[aprssig] Deviation (the new narrow) comments

Wed Mar 17 01:19:11 EDT 2010

> Steve Bunting, in part, wrote:
> ... what do I set the deviation at for ... a 12.5KHz channel? 
and got
>... several different answers (ranging from 0,9 to 2.4KHz).

2.4 is out of the question and .9 is hard for me to say for sure, but seems low.  After FEBO, Read on...

Two more cents.  
The "It depends." answer is more correct than one might imagine.

In that FEBO article you recall that it says to stay out of clip - this is the bottom line.  You basically want the most deviation that is just out of (below) clip.    Unfortunately, depending on just how any particular radio is designed, this can vary.

Short answer may vary well may to set it at 2/3 the peak deviation of the system.  Since different radios may have different clip points, this may just work well for all.  If some clip a little, but only a little, and others none, that "should" be ok.
Looks like 1.66 is 2/3.  

There is a suggestion at the bottom.

    HOWEVER, here comes the disclaimer.

Some background:
In commercial two-way (I'm a retired Motorola Engineer who has designed this type of circuitry) our standard for things like this was 3/2 of the peak deviation.  The reason for this is to STAY OUT OF CLIP in the deviation limiter.  The limiter is simply a peak voltage limiter.

If you look  at the method of audio filtering/preemphasis/limiter/filter, The point at which the audio just starts to hit the limiter's limits on peaks gives about 2/3 the deviation compared to the peak deviation (when the audio was cranked up 20 dB [10x voltage] above that level).  

It works this way because the limiter is followed by a steep (2 to 4 poles depending on radio) 3,000 Hz low pass filter.  This is called a "splatter filter" and it prevents the higher frequency sidebands from exceeding the strict FCC limits for the energy on the skirts of the transmitted bandwidth.  Without it, the 'almost square wave', from the clipper, would have sidebends out the wazoo across the band.  

The clipper followed by 2 or more poles of lowpass filter has this 2/3 characteristic between onset of clip and large signal peak.  When a 1 kHz tone is set to the +20dB level, the end waveform has a characteristic single-shoulder shape that hits the peak (5 kHz).  It comes about from the square wave into the lowpass and phase relationships of harmonics, bla, bla, bla.

If Ham radios do not have this same type of filtering (we have NO real FCC sideband energy limits, right?), the 2/3 ratio will be a little different.  

SO...

With the (new) narrow bandwidth of the 12.5 kHz spacing, the characteristics can be different and I haven't worked on this, so I can't comment from experience, how good the 2/3 number is.

Suggestion:
So ... I suggest a good approach is to take a well designed 12.5  kHz radio and measure this ratio and use that as a standard.

My gut feel says that it could be a little different because this 12.5 kc system is really pushing things at 2.5 kc deviation with 3 kc audio (if it has that audio bandwidth).

73, Steve, K9DCI
That's 35+ years worth of gut feel, for what it's worth. (;-)

FYI the Tx filtering is typically in this order:

1 - PL (CTCSS) filter 300Hz sharp cutoff (-18 to -24 dB/o) high pass.
2 - 6db per octave pre-emphasis, sometimes called a high pass filter, but I think this is misleading since it is passing the 300-3000 desired band, just at a rising characteristic.
3 - Clipper
4 - 3,000 Hz sharp cutoff (-18 to -24 dB/o) low pass.

NOTE many Ham rigs don't try to keep the PL out - mmmmmmmmmmmmmmm.

-- 73, Steve, K9DCI

FINALLY, an up to date APRS Beginner Guide - 2009.
Direct link:
http://k9dci.home.comcast.net/APRS Beginner Guide - K9DCI Ver 5-1.pdf 
OR
  http://K9DCI.home.comcast.net 
    Click the APRS Beginner INFO link under Contact

More Details, Features and Downloads at:
http://home.comcast.net/~k9dci/site/?/page/Amateur_Radio_Control/

  http://k9dci.home.comcast.net/