SSB - why poor sound quality?
Dave Kirkby
to my mind, the sound quality is not as good as an AM transmission, using a 6KHz
bandwidth for AM. Why?
I realsise the SSB bandwidth is less than half the AM one, but both contain only the 300-3000Hz frequencies. Sure AM takes up more space, but an AM TX just sends two
copies of the sidebands. So SSB's more efficient use of bandwidth should not mean
a redection in quality.
Perhaps other people disagree, but I'm convinced SSB does not sound as good as an
amateur AM signal. Obviuosly one would not expect it to sound like a commercal AM
signal, where the bandwidth exceeds the 300-3000Hz.
My thoughts as to a possible explanation are:
1) Would a typical AM signal send from very low frequencies (say 20Hz) to 3KHz, since
there is no point restricting the lower ends? Since although the signal takes up less
bandwidth, its most unlikely anyone else can use the space between the carrier and the
lower limits of the sidebands. Perhaps then the 20Hz to 300Hz information makes the
signal sound better.
2) Its very difficult to re-insert the carrier at precisely the correct frequency at
the receiver. I've recently used a 2m transceiver as a tuneable IF for an HF receiver
I'm biliding, The fact the tranceiver only steps in 100Hz steps definately causes
poor quaility audio, but I've noticed this with VFO's too.
3) Perhaps there need to re-inser the carrier at the correct amplitude, or
in the correct phase. I've never looked at the maths to say if this is true.
Any thoughts?
dave kirkby G8WRB.
Scott Dorsey
>Like many hams, I predominately send (when I had a TX) and receive SSB. However,
>to my mind, the sound quality is not as good as an AM transmission, using a 6KHz
>bandwidth for AM. Why?
Mostly this is a result of phase distortion, and massive group delay across
the passband, both on the transmit and the receive side.
I would suspect that something could be done about this by using very high
speed DSP in place of the usual phasing networks, but I haven't heard anyone
actually trying this. It could be an awful lot of fun, though.
--scott
not that cranking the ARC-5 up on AM isn't a lot of fun, too....
--
"C'est un Nagra. C'est suisse, et tres, tres precis."
Bill Kirkland
Dave Kirkby <da...@medphys.ucl.ac.uk> wrote:
>Like many hams, I predominately send (when I had a TX) and receive SSB. However,
>to my mind, the sound quality is not as good as an AM transmission, using a 6KHz
>bandwidth for AM. Why?
>
I don't know about AM but I do find SSB speech quality very poor.
Note: Before digital transmission SSB used to be one way of multiplexing multiple
phone calls over a telephone cable.
>My thoughts as to a possible explanation are:
>
>1) Would a typical AM signal send from very low frequencies (say 20Hz) to 3KHz,
since
>there is no point restricting the lower ends?
I doubt it. I do like the phase distortion theory that is presented in a reply
to this article. Traditionaly not much attention has been paid to the phase
distortion introduced by filters, the concern has generally been on magnitude
response. Phase is very important especially when using digital transmission
modes (QAM, PSK)
>
>2) Its very difficult to re-insert the carrier at precisely the correct frequency at
>the receiver. I've recently used a 2m transceiver as a tuneable IF for an HF receiver
>I'm biliding, The fact the tranceiver only steps in 100Hz steps definately causes
>poor quaility audio, but I've noticed this with VFO's too.
>3) Perhaps there need to re-inser the carrier at the correct amplitude, or
>in the correct phase. I've never looked at the maths to say if this is true.
>
Amplitude doesn't matter, except in obtaining proper performance of the
demoduator, i.e. too much/too little carrier power. In my text books
they always perform synchronous detection, i.e. the phase of the carrier
at the receiver is identical to that at the transmitter. If the phase
is not identical then when demodulation takes place the amplitude of
the demodulated signal is weighted by the cos/sin of the phase error.
This requires a phase tracking loop (& frequency) which greatly increases
the complexity of the receiver as opposed to a simple VFO. I have never
understood how we get rid of the "phase error" with our equipment - unless
it is done by "manually tuning in" the signal. Anyway, a resolution of
better than 10 Hz is needed to get satisfactory performance (unless using
a channelized system, e.g. like the 5 kHz steps on the 2 mtr band).
Any thoughts on the phase problem ?
Bill Kirkland
Michael Barts
> Any thoughts on the phase problem ?
