UK Vintage Radio Repair and Restoration Powered By Google Custom Search Vintage Radio Service Data

Go Back   UK Vintage Radio Repair and Restoration Discussion Forum > General Vintage Technology > General Vintage Technology Discussions
Register FAQ Forum Rules Calendar Mark Forums Read

Notices

General Vintage Technology Discussions For general discussions about vintage radio and other vintage electronics etc.

Closed Thread
 
Thread Tools
Old 29th Mar 2010, 1:06 am   #21
Synchrodyne
Octode
 
Join Date: Jan 2009
Location: Mt. Maunganui, New Zealand
Posts: 1,070
Default Re: FM "deviation"

I have a recollection that one of the Japanese manufacturers did some years back offer an FM tuner that used an SSB stereo decoder. Apparently the rationale was that using the lower sideband only of the subcarrier offered an improved signal to noise ratio. What the actual technique was I don’t know, although I doubt that it was done with filters. Perhaps it was done by quadrature demodulation followed by phase-shifting and summing-and-differencing. I do not recall who the manufacturer was, and I haven’t kept the reference, but it sounds like something that Pioneer, Technics or Sansui might have done. Certainly Sansui used similar techniques, albeit for AM reception not FM stereo, on its 1980s multisystem AM-stereo tuners and for sideband selection on its TU-X1.

Cheers,
Synchrodyne is offline  
Old 29th Mar 2010, 3:11 pm   #22
G8HQP Dave
Nonode
 
Join Date: Sep 2008
Location: St Neots, Cambridgeshire, UK.
Posts: 2,782
Default Re: FM "deviation"

If you don't use filters then you would have to use something like a phasing sideband receiver, but again this introduces phase shifts where you really want none. To do this you need an accurate 90 degrees over 23-53kHz, or equivalently P and P+90 in two paths, combined with an accurate P over 0-15kHz. I guess the Japs were smart enough to do this, but what about component ageing? You gain a little signal-to-noise, but potentially lose stereo separation and image clarity. My guess is that it might measure well and could have impressive advertising, but sound awful with instruments moving around for different notes. Not like other 1980's hi-fi then!
G8HQP Dave is offline  
Old 29th Mar 2010, 10:48 pm   #23
dominicbeesley
Octode
 
dominicbeesley's Avatar
 
Join Date: Nov 2004
Location: Hebden Bridge, West Yorkshire, UK.
Posts: 1,470
Default Re: FM "deviation"

The more I think about the stuff the more it makes my brain hurt...fascinating though! I wish I had more time...I'd dig out my maths books and get my head back around all those "identities" for transforming complex cos/sin/e functions and try and work out a clever decoder I wonder would a 52, 26 or 13uS delay line help....just an idle though

Dom
dominicbeesley is offline  
Old 30th Mar 2010, 12:05 am   #24
Synchrodyne
Octode
 
Join Date: Jan 2009
Location: Mt. Maunganui, New Zealand
Posts: 1,070
Default Re: FM "deviation"

Quote:
Originally Posted by G8HQP Dave View Post
If you don't use filters then you would have to use something like a phasing sideband receiver, but again this introduces phase shifts where you really want none. To do this you need an accurate 90 degrees over 23-53kHz, or equivalently P and P+90 in two paths, combined with an accurate P over 0-15kHz. I guess the Japs were smart enough to do this, but what about component ageing? You gain a little signal-to-noise, but potentially lose stereo separation and image clarity. My guess is that it might measure well and could have impressive advertising, but sound awful with instruments moving around for different notes. Not like other 1980's hi-fi then!
Perhaps longer-term stability was less of a problem with 1980s technology? In the HF world, I have a couple of Sony ICF2010 receivers, and a Liniplex F2 (excellent piece of equipment) that use the phasing method for sideband selection, and none seems to have suffered any deterioration over time in terms of sideband separation.

Quadrature modulation of the FM Stereo 38 kHz subcarrier was also proposed in respect of some of the various quadraphonic (or should it be tetraphonic to be homologous with stereophonic) broadcasting ideas, and also in connection with the FMX noise reduction scheme.

