The AM Forum

Why bother with audio bandwidth limiting? I've been looking at curves of the male voice. Most show 20 - 30db down at 5kHz and going down from there. I mean really. Is there enough power in a sideband out there to actually worry about?

I have a bandwidth filter in line which cuts things off sharply at 5 kHz, The only complaints I've ever gotten when not using it are from SDR users who see glimmers out there and complain about them.

Interestingly, some opera singers don't obey the curves. They have a "singer's formant," a very significant peak in the range 2500 - 3500 with real power out above 5 kHz. Placido Domingo needs a bandwidth filter. But do I? Do you?

Personally, I think it's important to control _excessive_ bandwidth (high end).

Some sounds in normal speech such as esses  "ssssssssssssss" and things like that do create one heck of a footprint.  Also, the harmonics and other products of normal speech can be fairly high frequency in nature, generating products that have no useful purpose.

Now, just to make it clear and for the record, I am *not* NOT    N O T   suggesting any sort of rule, regulation or mandatory bandwidth limitations place on phone operations.  This would be bad.  But a "good neighbor" filter might be a good thing to have.  I have one such filter here, which brings in another 8 poles of filtering with a 5.7 kHz corner.  I use it regularly at night (but sometimes I forget to turn it on, and I have received reports from adjacent stations when I do forget)...  :D

A 5kHz sharp filter is audibly noticeable, at least on my receiver.

It would be nice ifin there was a on-line receiver that would graph ones signal.


klc

Well said, Steve.  Rice boxes - no problem but if one has gone to the effort of HIFI audio, BW needs to be controlled "according to good engineering principles"   --- that's the key.  FCC intrusts us with the knowledge and motivation to practice good engineering techniques.  We do NOT want to be lead down the BW path with a ring in our noses by the FCC because we have no control over our BW.

Let me get up on the soap box and say that a few, just a few HIFI stations have issues with sibilants which may be the result of upping the upper end with no control in that frequency band (limiting the upper limit BW and the level of modulation).  Yes, you can over modulate in the upper register and it can really produce some very interesting "buckshot" problems.

Al

That's absolutely correct for a voice print average without any EQ or compression, but once either of those are added, things change dramatically. Steve QIX noted what happens with the "sss" sound. It's almost expected to use some amount of high frequency boost (pre-emphasis) on AM to overcome the corresponding rolloff in most receivers. Add compression and those sibilance frequencies get pushed up to 100% mod even though they're 6-10 kc out. Multiband processors, which were designed for music, ensure that each of the frequency bands are maxed regardless of actual input energy. That's how a thumping bass drum has little effect on higher frequency instruments resulting in denser/louder audio. Even background hiss can create havoc, so multiband processing requires extremely careful set up.

Then there's clipping. Widely used in broadcast processing to ensure maximum loudness, clippers generate voluminous amounts of harmonic energy. Without some form of LPF to limit bandwidth, tremendous spectral energy could be transmitted far beyond what any receiver is actually receiving.

AM can sound magnificent when set up with care, but ops have a lot of factors to consider to achieve great sound while keeping peace on the bands even when simply speaking into a microphone.

http://dl.dropbox.com/u/17340243/W2ICQ-and%20the%20SENTINEL.mp3

 Then again here is Bob W1ICQ,  "speaking into a microphone".. No "processing", no group delay low pass phase distortion, "smart clippers", or all-pass symmetry magic.. Just a well designed transmitter and a D-104. Listening fatigue? Nope.
 
Caw Mawn

KISS works

Lest we forget...

There are really TWO types of audio "limiting"  when discussing bandwidth.

One, is simply the program audio range, like 20 Hz to  6 KHz - simple.

Two, is the bandwidth created by distortion products - complicated.

The first is easily controllable with audio filtering.  The second is VERY difficult to accomplish unless the operator makes an extra effort to design and test a modulation system that works well and then actually runs it in a linear fashion, keeping it under control using accurate spectrum monitoring.  We need a clean high level modulator, proper drive levels, regulated voltages where needed, clean linear amplifier, tubes with good emission, no arcing, etc., etc.

I hear it all the time - confusion between the two subjects on the air. Bottom line is we first need to test our rig to insure it is truly say, -30dB 3rd order IMD or better.  Then and only then does discussing audio filtering and limiting our voice highs make any sense.

