FM multipath issues

[stevehertz]

I know what multipath distortion is wrt FM reception, I don't need a basic primer, but here's a particular quirk that I would like to understand more about.

A Trio 600T hifi tuner has the normal signal strength and tuning meters, but it also has a multipath meter. With a huge power station nearby, we do suffer from multipath issues. On the stations that are showing multipath signal reception - and believe me, that MP meter can fair dance around at times - I find that if I tune in carefully I can limit the amount of multipath being displayed by the MP meter. However, the tuning meter is invariably off centre for this to be the case. Does this suggest that the multipath signal (of the same station?) is somehow being received at a slightly different frequency? What is happening?

Also, as I understand it, best quality stereo ie least distortion is achieved when, on a properly set up and aligned tuner, the tuning meter is in the centre. TBH, unless the tuning meter is considerably off centre I cannot detect any appreciable increase in distortion when tuned this way, although I concede that theory says that that is the case. So tuning slightly off centre to reduce MP input does not seem to be a bad thing. I guess it's all a compromise once you start tuning off centre.

In the photo the meters are L to R, Signal strength, Tuning and Multipath. NB the Multipath meter can be switched to be a Deviation meter, but it's in MP mode in the photo. Tuning is on centre as you can see, and the MP signal level is considerable.

[Nuvistor]

Non technical answer, and could be very wrong.
If the tuner is correctly aligned then tuning off centre will alter the levels and possibly phase of the multi path signal. If these reduce enough the audio will be much cleaner sounding.

Is the multi path reduced if using mono rather than stereo, theory says it should?

[G8HQP Dave]

The multipath meter is probably looking at the amplitude modulation caused by multipath components beating with the main signal as the FM modulation sweeps everything up and down. Tuning off-centre will modify both the amplitude and phase of everything, but not all in the same way. This will affect the multipath meter, while not necessarily affecting audio distortion on the same way. Therefore I would not assume that minimising the multipath meter deflection by off-tuning will necessarily improve audio. Minimising the meter by adjusting the antenna should improve audio; that is what the meter is there for.

[stevehertz]

Thanks Frank, Dave. I'll try the mono thing. My single dipole aerial is vertically polarised by choice; it doubles perfectly as my DAB aerial via a disty unit and I am also a fan of picking up FM stations far and wide. Hence the need for 360 degrees. No, I don't need or want a rotator anyone!

No, the main 'upgrade' or fix re MP will hopefully be achieved when Rugeley power station is finally pulled down next year. It has been out of service for a couple of years now. The MP problems it causes - particularly in the case of analogue TV signals (ghosting) - are well known in the area.

[G8HQP Dave]

Mono will not reduce multipath, but it may reduce the audible effects of multipath. What effect, if any, it has on the multipath meter depends on circuit details - but no effect is quite possible.

[paulsherwin]

You will certainly get more multipath with an omnidirectional aerial. The only real solution for you would be a directional aerial with a rotator. I appreciate that this is probably overkill for you. Hopefully things will improve when the cooling towers come down, but these things can take a long time - they've been demolishing Didcot A for nearly a decade, and half the cooling towers are still there.

[Nuvistor]

Originally Posted by G8HQP Dave

Mono will not reduce multipath, but it may reduce the audible effects of multipath. What effect, if any, it has on the multipath meter depends on circuit details - but no effect is quite possible.

Quite correct, I should have said the audible MP distortion should be less.

[stevehertz]

Thanks everyone.

The particular advice I was/am looking for was re how tuning slightly off centre seems (?) to remove or reduce the MP signals. Technically what is going on? Surely a bounced signal is the same frequency as the main signal but separated by time? So how does detuning reduce it? Maybe the main signal, being much stronger can stand being detuned a little, whilst the the MP signal being of lesser strength cannot - does that make any sense. Probably not..

I know 'all about' vertical vs horizontal (had em), rotators (had em), yagis (had em) etc etc. I'm not needing advice re an aerial.

[Nuvistor]

An interesting white paper from the BBC, I have to be honest the maths are beyond me but there are diagrams and text to explain some problems with multi path.

http://downloads.bbc.co.uk/rd/pubs/w...les/WHP184.pdf

[stevehertz]

Thanks Frank, interesting paper. I just skim read it and got the gist of things generally. Like you say the maths is quite intense.

[G8HQP Dave]

It is all to do with how the amplitude and phase varies across the IF passband. The more perfect the IF, the less does multipath create problems.

[Nuvistor]

A tuner with excellent AM rejection will reduce the effects of amplitude MP but can not do anything for phase problems.

