Filter Failure!

Why your super-duper-fantastic-cutting-edge-top-of-the-line rig is not filtering the guy next to you


One of the common misunderstandings frequently debated on the various reflectors is how to better filter the SSB guy that's just up or down the band from you.  And speculation on the inadequate level observed because filtering just does not seem to make as much of a difference as it should. 

How can that be?

But it really should not be a sup rise at all that the because the biggest problem on the bands today is the sideband content generated by the TRANSMITTER - not due to overload on the receiver.  NC0B Rob has been making this point for at least 10 years and he's by no means the first.

Below is a short discussion of the 3 major types of interference that you hear, and how filtering does or does not help - and why.



These are the signals that you hear which are not really there on the antenna end; they are mixing products made some where inside the RX chain by the presence of strong signals.  These strong signals combine to generate phantom signals (IMD products) that can sound real enough - but that are generally unintelligible for reasons you never can quite nail down.

There are a couple of fixes here. The expensive one is to toughen up the entire front end of the radio so that even with some big signals banging away, the rig is not generating IMD products.  PA3AKE has an extensive write up to give the curious a flavor for just how difficult this task can be.  

Because making the front end tough is an expensive and difficult task, the cheaper - and as a result more common - fix is just to put a narrow filter right behind the 1st mixer and hope to chop away all the strong signals but the one that you want.  We call this a roofing filter because it offers protection to the down-stream components with the idea that the roof takes the beating and everything else can be relatively light weight from a dynamic range standpoint.  Most modern rigs will generate IMD levels louder than the MDS at around an S9 input signal level.  

KEY POINT:  a roofing filter only prevents YOUR RECEIVER FROM GENERATING MIXING PRODUCTS as a result of overload.  It is not the primary selectivity tool in most rigs.




This interference case is the guy who starts up right next to the guy you are working on CW, and who is say just 100 Hz away.  If his signal is not too much stronger than your QSO buddy, good filtering (or maybe a notch) may do the job to cut his signal and free you from brain DSP filtering duty.

In the case of CW or PSK, with moderate signal levels especially, the guy is essentially occupying a very narrow slot of band, and you are able to move your filter passband around so the guy's signal is not in the passband.  Just like the pictures in the book suggest.

KEY POINT: Filters work IF you are able to adjust the filter passband so that it's passing your QSO buddy's signal, and at the same time, puts the filter stop band on the QRM frequency.




In this last case, the curtain is pulled back and we can see the wizard in action. Let's take the two most common cases separately.  The principle is the same but how we hear the effects is different enough that we may not realize it's two sides of the same coin.


Example 3A - CW.  Why your rig is not filtering key clicks on the guy next to me


There has been a lot of attention brought to the issue of occupied bandwidth/key clicks on rigs in the last year, and so most guys will probably have heard something by now (if they did not already know) that running slower rise times generates less adjacent channel QRM.  Here's a spectrum analyzer shot that illustrates this point.























Let's say your new rig is the ULTIMATUM MAXIMUS FILTER-MAGIC 1000.  And thanks to the combination of roofing filter, DSP and other technology (some smuggled out of Area-51), the filter passband of the rig looks like the red rectangle-ish shape in the plot above.

And let's also say that you are across town from a fictitious powerhouse station KI6QRM who bangs your S-meter at +60 over making you appreciate the meter stop keeping the thing from being bent.  You take note that even though you are now 750 hz away and your MAXIMUS FILTER-MAGIC 1000 has the filter slopes of the Radio Gods, you still have an S9 indication on the meter. 

Now how can that be?  MY FILTER IS NOT WORKING!!!

Well the filter is working just fine. It just seems as if it's not because KI6QRM's signal is NOT ONLY located 750 hz away.  Thanks to his keying sidebands, his audible signal actually extends quite a few KHz either side of his primary note that you identify as his QRG.  

We tend to think of CW as being a zero bandwidth mode - but that's not really the case.  A constant unkeyed tone can be very narrow - but the instant you start modulating that tone, then you get a wider bandwidth as part of the deal.  Selecting a slower rise time (either by rig mod or menu setting) was the best way to cut down on occupied bandwidth until the advent of DSP techniques (which are by no means standard - yet).

So how much of a difference does it make? 

In your QSO with KI6QRM you note the meter reading as he slides his rise time setting from 3 mS to 10 mS.  And are amazed to see your S-meter drop 20 dB!   We see that in the graph - the narrow plot is the 10 mS case; the wider one is the 3 mS case.   [OK, OK, not all rigs will look JUST like this graph.  There are a lot of factors that go into the occupied bandwidth of a given transmitter, but all things being equal a slower rise time yields a more narrow transmitter footprint.]

Wow, THAT IS COOL you remark and thank KI6QRM for his consideration.

