FM Deviation
#1
Hi All

The other day I was listening to a QSO on 2M FM while mobile,  One of the stations had the audio level so low that you could not tell they were talking.  Only by turning up the volume to maximum did you stand any chance of making out some of what they were saying. Yet no one participating in the QSO ever said anything about it.  

It was the last straw for me.  

It resulted in me I posting the an article on the local VK6 user group to address the low deviation issue.  Below is an updated version of that.


FM Deviation

With FM you should not ever need to adjust to volume level between different stations regardless of their signal level. 
FM Modulation has the ability to provide good communications as long as sufficient signal level is present so as that the proper operation of the FM detector is achieved.  While this condition is met a constant audio level is present regardless of signal strength.

A 12dB S/N ratio is used as the reference point for comparing receiver weak signal performance.  In the past a 20 dB quieting level was used as this allowed the use of simpler test equipment.

The receiver bandwidth determines the FM deviation level it is intended to be used for.  This is determined by the receive filter in the receiver.  The deviation present determines the volume not the signal strength. 

To obtain best audio S/N (signal to noise) ratio on FM the deviation needs to make full use of the receiver bandwidth available. 

Deviation outside receiver bandwidth results in a distorted and/or loss of audio signal.  On the other hand if full receiver bandwidth available is not made it results in an inferior S/N ratio than that possible. Wide deviation provides for a high S/N ratio and audio bandwidth but this comes at the cost of requiring a higher signal level for satisfactory operation.  Narrow deviation on the other hand enables reception of weaker signals but at the expense of a lower audio S/N radio.  You can’t have it both ways. 

The overall effect of over deviation varies depending on a number of factors such as the deviation level, audio frequencies involved, signal strength and mute setting.  No simple answer is possible.   

Some quick tests on my IC-2720 produced the following results:

RF Level/AF Freq                   Audio Lost*             Audio Lost*            Mute Status

-120dBm/1000Hz                       7.8kHz                     ~17kHz                      muted
     0dBm/1000Hz                       8.4kHz                     ~200kHz**                unmuted           
-120dBm/2700Hz                       6.2kHz                     ~8.7kHz                     muted
     0dBm/2700Hz                       7.1kHz                     ~200kHz**                unmuted

 *   level lost occurred was with audio with some loss in quality
**  max deviation possible from my communications test set

With strong signals a higher level of deviation may be usable.  This is probably due the receiver filter appearing wider with its skirt selectivity coming into play.

FM transmitters normally include some form of deviation limiting circuitry and it is relatively easy to set the maximum deviation level.  But you cannot take this for granted.  I found that the Ranger RCI-2950/2970 series does not include any form of deviation limiting.  Deviation on this is entirely dependent on the audio level going in.  It is a somewhat poor design.
Broadcast FM stations use wide band FM (+/- 75kHz) so as to be able to provide a wide audio level and wide frequency range at a low noise level for music programs.  They compensate for the higher signal level requirement by using high power levels.

Two way radio communications use narrow band and can achieve the similar coverage as FM Broadcast transmissions but lower power levels.  They only provide voice audio bandwidth.  These normally use (+/- 5kHz) in amateur service (commercial services now use +/- 2.5kHz as it provides for more usable channels).  Unfortunately a down side to low deviation levels is that while the sensitivity is increased the S/N ratio achievable is reduced and it becomes more susceptible to impulse interference.

The transmitter deviation needs to be controlled so that the maximum deviation does not exceed that required for the system to avoid distortion.  At the same time the deviation should be kept close to the maximum allowable so as to maintain received audio S/N ratio as high as possible.  Maximum deviation is usually limited by using a clipper with audio filtering after it.  Sometimes this is achieved using fast acting compression.  A slight amount of audio clipping and/or additional compression can be used to keep the average audio level up and increase talk power (speech processing as often used on SSB).  The amount of "speech processing" you can achieve will depend on the limiter used.

My observation of amateur FM radio communications is that the deviation of many stations is rather low resulting in poor communications quality.  My base radio data port feeds a meter calibrated to indicate the deviation level present along with an oscilloscope so it easy to compare transmissions.

With a 5kHz deviation system, the goal should be to maintain a deviation around the 3kHz on speech so as to achieve maximum talk power.  Unfortunately the deviation on many stations is around 0.5kHz or lower.  This corresponds to the drop in audio level of some 15 dB on what it should be.  Would many SSB operators deliberately only operate their 100W radio so that is effectively only operates at the 5W level?  I don’t think so.

The low deviation could be due to the radio not been set up correctly.  Having a new radio from the factory however does not necessarily mean it is set up correctly. 

In my case, many years ago, I bought a new Icom IC- 2720, loaded channels into it and as no one ever commented on the audio continued to use it for many, many years until one day Will 6UU commented that I had low audio.  I found this interesting and eventually checked the deviation.  To my surprise I found that the maximum deviation was only 1.5kHz not 5kHz.  