Cheers,
Synchrodyne is offline  
Old 30th Mar 2010, 9:33 am   #25
G8HQP Dave
Nonode
 
Join Date: Sep 2008
Location: St Neots, Cambridgeshire, UK.
Posts: 2,782
Default Re: FM "deviation"

Quote:
Perhaps longer-term stability was less of a problem with 1980s technology? In the HF world, I have a couple of Sony ICF2010 receivers, and a Liniplex F2 (excellent piece of equipment) that use the phasing method for sideband selection, and none seems to have suffered any deterioration over time in terms of sideband separation.
There is a world of difference between maintaining accurate P and P+90 over 300-3kHz for communications purposes, and the requirements of a decent stereo receiver. Probably the best method for communications is polyphase, but that would be useless for stereo as P is very frequency dependent. Getting accurate phase and amplitude tracking of three signals at F, 38-F and 38+F is hard! Nowadays I suppose you could do it with a digital filter.

Quadrature modulation would be OK, provided that it was an accurate 90 degrees in both transmitter and receiver in order to avoid crosstalk.
G8HQP Dave is offline  
Old 30th Mar 2010, 8:24 pm   #26
Hybrid tellies
Octode
 
Hybrid tellies's Avatar
 
Join Date: Jan 2004
Location: Bristol, UK
Posts: 1,794
Default Re: FM "deviation"

I never did FM theory at coolege. But I think this topic has cleared up a few things for me. With FM the amplitude is constant, deviation is limited to 75Khz, so the modulation frequency varies the carrier so many times +- 75Khz. Ok clear to this point but still not sure what de emphasis is all about and how is the different levels of amplitude or loudness of the modulated signal effect the FM signal. Sorry if this is obvious to some but have never been totally 100% with FM theory.
__________________
Simon
BVWS member
Hybrid tellies is offline  
Old 30th Mar 2010, 10:36 pm   #27
kalee20
Dekatron
 
Join Date: Feb 2007
Location: Lynton, N. Devon, UK.
Posts: 3,004
Default Re: FM "deviation"

Maximum deviation for FM broadcast is 75kHz as you say. But if the programme material is quiet, (low audio amplitude), then the carrier frequency gets deviated a lot less than 75kHz.

Pre-emphasis and de-emphasis is just about treble boost of the audio before it gets to the modulator in the transmitter, and treble cut after the discriminator in the receiver. This improves the signal-to-noise ratio, because the treble cut also cuts down the background hiss.
kalee20 is offline  
Old 31st Mar 2010, 4:34 am   #28
Synchrodyne
Octode
 
Join Date: Jan 2009
Location: Mt. Maunganui, New Zealand
Posts: 1,070
Default Re: FM "deviation"

FM has an inherent triangular noise characteristic, with noise increasing at 6 dB/octave with increasing modulating frequency. One way of looking at this is that for any given deviation level, as the modulating frequency increases, the modulation index decreases. And the higher the modulation index, the better the signal-to-noise ratio, and vice versa.

To achieve a rectangular noise characteristic, as for example occurs with AM, then the use of a 6 dB/octave AF roll-off in receivers is necessary. In turn this requires mirror-image boosting (pre-emphasis) at the transmitter.

Some narrowband FM communications systems simply use 6dB/octave slopes across the voice frequency communications band (300 to 3400 Hz). Given that in this case heavy limiting and peak clipping are acceptable, such an approach is workable. However, for broadcasting, program energy distribution does not allow the application of a simple 6 dB/octave constant slope across the whole audio band, say 40 to 15 000 Hz. The compromise is to select a mid-frequency turnover point beyond which the 6 dB/octave slopes apply, and below which, where the response is flat, the noise is not obtrusive anyway. The two standard pre-emphasis/de-emphasis curve numbers are the “European” 50 microseconds (3.18 kHz turnover) and the “American” 75 microseconds (2.12 kHz turnover point). 25 microseconds was also proposed for use with Dolby B noise reduction, but I don’t think its use ever became widespread.

