - implying a hetrodyne system) - is
that with a 40Khz passband (+/- 3db points - ooops gave away the answer
to the above question)- unwanted "stuff" pours through on unwanted
hetrodynes as well. But that's another issue.

best regards...
--
randy guttery

A Tender Tale - a page dedicated to those Ships and Crews
so vital to the United States Silent Service:
http://tendertale.com

Patrick Turner wrote:

Would not the use of pink noise through a low pass filter
and used as the carrier signal modulation be a better way to see the
frequency contour on an analyser, why noise + piano?

Absolutely - then you "know" what you're looking at.

Actually the NRSC (I know Patrick -you guys don't have to fool with
such) specifies a white noise source (equal energy at all frequencies)
filtered* then gated at 2.5hz with a 12.5% duty cycle. This is felt to
best simulate "real world" broadcasting. Again see the NRSC-1 spec I
noted yesterday.

*filter is 100hz high pass 6db/octave and 320hz low pass 6db/octave.
(yes that's -36db @ 10.24khz)

best regards...
--
randy guttery

A Tender Tale - a page dedicated to those Ships and Crews
so vital to the United States Silent Service:
http://tendertale.com

the J.W. Miller
TRF receiver



I did a cursory check on the Internet, but did not yet find any
schematics for the mentioned receivers. Are they online somewhere?
Anyone?



Just posted a schematic of the Miller TRF receiver, with the "secret"
inductance values filled
in. a.b.p.radio

RDH4 says most AM BCB radio makers tried for a final IF bandwidth response of
3.5 kHz
That was in 1955/
Since then, the BW has shrunk in many sets to even less than 2 kHz, especially
in solid
state gear, giving horrid state AM listening.
No good turning up the treble control knob, there is no treble there to boost.


I modified a fairly generic Radio Shack Optimus receiver's AM section by
removing the narrow
bandwidth ceramic filter and using a set of overcoupled IFs in its
place. See:
http://pw2.netcom.com/~wa2ise/radios...tml#solidstate
Sounds a lot better on local stations, though DX will have a lot of
monkey chatter.




The FCC limits interference only partly by bandwidth restrictions. Mostly,
it uses geographic seperation and power restrictions.

By ear, I think most stations go to about 7 or 8 kHz audio. Many of the AM
stations are talkers, but the ads can really sparkle. There's one I hear
which sounds like it goes to the 10 kHz audio max.



Much AM is talkback from mobile telephones, and its pretty dreadful....






Of course talk shows using telephone lines will be limited by the
quality of the phone system.
But you should hear better quality from the talk show host and
commercials played on station
equipment (or via satellite), as mentioned above. Audio from digitally
compressed cell phones sounds the worst.
If someone uses a cell phone to do a remote (like a high school football
game) be sure to
use an old analog cell phone (the kind that one could easedrop on with
an FM scanner radio).
But that assumes that the phone system doesn't do compression at the
cell tower site to send
it down the landlines.

In article , Patrick Turner
wrote:

John Byrns wrote:

In article , Patrick Turner
wrote:

John Byrns wrote:

Patrick, you are missing the point, the issue was the merits of a 2.0 MHz
IF frequency vs. a 455 kHz IF frequency with respect to
bandwidth/selectivity, my point was that for the sort of bandwidths we are
talking about for audio, a 455 kHz IF can provide virtually identical
"pass band and attenuation out of band" with exactly the same number of
IFTs as a 2.0 MHz IF frequency. The loaded Qs result from the design
specifications in both cases, and are what they are.

Sure. But the same Q would give wider BW at 2 MHz.
I have not ever done this, so I guess at what the final response could be.

But so what, I thought we were talking about IFs for audio here, not video
IFs? For an audio receiver I would think at the most we would want a 40
kHz bandwidth, more likely 30 kHz, or even 20 kHz in the US where the FCC
effectively limits the audio bandwidth to 10 kHz? What exactly do you see
as the advantage of a 2.0 MHz IF in an AM broadcast receiver?

The way I see it both 455 kHz IFs and 2.0 MHz IFs can be built with the
bandwidth necessary for High Fidelity AM audio reception. The stage gains
will be virtually identical for both the 455 kHz IFs and 2.0 MHz IFs of
similar bandwidth, with the exact stage gain depending somewhat on design
choices and practicalities. The wideband 455 kHz IF will have lower stage
gain than a normal narrow 455 kHz IF, but the 2.0 MHz IF also suffers from
lower stage gain. The wideband 455 kHz IF has the advantage that standard
RF front-end components like tuning capacitors and oscillator coils can be
used, while the 2.0 MHz IF will require special RF components.