Mmm, for a radio straight from the factory???

After much searching on the internet I found clues as to what it could be.  I, of course, like all other Amateurs had read the user instruction manual – well the parts that were relevant to me at the time, or at least I thought so.  I however had missed the one on page VIII of the Icom IC-2720 Instruction Manual.

"Although you have purchased a brand new transceiver, some settings may be changed from the factory defaults because of the QC process. Resetting the CPU is necessary to start from factory default.”

Now I had always assumed that the QC process is supposed result in a better product going out from the factory door.  Obviously I was mistaken.  I checked a later Icom radio manual and found a similar statement present using slightly different wording.  I don’t know if this still is the also case with other Icom radios or with to other brands but it is something to consider when setting up the radio. 

= Don't Assume anything -

-
I saved the IC-2720 memory channels, reset the radio and reloaded the memory channels.  And, guess what, the max deviation was now where it should have been.
Now getting back to the low deviation levels on many amateur FM stations this may be due to: 

1                    Factory reset requirement.
2                    Deviation level not set correctly (beware of 5/2.5 kHz option trap).
3                    Deviation level set correctly but insufficient mike gain.

I suspect that it is more likely that most of the amateur stations probably have the deviation level set correctly but are suffering from low mike gain.  Many are probably using mobiles as base radios.  Mobiles are set up to expect a reasonably high audio voice level as they are used in a noisy car environment.  However in the quieter base station environment one naturally talks at a much lower level.  If this is the case, the mike gain needs to be increased (if possible) to compensate for this, or failing that remember to always talk at a higher level.

Handhelds are another issue.  They are rather difficult to set up as they are used in a very variable environment.

So, if you have a QSO with someone that has low audio.  

Let them know

It will give them the opportunity to do something about it.

With FM you should never need to adjust the volume level between stations.

73
Igor
VK6ZFG
Reply
#2
Igor,

Another thought : some (commercial) transceivers are supplied with two transmit (and receive) options - Wide and Narrow.  Users may not recognise what the correct setting is and unknowingly leave them set to Narrow.  The Narrow option arrived when some countries implemented either a 10 or 12.5 KHz channelling plan as user growth was expected, but my appraisal is that it simply didn't happen in the amateur service in VK. What happened in lieu was the arrival of the various digital voice modes (DMR, etc) so we still ended up with 12.5 KHz digital channelling interspersed with 25 KHz voice-only channelling !

Historically we all set up our 6M, 2M and 70CM transmitters for 5KHz FM deviation but nowadays almost no one actually posesses the gear to measure it and it is pure guesswork if it is close to optimum.  The chances are that if you overhear someone with low voice deviation, that particular radio may be set up for "Narrow" operation..

There is no easy fix for it - except education.

Doug

PS I have a Marconi Modulation Meter so I can directly measure FM deviation, one of the few I guess  !
Doug VK4ADC @ QG62LG51
http://www.vk4adc.com

This Forum is only going to be as interesting as the posts it contains. 
If you have a comment or question, post it as it may trigger or answer the query in someone else's mind.
Reply
#3
HI Doug

With regards to the 5.0/2.5 kHz option I did allude to this trap but did not expand on this.

I originally did not go too far into the over deviation issue as I was only trying to address the low deviation side of things.  However someone responded to my other post so I added some more to that post on this. 

Many, many years ago I organized the migration of all the fire services in WA to the 2.5kHz channeling into a continuous block of channels (by jumping in very early) with all fire agencies being fitted with them in common thus ensuring cross agency communications.  However I failed in my goal to have them move away from mid band 70-85 Mhz.  This is because one agency refused to change bands. The reasoning behind my band change proposal was limited availability of equipment on mid band due its limited use worldwide, susceptibility to electrical interference and, poor handheld performance.  In the long run I have proven to have been on the right path.  They have now have migrated away from mid band.  But now, instead of a nice continuous block of channels, they now have channels all over the place (dependent on where free channels could be found for that area).

Years ago there was some work carried out on the use of SSB in place of FM for mobile communications as a way to generate more channels.  As frequency stability, unlike now, was an issue then, it used a pilot carrier with the transmission.  However this has since been overtaken by digital modulation developments and as result it did go progress much further than the testing phase.

In my early days 15 kHz deviation was used (commercial 60kHz channeling), but when the commercial service migrated to 5 kHz (30 kHz channeling) and lot of surplus commercial was available.  In time old 5kHz surplus commercial equipment becoming available and amateur FM also eventually changing to 5 kHz.  Commercial services have since moved to planned split frequency and simplex blocks on all bands using 2.5 kHz deviation and 12.5 kHz channeling.  UHF used 25 kHz channeling but the frequency blocks splits were rearranged so as to fit in with more commonly used international ones. 

2.5 kHz deviation is about as low as it is viable for FM.  Digital modulation has now however enabled better use to be made of bandwidth and this is the area where progress is being made to achieve more efficient use of spectrum.  It is unlikely that in the future FM will change much if at all from where it is now.

