Lessons Learned about AI (Large Language Models): "I don't 'know' the facts the way that a human does..." Let's do an experiment...

Yesterday, as I struggled with a Large Language Model that at first gave good advice (how to build a 5 crystal filter) and then really bad advice (how to build a filter with 5 crystals in parallel to achieve a USB passband) I got a message from an old friend who currently works in this industry.  

He advised me to be careful about advice from LLM AIs.  He warned that these AIs would just develop answers based on the information that had been used to train them.  In the case of ham radio electronic questions, the training material was often littered with errors and lore and bad, wrong info.  Often this bad information would be repeated over many years.  This would make the AI more likely to use it in their answers. 

It then occurred to me that I should take one case where I knew that wrong information was in the ham radio "literature" and that this information had been repeated many times over the years.  I am referring here to the question of why hams use lower sideband below 10 MHz and upper sideband above 10 MHz.  I had explored this question in many blog posts.   Many explanations were offered -- all of them violated the Hallas rule.  All of them failed to answer the question.  I wondered, would AI be wrong?  

Well, AI was wrong.  AI said:  

Subtracting in heterodyne mixing inverts the frequency spectrum. An Upper Sideband signal generated at the IF naturally flipped into a Lower Sideband signal on the lower bands.

That is wrong.   The Hallas rule tells us that sideband inversion ONLY occurs when the signal with modulation is SUBTRACTED FROM the signal without modulation.  

I called AI on this error in its answer.  I asked it why it had given me bad information.  Its responses were telling. 

Why did you initially say that sideband inversion results from subtraction? Is this because of the material on which you had been trained?

You're asking a fair question about where that explanation came from.

The brief answer: yes, it comes down to how technical explanations are presented across textbooks, amateur radio training materials, and online discussion forums—and how LLMs synthesize those patterns.

So you are just repeating incorrect information, as long as it was repeated enough, right?

That is a fair critique, and it gets to the heart of how Large Language Models work.

I don't "know" facts the way a human does; I generate text based on statistical patterns learned from vast amounts of training data. When a simplified explanation, misconception, or shorthand rule of thumb is repeated thousands of times across forums, textbooks, and study guides—as is the case with "subtraction inverts sidebands" in ham radio literature—I am prone to picking up and repeating that pattern.

-------------------------

I think we can draw three conclusions from all this: 

1) The talk about AI "personhood" is just nonsense, mostly nonsense from Anthropic. 

2) The concerns about data poisoning are real.  The ham radio "literature" is already poisoned, and look what happens. 

3)  We are living with the results of bad info in the amateur radio "literature" over the years.  AI is just spitting it back at us.  And I think it is getting worse.  Now everyone with a callsign and camera has a YouTube channel. The quality of content is low and dropping.  Soon AI will be spitting this stuff back at us. 

---------------------------

Let's do an experiment.  Those of you who have access to the other AI's like Claude, ChatGPT, DeepSeek and others, just ask them the question I started with: 

"Why do ham radio operators use LSB below 10 MHz and USB above 10 MHz?"  Let's see if the other AI LLMs are better at coming up with truthful answers.  Of course, they may see this and not repeat the error. 

SolderSmoke Podcast #255 -- Accept the HB Challenge!, DeMaw SSB, Brilliant TR-3, Tube Talk, Ground Truth, Tales of Woe, SDR RX, Pico Balloons, MAILBAG

Mythbuster II -- 20 meters only

SolderSmoke Podcast #255 is avalable: 

Audio:      http://soldersmoke.com/soldersmoke255.mp3

Video:       https://www.youtube.com/watch?v=c54qiITMxwM

--   First: Happy Holidays!   I have on a Santa Claus hat!  

-- December 18.  Pete completes another orbit.  Happy Birthday Pete.  Please send him birthday greetings. 

 --  Bill was on Ham Radio Workbench: https://soldersmoke.blogspot.com/2024/11/bill-n2cqr-appears-as-guest-on-ham.html  Our challenge to HRWB.  Gauntlet thrown down... OUR CHALLENGE HAS BEEN GRACIOUSLY ACCEPTED!     We now extend the challenge to the entire SolderSmoke community: Build one of these:  https://hackaday.io/project/190327-high-schoolers-build-a-radio-receiver

Homebrewing is not for the faint of heart!  Accept the challenge!  Build stuff! 