What exactly are the advantages of a 2.0 MHz IF from a
selectivity/bandwidth point of view?

There may be
architectural advantages to using one or the other IF frequency in a
radio, but so far only the bandwidth/selectivity has been mentioned and in
that regard an IF of 2.0 MHz offers no significant advantage over a 455
kHz IF for the reception of the full audio bandwidth.

I supect it might, and one article in Wireless World refered to using
10.7 MHz.

Certainly a high IF frequency will have advantages in image response, but
if the bandwidth is the same, the audio quality should be similar. What
exactly did Wireless World say was so great about using a 10.7 MHz IF for
a MW AM receiver? Wireless World is a hobbyist magazine and all their
authors are not necessarily up to speed, although in the old days they
often did have articles by people who knew what they were talking about
with respect to radios. I suspect that the reason Wireless World might
have used a 10.7 MHz IF in a MW AM broadcast receiver is because it was an
easy way for a hobbyist, who both doesn't have a clue what he is doing,
and doesn't have the necessary test equipment, to get a super wide
bandwidth.

To illustrate this consider the example of the following calculated
response curves for both a 455 kHz IFT and a 2.0 MHz IFT:

The only advantage the 2.0 MHz IFT shows is marginally better symetry of
responce about the ceter frequency, the response of the two IFTs is
virtually identical.

The equality in performance depends on a large Q difference, with
544 kHz Q much lower than 2MHz Q to get the same BW.

Yes, although I have some reservations about the use of the term "Q", that
is obvious, but so what, what difference does it make?

The Q of a typical 455 kHz IFT is higher than you have indicated, because
the impedance of the LC circuit at Fo is required to be high to suit
pentode loading, and to get high gain.

You also are going to sacrifice stage gain in the same way with a 2.0 MHz
IF, so this is no more of a problem for the wideband 455 kHz IF than for
the 2.0 MHz IF.

If the Q was real low, and hence the Fo impedance, you
would probably need 3 IFTs.

This is a consequence of the wide bandwidth, not the IF frequency, the
problem is identical at 2.0 MHz.

I have never tried 3 very damped IFTs.

The fact that you haven't tried something doesn't prove anything one way
or the other. Also, what does "damped" mean in this context? I would
have to do some research, but I suspect that "damping" is more related to
filter bandwidth than to the center frequency, and both filters are aiming
for the same bandwidth.

What I said was what I said.
You are confused.

Maybe, in what way are you suggesting I am confused? I would suggest to
you that you don't understand how to design an IF filter, and don't
understand what can be done at 455 kHz.

Build a radio with 2MHz and measure it, maybe it works better.

You are the 2.0 MHz IF advocate not me, you still haven't suggested any
reason why it might work better from a bandwidth/selectivity standpoint?

Just don't knock the idea before trying it, or condemn the idea
with postulations about what might be.

I'm not, I know it would work, what I don't understand is what the
advantages are over a 455 kHz IF of the same bandwidth? You are not
explaining yourself, cite some concrete facts.

These things must be tried and measured, to really know.

While I can't claim to have designed the filter I used, I have actually
built a transistor superhetrodyne AM tuner using a 455 kHz block filter
with a 30 kHz IF bandwidth. Will the 2.0 MHz IF work better than this?
Have you tried a properly designed wideband 455 kHz IF filter to see how
it worked? The filter I used came out of a 2-way land mobile radio and I
think it was about an 8 pole filter. Back in the old days of land mobile
here in the US, wider channels with greater bandwidth were used than are
used today. Over time the channels were squeezed down to accommodate
additional channels in the same space, and block filters of several
different bandwidths were available to suit the changing allocations and
operating frequencies.

I have also built wideband single frequency TRF receivers using modified
double tuned IF transformers.

So what it boils down to is that you haven't tried a wideband 455 kHz
filter while I have, and I haven't tried a 2.0 MHz IF filter, which you
may or may not have done. I at least have cited some concrete facts about
IF filters, while you have only muttered about Q, without indicating how
it actually relates to the problem. I am not a "filter jock" (tm) but I
think it is generally desirable that the Q of the components used in a
filter be high, especially when we get beyond simple double tuned
transformers. What you are calling Q is more related to how the filter is
terminated, which is a different matter than the Q of the components that
make up the filter.