>Bill Kirkland
I've had the same thoughts about group delay distortion through the
crystal filters used for SSB. In Saben's book on SSB he mentions that
group delay distortion can result in distorted audio since speech is
made up of many frequency components, if the individual frequency
components do not arrive in time synch, due to group delay distortion,
the result is distorted audio. That's roughly the explaination from
memory. I've been wanting to get my hands on some crystal filters to
measure group delay through them to get a feel for the effect. Anybody
done these measurements already so I don't have to re-invent the
wheel? The real question is whether or not the group delay distortion
problem is on the order of the same magnitude of the non-linearities
in the circuit. A friend and I have had this discussion. He thinks the
circuit non-linearities are greater but I'm still waiting to be
convinced.
Mike Barts
KB4NT
mba...@vt.edu
Scott Dorsey
>kirk...@bnr.ca (Bill Kirkland) wrote:
>> Any thoughts on the phase problem ?
>
>>Bill Kirkland
>
>I've had the same thoughts about group delay distortion through the
>crystal filters used for SSB. In Saben's book on SSB he mentions that
>group delay distortion can result in distorted audio since speech is
>made up of many frequency components, if the individual frequency
>components do not arrive in time synch, due to group delay distortion,
>the result is distorted audio. That's roughly the explaination from
>memory.
You got it. And, this is pretty much going to happen if you use any
sort of infinite impulse response filter. Note that also the phase-shift
networks on the transmit side do not have an even phase response, and they
are also responsible for a lot of the distortion.
>I've been wanting to get my hands on some crystal filters to
>measure group delay through them to get a feel for the effect. Anybody
>done these measurements already so I don't have to re-invent the
>wheel?
Yes, there is a bunch of stuff in the IRE journals in the forties and
fifties on the subject, mostly talking about filters for AM listening
but which still should be useful. Remember that the group delay at
the edges of the passband is directly related to the slope of the
cutoff. If you think the crystal filters are bad, you should just
listen to something like the Collins receivers with the mechanical
filters. They do an amazing job of cutting out interference, but
the group delay is awful. Incidentally, has anyone out there tried
building an all-pass network on the audio output to compensate for
somet of this?
The real question is whether or not the group delay distortion
>problem is on the order of the same magnitude of the non-linearities
>in the circuit. A friend and I have had this discussion. He thinks the
>circuit non-linearities are greater but I'm still waiting to be
>convinced.
The nonlinearities are significant. The group delay distortion is
also significant. Which one is dominant depends on your transmitter
and your receiver, but with good equipment, the group delay is dominant.
Tight filters are good things in that they reduce interference, but
they are bad things in that they induce phase distortion. Likewise
the phasing networks on the transmitter are essential to get proper
modulation, but building good ones is expensive and building perfect
ones is impossible.
If you want to do a measurement on a crystal filter yourself, just get a
colorburst crystal from your local TV supplier, a signal generator, and
a scope, and rig it up as a series resonant filter. (Yes, I know that
isn't an optimum configuration, but it's easy to do and it works well
enough for a demonstration).
--scott
Paul H. Bock
in SSB voice transmission might want to consider that in FM stereo
broadcasting the L-R audio is broadcast as a DSBSC (double sideband
suppressed carrier) signal centered at 38 kHz. In the receiver, this
DSBSC signal is demodulated (using the 19 kHz pilot tone as a frequency
and phase reference) and the resulting L-R audio combined with the L+R
audio from the main channel, resulting in separate L & R signals. The
FCC technical standard for phase coherence between the main channel
and stereophonic subchannel is +/- 3 degrees (which is satisfied by
achieving a stereo separation of 29.7 dB) (Paragraph 73.322(m) of
Part 73).
The key differences are (a) a DSB signal requires no filtering, and
(b) in FM broadcasting a 19 kHz pilot tone is transmitted which must have
a frequency accuracy of +/- 2 Hz. In addition, there is a phase-coherency
requirement which is "built in" to the transmitter, and requires careful
attention throughout the entire audio chain since the FCC specification
applies to the entire path from microphone input to transmitter output.
As far as determining which of the two components (filter group delay
or TX to RX phase incoherency) is the biggest culprit, that might be
difficult to pin down. I would be tempted to try DSB with a SSB transmitter
by perhaps bypassing the filter and doing an "empirical audio quality
comparison." This would take filter group delay out of consideration.
Trying to come up with some kind of phase-coherent detection in your average
ham rig would, I suspect, be a much nastier problem to attack.
73,
Paul, K4MSG
Paul H. Bock
>memory. I've been wanting to get my hands on some crystal filters to
>measure group delay through them to get a feel for the effect. Anybody
>done these measurements already so I don't have to re-invent the
>wheel? The real question is whether or not the group delay distortion
>problem is on the order of the same magnitude of the non-linearities
About 3 years ago we did a study on this problem as related to the
use of VHF/UHF receivers for signal collection. We examined a number
of different filter types including Gaussian, 6 and 12 dB transistional,
Chebychev, Elliptic, and Equiripple Phase 0.5 degree. The last type had
the best combination of group delay and ringdown characteristics (the
latter being of most importance when dealing with pulse signals).