Our question:  Did Doug DeMaw ever build an SSB transceiver?  Starting in September 1985 he wrote a five part series on an SSB TRANSMITTER for QST.  But he prefaces it by asking, "Why would anyone build an SSB transmitter today?" He says it would be fun "for the experience and understanding it would provide." But not for use, you see... And it is not a transceiver. 

Bill's theory about DeMaw, SSB,CW and sideband inversion:  He was a CW guy so sideband inversion did not really matter. He could get it wrong and still make it work.  

Pete's Bench: 

Brilliance and the TR3. https://www.youtube.com/watch?v=2C-eYB8yFzg&t=13s

A tale of woe.  Done in by a light bulb.

Thanksgiving dinner and SSB transceivers. https://n6qw.blogspot.com/2024/11/11292024-how-to-homebrew-thanksgiving.html             https://www.pastapete.com/

Hybrid plans:  https://www.youtube.com/watch?v=zKUjHZMf3Fs&t=5s

Dean's Bench:

Travelogue – Falcon 9 Launch

Building a homebrew T Match tuner for the end-fed long wave – sourcing the parts, winding the coil – taps, testing 

VWS Makers Projects

SDR Receiver Project – starting in January

VWS Pico Balloon – Traquito - Traquito - WSPR Pico Balloon

Revisiting the 10M DSB rig

Soldersmoke listener challenge – build a DCR with KK4DAS – overview then one board a week

SHAMELESS COMMERCE DIVISION:  Mostly DIY RF.    Please subscribe to our YouTube Channel.  And use the Amazon link on our blog.  Become a Patron via Patreon (on the blog). SolderSmoke is now on Blue Sky and Threads -- follow us or at least like us there. Please turn on automatic downloads on your podcast app -- most podcast apps will only store a few episodes. This will help bump the numbers, which will improve visibility.  Please give the show five stars and, if possible, a nice review on your podcast service,  That will help with the discovery rate for people looking for new podcasts.

Bill's Bench:

Is "The Ground" a Myth?   ARRL VP says Ground is a Myth: https://www.youtube.com/watch?v=KdX-978tvkY.  Bill disagrees.  Helicopter story.  Refrigerator story.  Original single wire telegraph system. https://en.wikipedia.org/wiki/Single-wire_earth_return

Front Panel and Freq Counter for the Mythbuster II. 

Another Tale of woe:  A mysterious audio problem on the 15-10 II rig.  Done in by.... Duh!   Comparing sideband suppression with Mythbuster I.  Differences in Hfe?  Notes on Mythbuster II build. 

TinySA Ap -- A cure for Fat Finger Syndrome? : http://athome.kaashoek.com/tinySA/Windows/  How to get and load the Ap (you might want to start watching at 1:11) https://www.youtube.com/watch?v=zu4X5dyUlpo&t=2s  General info video on the TinySA Ultra: https://www.youtube.com/watch?v=6C24RnYNOWQ&t=1143s

For the DR shack -- I got a Swan SWR-1A on E-Bay. 

 MAILBAG:   

-- Scott KQ4AOP trying to track down the DeMaw SSB transceiver mystery.  On the DC Receiver: This was my first receiver build and, it was great fun. When you finish the build and prove you are able to tune through the band, you are welcomed into the secret society! The build is the initiation.I am happy to print and ship the PTO if needed. 

-- Bill WA5DSS has built a High School Direct Conversion Receiver! 

-- Grayson KJ7UM liked the 1971 video on old THERMATRON AM radios: https://soldersmoke.blogspot.com/2024/11/basic-radio-circuitry-1971-film.html

-- Chris KD4PBJ sent nice electronic care package. 

-- Walter KA4KXX is honored that Dean named his dog for him (Walter was just kidding) 

-- Thanks to Bob W8SX for FDIM 2024 interviews.

-- Tony G4WIF insomnia driving him to podcasts.  Amazed by quantity of food eaten on Thanksgiving. 

-- Nice comment from Trigger about the podcast.

-- Clint says "valves" when he means THERMATRONS.   Kindly asks about "Oooo Thats Awesome"

-- Eric 4Z1UG faced with a new challenge.  Get well soon OM.

-- Sam WN5C and his Chat GPT AI Elmer. 

-- Paul VK3HN on using AI for electronic design.  I dunno... Apocalypse Now in the DR?

-- Tommy SA2CLC FB old military gear on QRZ site. Helps with HP8640B repair. 

-- Mike WN2A nice comments on Chappy Happy's FB Tezukuri DC RX https://soldersmoke.blogspot.com/2024/11/tezukuri-and-chappy-happy-amazing.html 

-- Allison KB1GMX.  Good info on ground truth. 

