Podcasting since 2005! Listen to Latest SolderSmoke

Tuesday, January 4, 2022

Exorcism Completed! Getting Rid of the Spur in my 17 meter SSB Transmitter using a TinySA (video)

To re-cap:  The problem became evident when trying to "net" or "spot" my transmitter onto my receive frequency.  Around 18.116 MHz, I could hear at least two tones in the receiver as I moved the transmitter frequency.  I needed to get rid of the extra tone. 

First, thanks to all who sent in suggestions.  They came in literally from around the world, and this is a demonstration of the IBEW in action.  I used or at least tried all of them.  They were all good ideas. 

Following Vasily Ivananeko's pseudonymous suggestion I rebuilt the carrier oscillator (apologies to G3YCC).  I used the carrier oscillator/buffer circuit from Farhan's BITX20.   

Henk PA0EME said I should look at the signal level at the input ports of the NE602 mixer.  Henk was right --- the VXO input was far too high.  I lowered it, but the problem persisted. 

At first, I thought that the spur in question was so small that it would not show up on the air.  I could not see it in the TX output using my TinySA spectrum analyzer.  That was good news and bad news:  Good that it was not showing up on the air, bad that I could not see it in the TinySA and use that image in the exorcism. 

At first I thought that the spur was being caused by the 10th harmonic of the carrier oscillator and the third harmonic of the VXO.  This seemed to fit.  So, following VK3YE's sage advice, I built a little 69 MHz series LC trap (using a coil sent by AA1TJ, on a board CNC'd by Pete N6QW).  That trap succeeded spectacularly in crushing the 10 harmonic.  Look at these before and after shots on the TinySA: 

Before Trap

After Trap

Spectacular right? But guess what?  The problem was still there.  

I scrutinized the situation once more. I realized that the spur would be more visible if I put the TinySA on the input of the transmitter's PA (a JBOT amp designed by Farhan) as opposed to putting it on the output.  Watching the spur and the needed signal move in the TinySA as I tuned the VXO, I realized that they were moving in opposite directions.  This indicated that the spur was the result of a carrier oscillator harmonic MINUS a VXO-generated frequency (as the VXO frequency increased, the spur frequency decreased).  Looking at my EXCEL spread sheet, I could see it:  8th harmonic of the carrier oscillator MINUS the main output of the VXO. 

To confirm this, I plugged the values into W7ZOI's Spurtune program.  Yes, the spur popped up and  moved as predicted.  

For further confirmation I shut down the carrier oscillator by pulling the crystal from the socket, and then just clipped in a 5.176 MHz signal from my HP-8640B signal generator (thanks KB3SII and W2DAB). Boom!  On the TinySA, the spur disappeared.  Now I at least knew what the problem was:  a harmonic from the carrier oscillator.  

Following good troubleshooting practice, I turned off the gear and went to bed.  When I woke up, an idea came to me:  Before launching into a lot of filtering and shielding, just try running the carrier oscillator at a lower voltage, seeing if doing so might reduce the harmonic output.   I disconnected the carrier oscillator board from the main supply and clipped in a variable voltage bench supply.   Watching the signal on my TinySA, I watched as the spur completely disappeared as I reduced the voltage from around 13V to 10V  (see video above).  The main signal frequency level did not change much.  I tested this by listening for the hated extra tones.  They were gone.  Exorcised.  

Key lessons: 

-- Spur problems are difficult to troubleshoot.  Armstrong's superhet architecture is, of course, great, but this is definitely one of the pitfalls.  Single conversion makes life easier.  IF selection is very important. Choose wisely! 

-- When looking at the TinySA as you tune the rig, pay attention to which way the spur is moving.  This provides an important clue regarding the combination of harmonic you are dealing with. 

-- The TinySA is a very useful tool.  It seems like it is easier to use than the NanoVNA (which is also a fantastic tool). 

-- It can be fun and rewarding to re-visit old projects.  In the years between original construction and the re-look, new test gear has become available, and the skill and experience of the builder has improved.  So problems that once seemed insurmountable become fix-able. 

-- Thinking through a problem and thinking about possible solutions is very important.  It pays to step away from the bench to think and rest.  Rome wasn't built in a day. Here's a rough block diagram that I drew up (noodled!) while trying to figure out this problem: 

5 comments:

  1. And don't forget persistence as an essential trouble shooting ingredient.

    Superhet radio projects are combinations of a number of interworking modules. Progressively isolating, replacing or measuring each module in turn becomes a game of eliminating variables. We've all seen oscillators and gain stages become non linear when pushed. A reminder to only run the maximum Vcc necessary, and no more!

    Thanks for telling this story, we all learned from its resolution.

    Paul VK3HN.

    ReplyDelete
  2. Bill -

    Amazing detective job.

    FYI, Spurtune isn't on the EMRFD CDROM as far as I could tell.

    A web search says it was included with the W7ZOI book: "Introduction to Radio Frequency Design".

    Now to find my copy of that book and accompanying computer media.

    73,

    Chuck, WB9KZY

    ReplyDelete
  3. Sadly SPURTUNE.exe does not seem to work on modern windows 10 machines. Not even in combability mode.
    At least the version distributed in the irfd.zip package available from arrl.org.

    ReplyDelete
    Replies
    1. A thousand thanks for the tip ! I had found my copy of IRFD but the media envelop on the back flap of the book was empty.

      I've had pretty good luck running DOS programs on Windows 10 using DOSBOX, hopefully the IRFD programs will also work.

      73,

      Chuck, WB9KZY

      Delete
    2. Replying to myself :) Yes, spurtune does seem to work under dosbox.

      But Bill is using spurtune08 which was from 2008 (my copy of EMRFD is from 2003 ?) and spurtune08 does run under Windows 10. The program can be found on the wayback site for w7zoi.net on the emrfd errata page for the 3rd edition.

      Delete

Designer: Douglas Bowman | Dimodifikasi oleh Abdul Munir Original Posting Rounders 3 Column