Regards,

John Byrns


Surf my web pages at, http://users.rcn.com/jbyrns/

John Byrns wrote:
What exactly are the advantages of a 2.0 MHz IF from a
selectivity/bandwidth point of view?

Both pros and cons to this John.

Since the bandwidth is a percentage of the center frequency,
the shape of the bandwidth will change based on the distance
from the center frequency (as a percentage.)

Assuming just for the moment +/- 5 KHz.
At 455 KHz that's about 1% above and below.
At 2 MHz, that's now only .25% above and below.

As you get further from the center frequency, percentage
wise, the shape of the curve as it transitions from inside
to outside of the band pass is going to look different at
the upper frequency than it does at the lower frequency.

Certainly a high IF frequency will have advantages in image response, but
if the bandwidth is the same, the audio quality should be similar.

With the notable exception of the difference in shape of the
roll off above and below the center frequency.

In the world of designing filters (and overall system performance)
this is called group delay. A shorter, perhaps more recognizable
term would be linear phase shift over the entire band pass of the
filter.

Wireless World is a hobbyist magazine and all their
authors are not necessarily up to speed

They were under the same constraints as the Weekly World News.
"If it wasn't true, they couldn't print it." Note smiley face
here. ;-)

Yes, although I have some reservations about the use of the term "Q", that
is obvious, but so what, what difference does it make?

Back to the original comments about Q. In a perfect world, it
would only be a matter of the LC ratio setting the bandwidth
of a tuned circuit. Of courses, there are other things that
get in the way to reduce the overall Q of a circuit. Nasty
little things like the series resistance of the coils, dielectric
losses in both the coil forms and capacitor insulation material.

Back to the original "ideal" values of Q. 15 at 455 KHz and
67 at 2 MHz. It is physically "more challenging" to get higher
Q at a higher frequency. All of the various losses of the
components tend to get in the way. Wire losses, dielectric
losses and any losses of the ferrite used in the core materials.

Also, what does "damped" mean in this context? I would
have to do some research, but I suspect that "damping" is more related to
filter bandwidth than to the center frequency, and both filters are aiming
for the same bandwidth.

"Damped" means adding some form of resistance across the reactive
components of a circuit. As an example, if you were to assemble a
nice 455 KHz IF transformer and found that the bandwidth was too
narrow, a fast method of widening it would be to place parallel
resistors across the windings.

Another point about "damped" is that if a tuned circuit has too
high a Q, a sudden transient will tend to make it oscillate.
In communications receivers, this is obvious that a signal sounds
more like you're ringing a bell, than simply turning a tone on
and off. (Kind of like using the sustain pedal on a piano.)

Maybe, in what way are you suggesting I am confused? I would suggest to
you that you don't understand how to design an IF filter, and don't
understand what can be done at 455 kHz.

Lets not go in that direction.

An IF transformer is simply a two pole butter worth filter.
That it can have different input and output impedance just
makes it really convenient for taking the source from a plate
and connecting it to a grid for a load.

By definition, a butter worth filter has a smooth curve with
only one peak (in the middle.) And the shape (steepness) of
the band pass is related to the overall Q of the circuit.

The next type of filter, would be Chebychev, This is no more
than a "predistorted" butter worth filter network. By allowing
a certain amount of ripple in the pass band, the shape of the
rejection can be made sharper. The obvious trade off is the
amount of distortion to the signal within the pass band.

A simple example of this would be stagger tuned IF coils.
Two or more peaks, and a dip (or dips), ripple, in the middle.

While I can't claim to have designed the filter I used, I have actually
built a transistor superhetrodyne AM tuner using a 455 kHz block filter
with a 30 kHz IF bandwidth. Will the 2.0 MHz IF work better than this?
Have you tried a properly designed wideband 455 kHz IF filter to see how
it worked? The filter I used came out of a 2-way land mobile radio and I
think it was about an 8 pole filter.

The point you've probably overlooked in land mobile operations is that
it was NEVER designed as a "hi-fi" system. There's a reason for the
term "voice grade." Having as much a 3 dB of ripple in a band pass
filter is meaningless especially when the filter is in the midst of
a limiting IF strip for FM recovery, and on AM demodulation. What
really matters here is limiting the bandwidth of the received signal to
ONLY include that of the wanted (in channel) information and none of
the unwanted (adjacent channel) information to get to the discriminator.