-- Phil W1PJE Had to throw out 15 test leads.  Fake wire!  

--Todd K7TFC Thoughtful comments on AI and ChatGPT, Help with TinySA Ap 

-- Steve KW4H Boatanchor guy.  Likes that we often scratch our heads trying to understand. 

-- Nick M0NTV built a 40 meter DC receiver: https://soldersmoke.blogspot.com/2024/12/a-40-meter-direct-conversion-receiver.html

-- Dave AA0KU asks about CCI amp (AN762)  Also woking on Drakes. 

-- Jack (Dhaka Jack!) F4WEF/AI4SV  Good thoughts on how to bolster SolderSmoke's ratings.

--Tobias thinks the decline IS ALL HIS FAULT!   

-- Tony VE7JUL building a TJ DC RX.  Go Canada! Dean says: 3D print PTO former at 110%

-- Jim KI4THC getting his uBITX on the air. 

10S November 23, 2024 1517Z SV1AER Kostas in Athens.  Said a very sincere “Oh my goodness! Congratulations!  That is not a very common thing!” when I told him rig was homebrewed.  Nice fellow.  Great response. 

Mike WU2D's Video on the SimpleX Super Receiver -- Part II

 Another FB video from Mike WU2D.   

But you know,  I too find myself kind of opposed to front panel on-off switches.  I power my rigs with small DC supplies.  I just turn on the supply when I want to use one of the rigs.  I don't have or need a switch on the front panel of the rig.  

I especially liked Mike's use of the gate dip meter and, of course, the Q meter.  FB OM. 

I must say I have a preference for the first version, but only because I dislike the regenerative circuit in the second version.  I do like the newer-style coils -- I have one in the BFO of the Mate for the Mighty Midget receiver.   

Thanks Mike for the sideband inversion factoid in Part 1!   The Hallas Rule -- words to live by. 

One word of caution.  I used 6U8s on my Mate for the Mighty Midget receiver.  I had good results, but WA9WFA had a lot of trouble.  We eventually concluded that the 6U8s didn't age well.  And they were quite long in the tooth.  We found (from the tube guys) that 6EA8s aged better and were a good and easy sub for the venerable (perhaps TOO venerable) 6U8s.  I switched tubes in my rig and it did seem to work better.  BTW, this is the receiver that I use to listen to the Old Military Radio Net on Saturday mornings.  

Here is the story of our switch from 6U8s to 6EA8s: 

https://soldersmoke.blogspot.com/2022/02/6ea8s-in-mate-for-mighty-midget-and.html

SimpleX Super Superhet Receiver -- A Great Video from Mike WU2D

Here is another great video (and project) from Mike WU2D.  I'm a big fan of homebrew superhets.  And wow, Mike presents a band-imaging superhet!  Two bands for (almost) the price of one!  I have FIVE homebrew dual-band band-imaging transceivers around me.  Believe me, once you have the experience needed to build an SSB transceiver, a dual-bander is the way to go.  Five bands seems like a bit too much.  But two seems to be at the sweet spot.  

I wrote to Mike reminding him to talk about the sideband inversion problem.  This rig will invert the 75 meter signals,  but this is easily resolved by just shifting the BFO frequency.  I also pointed out that many of today's builders will be detered by the need to scrounge for parts.  Where oh where is the BOM OM? 

Thanks Mike! 

Band Imaging Rigs (Receivers and Transceivers) -- Video from WA7MLH

In the video above (from 16 years ago) we see Jeff Damm, WA7MLH's  band-imaging receiver for 75 and 40 using an IF of 1.750 MHz and a VFO of 5.2 - 5.7 MHz,  For a signal at say 3.579 MHz (!) you subtract the signal from the VFO and you end up at the IF.  For a signal at say 7.030 MHz you subtract the VFO frequency from incoming signal and get to the IF.  (By the Hallas rule you get sideband inversion on 75/80 meters, but Jeff was on CW so this doesn't really matter.)  

Sixteen years ago this receiver was a work in progress and Jeff was having some trouble with the bandpass filters. I had similar trouble with bandpass filters. Like Jeff, I eventually got this sorted.  

I was happy to see a comment from my friend Joanthan-san on Jeff's old video.  