[Radio Wrangler]

That is an awfully nice paper from the BBC. He avoids exploding the whole thing in a mass of Bessel function terms and chucks a model at a spreadsheet.

Offsetting your tuning is not good for two reasons:

Firstly, the discriminator will have been engineered to give as flat a section as possible in the middle of an otherwise S-shaped curve of frequency to voltage translation. Offset your tuning and the peaks of deviation will run outside the flattest zone and into the curved section.

Secondly, you do not want to offset the signal within the bandwidth of your IF filter. Such filters are designed for linear phase as much as generally possible, but what is possible is limited and the designer needed to compromise to get some adjacent channel rejection for people in areas where the band is heavily populated. The consequence is that although the group delay response is flat and fairly minimal in the centre, it still has peaks in the filter's roll-off region. Maybe not as bad as they could have been, but not good either. THe group delay peaks are another way of saying these are regions of non-linear phase response. Non-linear phase response in a circuit carrying an FM signal equates to adding distortion the the baseband whenever it is demodulated.

For these reasons you want to make sure your tuner is aligned well, and that the centre of your discriminator (tuning meter centred) corresponds to the centre of the passband of your filter (midpoint of the response seen on your signal strength meter) And then you normally want to make sure it is centred on the incoming signal.

But what about multipath effects?

THe addition of two versions of a signal, each taking a different path length, creates a comb filter, with a series of amplitude nulls right across the frequency range, having a spacing of 1/(the time delay difference of the paths) Hertz.

The FM signal has a complex spectrum, each component tone in the baseband created a series of sidebands spaced apart by that component's frequency. Some fall into a null, some don't. The full set of these sidebands was needed, though to recreate an undistorted baseband signal in your demodulator. Frequency and phase modulation mean that the carrier vector is waggled in phase, so its endpoint moved ahead and behind in its circular path. Sidebands are additive vectors, and while a simple pair are sufficient to make the AM effect of a lengthening or shortening carrier vector (adding together to make a kite shape vector sum diagram, FM is more complex, a whole infinite series of progressively smaller and smaller vector components (sideband components) need to be added to create the circular motion. With finite channel bandwidth, the higher frequency ones (fortunately smaller ones) get dumped. But multipath can plant its nulls anywhere, even on strong components.

In offsetting your tuning, I think you may be acting to limit the damage by shifting your filter passband to one side of the signal, and sacrificing the more mutilated side, where the null happens to be.

THe multipath comb happens and affects the signals as seen by your antenna, however your tuner is tuned, and even when it's turned off. You aren't changing the multipath situation, you're altering your perception of it.

In strong signal areas it's common to fit highly directive antennae, and then sometimes fit attenuators to offset their high gain if it becomes excessive.

Note how well the Studer tuner comes out on distortion in that BBC paper. The Studer and Revox machines took a few interesting design choices. They don't use piezo-ceramic resonator IF filters, they have LC sections with high-Q coils which are still lower Q than the ceramics, but can be aligned properly for a more linear phase response. THey also use coax delay line discriminators which can be very linear. And they eschew the usual stereo decoder chips.

I've recently restored an ex-BBC B261 which had seen a hard life in a rack as either a quality monitor or as a fall-back programme source at an outlying site. Checking the alignment was a fairly delicate task with a vector network analyser to get the filter spot-on.

David

[stevehertz]

Thanks David for that in-depth and very interesting explanation. I'm wiser now.

[Radio Wrangler]

I got involved with FM microwave link systems towards the end of their era and into their replacement with digital modulation styles.

Think of a 40km path between two towers with an 8 or 11GHz transmitter and receiver carrying FM with a baseband of 12 or 18MHz. The baseband is 900 0r maybe 1800 phone calls all turned into single-sideband RF and spaced in a hierarchical channel structure. Add in a few pilot tones for fun.

To say this needs ultra-linearity is an understatement. Intermod and harmonic distortion as well as simple noise floor fill in any quiet slots.

Multipath fading happens. Lundgren and Rummler wrote a paper on multipath simulation. I don't know why they chose a 6.3ns path difference. it seems too short for real world conditions were they inspired by valve heater voltages?

One of the things we had were a few reels of 16mm cine film of the screen of an HP microwave link analyser monitoring a long over-sea link from the American government's Point Magu naval air station/ missile test range, reckoned to be one of the worst links ever for selective fading. The film cans bore large graphic labels "Point Magu Missile Test Range" with a cartoon-like illustration. How Geoff got them out of the US, I'll never know!

Anyway this sort of case has been heavily studied and the maths is out there. It's not terribly difficult stuff, there's just a lot of it to wade through, but we used to have to do it the hard way. You have to have done this in order to appreciate the lighter approach in that BBC report.