So what it boils down to is that you haven't tried a wideband 455 kHz
filter while I have, and I haven't tried a 2.0 MHz IF filter, which you
may or may not have done. I at least have cited some concrete facts about
IF filters, while you have only muttered about Q, without indicating how
it actually relates to the problem. I am not a "filter jock" (tm) but I
think it is generally desirable that the Q of the components used in a
filter be high, especially when we get beyond simple double tuned
transformers. What you are calling Q is more related to how the filter is
terminated, which is a different matter than the Q of the components that
make up the filter.

You should take the time to read up on "filter jockeying" John.
You're making a lot of incorrect assumptions on how they work.

The primary requirement on the Q of individual components in
filter design is only such that their value of Q be high enough
to not materially effect the overall Q of the circuit. As an
example, (and without getting into cryogenic treatments and
styrofoam cups) A speaker system sounds better through 25 feet
of #12 AWG wire than it does through 25 feet of #18 AWG wire.
And that's strictly due to the resistive loss of the wire in
comparison to the losses in the actually speaker design and
implementation.

I had a electronics instructor in college that would show you
"The secret of electronics" that he kept hidden, and locked,
inside a small jewelry box if you "caught on" during his course.

With some fanfare, he would slowly open the box and you would
see an inductor, a resistor and a capacitor.

And it's really just that simple. What gets complicated is when
you forget that all three items have hidden values of the others
contained within them. (I.e. the difference between practical
and theoretical parts.)

Jeff



--
"They that can give up essential liberty to obtain a little temporary
safety deserve neither liberty nor safety." Benjamin Franklin
"A life lived in fear is a life half lived."
Tara Morice as Fran, from the movie "Strictly Ballroom"

Randy and/or Sherry wrote:

Patrick Turner wrote:

Would not the use of pink noise through a low pass filter
and used as the carrier signal modulation be a better way to see the
frequency contour on an analyser, why noise + piano?

Absolutely - then you "know" what you're looking at.

Actually the NRSC (I know Patrick -you guys don't have to fool with
such) specifies a white noise source (equal energy at all frequencies)
filtered* then gated at 2.5hz with a 12.5% duty cycle. This is felt to
best simulate "real world" broadcasting. Again see the NRSC-1 spec I
noted yesterday.

I thought white noise had a rising amplitude as F rose.
Pink noise is white noise filtered at a slope of 3 dB/octave,
and thus giving a flat average level amplitude response for any single F
filtered
out of the pink noise, and is thus used for speaker testing etc....

The pink noise gene I made has such a filter applied to a white noise
source.
My bandpass filter for speaker tests has a Q of 12 for any part of the
audio band, and the amplitudes
of the noise bands filterd out from the noise signal is the same between
20 Hz and 20 kHz.

Patrick Turner.



*filter is 100hz high pass 6db/octave and 320hz low pass 6db/octave.
(yes that's -36db @ 10.24khz)

best regards...
--
randy guttery

A Tender Tale - a page dedicated to those Ships and Crews
so vital to the United States Silent Service:
http://tendertale.com

John Byrns wrote:

In article , Patrick Turner
wrote:

John Byrns wrote:

In article , Patrick Turner
wrote:

John Byrns wrote:

Patrick, you are missing the point, the issue was the merits of a 2.0 MHz
IF frequency vs. a 455 kHz IF frequency with respect to
bandwidth/selectivity, my point was that for the sort of bandwidths we are
talking about for audio, a 455 kHz IF can provide virtually identical
"pass band and attenuation out of band" with exactly the same number of
IFTs as a 2.0 MHz IF frequency. The loaded Qs result from the design
specifications in both cases, and are what they are.

Sure. But the same Q would give wider BW at 2 MHz.
I have not ever done this, so I guess at what the final response could be.

But so what, I thought we were talking about IFs for audio here, not video
IFs? For an audio receiver I would think at the most we would want a 40
kHz bandwidth, more likely 30 kHz, or even 20 kHz in the US where the FCC
effectively limits the audio bandwidth to 10 kHz? What exactly do you see
as the advantage of a 2.0 MHz IF in an AM broadcast receiver?

Its a lot easier to get a wider pass band of 30 kHz with 2MHz IFTs than with
455 kHz IFTs.
Try it some time, and then you'll know.