The tradeoff for the more common filter architectures is that the
better the group delay and ringdown, the worse the shape factor. The
Chebychev is nice and sharp but is by far the worst in terms of group
delay. Since USB and LSB filters tend to have a very sharp slope on
the side adjacent to the center frequency (for unwanted sideband
rejection) I'm confident that their group delay characteristics are
horrendous.
BTW, regarding the "tradeoff" noted above, in applications where
a low ultimate rejection ( <50 dB ) and high insertion loss can be
tolerated, a SAW filter is the best choice; near-vertical skirts and
nice, flat group delay.
Of course, none of this answers the question of "what's doing the
most damage"..............
Paul, K4MSG
Richard Karlquist
>kirk...@bnr.ca (Bill Kirkland) wrote:
>> Any thoughts on the phase problem ?
>
>>Bill Kirkland
>
>I've had the same thoughts about group delay distortion through the
>crystal filters used for SSB. In Saben's book on SSB he mentions that
>group delay distortion can result in distorted audio since speech is
>made up of many frequency components, if the individual frequency
>components do not arrive in time synch, due to group delay distortion,
>the result is distorted audio. That's roughly the explaination from
The distortion from unequal group delay affects modems (which is one
reason why they have adaptive phase equalizers) but is not audible
by ear when listening to speech. What *is* audible vis a vis filters
is amplitude ripple, at least if it is bad enough.
The SSB filters the phone companies used in the old analog FDM
systems had flat amplitude response, and lots of phase distortion,
and they sounded much better than anything you hear on the air,
except for SWBC station point to point SSB links, which sound as good
as AM on a good receiver.
I suspect the main culprit is the nonlinear distortion
in the average SSB transmitter.
Rick Karlquist N6RK
rka...@scd.hp.com
Richard Karlquist
Paul H. Bock <p...@syseng1.melpar.esys.com> wrote:
>mba...@vt.edu (Michael Barts) writes:
>
> BTW, regarding the "tradeoff" noted above, in applications where
>a low ultimate rejection ( <50 dB ) and high insertion loss can be
>tolerated, a SAW filter is the best choice; near-vertical skirts and
>nice, flat group delay.
You can also do this with digital (FIR) filters. I have listened
to such FIR filters and they don't sound any better than the old
phone company analog filters, because group delay distortion is generally
inaudible. They are, of course, kinder and gentler to modems.
Rick Karlquist N6RK
rka...@scd.hp.com
Richard Karlquist
Scott Dorsey <klu...@netcom.com> wrote:
>the group delay is awful. Incidentally, has anyone out there tried
>building an all-pass network on the audio output to compensate for
>somet of this?
I once built an adjustable audio all pass network, where I could
adjust the center frequency from 20 Hz to 20 kHz. with a pot.
I put music thru it (from a decent stereo system) and listened
to it with high quality headphones. I A/B'ed the music with and
without the network at various frequency settings, and was never
able to detect any difference. I took it into work here and hooked
in up to a function generator and tried square waves, triangle waves,
etc. No audible effect. On the other hand, the waveform, as
displayed on a scope went totally nuts, as you'd expect.
I let some of the guys here try it too and they couldn't hear any
difference either.
The all pass in question was a first order type:
s-a
F(s) = -------
s+a
The conclusion is that you can't hear delay distortion. I later
found out about a JAES paper that essentially agreed with my findings.
Rick Karlquist N6RK
rka...@scd.hp.com
WB3U
>I realsise the SSB bandwidth is less than half the AM one, but both contain only the 300-3000Hz frequencies. Sure AM takes up more space, but an AM TX just sends two
>copies of the sidebands. So SSB's more efficient use of bandwidth should not mean
>a redection in quality.
Do you mean that AM sounds better when you listen to a commercial broadcaster using a
modern receiver with a 6 KHz xtal filter, or that Amateur AM in the old days sounded better?
I'm asking this because I agree with you, but only as it relates to the old rigs. I think the
difference occurs for two reasons. First, older transmitters (AM) used audio filtering to
rolloff the audio outside the 300 - 3000 range. In most cases, the rolloff was *very* slow
and still had significant energy outside this band. Second, older AM receivers had a
passband much broader than 6 KHz, and with kinder, gentler rolloff than today's filters.