You only have to have seen the spectrum of an FM sig gen as you play around with the deviation and modulation frequency to understand how what sounds simple produces a very complex spectrum. And understanding this as well as how all the terms add up to a series creating motion around a circle to give FM with no AM content and you get a good visualisation of what really goes on in an FM radio and what selective fading (=multipath) can do to it.

David

[MotorBikeLes]

I bought my first decent FM radio (NordMende Transita Spezial) in the late '60s, before I had any real understanding (which I still don't have). At the time I lived in a N-S valley with quite steep slopes on E and W sides. If an aeroplane flew over, the sound would start to flutter, get worse, (sometime better briefly when pretty well overhead) then slowly improve until well away. I would hear the first mild flutter and knew it would be un-listenable until past, which could take a while. One foggy morning, I switched it on, and there was the flutter, but it persisted for a long time, so bad I switched off. Later it was all OK again. This happened again a few times. Eventually it clicked, only when it was cold. I put it in a plastic bag (Yes, we had them then) and in the fridge overnight. Next day, it misbehaved as expected, so I opened it up and started poking around. I think it was warm finger and thumb that eventually alighted on a disc ceramic tacked on the tags of the FM tuner.
My first proper transistor radio repair.
Very shortly afterwards, I got heavily into playing with them, and bought a new Russian scope (CY5??) and a very early Cossor Band 1 TV alignment generator (1324?). Working on harmonics, I learned how to wobblulate, and took pleasure in alignment check on everything that came my way. I quickly realised that the German stuff, with their extra IF stages, could give very god results. They all had the extra little twiddle pot to optimise the AM rejection after the S curve was established. Their manuals advise running a little commutator motor to produce bad AM interference, making that last adjustment so much easier.
It kept me out of mischief at least.
Les

[Synchrodyne]

Originally Posted by Radio Wrangler

You only have to have seen the spectrum of an FM sig gen as you play around with the deviation and modulation frequency to understand how what sounds simple produces a very complex spectrum. And understanding this as well as how all the terms add up to a series creating motion around a circle to give FM with no AM content and you get a good visualisation of what really goes on in an FM radio and what selective fading (=multipath) can do to it.

A simplistic and probably only approximate way to look at FM sidebands is to consider that the first set, whose sum vector is orthogonal to the carrier, not only provide the frequency swing but also cause some AM. The AM is cancelled by a second set of “AM” sidebands (whose sum vector is in-phase with the carrier) at twice the modulating frequency. But in cancelling the AM, these distort the original FM, because their rotation circle is not equally divided between net increase and net decrease of the carrier amplitude. This distortion may then be corrected by another set of “FM” sidebands at three times the modulating frequency, and the AM that they then introduce is cancelled by an “AM” sideband set at four times the modulating frequency, and so it goes on ad infinitum, with the required corrections getting smaller and smaller. The odd-numbered sidebands are “FM” (orthogonal) and the even-numbered sidebands are “AM” (in-phase).


Cheers,

[SteveCG]

I appreciate that stevehertz does not want advice on VHF/FM aerials. However I was wondering whether a log periodic design (with its smooth sidelobe performance) may be better than a yagi (eg even a decent Antiference 5 element design) with their sidelobe peaks and troughs. This would be for the situation where MP is a problem even with a strong signal - so the lower forward gain of an LP is not a problem.

[emeritus]

In the 1970's, before the FM band filled up with local radio, I could usually receive a number of continental FM stations in stereo at my house in outer East London. I used a loft-mounted simple horizontal dipole cut for 100 MHz made from 15mm copper pipe, originally mounted on a home made rotator to use the dipole's end null to null out local stations for long range reception. It was a bit of a faff, and I ended up discarding the rotator and making a second dipole for long-range use mounted to null out the BBC, fed by a second feeder, and simply plugged in the appropriate feeder. Luckily, no tall structures or higher ground locally to cause multipath.

[SteveCG]

As I read the BBC white paper (thank you Frank for the link) the signal into the tuner ultimately determines the distortion the listener will hear. And depending upon the surrounding terrain and the path to the transmitter the choice of aerial can assist here. So where I live the direct path is somewhat blocked by a neighbouring house and further away by the low line of a hill. In contrast behind me I have, about a mile away, a higher ridge of hills. Hence my MP, wanted to unwanted signal, ratio is not as good as I would wish. I'm not short of signal strength, but I suspect the MP distortion is higher than it could be. Currently I'm using a Quad FM3 and 5 element HP Antiference
Trumatch.

I was interested to read of the particular Trio tuner that stevehertz is using. I wonder whether there were other makes of tuner sold that had a MP distortion readout?