Jeff has an awesome and rececntly updated QRZ site:  https://www.qrz.com/db/wa7mlh

Band imaging like this is an old idea, and a very good one:  I used a slightly different scheme:   Start out planning on using a single conversion design.  Pick two bands you are interested in.  Select an IF midway between the two.  Build a single VFO that --when added to the incoming (or the outgoing) signal will get you to one of the bands, and when subtracted from the signal will get you to the other one.  Bob is then your uncle.  Two bands, with minimal switching. 

I got started with band switching with my Mythbuster rig:  I would get 75 and 20 meters.  The IF was midway between the two at 5.2 MHz.   My VFO (from an old Yaesu FT-101) ran around 9 MHz.  Boom, it worked, with the added benefit of receiving and transmitting LSB on 75 and USB on 20 with no switching of the BFO/Carrier Oscillator.

Then I did 17 and 12 meters.  Kind of a WARC-band special.  IF was at 21.4 Mhz.  VFO ran around 3.5 MHz.  So by adding the incoming modulated signal 18 MHz signal and the VFO, you get to 17 meters.   By subtracting the VFO from the incoming 24.9 MHz signal you get to 12 meters.  And both are on USB (apply the Hallas rule), so again, no switching of BFO/Carrier frequencies are required. 

Finally,  at solar max, I built rigs for 15 and 10.  Here the IF was 25 MHz.  Again the VFO was around 3.5 MHz.  Adding the incoming 21 Mhz signal to the VFO gets you to 25 MHz, subtraction of the VFO frequency from the in coming 28 MHz signal takes you to 25 Mhz and thus 10 meters.  Again, no sideband inversion (Hallas rule).   Both signals are USB and stay on USB. (I built two versions of this rig -- one stays in Virginia, the other is heading to the Dominican Republic.) 

In the ARRL book QRP Classics, there is an article from the 1990 Handbook entitled "A Band-Imaging CW Receiver for 10 and 18 MHz."  The article may have been based on a receiver built by Dave Newkirk AK7M (Rod Newkirk's son). Unfortunately in the write-up for the ARRL handbook, the drafters repeat the oft-repeated myth about how 9 MHz IF and a 5.2 MHz VFO would supposedly produce LSB on 75 and USB on 20.  This just doesn't work.   But if you put the IF at 5.2 MHz and the VFO at 9 MHz, it does work, as demonstrated by my Mythbuster rig. 

But why? Why Can't I Listen to DSB (or AM) on my Direct Conversion Receiver?

I've said this before:  I just seems so unfair.  We just should be able to listen to DSB signals with our beautifully simple homebrew Direct Conversion receivers. I mean, building a DSB transmitter is a natural follow-on to DC receiver construction.  And we are using AM shortwave broadcast stations (Radio Marti --I'm looking at you)  to test our DC receivers for AM breakthrough.  But when we tune these stations in, they sound, well, awful.  So unfair!  Why?   Unfortunately it has to do with laws.  Laws of physics and mathematics.  Blame Fourier, not me.  

Over the years there has been a lot of handwaving about this problem.  From Doug DeMaw, for example: 

In his "W1FB's Design Notebook," Doug wrote (p 171):  "It is important to be aware that two DSSC (DSB) transmitters and two DC receivers in a single communication channel are unsatisfactory.  Either one is suitable, however, when used with a station that is equipped for SSB transmissions or reception. The lack of compatibility between two DSSC (DSB) transmitters and two DC receivers results from the transmitter producing both USB and LSB energy while the DC receiver responds to or copies both sidebands at the same time."

That's correct, but for me, that explanation didn't really explain the situation.  I mean we listen to AM signals all the time.  They produce two sidebands, and our receivers respond to both sidebands, and the results are entirely satisfactory, right?  Why can't we do this with our Direct Conversion receivers?  I struggled with this question before: https://soldersmoke.blogspot.com/2015/07/peter-parker-reviews-dsb-kit-and.html  You can see in that post that I was not quite sure I had the answer completely correct. 

It took some discussion with a fellow Vienna Wireless Society member, and some Googling and Noodling for me to figure it out.  But I think I've got it: 

Imagine a station transmitting a DSB signal at 7100 kHz with a 1 kHz tone at the AF input.  There will be signals at 7101 kHz and at 7099 kHz.  Assume the carrier is completely suppressed. 

We come along with our DC RX and try to tune in the signal. 

Remember that they heart of the DC RX is a product detector, a mixer with the VFO (or PTO) running as close as we can get it to the suppressed carrier frequency (which we can't hear).  