Even so lack of treble headroom could be a problem. In an earlier version of this world, before the Optimod era (optimum for whom, I wonder?) I seem to recall that the BBC used variable pre-emphasis limiting (on Radio 3 at least) as an alternative to compression.

Cheers,
Synchrodyne is offline  
Old 31st Mar 2010, 11:09 am   #29
G8HQP Dave
Nonode
 
Join Date: Sep 2008
Location: St Neots, Cambridgeshire, UK.
Posts: 2,782
Default Re: FM "deviation"

Is this the right point in the discussion to bowl in a googly? Namely, that in FM the carrier amplitude reduces with modulation (although the total signal amplitude remains constant) while in AM the carrier amplitude remains constant (although the total signal amplitude varies). I remember being quite shocked, then amused, when I first realised this.

Just when you all thought you were beginning to understand it!!!
G8HQP Dave is offline  
Old 31st Mar 2010, 11:29 am   #30
dominicbeesley
Octode
 
dominicbeesley's Avatar
 
Join Date: Nov 2004
Location: Hebden Bridge, West Yorkshire, UK.
Posts: 1,470
Default Re: FM "deviation"

I'm going to have to get this Spectrum Analyser finished - then I can look at all this stuff!

Not sure I'm understanding what you mean above - do you mean the power at the centre frequency reduces (goes to the side bands). i.e. if I modulated a DC or very low frequency there would be no power at centre frequency the power would shift all into the sideband?

Cheers

Dom
dominicbeesley is offline  
Old 31st Mar 2010, 1:45 pm   #31
G8HQP Dave
Nonode
 
Join Date: Sep 2008
Location: St Neots, Cambridgeshire, UK.
Posts: 2,782
Default Re: FM "deviation"

Yes. In FM some of the power shifts into the sidebands. At a modulation index around 2.41 (e.g. 1kHz baseband, 2.41kHz deviation) the carrier disappears, then at higher values of m it goes antiphase. You would get a greater effect at very high mod index, as you suggested (e.g. 100Hz baseband, 75kHz deviation gives m=750). This is because Bessel functions oscillate but get smaller at higher values. See Wikipedia, and the article on Bessel functions it links to. You could skip the Bessel maths and just look at the graph of the first few Bessel functions J0, J1 etc. The carrier amplitude is given by J0, first sideband by J1 etc.

I don't know what the minimum bandwidth of your analyser will be but if it is sufficiently narrow you could see this effect.
G8HQP Dave is offline  
Old 31st Mar 2010, 1:47 pm   #32
jimmc101
Heptode
 
jimmc101's Avatar
 
Join Date: Mar 2009
Location: Portsmouth, Hampshire, UK.
Posts: 520
Default Re: FM "deviation"

This is FM, nothing is that simple!
The carrier amplitude is not a simple function of the modulation index (m), it varies as a Bessel function.
At certain values of m the carrier disappears, the so called Bessel nulls, just past these nulls the carrier increases to some value before again decreasing to zero at the next null. The values of m for the first 3 nulls are 2.405, 5.520 & 8.654.
Incidentally this provides an accurate method of calibrating deviation if you have a spectrum analyser.
See for example http://cp.literature.agilent.com/lit...988-5677EN.pdf (hint 6)

Oops Dave beat me to it.
I should add that for FM the total power is constant so as Dave has said the total sideband power must come from a reduction in carrier power.

Jim

Last edited by jimmc101; 31st Mar 2010 at 1:52 pm. Reason: Last para
jimmc101 is offline  
Old 31st Mar 2010, 2:58 pm   #33
jimmc101
Heptode
 
jimmc101's Avatar
 
Join Date: Mar 2009
Location: Portsmouth, Hampshire, UK.
Posts: 520
Default Re: FM "deviation"

If anyone is interested in plumbing the murky depths of Bessel functions and FM sidebands, Excel has the Bessel function as part of the analysis pack. ( It's not installed by default. For Excel 2000 it's Tools > Add-Ins > Analysis ToolPak)

I've attached a simple spreadsheet for calculating the amplitude of the first 10 sidebands relative to the unmodulated carrier for a number of values of modulation index. (Sheet1 absolute, Sheet2 dB)