I can still remember how bad they sounded on AM, relatively speaking, when the crystal
filter (usually only two crystals) was activated. Try this sometime with an SX-28 and
you'll see what I mean.
>Perhaps other people disagree, but I'm convinced SSB does not sound as good as an
>amateur AM signal. Obviuosly one would not expect it to sound like a commercal AM
>signal, where the bandwidth exceeds the 300-3000Hz.
To say that it doesn't sound as good is an understatement. Personally, I find it extremely
tiring, which is just one more reason I work more CW than 'phone <g>.
>My thoughts as to a possible explanation are:
>
>1) <snip> Perhaps then the 20Hz to 300Hz information makes the signal sound better.
Yep.
>2) Its very difficult to re-insert the carrier at precisely the correct frequency at
>the receiver. <snip>
>
>3) Perhaps there need to re-inser the carrier at the correct amplitude, or
>in the correct phase. I've never looked at the maths to say if this is true.
I don't think 2) and 3) have a significant impact. As evidence of this, I recall several
friends of mine in the 60's who converted their ARC-5 transmitters to DSB and who
achieved very respectable carrier suppression. Using a receiver at the time which
couldn't even keep 100 Hz accuracy (let alone 10 Hz), I can tell you that the fidelity
of their signals was *almost* as good as standard AM.
73,
Jack WB3U
Gary Coffman
>>
>>2) Its very difficult to re-insert the carrier at precisely the correct frequency at
>>I'm biliding, The fact the tranceiver only steps in 100Hz steps definately causes
>>poor quaility audio, but I've noticed this with VFO's too.
Bingo! It's the frequency translation error that gets you. If you have
any frequency translation error, the harmonic relationships of the
speech (or music) components no longer line up. So instead of nice
harmonies, you have dissonance which is very grating to the ear.
It's like listening to a musical instrument that isn't tuned on a
well tempered scale.
Gary
--
Gary Coffman KE4ZV | You make it, | gatech!wa4mei!ke4zv!gary
Destructive Testing Systems | we break it. | emory!kd4nc!ke4zv!gary
534 Shannon Way | Guaranteed! | ga...@ke4zv.atl.ga.us
Lawrenceville, GA 30244 | |
John Seboldt
makes sense to me. Other reasons: SSB transmitters may be less
concerned with low audio distortion because they know a lot of the
distortion components will end up filtered out. This is true of harmonic
components, of course, but still leaves various intermodulation
components audible.
Also, I'll bet many of the old AM transmitters did not filter the speech
bandwidth quite as sharply as a modern filter-type SSB rig; thus a little
more lows and highs creep through, despite the nominal frequency response
of 300-3000 Hz.
It's a real kick to use my Campbell R2 receiver design -- the Jan. 93 QST
direct conversion RX that phases out the other sideband. All the audio
is low distortion -- and if you switch out the audio filters, you have a
nice broad response that gently rolls off outside the speech bandwidth,
making for very good zero-beat broadcast reception, and a reasonably
objective evaluation tool for audioquality evaluation. I can't wait to
hook up the companion T2 phasing exciter -- I could run it on AM without
filters and go true hi-fi :-)
: John Seboldt rohr...@netcom.com / CW: It don't mean a thing
: Amateur radio K0JD... / if it ain't got that swing!
: Church of the Annunciation, / Di dah, di dah, di dah, di dah...
: Minneapolis / (sorry, Duke!)
Gary Coffman
>In article <3q510p$a...@solaris.cc.vt.edu>, Michael Barts <mba...@vt.edu> wrote:
>>
>>I've had the same thoughts about group delay distortion through the
>>crystal filters used for SSB. In Saben's book on SSB he mentions that
>>group delay distortion can result in distorted audio since speech is
>>made up of many frequency components, if the individual frequency
>>components do not arrive in time synch, due to group delay distortion,
>>the result is distorted audio. That's roughly the explaination from
>
>The distortion from unequal group delay affects modems (which is one
>reason why they have adaptive phase equalizers) but is not audible
>by ear when listening to speech. What *is* audible vis a vis filters
>is amplitude ripple, at least if it is bad enough.
Group delay is audible if it is bad enough. This is the *click-boom*
effect that plagues some loudspeaker designs, IE the path for the
high frequency components is a different length than the path for low
frequency components with the high frequency components typically
arriving first. This causes the "click" of a percussive drum strike
to arrive before the "boom" of the drum. Instead of kaboom, you hear
click <pause> boom. It's less noticable on speech, but it's there,
and adds to listener fatigue. However, the group delay through an
*RF* filter is so small compared to the time of an *audio* cycle
as to be insignificant. You can't hear it.
The much more serious effect for SSB is frequency mistranslation.
