Real On-Air Spectrum Analyzer Test Videos - Put 'em on the table mate!

[K1JJ]

I was looking over some of the recent Amfone threads and was happy to see how many discussed bandwidth and improving signal cleanliness. Even a year ago, this was barely mentioned. My posting goal here is to make us even more aware of our own signal housekeeping obligations as Amers.

As HUZman will testify, I've been working diligently on both Fabio and Dr. Love to make them as clean and bandwidth controllable as possible. Rob, W1AEX was nice enough to make two videos of my RF spectrum display as he tuned across both rigs - both live, ON the air..  

First, Fabio, the 4X1 AM rig:   Fabio was on 3875, running full power and had the audio LP filter  set to 6.5kc.  As shown in the video, the bandwidth on one side was clean and dropped off sharply at about 6-7kc up on 'S' peaks.  For reference, there was an AM QSO on 3885 and K1MAN was on 3890. Each vertical division is 5kc. The drop-off point was down more than -40db. You can see that the adjacent QSO and Baxter were not touched.  Later, I took out the LP filter and the bandwidth was 9-10kc up. I then set it at 4kc and the spectrum matched 4-5 kc reasonably well. So, the rig appears controllable at whatever the audio bandwidth is set at. That's a good thing to test to be sure it is clean.
Fabio -  4-1000A modulated by a pair of 4-1000A's with LP audio filter set at 6.5kc.
http://www.youtube.com/watch?v=7Zd_NOIiSFo


(You may have to sign in with your YouTube password – let me know  if problems)


Next, Dr. Love - 8877 (3CX-1500A7) homebrew linear amplifier running 1500W output 3785. Dr. Love was running full power on LSB. The FT-1000D TX driver has a 2.8kc filter, so the ssb signal should roll off sharply at ~ 3.5kc. The 5kc divisions show the signal is contained well. Notice the lack of side AUDIO crud after 3kc as Rob tuned off to the sides. This is a good general test.
Watch the frequency tuning numbers:

http://www.youtube.com/watch?v=pkty-gwPXLA


My reasons for showing these spectrum videos is:

1) I’ve worked very hard to get these rigs clean. I had lots of bumps in the road, but it all finally worked out. It was hard-earned. I’m proud of the results.
2) Others will see the results and possibly inspire them to have their rigs tested by someone with an on-air spec analyzer. The reason for on-air is to include the antenna system and have the rig set at parameters that mimic real world situations. Antennas and connectors can arc and wreak havoc, of course, so include everything, even your own voice and mic.
3) It would be good to see posts of other rigs on the band so we can get a better understanding for various situations and equipment. We’re all learning here.

The Goals:
1)     Get our transmitter running as cleanly as possible and controllable to what audio we put thru it. If we put 4kc audio thru, make sure it’s 4kc coming out, not  6, 7kc or more from generated crud, IMD, etc.
2) Be sure to actually USE our LP filter or whatever we use to limit bandwidth during crowded times. For example, when we are packed in every 5kc from 3870 to 3895 during prime times, be able to dial in a narrower bandwidth and KNOW it works.
3) Find out the maximum power and sweet spot for our transmitters that passes #1 and #2 above.
4) Once confident the transmitter is clean, we will have knowledge that we are good neighbors on the bands.  Forget worrying about our own splatter and have fun!


Hope this helps and gets us all thinking more about bandwidth and cleanliness. I personally feel that more goodwill is done in the long run for AM by practicing audio bandwidth control on AM from ~ 4.5 to 7kc (per side) when needed, rather than chasing away QRM with intentionally generated wide signals, etc. Very few specialized ham groups bother with this subject. If we wish, we can be the exception and let it be known.

Tom, K1JJ

[Jim, W5JO]

It is a shame others don't have your diligence.

[Steve - WB3HUZ]

Mr. Vu runs a clean ship. He should be proud and the rest of us should be happy - no splatter when he's up/down the band. It's something to aspire to.

[k4kyv]

Another thing that helps is to keep all stations in a QSO close to zero-beat with each other.  Carrier frequencies scattered up and down the band cause the QSO to occupy excessive bandwidth even if all the  signals are clean.

Of course, this is impossible if two or more of the stations in a QSO are rock bound, but this is the exception not the rule.