A really clean rig (-35dB 3rd order or better) running 6KHz audio sounds like gentle wisps as we tune up the band. A beautiful sound.  In contrast, a rig with nasty -15dB 3rd order products sounds like a buzz saw when tuning up the band 10 KHz, even when running restricted 2.5 KHz audio. This is due to audio/RF mixing products generated within the rig.

Clean, tested, verifiable linear audio is what it's all about before we put the audio filtering to it.

T

May I suggest a good engineering principle:

Transmit audio bandwidth should not exceed received audio bandwidth.


Also, take a look at the peak to average figures across the speech spectrum. There are very large amounts of energy at the higher frequencies, although transient.

Those large amounts of energy equal power better concentrated on the desired speech spectrum.

I suggest that it's good to be flexible -- take advantage of wider audio bandwidth when the spectrum is free (and, yes, others should respect that and give you your room if you were there first, even if they have hearing loss or inherently narrower gear), and use narrower audio when necessary to avoid causing interference. I use brick wall filters and strive to be very clean too, to be a good neighbor.

7.5 kHz of audio bandwidth is my default if the space is there, and I start to lose interest in operating AM if there is only room for 4 kHz (near-stock-SSB sound) audio or less. In my 45 years of amateur radio, I have always modified my receivers for wider bandwidth when necessary -- 7.5 kHz sounds more than twice as good as 4 kHz; to me it sounds 100 times better, and is SO worth the small amount of extra noise introduced.

The question of neighborliness always includes the question of whether we should restrict our audio bandwidth if someone moves in on US. (We should always restrict it to avoid causing QRM to someone already there before us.) But if someone moves in very close to my existing QSO, should I restrict my audio bandwidth to accommodate him?

73,

Kevin, WB4AIO.

Definitely need to limit audio bandwidth.
Some guys run 10 and 12 kHz wide.
If you even suggest in the nicest way, that they are buckshoting up the band, you get flamed and told that if you don't like it, move to another frequency.
Bad operating practice and Sooo un-neighborly. 

Here's my IMD and HD testing thread about how to determine how clean a rig is. (Inter-Modulation Distortion and Harmonic Distortion) This is a chain of homebrew linear amplifiers starting with a 1 watt lab amp to a 4CX-350J, to an 8877, to a pair of 8877s all running in class A or class AB. In the SDR IMD picture below, you will see results showing -55dB 3rd order IMD at 1500W output, which is unusually clean. It took a lot of work to get there. This is using old school, open loop, without pre-distortion software. There is also a link showing the testing of my class E rig in comparison.

These tests can be done with any plate modulated rig, class E, SDR chain, ricebox/ linear rig, etc.

http://amfone.net/Amforum/index.php?topic=33505.0


An important thing to remember:  Rig cleanliness is everything. This is where we start. For example, we can run + - 7 KHz audio into a super clean rig and it will probably not bother anyone.  The skirts will have some nice "ss" wisps. The signal will be well behaved. But start with a dirty rig and even + - 2.5 KHz audio will tear up the band and cause complaints.

Every AMer should eventually run 2-tone IMD tests on his own rig and actually know what his 3rd order IMD products are. Otherwise, we are driving blind and hoping for good results without knowing for sure.  Limiting the audio bandwidth will not solve the problem.

Many times a stock Valiant or DX-100 running only 100 watts can get away running poor IMD products, but if conditions get really good and signals are super strong, even they can tear up the band. Many boatanchor rigs I've tested are good for maybe -20db 3rd or worse. Some need work. In this day and age with used SDR receivers costing $30 and a 2-tone generator easy to build or buy, we can be on top of it.  

"Buckshotting" is almost always a sign of severe distortion generated from within the rig from poor design, faulty components or improper operation parameters (flat-topping, over-modulation, overdriven audio stages, etc) - and almost always has very little to do with the original voice bandwidth.  The difference can be heard as harsh, raspy garbage sidebands - vs: smooth, ssss wisps.  The poorer the inherent rig's IMD, the more we need to reduce the audio bandwidth and modulation % to compensate and hide the IMD and HD flaws... just an infected Band-Aid solution.

T

Which two tones? Maybe it's in your thread but I didn't spot it.

A pair of tones that are not harmonically related like 800 Hz and 1800 Hz of equal amplitude will work fine.