This causes harmonic dissonance, and the ear is very sensitive
to that. For example, a speech component that has a 16 Hz, 32 Hz,
and 64 Hz harmony, will be dissonant if the frequency translation
error is 10 Hz so that you now have the sequence 26 Hz, 42 Hz,
and 74 Hz. The tones are no longer harmonically related, and
sound dissonant to the ear. This is still noticable to the ear
at higher frequencies where the error is a smaller percentage.
You have to get the error down under a couple of Hertz for the
ear not to detect it. Otherwise, the harmonic "beat" will cause
a warble that's noticable to the ear. An example is a guitar that
is slightly out of tune, you can notice the beat even though it's
at a frequency below the audible range.
>I suspect the main culprit is the nonlinear distortion
>in the average SSB transmitter.
This is certainly a factor with amateur gear. The frequency
response of the entire chain, microphone, speech amp, compressor,
DSB generator, filter, and linear amplifiers, all have an effect
on speech quality. And similarly, all the stages of the receiver
also contribute. Aside from frequency mistranslation, the most
striking effects occur in the audio stages at both ends. For some
reason, amateur manufacturers seem to think 10% THD is a *desirable*
specification for the audio sections of their radios.
One of the best things you can do to improve the fidelity of your
radio is to improve and flatten the response of your audio stages. A
receiver audio stage that has flat response from 20 Hz to 20,000 Hz,
and that has 0.01% THD, will sound worlds better than one that is
constricted and distorted, and will lead to *much* less listener
fatigue. The same can be said for the transmitter audio stages.
In this regard, the use of audio filters on receivers should
strongly be discouraged. The group delay through such filters *is*
significant compared to an audio cycle, and *is* audible. Do your
selective filtering at RF where the group delay is an insignificant
portion of an audio cycle.
Robert Everitt Heiss
box used by professional audio engineers to add distortion to certain
musical tracks before the mix. Note also that some ham transmitters are
reported to have subjectively better audio with the processor turned on.
The gently saturating kind of distortion is similar to that produced
in the ear itself and is perceived as loud, not distorted.
Phase distortion is not audible, but may change the character of
subsequent nonlinear distortion. The result may be an unfamiliar
sound which is perceived as rude or ugly.
Another important kind of distortion, which I did not see mentioned, is
unintentional angle modulation. The worst example is an old modulated
oscillator AM rig, which has FM as well as AM. Audiophiles know about
this and prefer a separate tweeter to a whizzer cone mounted on a
woofer. You don't want the tweeter moving back and forth with the bass
sounds and changing the delay of the treble sounds. In a transmitter we
don't want microphonics or poor supply regulation to chirp one of the
oscillators or reactance-modulate a narrowband amplifier. What happens
to SSB theory when the transmitter carrier is not frequency and phase
stable?
Nitpicking aside, I disagree that SSB has to sound bad. Like many hams
I sometimes use a receiver with good audio, so I can hear the difference
between transmitters. Some are at least as clear as a digital telephone
connection -- am I too easily pleased? If it matters, the receiver has
a broad four-pole crystal lattice filter, no synthesizer, external DC
power, and cheap LM380 class-B audio output.
--
73, Rob KO6KA
rhe...@tuba.calpoly.edu
Scott Dorsey
>Bingo! It's the frequency translation error that gets you. If you have
>any frequency translation error, the harmonic relationships of the
>speech (or music) components no longer line up. So instead of nice
>harmonies, you have dissonance which is very grating to the ear.
>It's like listening to a musical instrument that isn't tuned on a
>well tempered scale.
Maybe I just don't see this as very much of a problem, since I use a
continuously tuned receiver with a BFO. If the tuning isn't spot on,
the frequency translation error is very audible and offensive, but
even with it set perfectly, the audio quality is poor.
Now, I am assuming that there is some sort of error such that there
is still not linear translation of frequencies across the band, and
that with the BFO set for 1:1 translation of midrange frequencies, the
high and low end may still be offset somewhat.
However, I don't have my copy of Terman handy here, or the ability to
do the math without. But are we getting close here?
WB3U
>Another important kind of distortion, which I did not see mentioned, is
>unintentional angle modulation. The worst example is an old modulated
>oscillator AM rig, which has FM as well as AM. Audiophiles know about
>this and prefer a separate tweeter to a whizzer cone mounted on a
>woofer. You don't want the tweeter moving back and forth with the bass
>sounds and changing the delay of the treble sounds.