Lets assume that we can somehow get our VFO or PTO exactly on 7100 kHz.  The incoming signals will mix with the VFO/PTO signal.  We are looking for audio, so we will focus on the difference results and ignore the sum results of the mixing.  

The difference between 7101 and 7000 is 1 kHz.  Great! And the difference between 7099 and 7000 is 1 kHz also.  Great again, right?  We are getting the desired 1 kHz signal out of our product detector, right?  So what's the problem?  

Here it is: SIDEBAND INVERSION.  Factoring in this part of the problem helps us see the cause of the distortion that plagues DSB-DC communication more clearly. 

Remember the Hallas Rule:   Whenever you subtract the modulated signal FROM the unmodulated signal, the sidebands invert.  So, in this case, we are subtracting that 7099 "lower sideband" signal FROM the 7100 VFO/PTO signal.  So it will invert.  It will become an upper sideband signal at 1 kHz.  We will have two identical 1 kHz signals at the output.  Perfect right?   Not so fast. Not so PERFECT really.  

The perfect outcome described above assumes that our VFO/PTO signal is EXACTLY on 7100 kHz.  And exactly in phase with the suppressed carrier of the transmitter.  But if it is even SLIGHTLY off, you will end up with two different output frequencies, signals that will move in and out of alignment, causing a wobbling kind of rapid fade-in, fade-out distortion.  You can HEAR this happening in this video by Peter Parker VK3YE, starting at 6:28:

And you can see it in this LTSpice simulation.  

This LTSpice model just shows two diode ring mixers.  The transmitter is on the top, the receiver is on the bottom. The transmitter has RF at 7100 kHz at L1 and audio at 1 kHz at R1.   The receiver has the VFO at 7100.001 L7,  DSB from the transmitter at L12 with audio appearing at R4.  It is instructive to watch the output as you move the VFO frequency.  If you move the VFO freq away from the transmit carrier osc frequency you will see the distortion.  Here is the netlist for the LTSpice simulation: 

http://soldersmoke.com/DSBDC.net

On paper, using simple mixer arithmetic, you can tell that it will be there. With the VFO/PTO just 1 Hz (that's ONE cycle per second) off, you will end up with outputs at 1.001 kHz and at .999 kHz.  Yuck.  That won't sound good. These two different frequencies will be moving in and out of alignment -- you will hear them kind of thumping against each other.    And that is with a mere deviation of 1 Hz in the VFO/PTO frequency!  We are scornful when the SDR guys claim to be able to detect us being "40 Hz off."  And before you start wondering if it would be possible to get EXACTLY on frequency and in phase, take a look at the frequency readout on my PTO.  

Now consider what would happen if the incoming signal were SSB, lets say just a tone at 7101 kHz.  We'd put our VFO at around 7100 kHz and we'd hear the signal just fine.  If we were off a bit we'd hear it a bit higher or lower in tone but there would be no second audio frequency coming in to cause distortion.  You can hear this in the VK3YE video:  When Peter switches to SINGLE Sideband receiver, the DSB signals sound fine. Because he is receiving only one of the sidebands. 

The same thing happens when we try to tune in an AM station using a Direct Conversion receiver:  Radio Marti sounds awful on my DC RX, but SSB stations sound great. 

My Drake 2-B allows another opportunity to explore the problem.  I can set the bandwidth at 3.6 kHz on the 2-B, and set the passband so that I will be getting BOTH the upper and the lower sidebands of an AM signal. With the Product Detector and the BFO on,  even with the carrier at zero beat  AM sounds terrible.  It sounds distorted.  But -- with the Product Detector and BFO still on --  if I set the 2-B's  passband to only allow ONE of the sidebands through,  I can zero beat the carrier by ear, and the audio sounds fine. 

There are solutions to this problem:  If you REALLY want to listen to DSB with a DC receiver, build yourself a synchronous detector that gets the your receivers VFO EXACTLY on frequency and in phase with the transmitter's oscillator.  But the synchronizing circuitry will be far more complex than the rest of the DC receiver. 

For AM, you could just use a different kind of detector.  That will be the subject of an upcoming blog post. 

Please let me know if you think I've gotten any of this wrong.  I'm not an expert -- I'm just a ham trying to understand the circuitry. 

