In article , tony sayer

wrote:
In article , Jim Lesurf jcgl@st-
and.demon.co.uk writes



The problem isn't the generator, it is the effect of a finite
transmission bandwidth on an FM signal. This means that no matter how
good the generator or RX, there will be a lower limit to the distortion
for high levels of distortion which will be way above the values you
quote.

My recollection of this may be hazy as it is some years since I did
this, however it is that the levels of nonlinearity due to the finite
transmission bandwidth rise, and are particularly a problem for the L-R
component due to the subcarrier components at HF.

Yes so it is, but its to a degree.

Well, the "degree" can mean levels of distortion well above 1 percent. My
point is that I am talking about the nonlinearity produced by the finite
transmission bandwidth for the FM stereo system. Hence this is nothing to
do with tuner design as such.

No of course FM isn't perfect by any means but I think it puts up a very
good performance against the current implementation of DAB in the UK..

I agree that it does/can - but on the basis I explained. i.e. that the
modilation level is generally kept well below the permitted peak deviation.
The problem is that this tends to eat away some of the dynamic range
available to FM.



*300Hz 30% mono THD. However if you do this, you find you have
increased the
distortion for some other form of input modulation pattern. My perhaps
unreliable recollection is that trying to get a genuine (i.e. with a
band limited input) result much below 0.1 to 0.2 percent for mono was
misleading as real stereo TX modulations would end up being worse than
this, and you might make the stereo performance worse by tweaking to
get
an apparent low distortion for a mono test signal.

I think receivers may have come along a bit since that!..


My point was that I am not referring to the levels of nonlinearity which
are determined by tuner design. Thus this isn't a matter of having a 'more
linear' tuner. It is that some designers may have made the tuner
nonlinearity to null a specific signal's distortion, but this would be at
the expense of other signal waveforms becoming more distorted.


[snip]


The problem here is that the R3 engineers have a more limited dyamamic
range for FM in reality than is available for CD. Thus the tendency to
keep the modulation well below peak to avoid noticable peak/HF
distortion, and then the temptation to level-compress to avoid noise or
loss of audience.

I don't think its done for that reason, its done for people in cars and
other compromised listening environments..

To some extent, yes. But the problems remain even for those who do not.

[snip]


What depth do you mean in an FM system?..

Modulation depth. The tradition was to use 30% in magazine reviews
(when they bothered to actually measure tuners). Full mod (100%) would
correspond to 75kHz, nominally.

It have a measure up with a good exciter and couple or three tuners just
to see what is the current standard.

If you are going to do some measurements then I'd suggest:

1) Ensuring the output from the generator is band-limited to give no
sidebands outwith the 200kHz nominal band permitted.

2) Try L-R and Ronly and Lonly for high modulation depths, preferrably as
HF intermod.



I suspect that. like myself, you have measured the distortion of
more than one FM RX whilst trying to align or tweak it, or just to
see if it was working as it should. it is easy enough to get THDs of
the order of 0.2 percent for 300Hz 30 percent mod mono. But when you
then measure higher (signal) levels, etc, the results can be
somewhat different. Ditto for HF intermod or L+R and L-R intermod.

Haven't done that for quite somewhile but a Denon we're using as an
RBR receiver was down to .06% at 10 K at 50 K dev a while ago....

Erm... The second harmonic of 10k is 20k. What components were you
expecting to get through the MPX filtering? Or were you including
intermod with the pilot tone, etc?

Well the residual was in the noise in that instance..

But for even a tuner whose front end was dreadful, the harmonics of 10k
would have been removed by a good MPX filter. If you want to problem the HF
performance then you would need to use intermod, or the methods I outline
above.



Thus even though such tuners are indeed excellent ones, the performance
in real use is perhaps not as impressive as the bench measurements may
indicate.

I think that tuner design has come some what I recent years.

They probably have. :-) However the Bessel functions remain the same, as
does the sideband pattern created by a given FM modulation... ;-


One does wonder about the sideband issue and the practical effect it has
compared to the encoding of DAB...

Provided the modulation level (primarily for the L-R) is kept well below
75kHz then the distortion can be expected to be too low to be audible.
However it may well affect musical peaks. The problem is that if you then
use modulation levels well below 75kHz and want the quiet levels to be well
above noise you start having to avoid a large part of the 70dB-ish range
available for FM.

Slainte,

Jim

--
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Audio Misc http://www.st-and.demon.co.uk/AudioMisc/index.html
Armstrong Audio http://www.st-and.demon.co.uk/Audio/armstrong.html
Barbirolli Soc. http://www.st-and.demon.co.uk/JBSoc/JBSoc.html