This is also known as Doppler distortion, and I strongly question whether
whizzer cones suffer because of this, or rather because this type of
tweeter simply can't reproduce high frequencies. Despite assertions by
the "experts", I've never been convinced that this type of distortion is
discernible as "bad". Some of the early planar speakers had a superb
sound quality and only used a single diaphragm to cover a very wide range
of frequencies. Also, microphones don't have separate elements for
different frequencies - the highs are constantly modulated by the lows. As
far as I'm concerned, the only reason to break up the reproducers into
different units is to better match the frequency response and power
handling capability of each type speaker.
>Nitpicking aside,
Not before I mention that my personal favorite is Transient Intermodulation
Distortion. Yeah, that must be the problem with SSB, T.I.M. . . .
>If it matters, the receiver has a broad four-pole crystal lattice filter,
*That's* why it sounds better. Now, if it only had some output tubes instead
of that LM380 . . . <g>
73,
Jack WB3U
Unknown
> Not before I mention that my personal favorite is Transient Intermodulation
> Distortion. Yeah, that must be the problem with SSB, T.I.M. . . .
>
response of a typical SSB sideband reject filter...what do the overshoots
from fast wavefront voice peaks do to the following stages?
Joe
WA3CMQ
Unknown
Zack Lau (KH6CP)
: I suspect the main culprit is the nonlinear distortion
: in the average SSB transmitter.
I agree with this. Since I talk with lots of hams and
have a distinctive voice, I decided to rely on voice
recognition, rather than speech processing, to help make
lots of QRP phone SS contacts. I remember getting at least
one report last year that the audio "was just like me."
I used a class A FET final to reduce IMD. It looked
awfully good on a spectrum analyzer.
As a rule of thumb, for a clean signal a SSB amp ought to
be run at under 50% of maximum output on voice peaks...
--
Zack Lau KH6CP/1 2 way QRP WAS
8 States on 10 GHz
Internet: zl...@arrl.org 10 grids on 2304 MHz
Scott Dorsey
I actually did once try measuring the impulse response of the Collins
mechanical filters, and it was so incredibly nasty that I just packed
the whole project in. You've never seen so much ringing in your life...
Ted Viens
>
>
>> Not before I mention that my personal favorite is Transient
Intermodulation
>> Distortion. Yeah, that must be the problem with SSB, T.I.M. . . .
>>
>Maybe Jack's on to something here...has anyone characterized the
impulse
>response of a typical SSB sideband reject filter...what do the
overshoots
>from fast wavefront voice peaks do to the following stages?
>
>WA3CMQ
Some of us have been lucky enough to play with receivers using IF
DSP's. When you have available highly adjustable digital IF Bandpass
filters, digital demodulation, and one Hertz tuning steps, SSB sounds
like a guy talking across the table (given a decent multi-thousand
dollar transmitter.) In just 5 or 10 years, when IF DSP technology is
affordable on any decent receiver, we will look back on this discussion
and laugh about the good ole days...
--
Bye... Ted..
Deep in the Heart of the Armpits of Houston, Texas...
Joanne Dow
>Like many hams, I predominately send (when I had a TX) and receive SSB. However,
>to my mind, the sound quality is not as good as an AM transmission, using a 6KHz
>bandwidth for AM. Why?
>
>
>copies of the sidebands. So SSB's more efficient use of bandwidth should not mean
>a redection in quality.
>
>amateur AM signal. Obviuosly one would not expect it to sound like a commercal AM
>signal, where the bandwidth exceeds the 300-3000Hz.
>
>
>bandwidth, its most unlikely anyone else can use the space between the carrier and the
>signal sound better.
>
If you include the full audio bandwidth SSB will sound more "natural". It will
not perform as well under noisey conditions, though. As it turns out the
majority of the "intelligibility" of speech is in the higher frequencies of the
speech spectrum, above 1kHz. So if you accentuate them slightly you may get a
harsher sounding result; but, that result will cut through noise much better
than one with most of the power wasted in bass notes. Most users of SSB intend
to get their signals through rather than sound natural. "Natural sound" is an
"FM thing." (Which should bring a few people down on me I bet. {^_-})
Another thing that will degrade the "fidelity" of SSB speech is the MUCh higher
level of speech processing that is usually applied to the signals. ALC is only
part of the game. Another part performed by some radios involves clipping the
generated SSM speech and then refiltering to remove the clipping generated
frequencies outside the signal passband. This can also significantly increase
the percentage of the PEP used for the speech. Again, sounding better is not the
usual design goal. The usual design goal is intelligibility rather than
"fidelity".