Here is a link to "SigJenny"  -    a free computer software audio generator.

http://www.dxzone.com/dx17728/sigjenny.html

There are plenty of good IMD testing articles on the web. Here's one about testing linear amps with ssb:

 http://www.ab4oj.com/test/docs/ssb_im.pdf

T

That's where the SDR and panadapters come in handy. You can plainly see what's going on. To emphasize those higher freqs to get those ssssssssssss's out there is annoying, during crowded band conditions. The side band slap back is not goodness.
The FM sound some people try to replicate on AM is fine for the hams who are within 50 miles of each other. Once atmospherics and HF crap hit the fan, then the B/W needs to be reduced on the receiving end. And the person transmitting ultra Hi-Fi is wasting time.

We're not broadcasters...Hams who have disposable income for the latest Orban box are really out there.

Nice good ole AM audio, with some minor processing and EQ....A D-104 and a DX hundred is really nice on the ears.

Have to disagree with you there, Fred. It is the crowded band conditions that are annoying, not ultra-hi-fi. Ultra-hi-fi is wonderful. If two dozen useless no-traffic traffic nets have to die to make room for it, then let them die!

Kevin, WB4AIO.

Like the one that is parked on the 40 meter AM calling frequency running lower side band?
No traffic, traffic net...really?

Yes, that is one -- of many.

One of my proudest moments in ham radio was when Bob Heil, K9EID, while visiting No Money Mike, KO6NM's station declared, "That's the best sounding Ranger I've every heard!" Mine. It was mine. With a D104. That was years ago but I cherish the memory. I must admit though, I had modified the heck out of it along the lines of WA1HLR's recommendations but still.

These days, of course, I'm usually using a fancy microphone and a fancy audio chain with de-essing, parametric equalization, bandwidth limiting, a fancy schmantz class E (thanks Steve for your great work), all kinds of monitoring equipment. However, there is a lot to be said for a D-104.

Or the related one that parks on 3.878 megacycles?


Phil - AC0OB

Here's a novel traffic net idea... open the net asking if any station -DOES- have traffic. Hearing none, net done.  Golly, how efficient!

That'll wake them up

Good point, Kevin.THEY moved in on an ongoing QSO, and shudda known better!

What if:
Stations start up a QSO on a "clear" x frequency.
At the same time, stations 1500 miles away start up a QSO on the same "clear" or close to same x frequency.
Bandwidth of both sets of QSO's might vary from 3 to 8 KHz depending on mode.
Roughly 20 to 30 minutes into the QSO's  on this "clear" frequency area, propagation changes.
Now both sets of QSO's can hear each other causing interference to each others QSO.
What do you do?
Do you:
all just piss and moan about the interference;
your QSO moves;
they move or you ask them to move or they ask you to move;
each claim the other is causing intentional interference and no one moves;
piss and moan about each others bad operating habits and no one moves;
threaten recording tapes and notification to the FCC or to anyone who will listen to each others piss and moan;
What do you do?

If two QSOs start at the same common time, such as UTC, then it would need to be worked out in a gentlemanly way at the time when propagation causes them to conflict.

If the coincidental beginning at the same common time happens nightly, then a longer term solution should be worked out in a gentlemanly way.

If a QSO begins at 2230Z and a net begins at 2330Z on the same frequency at a time of year when both groups can hear each other, then the solution is obvious. The harder question arises when the 2330Z net refuses to QSY and says they have priority because:

A) "we are an established net"
B) "our frequencies are published on our website"

That brings us back to "What would you do?" The attached MP3 gives a real-world example...

What would you do?

You should take this conversation to the QSO section. It's no longer a Technical one.

1500 mi away and causing interference would tell me that SOMEONE is on the air with 10KW of power!!

Indeed, but to bring it back, testing my latest killer transmitter, an AT-1 with a cathode modulator that is  only -3 dB at 5 kHz  and -6db at 8 kHz, but with a D-104 with its falls-off-a-cliff high frequency response yielded essentially nothing past 5 kHz out. This test wasn't terribly scientific. I was just watching my signal on a WebSDR waterfall display during an on the air test QSO, but that's the way it looked. A buddy was listening and could not hear anything out there either. I was driving a linear, 100 watts carrier. My signal was strong at the SDR. Unfortunately, I didn't note the S meter reading which would have made the test a touch more scientific. Not in the mud anyway.

Maybe it's my foggy voice but there you have it.

Compare the response curve of the D104 (and Shure 444) with my usual microphone, a Heil PR-40. With the D-104, I can get away with very little bandwidth limiting in the audio chain.  The PR-40 is a different story. It actually has a peak at 10 kHz.