But I am not immune to shifting off frequency ("exit stage left") when the interference is deliberate, or switching to my wide filter when another station fires up 2 kc/s away from my carrier frequency even though there are 15 kc/s of vacant band space on both sides of my QSO frequency.

[kb3ouk]

I'd liked to have seen what that insulator of mine looked like in a spectrum analyzer when i kept beating it with a screwdriver. I tuned around and could hear it as far up as 3885 but it just happened to be strongest on 3878. Actually wonder what 3870 looked like tonight to someone looking with an sdr or something like that that could see the signals, especially when everyone was banging on something, i guess i kinda got everyone started on that with the infamous ceramic insulator.
Shelby

[W1DAN]

Tom:

Excellent report!

It would be nice to see the grass in the spectrum analyzer better. Also a swept sinewave would show your bandwidth more easily.

I too believe in good bandwidth control and have for a couple of years run a brick wall eliptical low pass filter..something like 36db/octave at 6kc and at least 50db down at cutoff. This modulates my PDM class E rig which has a 3 stage PDM reconstruction filter. Previously I ran an Optimod that had a brick wall filter at 4.5kc.

Al W1VTP grabbed an image of my spectrum. I am good to +/-7kc or so. I may reduce this. See attached image of my signal from Al.

Also a better reading is a time averaged RMS from a station in the same town in the daytime so that skywave does not affect the reading. The best tap off is from an RF sample from the output of the transmitter.

I agree...lets be good neighbors and stay linear. And thanks to Al for the image!

73,
Dan
W1DAN

[W1AEX]

Tom,

Are you using Stu's cascaded audio filter design with the 4X1 rig?  

I was very impressed with what I was seeing at the time you were on. Other signals that were averaging 10 db below yours in strength were making consistent bandwidth excursions that extended as far as 12 kc (see screenshot) on each side of center frequency (3875). Obviously, this makes it fairly unpleasant for anyone else operating between 3865 - 3885. At any rate, that's some very fine engineering you have done there!

I would add that about 2 months ago an old friend of mine on the bands sent me an email and discreetly mentioned that I was spraying artifacts up and down the band with my old Viker. I really appreciated that and after going through things I found a pretty obvious gain distribution issue and also found that my audio had flipped out-of-phase after inserting an additional piece of equipment in my audio chain causing me to hit the baseline very heavily. That incident reminded me how important it is to run with the monitor scope on. Even with things adjusted correctly, and the audio brickwalled at 5 kc, I have no doubt that my little Viker is wider than your 4X1. Gives me something to think about when I'm chatting on 7295...

Rob W1AEX

[kb3ouk]

Anyone happen to see what the Tron's SBE looked like on an analyzer. I bet that was something interesting to watch as it wobbled around.

[K1JJ]

Tnx much for the comments, guys!

Dan, that's a great idea - to use a swept sine wave to show a more stable bandwidth. I wonder if I can get that off the computer somewhere.  I also plan to generate a triangle audio wave for testing too. Between these two it ought to make this work easier.

Rob:  Yes, this is Stu's cascaded digital LP filter that does -96db per octave.  It's an amazing filter to use on the fly. I use it all the time. When someone moves in close I just narrow it down to 4kc - and when it's quiet, I do 6.5kc. Works like a charm. Good war story about your Viker I.. :-)   Thanks again for doing those spec tests, OM.

Ya know, Jim, what really motivated me to do this extra work on the rigs is I got tired of being beat up over the years for building rigs that splattered...  We can get away with a lot in the AM window area cuz we all have a high tolerance for extra bandwidth and the ssb guys generally stay away. But when running a big linear down in the quiet DX window where locals are S9+60 over and the DX is S5, that's when it gets sporting. It doesn't take much for them to send out a lynch mob down there.  I've tried SO many different tubes and combinations of GG, grid driven amp configs. Most were just OK. Some where terrible. Same for AM. I've had a few plate modulated rigs that every so often a friend would complain to me I was taking out a QSO down the band. Does anyone enjoy QRMing his friends? I think not.