A third factor is the number of hams who overdrive their rigs way beyond what is
necessary. A pegged ALC meter is NOT a sign of good operating practice. {^_-}
>2) Its very difficult to re-insert the carrier at precisely the correct frequency at
>the receiver. I've recently used a 2m transceiver as a tuneable IF for an HF receiver
>I'm biliding, The fact the tranceiver only steps in 100Hz steps definately causes
>poor quaility audio, but I've noticed this with VFO's too.
>
Over time you generate an ear for it. I have noodled out a design that would
work to make the speech harmonics actually appear as harmonics of each other.
But I figured this is more circuitry than is really needed, especially with
modern synthesized radios. (If they are calibrated well and have good TCXOs
driving them being 10Hz off is kinda "bad" at 14MHz or even 28MHz. And at 10Hz
off things do not sound particularly bad. Even at 100Hz off intelligibility does
not suffer much.) Reinjected carriers, even 20dB down, is a workable idea. It
has never caught on because there are enough hetrodynes, sideband splatters, and
other junk floating around on 20 meters, for example, without having your
neighboring hams tossing in their own for good (bad?) measure.
>
>3) Perhaps there need to re-inser the carrier at the correct amplitude, or
>in the correct phase. I've never looked at the maths to say if this is true.
Depends on how "hi-fi" you want things. Again, the goal is intelligibilty not
fidelity. I've performed enough tests myself and with others to know that these
two concepts are at best "loosely" related and at worst inversely related.
>
>Any thoughts?
>
>
>dave kirkby G8WRB.
>
{^_^} Joanne Dow The Wizardess
Joanne Dow
...
> Amplitude doesn't matter, except in obtaining proper performance of the
> demoduator, i.e. too much/too little carrier power. In my text books
> they always perform synchronous detection, i.e. the phase of the carrier
> at the receiver is identical to that at the transmitter. If the phase
> is not identical then when demodulation takes place the amplitude of
> the demodulated signal is weighted by the cos/sin of the phase error.
> This requires a phase tracking loop (& frequency) which greatly increases
> the complexity of the receiver as opposed to a simple VFO. I have never
> understood how we get rid of the "phase error" with our equipment - unless
> it is done by "manually tuning in" the signal. Anyway, a resolution of
> better than 10 Hz is needed to get satisfactory performance (unless using
> a channelized system, e.g. like the 5 kHz steps on the 2 mtr band).
>
>
>Bill Kirkland
Phase of the LO is not nearly the problem you'd like to make of it. I have,
under strong signal conditions with neighbors, had them crank off on their audio
gain, turn off speech processors, and run a more "linear" opreation. Then I have
tuned carefully. When within a few Hz of their suppressed carrier things sound
pretty decent. Once I toss in a little audio shaping I can make it sound pretty
much like being face to face with them. This does not "get out" worth chicken
droppings. Once he cranked it back up to normal things sounded much more "ham"
than "broadcast." And once he turned on the processor things really went to
<brown steaming stuff>. But THAT signal performed very well on 20. It cut
through the interference and noise VERY well.
Joanne Dow
> Those wishing to pursue the "whys" of the problem of poor fidelity
>in SSB voice transmission might want to consider that in FM stereo
>broadcasting the L-R audio is broadcast as a DSBSC (double sideband
>suppressed carrier) signal centered at 38 kHz. In the receiver, this
>DSBSC signal is demodulated (using the 19 kHz pilot tone as a frequency
>and phase reference) and the resulting L-R audio combined with the L+R
>audio from the main channel, resulting in separate L & R signals. The
>FCC technical standard for phase coherence between the main channel
>and stereophonic subchannel is +/- 3 degrees (which is satisfied by
>achieving a stereo separation of 29.7 dB) (Paragraph 73.322(m) of
>Part 73).
>
> The key differences are (a) a DSB signal requires no filtering, and
>(b) in FM broadcasting a 19 kHz pilot tone is transmitted which must have
>a frequency accuracy of +/- 2 Hz. In addition, there is a phase-coherency
>requirement which is "built in" to the transmitter, and requires careful
>attention throughout the entire audio chain since the FCC specification
>applies to the entire path from microphone input to transmitter output.
>
>73,
>
>Paul, K4MSG
If you play with the trigonometry involved you'll find that DSB is a *VASTLY*
different kettle of fish than SSB. With DSB a modest phase error can translate
to a significant amplitude error on demodulation. With SSB this is most
defintely not the case.
Again, the BIG difference in fidelity between point to point SSB stations, which
are basically feeds for remote "commercial" broadcasts as often as not, and that
used by the military and hams comes from the demands of the respective services.