The station mic here is an Astatic D-104 crystal element, which as seen in the curves is 12db off the peak at 5kc, and more than 30db down at 10kc.   So long as all is kept linear at voice peaks, all is well in my book.

As for Pete's scenario, If I felt "I was there first" I'd ask the "new" guy to move over a little, but if he thinks it's HIS fire hydrant, I'd just move, not happy, but life's too short to whine about such inconsequential things.   

Now if the other guy is obviously over modulating, or has other significant distortion products, I'd probably let him know, but beyond that, there's lots of room on most the bands.       

I always have a bandwidth limiting arrangement with a pretty sharp cutoff at 3.8KHz or so, and as clean a modulator as it can be run.

The reports always indicate no buckshot or splatter. I can't stand to cause noise outside my intended bandwidth. I feel like it is shameful to transmit distortion outside one's chosen bandwidth and I hope others feel the same way.

I totally agree -- stay clean and use brick wall filters, definitely. Sometimes, of course, my chosen bandwidth is +/- 7.5 kHz -- fairly often, actually. At other times it might be less or more -- though I much prefer operating when and where it can be fairly wide. When the AM passband gets significantly narrower than +/- 5 kHz, I feel like the quality is so low that I might as well be running filter-type SSB.

73,

Kevin.

It depends on how you want to be perceived. This would work much better in an open lot...

(http://amfone.net/Amforum/index.php?action=dlattach;topic=42592.0;attach=54949;image)

Great illustration.

Yes, the "three kHz audio is always best" and "every three kHz" channelized mentality is a plague on amateur radio (in addition to being very questionable from a best intelligibility point of view), and should never have been instituted or taught.

Just like those narrow parking spaces -- so ridiculously narrow that even compact cars' doors can't fully open without striking other cars! -- should never have been designed that way.

73,

Kevin, WB4AIO.

This happens too.... :P

The guy in the Ferrari picked a place away from everyone, and occupied two spaces (channels).

Than some person decides to park right next to him, and, and occupies one space (channel).

This happens on the radio as well as in the parking lot.

I recall once as a young man, doing the same thing with my 73 Dodge. I was within inches of a diagonally parked Corvette. My entire care was within the lines of 1 parking spot. The Vette driver was trapped, and was waiting for me.... :-[

Jim
Wd5JKO

Around here the red car would be more likely than the others to be accidentally damaged. Someone with a truck might be confused about which parking slot it was in and try to pull in to the perceived empty slot.

As for narrow slots, unless it is marked 'compact car only' (which is ignored..) the state requires the slot to be minimum 96" wide (502.2 Vehicle Spaces).

The only 'narrow slots' or slots of any kind? on the ham bands are those 60 meter SSB channels.

At least the red and blue cars are clean, (see K1JJ's post). Some of the signals I encounter on the band are more like this:

Is that a plate modulated rig or PWM?  After studying the photo, from the looks of the front end alignment, that might be Tim's (WA1HLR) SBE (Slop bucket eliminator).

That jalopy there is what's known as a rat.
https://en.wikipedia.org/wiki/Rat_rod

truck version: (equivalent to CB with 5KW leen yar)
https://encrypted.google.com/search?q=rat+truck&tbm=isch&imgil=GpJntqLozaYTaM%253A%253Bq2y-Jdz5KPLgQM%253Bhttps%25253A%25252F%25252Fwww.pinterest.com%25252Fpin%25252F546131892287351022%25252F&source=iu&pf=m&fir=GpJntqLozaYTaM%253A%252Cq2y-Jdz5KPLgQM%252C_&usg=__n2S6m6tLhllOrFZw9amYhDGOSZg%3D&biw=1440&bih=698&ved=0ahUKEwjWo9-BocjSAhVDeCYKHXnjDaIQyjcIJA&ei=E67AWJbOOcPwmQH5xreQCg#imgrc=GpJntqLozaYTaM:

You have to admire the creativity and rebelliousness.  Still I wouldn't want to be driving next to one, especially that first truck,  5K linear, over driven, no filter network at all:

Ah, yes, gotta love the Rat Rods!  Check out the one below... how does the guy see the street? Imagine how loud those open headers are.

I've got my heart set on a blue '67 GTO with an off-chassis restoration, 4-speed, black interior and even some pro-stock mods. Used to own two '64 GTOs in high school, but a '67 was always my favorite dream car.  

My second choice is a marina blue '66 Chevelle SS-396.

T

                                                                            ?
Looks like the "Mystery Date" game.


klc