The way I see it both 455 kHz IFs and 2.0 MHz IFs can be built with the
bandwidth necessary for High Fidelity AM audio reception. The stage gains
will be virtually identical for both the 455 kHz IFs and 2.0 MHz IFs of
similar bandwidth, with the exact stage gain depending somewhat on design
choices and practicalities. The wideband 455 kHz IF will have lower stage
gain than a normal narrow 455 kHz IF, but the 2.0 MHz IF also suffers from
lower stage gain. The wideband 455 kHz IF has the advantage that standard
RF front-end components like tuning capacitors and oscillator coils can be
used, while the 2.0 MHz IF will require special RF components.

What exactly are the advantages of a 2.0 MHz IF from a
selectivity/bandwidth point of view?

I suspect 3 x 2MHz IFTs would be easier to get a flat topped
pass band and sufficient steep roll off just outside the band.

I also suspect any old 455 kHz IFTs could easily
have about 3/4 of their turns removed, and retain the same
caps of 250pF.

For 250pF, to get 455 kHz, one needs 0.48 mH

For 250pF, and 2 MHz, one needs 0.025 mH.

To reduce L by 20 times, the turns would need reducing
by a factor of 1/4.47.

Thus the DCR would fall, and Q could rise.

I have used ex IFT windings with turns removed for
high Q RF input coils on my reciever, to get the range of
tuning required between 500 and 1750 kHz with a
20 pF to 360 pF tuning gang.
The ferrite slug is retained.

The wire is litz wire, with low RF resistance, hence it gives a high Q,
but for 2 mHz, solid round wire would probably be OK, like in 4.5 MHz
TV IFTs and 10.7 MHz FM IFTs.

HF IFTs are easier to wind than 455 kHz.



There may be
architectural advantages to using one or the other IF frequency in a
radio, but so far only the bandwidth/selectivity has been mentioned and in
that regard an IF of 2.0 MHz offers no significant advantage over a 455
kHz IF for the reception of the full audio bandwidth.

I supect it might, and one article in Wireless World refered to using
10.7 MHz.

Certainly a high IF frequency will have advantages in image response, but
if the bandwidth is the same, the audio quality should be similar. What
exactly did Wireless World say was so great about using a 10.7 MHz IF for
a MW AM receiver?

Wide AF response was easily achieved.

Wireless World is a hobbyist magazine and all their
authors are not necessarily up to speed, although in the old days they
often did have articles by people who knew what they were talking about
with respect to radios.

I differ. WW and what it became, Electronics World wasn't just an
amateur's magazine. It had cutting edge articles about electronics
from 1917 onwards, and I suggest you park yourself beside a
pile of all the old copies and have a good read.
Most of the info was only comprehensible by very well university educated
professionals, or intellectuals, and most ideas were backed up with mathematical
proofs which nearly all the general public couldn't understand.



I suspect that the reason Wireless World might
have used a 10.7 MHz IF in a MW AM broadcast receiver is because it was an
easy way for a hobbyist, who both doesn't have a clue what he is doing,
and doesn't have the necessary test equipment, to get a super wide
bandwidth.

I leave you to your suppositions.



To illustrate this consider the example of the following calculated
response curves for both a 455 kHz IFT and a 2.0 MHz IFT:

The only advantage the 2.0 MHz IFT shows is marginally better symetry of
responce about the ceter frequency, the response of the two IFTs is
virtually identical.

The equality in performance depends on a large Q difference, with
544 kHz Q much lower than 2MHz Q to get the same BW.

Yes, although I have some reservations about the use of the term "Q", that
is obvious, but so what, what difference does it make?

Build a receiver, and find out.



The Q of a typical 455 kHz IFT is higher than you have indicated, because
the impedance of the LC circuit at Fo is required to be high to suit
pentode loading, and to get high gain.

You also are going to sacrifice stage gain in the same way with a 2.0 MHz
IF, so this is no more of a problem for the wideband 455 kHz IF than for
the 2.0 MHz IF.

Use more stages if stage gain is low.
The EA design used 3 IFTs, with two j-fet IF amps, with quite
heavily damped 455 kHz IF coils.




If the Q was real low, and hence the Fo impedance, you
would probably need 3 IFTs.

This is a consequence of the wide bandwidth, not the IF frequency, the
problem is identical at 2.0 MHz.

I have never tried 3 very damped IFTs.

The fact that you haven't tried something doesn't prove anything one way
or the other.