Putting a Barebones Superhet on 17 Meters with an NE602 Converter (Video)

Armed now with a NanoVNA, I took a look at the passband of the 5 MHz filter in my Barebones Superhet (BBRX)  W4OP built it on a Circuit Board Specialist Board.  He put a 5 MHz CW filter in there;  I broadened the passband for phone by changing the values of the capacitors. Here is what the passband now looks like in the NanoVNA: 

This is what DeMaw would call an "LSB filter."  You would get much better opposite sideband rejection by using it with an LSB signal, placing the BFO/Carrier Oscillator slightly above the passband, in this case near 5.002 MHz. 

When I first built the down converter to get the 18.150 MHz signal down to the 7 MHz range (where I had the receiver running) I used an 11 MHz crystal for the NE602's local oscillator.  But this created a big problem:  18.150 - 11 =   7.150 MHz.  That is in the 40 meter band, but note:  NO SIDEBAND INVERSION.   Then in the BBRX  7.150 MHz - 2.150 MHz = 5 MHz  (the filter frequency) but again:  NO SIDEBAND INVERSION.   The signal started as a USB signal and remained a USB signal. 

I briefly tried shifting the BFO frequency to the other side of the filter passband.  If I could get it to around 4.985 MHz, it might work, but because the filter passband was so large, and because the crystal frequency was so low, I was unable to shift the crystal frequency that far.  In any case the results would have been less than ideal because of the "LSB" shape of the filter.  Back to the drawing board. 

I decided to cause one sideband inversion. 

At first I put a 25.175 MHz crystal module in my down converter.  This shifted the 17 meter phone band down to the 40 meter CW band.  It worked, but I cold hear strong 40 meter CW  signals being picked up by the wiring of the receiver (the box is plastic!).  I went back to the module jar in search of frequency that would move 17 meter phone to the 40 meter area (so I would not have to re-build the BBRX front end) but outside the actual 40 meter band.  

I ended up using a 25 MHz crystal in the down converter. 25 MHz - 18.150 MHz = 6.85 MHz WITH SIDEBAND INVERSION.  After checking on the NA5B Web SDR to see that there are no strong signals in the 6.835 to 6.89 MHz range, I retuned the output circuit on the converter and tweaked the input capacitor on the Barebones.  I shifted the VFO frequency down to 1.835 to 1.89 MHz and put the BFO at 5.002 MHz.   The receiver was inhaling on 17 meter SSB.  

One more change to the BBRX:  in his June 1982 QST article, DeMaw warned that trying to get speaker level audio out of the 741 op amp that he used would result in audio distortion.  And it did.  So I put one of those little LM386 boards I have been using into the BBRX box.  I just ran audio in from the wiper of the AF gain pot.  It sounds good.  

In effect this is my first double-conversion receiver.  I usually prefer single conversion, but this project has highlighted for me one of the advantages of double conversion for someone like me who eschews digital VFOs:   Starting with a crystal filter at 5 MHz,  with double conversion I could keep the frequency of the LC VFO low enough to ensure frequency stability.  That would have been impossible with a 5 MHz IF in a single conversion 17 meter rig.  But if I were starting from scratch for a 17 meter rig, I could stick with single conversion by building the filter at 20 MHz,  keeping the VFO in the manageable 2 MHz range. 

Now, on to the SSB transmitter.   The Swan 240 dual crystal lattice filter from the early 1960s needs some impedance matching. 

So Where DID the LSB/USB Convention Come From?

-- Bottom line:   I still don't know why ham radio adopted as a convention LSB below 10 MHz and USB above 10 MHz.  There are several theories. but so far there is no convincing explanation in favor of any one of them. And almost all of the people involved are probably Silent Keys by now; this makes it more difficult to gather first-hand information. 

-- I'm not even sure when the convention began to be observed in ham radio. Many of the early SSB books and articles make no mention of it. We don't see it in early ARRL Handbooks. The first mention of it that I found was in the 1965 issue of the ARRL's "Single Sideband for the Radio Amateur" page 8. This article claims that adding a provision for selectable sidebands would "add appreciably to the cost of the equipment. " It went on to say that,  "For this and other reasons there has been a species of standardization on the particular sideband used in the various amateur bands. Nearly all operations in the 3.5 and 7 Mc. phone sub-allocation is on lower sideband, while the upper sideband is used on 14, 21, and 28 Mc."  