Running a "safe" ricebox with a stock filter is easy. Buying an Ameritron and running it lightly is also easy. But pushing the limits with a big linear or big hi-fi homebrew AM rig takes extra care. Not only are we 5-10db louder than the weaker stations, but we have a wider bandwidth.  This means we have to be 5-10db cleaner just to blend in - never mind the wider fidelity to deal with.  So it's not easy. Add in that rigs break and change as well as human error causing us to forget to bias the modulators correctly or load the amplifer properly, causing splatter.

I used to listen to guys running clean narrow signals. They operated without a splatter care in the world and I wondered how nice it would be if someday I could too. There’s nothing worse than having to make double wide space for yourself to avoid bothering other stations – and have to keep checking to see if they are talking about your splatter… HA!  

I know most everyone feels the same general way and if they had a choice, there’s no contest which they wud prefer.  I think a good thread would be HOW to build, test, tune and operate a clean linear amp and plate modulated AM rig. Ie, What are the specific hints and kinks to a clean signal?

T

[kb3ouk]

And if your transmitter is really hi-fi you might even end up passing some really high pitched stuff that you dont even know is there and you cant hear it on frequency but someone 8 or 10 khz up or down can and you never knew it was there till you got to look at your signal in a spectrum analyzer.

[kb3ouk]

It would be similar to what was happening last night, i would hit the screwdriver off of the insulator and all that you could hear on frequency would be tink, tink, tink, but if you tuned up about 8 khz, where i could ntjear any of my carrier but when i hit that thing all you could hear was BANG, BANG, BANG.

[W1DAN]

Tom:

Got an audio generator? I use an old HP for testing. However, a swept sinewave is easy to create with a computer audio editing program. I can email a file or post it here if needed. The file can be played from a PC, iPod or burned to a CDR and played via a CD player. At the end of the day, you'd be doing an audio frequency response test through a transmitter and detector/receiver. If you want the file, what tone sweep do you want?

Triangle waves are good for transmitter linearity, but your LPF will round the tips. If you use this signal on the air, you will be wide without the LPF.

73,
Dan
W1DAN

[K1JJ]

OK, I see what you meant, Dan. With the spec analyzer running, manually sweep the audio signal generator thru its range and watch the pattern fill in, assuming it's integrating.

Yes, if you're willing to make up a computer file for a swept tone, that wud be great. I'll bet many here could use it as a standard test. Maybe Gary can put it into a new category area called "Testing Tools" or something like that for anyone to access later on.   I would think an automated sweep from 1 hz to 12khz or so would be fine. Though, maybe starting at > 25hz is safer for mod transformers. Not sure how to make this selectable without doing two files. If it's not too much extra trouble, a triangle wave could be included. I once asked QIX what freq he used for the triangle tone, and seem to remember down in the 200hz zone, but not sure. Maybe you know. I think the lower, the more difficult to pass. What do you think would be a good compromise triangle tone freq for either a decent tube/transformer rig and for a class E rig?

Think about the best way to do this and post back.

Yes, testing will be wide without the LP filter, but if I'm not using a dumy load, it will be brief for a spec analy test during the quiet day only.

Thanks, OM!

T

[KF1Z]

Anyone happen to see what the Tron's SBE looked like on an analyzer. I bet that was something interesting to watch as it wobbled around.

Yup.

I wouldn't say as much "interesting" as "annoying".

Do the same thing by wiggling the vfo knob back and forth.

[ke7trp]

Wait wait wait..  When I posted plots of stations using my Icom People on this forum BASHED me and the ICOM saying its totaly useless.  Mainly smug flex users... Where are they now? Now the icom scope is FB?  LOL

 
Nice clean signals Tom.. Good to hear you last night!  Sorry I had to run.. We where starving.  When I got back from dinner, I had rain and an odd 30DB over Pink noise on the RXers.. So I just shut the station down.

There really is something to be said about putting out ALL of the power in a short passband like that. 



Clark

[K1JJ]

Thanks, Clark -

Glad you checked in last night. Strong signal from AZ on 75M. You started a good run into California. It was still only about 10PM or so - very early for west coast. The east coast group here had lots of fun.

I wouldn’t say the ICOM was totally useless… It is a reasonable representation. Look at the picture Rob posted of the station with 12kc audio. That particular station took out a QSO up the band. After the picture, the QSO went QRT cuz they couldn’t carry on due to the wide signal. It showed up clearly on the ICOM.  Yes, I’ve received a few shots from Flex spec screens and they looked more detailed. Though, they told me basically the same thing as the ICOM did.