It means that what you or I am saying may not include all the facts about
the subject. Build and measure will give the facts.

Also, what does "damped" mean in this context?

Strapping resistance across the LC tuned circuit to reduce the Q.
The rate of attenuation just either side of the pass band becomes
much less, so more IF stages must be used.

I would

have to do some research, but I suspect that "damping" is more related to
filter bandwidth than to the center frequency, and both filters are aiming
for the same bandwidth.

Damping reduces Q, and increases BW.
But it also reduces Z at Fo, thus reducing gain in an amp
which must be a current source, like a pentode or j-fet,
to realise the best selectivity for the LC circuit.



What I said was what I said.
You are confused.

Maybe, in what way are you suggesting I am confused? I would suggest to
you that you don't understand how to design an IF filter, and don't
understand what can be done at 455 kHz.

I know enough about IFT design, after having built my own radio.


Build a radio with 2MHz and measure it, maybe it works better.

You are the 2.0 MHz IF advocate not me, you still haven't suggested any
reason why it might work better from a bandwidth/selectivity standpoint?

I refuse to repeat myself any further.



Just don't knock the idea before trying it, or condemn the idea
with postulations about what might be.

I'm not, I know it would work, what I don't understand is what the
advantages are over a 455 kHz IF of the same bandwidth? You are not
explaining yourself, cite some concrete facts.

I have already stated that for a given Q, the pass band for a 2MHz IFT
is naturally a lot wider than for a 455 kHz IFT.

Put it this way, if you make IFTs of 100 kHz, then its all the harder to get
a flat topped bandpass response which is 20 kHz wide, with
high sloped skirt response each side.



These things must be tried and measured, to really know.

While I can't claim to have designed the filter I used, I have actually
built a transistor superhetrodyne AM tuner using a 455 kHz block filter
with a 30 kHz IF bandwidth.

Ceramic filters are another way to achieve the same bandpass filter
that the IFT could do.
But they were never used in tube sets for the BCB.


Will the 2.0 MHz IF work better than this?

I suspect yes, but getting a 2 MHz cermic filter with 30 kHz of BW might be
unobtainium.


Have you tried a properly designed wideband 455 kHz IF filter to see how
it worked?

Yes, and trying to squeeze 20 kHz of flat topped BW was difficult with stock IFTs.

I have already said what my solution was, to use a variable distance coils and
some damping
on IFT no1, which allowed me to have only 2 IFTs, and 1 IF amp, a 6BX6, fixed
bias,
for low thd IF amplification.

The filter I used came out of a 2-way land mobile radio and I
think it was about an 8 pole filter. Back in the old days of land mobile
here in the US, wider channels with greater bandwidth were used than are
used today. Over time the channels were squeezed down to accommodate
additional channels in the same space, and block filters of several
different bandwidths were available to suit the changing allocations and
operating frequencies.

I have also built wideband single frequency TRF receivers using modified
double tuned IF transformers.

One of the Electronics Australia kit designs I have used
a two stage TRF design with highish Q LC, with stagger tuning
at the low F part of the band. This utilised having mutual capacitive
coupling of the Ls in their earthy ends to ground via one common 0.1 uF.
I couldn't easily reproduce the nice response curves of the kit set,
and it was not good enough to give selectivity between locals
here where I wanted to hear a 300 watt station which was only
45 kHz away from a 5,000 watt station.
But otherwise, the TRF was a fine performer.



So what it boils down to is that you haven't tried a wideband 455 kHz
filter while I have, and I haven't tried a 2.0 MHz IF filter, which you
may or may not have done.

I have tried getting 455 kHz IFTs to go wider, but I was dissapointed with
overall results, because I'd have needed 3 IFTs, and lots of damping.

I got 10 kHz of audio BW at low thd using simple methods of damping, sliding IFT1
coils closer,
and some RC boosting of audio HF.
I thus achieved the use of tubes, good AF BW, and excellent local station
selectivity, which allowed me to hear my wanted 300 watt station without
the 5,000 watt station able to be heard even though it is only 45 kHz away.

I at least have cited some concrete facts about
IF filters, while you have only muttered about Q, without indicating how
it actually relates to the problem. I am not a "filter jock" (tm) but I
think it is generally desirable that the Q of the components used in a
filter be high, especially when we get beyond simple double tuned
transformers. What you are calling Q is more related to how the filter is
terminated, which is a different matter than the Q of the components that
make up the filter.