2.5 kHz systems are slightly more sensitive than the 5 kHz ones, but are more susceptible to interference (some lower band ones even incorporate noise blankers), produce a lower S/N ratio and the audio frequency band is less.

Fortunately I have more than one communications test set so I am able to measure deviation fairly easily.  However this can only be achieved with direct connections or very strong signals.  It is no good for weak signals. For this I use a modified receiver setup.  This is as follows:

In my case I use an IC-2720 to receive signals.  I adapted a EA K7222 Noise and Distortion meter with the option to switch to a rear socket connected directly to the data port of an IC-2720.  It also has a BNC socket on the rear to permit viewing this on a CRO.  The audio level is preset trimmed at the N&D Meter so that 30mV corresponds to 3.0 kHz deviation.  A slight amount of frequency compensation is incorporated as the data was found to be close to, but not perfectly flat in frequency response.  I was chasing perfection.

Now you can do the same with any other radio that has a data port.  This should not require any frequency compensation unless you are chasing perfection.  Any audio level meter will do.

A normal audio line or speaker output is not suitable as in includes de-emphasis and will produce incorrect results other than on the spot audio frequency that is used to calibrate the setup.  Adding a emphasis compensation circuit to counteract the de-emphasis etc of the audio stages, while possible, would be extremely hard to actually achieve successfully. 

To calibrate the above system all you need is a signal with a know level of deviation.  1 kHz is usually used.  The level this is corresponds to the deviation reference level being received.  Deviation can then be directly related to this level  eg:  50% corresponds to half that level etc.  It is a linear relationship. Varying the modulating deviating frequency will give you an idea how flat the frequency response of the data port measurement setup is.

73
Igor
Reply
#4
Hi Steve

Yes, commercial mobile communication radios use 6 dB/octave pre-emphasis at the transmit end and 6 dB de-emphasis at the receive end to restore the signal back to the original one.  The pre-emphasis being used to improve the S/N ratio at the higher frequencies.  This is fairly straight forward unlike in the broadcasting application.  

The Broadcasting industry also plays games with speech processing especially with advertising.  According to the broadcasters they said they could do noting about the high audio levels on advertising when complains were raised about this years ago.  This is probably true in the sense that the advertising producers are the ones that implement the speech processing in their productions.  However eventually a drop advertising high audio levels did occur though it seems to be creeping up again.  

If I recall correctly a long time ago in the early of FM in amateur use, Japanese made amateur FM radios always seemed to sound somewhat different to the ex-commercial FM radios.  I suspect that this was probably because they did not use any pre-emphasis on transmit.  Be interesting to know if anyone else can recall this.

The original intent of my article was to address the issue of low deviation on amateur FM.  I tried to keep the original article as simple as possible without delving too deeply into the finer points of FM and boring everyone with a book.  Somewhat difficult to do as there are many factors that come into play.  

However it is the end result that counts - the audio level that you effectively hear.

73
Igor

On a side line, years ago I was involved with the world land speed record attempt on Lake Gairdner in VK5 (Aussie Invader) with the the communications and timing.  The communications from the site was HF SSB and the very early version of Mobilesat (satellite).  What I found was that the the codec used then by Mobilesat had trouble dealing with my children's voices and/or the XLY's (when she got excited telling them of if they were playing up).  On HF however the were no such problems were evident.  

I can also to this the claim to have built the fastest talk through repeater on wheels in the world.  Unfortunately I have not yet found viable market for this!
Reply
#5
Igor is NOT joking when he  said this ""I can also to this the claim to have built the fastest talk through repeater on wheels in the world.  Unfortunately I have not yet found viable market for this!""

He actually did and it all started on his kitchen table with some Graph paper and  other stuff to design the timing  mechanism from scratch for this world speed record attempt.
Reply
#6
Hi Steve

I had recollections that pre-emphasis had to be applied to one of them.  As I am not paid to think anymore I looked it up.

PM has inherent 6 dB/octave whereas FM does not.
To make FM the same as PM, 6 dB pre-emphasis needs to be applied to the FM modulating signal.  
The 6 dB/octave improves the S/N radio at the higher modulating frequencies. 

On receive, de-emphasis need to be applied to restore the incoming signal to the original source signal.  
This is separate to any signal/frequency processing that may have been applied to the original source signal transmitted.  

If the early Japanese radios used FM this would explain why they sounded different.  
I had always assumed that maybe they sounded different because they had been set up to better suit the Japanese language.  
But it looks like the radios just did not include any pre-emphasis. Overlooked during design maybe?

With crystal looked radios, PM is easier to implement on multi channel radios as it provides for consistent deviation between channels where as this is difficult to achieve with FM.  In addition FM can reduce transmit frequency stability due to the means used to achieve the FM.

73
Igor
Reply


Forum Jump:


Users browsing this thread: 1 Guest(s)