KEY POINTS:  There are a few on this one:

1. CW is not a zero bandwidth mode. With short keying rise times, the rig will occupy a lot of bandwidth compared to the longer keying rise time settings. That's why the guy is not only 750 Hz up from you, he is ALSO right on top of you - at least his key click sidebands are.

2. And the stronger he is, the stronger those key click sidebands will seem as well.  

3. And if the band noise is low, those key clicks can be heard quite a ways away.

4. Finally, even though you have the filter system NSA agents only dream about you cannot filter out key clicks.  Why?  Because that part of the guys signal which is IN THE PASSBAND OF THE FILTER.  




Example 3B - SSB.  Why the guy next to you seems to be "leaking through" into your QSO


In this example, your buddy KI6QRM is 4 KHz up from you and you have your ULTIMATUM dialed in to a 2.4 KHz bandwidth, bringing a staggering cascade of 24 ultra-high-Q crystals grown in zero G orbit backed up by 50 GFLOP DSP that uses water cooling and was designed by Cray. Of course we have zero phase nose thanks to a local oscillator that is so uniform and perfect that the NIST has tapped it as a backup to WWV.  You know it does work because the passband measurement broke the ARRL lab gear.   So surely the SSB case is going to be a lot better than the CW case.  In anticipation of recovering your reputation, your buddy's are standing around the shack now just as KI6QRM starts up his QSO with KN1QRN.  With KI6QRM's signal up 4 KHz and only 2.7 KHz or so wide, you expect silence. But all of the sudden a lot of SSB spits and nonsense comes out of your rig's speaker. 

How can that be?

Just as in the case of CW, SSB is not confined to the 2.4-2.7 KHz range that equipment makers claim.  In fact if you have spent any time with a bandscope, the graph below will be a familiar one.  Nice and tight with vertical side walls about -30 dB or so from the peak, and then flattening out with slopes that run on forever making you think of a scene out of Close Encounters.

The fact is, just as with the case of CW, even though the primary signal content is narrow, for various reasons there is tons of sideband signal content that extends a LONG WAY from the place the guy is actually tuned too. 

If KI6QRM had the transmitter shown in the graph, and he was running +60 over S9 when you had him tuned in, after you moved down 4 KHz he would still be about +20 over! How can he be that strong if your filter is working as it should? 

The answer is that you are not having filter leakage of his primary signal somehow slipping through, rather your filter system is doing it's job 100% right and passing the energy content in the passband.  It's just that you did not realize until now that SSB signals are really really wide - so what you hear is the tail ends of the guy's occupied bandwidth.   

A fair question to ask is - how far away would I have to move if the other guy was fat like this picture indicates, and is really strong, and the band noise is really low.  The answer is tens of KHz.  You will still be able to pick up a little bit of his sideband tail under those conditions.

One buddy asks - "why don't I notice guys that far away?" Because the combination of other signals on the band and the atmospheric QRM (band noise) combine to put a floor in what you can hear.  Just because KI6QRM is tens of KHz wide at -100 dB, if your noise floor is very high, you won't be able to hear the tails.  

Another buddy says  - "well my Model A (in black) from 1995 filters just fine. When I switch to a narrower band filter, it cuts the splatter more."  Yes, you can get that feeling but it's not for the reason you think.  What is happening is that the QRM is stronger on one edge of the filter passband than the other (in our graph below, it's the lower side of the filter). So by moving to a narrower filter you effectively lower the average signal level of the sideband in the passband; not because you are filtering out the signal in the traditional sense, but rather because you are moving a bit further down the slope of the side band signal strength curve.  An equivalent move would be to leave the filter passband width as it was, and slide the center frequency of the filter to the right as much as you can tolerate.

KEY POINT:  Just as with CW, in SSB the transmitter occupies a wide bandwidth with some amount of sideband energy.  And on quiet bands with a strong signal, you are going to hear that signal because it's right there in the middle of your passband.



And just as with the CW case, the skirts of various SSB transmissions have different sideband energy content.  Generally speaking most modern rigs are worse than most tube rigs, except for the ones running class-A.



So what can I do if the filtering thing does not work?  Several things:  QSY is the easiest and works 100% of the time.  If you can't QSY then slide your filter passband away from the other guy so you are hearing lower average strength sidebands.  And if you have a directional antenna, try to turn it so the other guy is in your beams null - which may be off to the side of the beam heading.



The RTTY case is about the same as the SSB case except you machine is doing the decoding and modern decoders are pretty good at getting the signal among the junk. But you should do your part and try to run a clean signal. The easy way to do that is with a quality audio interface and a modern terminal program that has waveform keying optimization (2TONE, for example).  

A RTTY myth is that FSK is more pure than AFSK and that's just not true.  Personally I hear a lot more of what look like fat FSK signals out of prior generation gear than I hear overdriven AFSK setups.  Ask a buddy to look at your transmitted sideband profile sometime.  You may be surprised.

© 2011-2016 AC0C - All Rights Reserved


Created with the SiteBuilder.