I leave you to wonder the full content of my mutterings,
and I do hope you spend some time soon in your shack with a soldering iron
and response meter.

The other advantage of a 2 MHz IF is that the filtering of RF from the recovered
audio is easier, because the C value is less, and the filter used has less effect
on recovered audio at 10 kHz, and at high amplitudes.

But don't let me mention it, I know you'd be aware of it already.

Patrick Turner.



Regards,

John Byrns

Surf my web pages at, http://users.rcn.com/jbyrns/

In article , Robert Casey
wrote:

Just posted a schematic of the Miller TRF receiver, with the "secret"
inductance values filled
in. a.b.p.radio


Hi Robert, that's certainly a cute little radio you have there. An
interesting point is that the separate "negative mutual coupling"
inductance, the one with the "secret" value, isn't even necessary and the
part can often be eliminated from the circuit. All you need to do is wind
L1 and L2 like a typical double tuned IF transformer, and if the coupling
coefficient is correctly chosen to yield the required value of mutual
inductance, and if the two windings are phased correctly to make the
mutual inductance "negative", then the separate coil like you used isn't
necessary, although you must retain the capacitor in the common lead of
"L1" And "L2", since that is part of the "secret". This scheme will work
in a circuit like the Miller "High Fidelity" Crystal Tuner where L1 and L2
are just single winding coils, there would obviously be problems applying
the idea to your circuit because of the extra winding you put on L1,
making it into a transformer by itself.

I listened to the WABC "jpg" you posted, and the tuner certainly has a
good bandwidth, although I wonder how much pre emphasis WABC might have
been using and how well your receiver matches it, I will have to listen to
it again to see how correct the de emphasis seems to me, there was also
some background noise at several points, I will have to listen again to
see if it was part of the audio at points, or if it was interference of
some sort. The thing I didn't like about it was that it had pretty
horrible levels of distortion, and while this could be WABC's fault, I
have found that it is typical of these so called "High Fidelity" crystal
receivers. I have a couple of J.W. Miller "High Fidelity" crystal tuners,
and they have the same distorted sound. I think this is because the
crystal detectors produce really horrendous distortion, unless you have a
big enough antenna to get the audio output level up to at least the 2
volts RMS that Patrick recommends. I would try for another 10 dB or so of
audio output above that level before being completely happy myself. But
no way do these simple crystal sets sound "High Fidelity" to me because of
all the nonlinear distortion. The distortion probably does help give them
their bright sound though, sort of a psycho acoustical trick if you will,
but it does wear on one.


Regards,

John Byrns


Surf my web pages at, http://users.rcn.com/jbyrns/

Patrick Turner wrote:

I thought white noise had a rising amplitude as F rose.

as "burnt toast" as I am tonight - someone could claim white noise comes
from Procol Harum and pink noise from Pink Floyd - and I'd agree.

Here is the quote from the NRSC-1 spec for bandwidth testing - as it
relates to "source"...

Section: 6.3.2 Use of Standard Test Signal. Audio bandwidth shall be
measured using a test signal consisting of USASI (United States of
America Standards Institute) noise that is pulsed by frequency of 2.5 Hz
at a duty cycle of 12.5%. See Figure 4. USASI noise is intended to
simulate the long-term average spectra of typical audio program
material. Pulsing of the noise is intended to simulate audio transients
found in audio program Material. USASI noise is a white noise source
[note 4](i.e. noise with equal energy at all frequencies) that is
filtered by (1) a 100 Hz, 6 dB per octave high-pass network and (2) a
320 Hz, 6 dB per octave low-pass network. too Figure 4. A pulsed USASI
noise generator is shown in Figures 5 and 6. Using the attenuator pad,
the ratio of peak-to average amplitude shall be 20 db at the audio
output of the pulser. [snip]

Note 4. Acceptable white noise sources include GenRad Models 1382 and
1390B; Bruel & Kjaer Model 1405; and National Semiconductor IC No. MM5837N.

[end NRSC-1 quotes]

If you can find specs on any of those generators or that IC - then
you'll find what they think white noise is.

Right now it's approaching midnight - just went through the emotionally
draining experience of watching a old family friend's funeral on TV...
and Sherry and I are toast - so someone else can look them up.

best regards...
--
randy guttery

A Tender Tale - a page dedicated to those Ships and Crews
so vital to the United States Silent Service:
http://tendertale.com