Point to point is demanding fidelity. So they work to achieve and maintain
fidelity. They get quite poor peak to average power ratios. So their signal's
"punch" must come from BIG amplifiers and antenna systems - and a nice clear
channel. For the ham and military and other COMMUNICATIONS use (as opposed to
"broadcast" use) the operative requirement is intellgibility. Without the big I
all the fidelity in the world does you no good at all. Processing the signal to
accentuate frequencies above 1KHz a few dB can make a wonderful difference in
how you get through at the expense of making you sound slightly "odd". (Er some
call it ridiculous.)
(W6SPK is a close friend of mine in the So Cal area. He has a nice deep voice.
He has had several different radios and mikes for use on the 440MHz repeater we
often use. The closer to "arog" he sounds the worse he "punches through" the
average road noise in my VW bug convertible. When he uses one of the
combinations that clips off most of the bass I have no problems at all
understanding what he says. But he surely does not sound "normal." And he rather
DISLIKES that. {^_-})
Paul H. Bock
>In article <phb.801507634@melpar>, p...@syseng1.melpar.esys.com (Paul H. Bock) wrote:
>> Those wishing to pursue the "whys" of the problem of poor fidelity
>>in SSB voice transmission might want to consider that in FM stereo
>>broadcasting the L-R audio is broadcast as a DSBSC (double sideband
>>suppressed carrier) signal centered at 38 kHz. In the receiver, this
>>
(snip)
>> The key differences are (a) a DSB signal requires no filtering, and
>>(b) in FM broadcasting a 19 kHz pilot tone is transmitted which must have
>>a frequency accuracy of +/- 2 Hz. In addition, there is a phase-coherency
(snip)
>If you play with the trigonometry involved you'll find that DSB is a *VASTLY*
>different kettle of fish than SSB. With DSB a modest phase error can translate
(snip)
All very true, m'lady.
>Point to point is demanding fidelity. So they work to achieve and maintain
>fidelity. They get quite poor peak to average power ratios. So their signal's
>"punch" must come from BIG amplifiers and antenna systems - and a nice clear
>channel. For the ham and military and other COMMUNICATIONS use (as opposed to
>"broadcast" use) the operative requirement is intellgibility. Without the big I
Well, I was waiting for someone to make that point, actually. In military
communications especially (with which I have a couple of decades of experience)
intelligibility is *all* that matters. Personally, I feel the same about
ham gear, but others may not - in fact, I'm sure there are hams who are searching for fidelity without regard to intelligibility under poor signal conditions.
My standard for a voice transmitter is "how well does it 'punch through'
under marginal signal conditions," and I take it that you use the same or
a similar benchmark. So, while I did kibitz a bit on the "fidelity" issue
I really regard it as irrelevant in a ham transmitter, but that's probably
due to my military background and the fact that, also being a a former AM/FM
broadcaster, I never regrded ham radio as being a "broadcast" medium (see
related threads on this in r.r.a.p ;-) )
Having said that, however, there is no doubt that the diffrence between
military and/or civil emergency comms and ham comms is that the ham can *make
a choice* about fidelity versus intelligibility, whereas the military et al
cannot. So, if you *want* to sound "really spiffy", go for it!
73,
Paul, K4MSG
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
* Paul H. Bock, Jr. K4MSG Principal Systems Engineer *
* E-Systems/Melpar Div. Internet: pb...@melpar.esys.com *
* Falls Church, VA Telephone: (703) 560-5000 x2062 *
* *
* If it's not Baroque, don't try to fix it; *
* if it is Baroque, make sure you can Handel it first. *
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
Charles Mccurdy
of any body ever useing am on 2mtrs just fm and ssb is what i like.
--
|Fidonet: Charles Mccurdy 1:106/1492
|Internet: Charles...@f1492.n106.z1.fidonet.org
|
| Standard disclaimer: The views of this user are strictly his own.
Dana Myers
>In article <3qa823$n...@netaxs.com> "ro...@netaxs.com" <Joe Rocci> writes:
>>
>>> Not before I mention that my personal favorite is Transient Intermodulation
>>> Distortion. Yeah, that must be the problem with SSB, T.I.M. . . .
>>>
>>Maybe Jack's on to something here...has anyone characterized the impulse
>>response of a typical SSB sideband reject filter...what do the overshoots
>>from fast wavefront voice peaks do to the following stages?
>
>mechanical filters, and it was so incredibly nasty that I just packed
>the whole project in. You've never seen so much ringing in your life...
Well, that's interesting, but not surprising. When will someone
invent phase linear mechanical filters?
;-)
---
* Dana H. Myers KK6JQ, DoD#: j | Views expressed here are *
* (310) 348-6043 | mine and do not necessarily *
* Dana....@West.Sun.Com | reflect those of my employer *