In fact, I’ve used my own second receiver with a wire stub for an antenna as an “analyzer.” I could easily determine my signal at S9+40 over was down to around S9 out 3kc on the ssb linear. So there’s many ways to skin a cat – some are simply more detailed than others. I like Dan’s idea to put a swept tone thru to show the curve even better. 

Yes, I too like the idea of putting a dense transmit audio into a bandpass of  100hz - 5 to 6khz or so. I certainly hear the difference when the stuff above 6kc is chopped off, but it’s so small as not to make much difference to me. 

T

[W1DAN]

Clark:

Thanks for posting your concern.

I am not basking the scope, but you cannot pull exact numbers off of it-much the same could be said for an SDR plot.

The Icom spectrum scope is an interesting item. I just do not know the weighing, levels, bandwidth, or saturation point. This makes it a limited tool as compared to say an Agilent or HP spectrum analyzer.

It is a good start and easily a lot better than nothing.

Thanks!

Dan

[ke7trp]

If you are going to use the Icom.. Turn on the peak hold feature of the scope.. This way, we can see the peak value and bandwidth. I just tested my T3 this way. While it looks great on the mode like the videos have... When I turn on the blue peak line and talk for a while. I can see there is energy out alot farther then the instant display shows..

C

[W1DAN]

Guys:

Here are some audio test files created in Adobe Audition

Some are made at 22khz, some at 44.1khz, all are mono WAV files. The level is pretty hot! -3dbfs.

The white noise signal can be used with an averaging spectrum analyzer to see power bandwidth.

Note:

Be careful with your levels. Initially Run the tests at less than 50% modulation to save tubes and transformers. Later you can run full modulation tests for short durations.

Dan
W1DAN

[W1DAN]

More files!

Dan
W1DAN

[K1JJ]

Beautiful, Dan!

I will certainly start using them.

How about a swept tone from 20hz to 12khz that sweeps very rapidly up and down so that we could generate a wide response curve - just like when aligning an FM IF with a sweep gen?

Thanks, OM.

T

[ke7trp]

Tom.. Go to http://www.spectraplus.com/

Download the free Trial.  This software includes a built in Audio generator.  Just hook the output the computer to your audio rack. 

Then Hook the output of your reciever to the input on your computer.

Now you can do your own bandwidth tests with one simple program. You can configure the audio gen for time and Freq span.  I use a fairly slow 20, to 10K. 

You can adjust the EQ so you are flat and you can see how your filters are working.  I used this software when I had the FT1000D to run scans. 

C

[W1AEX]

Hopefully, there are no "smug" ICOM Pro III owners anywhere out there Clark! It has a capable receiver, certainly not near the top of the list when compared with many others though. The spectrum scope or band scope is a nice indicator, with rudimentary calibration points. Each horizontal line represents 10 db and each vertical line (at the resolution used at the time) represents 5 kc. It's a fine indicator but not a lab instrument. Great for watching a 200 kc swath of 6 meters for band openings.

As Dan has shown, Audition has a number of tone functions that can be used to generate just about anything you can think of. There's also a lot of simple (and free) software that can be configured to sweep as well. For a quick look at things I use Sig Jenny just because it's uncomplicated (and free). It's about 2.5 meg in size and can even be installed on a little flash drive if you wish:    

http://www.natch.co.uk/downloads/SigJenny/SigJenny.html

The screenshot shows how simple it is to use. In addition to sine waves it will also generate triangle, sawtooth or square waves if desired.

(Edit note: Runs fine on WinXP SP2 Intel Dual Core 3.6 ghz processor with 3 soundcards - Can't speak to Vista or Win7)

[ke7trp]

Thats a neat little program. If it worked on modern PCs it would be great. But it does not.  It only supports older PCs with old operating systems

I was just poking fun about the icom.. I posted plots of peoples signals and some of the guys attacked the radio and me claiming it was useless.  Then Tom posted some and nobody said anything. I just thought it was funny

I have the icom Pro Like the one in the video. Its a fine radio with a great little built in scope

C

[W1DAN]

Tom:

Here is the multisweep-20hz to 12khz.

Each sweep is 1 second in length, repeated many times. 22khz sample rate.

Dan