-- We know that the informal convention was being followed as early as 1958.  Jim N2EY reports that in 1958, the manual for the Central Electronics 20A shows that LSB was the "sideband most commonly used" on 75, with USB preferred on 20:

-- Some cite a 1959 ITU recommendation on commercial multiplexed radiotelephony as the reason for the convention.  But I don't think this obscure and long-ago ruling explains the convention.  If this were the case, we'd  see follow-up FCC regulation, and at least some discussion of the ITU recommendation in the amateur radio literature.  But we see none of this.  And, as noted above, by 1958 hams were ALREADY -- on their own -- opting to use LSB on 75 and USB on 20.   The 1965 ARRL SSB book refers not to some hard-and-fast rule, but rather to  "a species of standardization" on LSB and SSB.  That ARRL book said nothing of the 1959 ITU recommendation. 

-- There is a widely held belief that this practice originated in the design of a rig that had a 5.2 MHz VFO and a 9 MHz filter.  According to this theory such a rig -- due to sideband inversion -- would produce LSB on 75 meters and USB on 20.  But, as we have demonstrated, this doesn't work, so this theory has to be discounted. 

-- Early SSB activity seems to have been concentrated on 75 meters, and there was a competition for space with AM stations.   SSB operators appear to have used the very upper band edge as their gathering spot.  Using LSB allowed them to operate very close to the upper band edge -- a lot closer than AM stations could go.  This may explain why LSB became the preferred SSB mode on 75.  But how do we explain USB on 20 and above?  That remains a mystery. 

-- It is important to remember that in the early days of SSB, for most hams there were only two important phone bands: 75 meters and 20 meters.  40 meters was CW only until 1952, and even after that was crowded with shortwave broadcast stations.   So a design that allowed for both 75 and 20 was twice as good as a monoband design. 

-- Early on there were designs and parts for phasing rigs.  You could take that ARC-5 VFO at 5 MHz, build a phasing generator around it, and then mix it with a 9 MHz to get on either band.  But with just a simple switch, this kind of rig could operate on USB or LSB on either band.  So the early popularity of this kid of rig does not explain the convention. 

-- There were a lot of surplus 5 MHz ARC-5 VFOs available. There were also FT-243 and FT-241 surplus crystals at both 5 MHz and 9 MHz that could be made into filters.  Later in the 1950s, 9 MHz commercial crystal filters became available.  If you used a 9 MHz filter with a 5 MHz VFO, there would be no sideband inversion in your rig.  If the SSB generator was putting out LSB on 9 MHz, you'd be on LSB on both bands.  So if there was a desire to have LSB on 75, why not just also have LSB on 20? 

-- But if you built a 5.2 MHz filter and a 9 MHz VFO,  you could have LSB on 75 and USB on 20 without having to shift the carrier oscillator frequency.  This would save you the trouble and expense of moving the carrier oscillator/BFO to the other side of the passband.  This desire to economize and simplify may explain why we ended up with LSB on 75 and USB on 20.  But this still begs the question: Why the desire for USB on 20?  

-- Both the manufacturers and the hams wanted there to be sideband standardization.  With monoband rigs, the manufacturers would be able to cut costs by building for only one sideband.  Hams also wanted to cut costs, and they did not want to have to figure out which sideband a station was on when trying to tune him in. 

-- By 1962-1963  Swan and Heathkit were selling mono-band SSB transceivers that used the "conventional" sidebands:  The rigs for 75 and 40 meters were on LSB while the 20 meter rigs were on USB.  There were no provisions for switching to the other sideband. This seems to have reinforced the practice of observing the convention.   (Heath later added sideband switching to the HW monobanders -- in view of the growing observation of the convention, they may have been better off sticking with their original design. Does anyone know why they did this?)  But again, why USB on 20 and above? 

--  In 1963, Swan, by then in Oceanside California, came out with the Swan 240.   Swan used a filter centered at 5174.5 kc. The VFO ran from 8953 kc to 9193 kc on 75 and 20.  The VFO ran from 12222 to 12493 on 40.  This gave the buyer 75 and 40 on LSB, and 20 USB with only one carrier oscillator frequency. (Swan offered a mod that allowed hams to install an additional, switchable carrier oscillator frequency.  I luckily acquired one such modified rig.)  But again, there is an explanation for LSB on 75, but why USB on 20 and above?

This is an important part of ham radio history.   There should be a clear answer.  We need to find it.   If anyone has any good info on this, please let me know.  

SolderSmoke Podcast #232 -- Mythbuster, Pete's Tube CW Rig, Pete's DC RX and Simple SSB Rig, NanoVNA and TinySA, Very FB Mailbag

SolderSmoke Podcast #232 is available -- Crank it in Robert!

http://soldersmoke.com/soldersmoke232.mp3

Featuring a guitar intro by Pete "Bluesman" Juliano,  playing his own composition: "Juliano Blues." 