Jim
Attached Files
File Type: zip FM sidebands.zip (8.8 KB, 23 views)
jimmc101 is offline  
Old 2nd Apr 2010, 2:49 am   #34
Synchrodyne
Octode
 
Join Date: Jan 2009
Location: Mt. Maunganui, New Zealand
Posts: 1,070
Default Re: FM "deviation"

Quote:
Originally Posted by G8HQP Dave View Post
There is a world of difference between maintaining accurate P and P+90 over 300-3kHz for communications purposes, and the requirements of a decent stereo receiver. Probably the best method for communications is polyphase, but that would be useless for stereo as P is very frequency dependent. Getting accurate phase and amplitude tracking of three signals at F, 38-F and 38+F is hard! Nowadays I suppose you could do it with a digital filter.
Agreed that the examples I have quoted, basis familiarity, would not necessarily scale up to wide band audio.

Still, the PLL, phase-shifting and matrixing circuitry used in the Sony ICF2010 was in fact taken direct from that which Sony developed, inclusive of dedicated ICs, for AM stereo hi-fi tuners with wideband audio (and which was also used in the SRF-A100 portable radio receiver). Phase-shifting and matrixing was required for full decoding of the Kahn-Hazeltine ISB AM stereo system. Sansui was amongst others who developed multiple system AM Stereo decoders inclusive of wideband phase-shifting and matrixing. It also used the same techniques in its landmark TU-X1 tuner for AM sideband selection. This predated its AM Stereo models, and evidently remains a much sought after model in the second-hand market. To a large extent that may be a reflection of its FM performance, although it is usually noted as having exceptional AM performance. Sansui TU-X1 original literature is available at several places on-line, such as at: http://www.goodsoundclub.com/Forums/...px?postID=4649

PLL/phase-shifting/matrixing circuits for AM sideband separation have also appeared in home construction magazine articles, including by Hershberger (Popular Electronics April, 1982) and Brook (Electronics & Wireless World September, 1989), both authors having, I think, good credentials. The Hershberger design included relatively elaborate three-stage phase-shifters that were said to be accurate to plus/minus 3 degrees over the band 50 to 12 000 Hz, and provided at least 31 dB of unwanted sideband rejection. The need for appropriate component selection and matching – including the avoidance of carbon composition resistors – was heavily stressed. The Brook design was simpler in this aspect, being of medium bandwidth, namely 7 kHz.

How well any of these systems worked or how well they maintained their performance over time I am not in a position to know. But it seems not unreasonable to suppose that those equipment makers who had developed what they believed, however erroneously, were adequate AM stereo decoding circuits inclusive of phase-shifting and matrixing following quadrature PLL synchronous demodulation of a (typically) 455 kHz carrier would not, had they so chosen, have applied the same techniques to similar treatment of the 38 kHz FM subcarrier with the objective of extracting the lower sideband. In fact PLL synchronous demodulation of the subcarrier had become pretty much normal from the early 1970s, following the availability of the RCA CA3090 IC and more so the Motorola MC1310 IC and its many derivatives. So moving to quadrature demodulation would not have been such a big step. In the FM stereo case, where the objective would be to reduce the influence of the noisier upper sideband, the crosstalk target might even have been milder than in the AM case, where either stereo separation was directly affected, or it was desired to suppress a sideband suffering from interference. Here’s an interesting question – clearly random and unequal phase-shifting errors in the respective I and Q demodulated signal paths would affect the relative amounts of the lower and upper sidebands that found their way into the nominally lower sideband (L-R) signal after matrixing, but would the net signal then have major phase errors that in turn would compromise the final matrixing with the baseband (L+R) signal? Or would it be simply the case that the LSB/USB mix in (L-R) varied somewhat over the audio band, but basic (L-R) integrity was retained.
.
I still have not been able to trace the make and model of the FM tuner that did use SSB subcarrier demodulation, but surely it will come to light sooner or later.

Cheers,
Synchrodyne is offline  
Old 2nd Apr 2010, 2:58 am   #35
Synchrodyne
Octode
 
Join Date: Jan 2009
Location: Mt. Maunganui, New Zealand
Posts: 1,070
Default Re: FM "deviation"

On FM theory, for those who don’t mind older publications, there is a nice, gentle treatment, with graphs and diagrams, in the book “V.H.F. Radio Manual”, by P.R. Keller, Newnes, 1957. I imagine that it would be typically available second-hand on ABE. It includes diagrammatic representation of the triangular FM noise spectrum and the effects of pre- and de-emphasis.

Cheers,
Synchrodyne is offline  
Old 6th Apr 2010, 5:31 pm   #36
G8HQP Dave
Nonode
 
Join Date: Sep 2008
Location: St Neots, Cambridgeshire, UK.
Posts: 2,782
Default Re: FM "deviation"

Quote:
The Hershberger design included relatively elaborate three-stage phase-shifters that were said to be accurate to plus/minus 3 degrees over the band 50 to 12 000 Hz, and provided at least 31 dB of unwanted sideband rejection. The need for appropriate component selection and matching – including the avoidance of carbon composition resistors – was heavily stressed.
That is pretty impressive performance from only three stages! I can understand why they wanted to avoid composition resistors.

Quote:
So moving to quadrature demodulation would not have been such a big step.
Yes, getting a 38kHz quadrature carrier is the easy bit - you just add a few extra transistors to your PLL chip.

Quote:
Here’s an interesting question – clearly random and unequal phase-shifting errors in the respective I and Q demodulated signal paths would affect the relative amounts of the lower and upper sidebands that found their way into the nominally lower sideband (L-R) signal after matrixing, but would the net signal then have major phase errors that in turn would compromise the final matrixing with the baseband (L+R) signal? Or would it be simply the case that the LSB/USB mix in (L-R) varied somewhat over the audio band, but basic (L-R) integrity was retained.
That is an interesting point - it may be why they could do it. Errors in the image-cancelling demodulator would, if not too bad, mainly just reduce the sideband rejection which would merely re-introduce some of the upper sideband noise which you are trying to reject. The wanted lower sideband signal would still be OK - so we mainly need accurate F and 38-F phasing, and not quite so accurate 38+F phasing. It is still hard, but perhaps not quite as hard as I first thought. A really clever designer might even be able to balance off errors in the two sidebands, as the main aim is to get an accurate L-R signal and noise reduction is secondary.
G8HQP Dave is offline  
Old 10th Apr 2010, 2:41 am   #37
Synchrodyne
Octode
 
Join Date: Jan 2009
Location: Mt. Maunganui, New Zealand
Posts: 1,070
Default Re: FM "deviation"

Thanks for that, Dave.

Something else that since occurred to me is that the degree of upper sideband rejection required would not be all that great. The noise increases at 6 dB/8ve, so across the subcarrier and its sidebands, there would be maybe an 8 dB increase in noise level from 23 to 53 kHz. Accordingly, 10 or so db rejection might be enough. Also, the rejection would not need to be “flat” across the sideband, if the objective is simply to push upper sideband noise level down to a level comparable to that of the lower sideband. Overall, that would seem to ease the phase-shifting burden.

But then why not simply put the subcarrier and sidebands through a mild filter with 6 dB/8ve downslope from 23 to 53 kHz and synchronously demodulate. In fact with FDM decoders, I think it is normal to apply the de-emphasis to the subcarrier ahead of demodulation using a bell filter, so it would just be a case of adding the tilt to that filter, and ensuring that the tilt, if not exactly 6 dB/8ve linearly across the band, was symmetrical about 38 kHz.

Cheers,
Synchrodyne is offline  
Closed Thread

Thread Tools



All times are GMT. The time now is 8:29 pm.


All information and advice on this forum is subject to the WARNING AND DISCLAIMER located at http://www.vintage-radio.net.
Failure to heed this warning may result in death or serious injury to yourself and/or others.


Powered by vBulletin®
Copyright ©2000 - 2014, vBulletin Solutions, Inc.
Copyright ©2002 - 2014, S&P Technology.