Upcoming GQRP convention and the N6QW rig
Frank Jones and the FMLA -- Possible Victory?
IBEW Stickers:  NASA, Johns Hopkins APL....
Cycle 25 Lookin Better Today:  SFI 93   SN 47
 
Pete's Bench:
Toobular!  A Tube Transmitter
SR-160
Simple SSB rigs around the world! 
KI7NSS's Pacific 40
 
Bill's Bench
The Mythbuster and the Struggle Against the Urban Legend
W2EWL's Cheap and Easy SSB
W4IMP's IMP. Articles in ER by Jim Musgrove K5BZH and Jim Hanlon W8KGI
The Spirit of Homebrew SSB. From Electric Radio K5BZH December 1991
Reduced Front End Gain on the DIGITIA
Back on 17!  HP3SS sells HBR receiver to Joe Walsh
Maybe another Moxon?
 
SHAMELESS COMMERCE DIVISION
 
Test Gear
NanoVNA -- Alan W2AEW helped solve mystery of why NanoVNA not providing accurate readout of circuit impedance.  Over driving.  Need attenuator. 
TinySA -- Limited Resolution Bandwidth.  But you can listen with it!  See video on blog.
 
MAILBAG
-- Google Feedburner to end e-mails from the blog :-(
-- Paul VK3HN -- TIA AGC? Farhan and Paul looking into options 
-- Ciprian's Romanian Mighty Mite
-- Dino KL0S SolderSmoke GIF and graphical presentation on sideband inversion
-- Allison KB1GMX helped me on 24 volts to IRF 510 issue.
-- Dave K8WPE Wabi Sabi and Martha Stewart. And thanks for parts!  40673s!
-- Steve N8NM building a 17 meter rig with 22.1184 crystals in a SuperVXO and a 4 MHz filter.  
-- Dean KK4DAS restoring an old Zenith.  One hand behind your back OM. 
-- Pete Eaton debating SSB or DSB for 17.  Go DSB Pete!
-- Richard KN7FSZ a FB HBer.  Asked about my solid-stating of Galaxy V VFO.  
-- Walter KA4KXX on benefits of no-tune BP filters like Farhan's   FB. 
-- Jack 5B4APL on Time Crystals and Homebrewing in the 4th dimension.  FB OM!  
-- Moses K8TIY listens to the podcast with his young son Robert.  Crank it in Robert! 
-- Farhan and the SBitx on Hack-A-Day
-- Also Tom's receiver from junked satellite rig on Hack-A-Day
-- Todd K7TFC sent in beautiful message about the spirit of homebrewing. On the blog.
-- Grayson KJ7UM was on Ham Radio Workbench with George Zaf
-- AAron K5ATG running a uBitx with a  homebrew tuner and antenna.  Hope I can work him 
-- Heard Mike WA3O last night on 40 DIGITIA.  Water cooled amplifier

Joe Galeski's 1960 "IMP" 3 -Tube Filter SSB Transmitter, and the Spirit of SSB Homebrew

Here is another important bit of SSB history.  In  May 1960, Joe Galeski W4IMP published an article in QST describing his super-simple SSB transmitter.  While Tony Vitale's "Cheap and Easy" rig was a phasing design, Joe came up with a filter rig.  He built USB filter at 5775 kc.  With it, he ran a VXO at around 8525 kc. This put him on 20 meter USB. 

Here is the QST article: http://marc.retronik.fr/AmateurRadio/SSB/A_3_tubes_filter_rig_%28SSB%29_%5BQST_1960_5p%5D.pdf 

In discussing how to put this rig on other bands, Joe got the sideband inversion question exactly right: 

Thank you Joe!  

Joe even provides an comment that seems to capture an important element of the homebrew SSB ethos.  Joe homebrewed his filter, but he mentioned the possibility of using a store-bought filter: 

That's the spirit Joe!  

Along the same lines, Jim Musgrove wrote in Electric Radio: 

Having built Lew McCoy's Mate for the Mighty Midget receiver (which also used just three tubes), I can't help thinking that an IMP-ish transmitter would be an excellent complement to the Mate for the Mighty Midget.  

Jim Musgrove K5BZH knew Joe Galeski and wrote about him in the January 1992 issue of Electric Radio.  Jim wrote that Joe was an optometrist by profession. When OE1FF wanted to know the cost of building an IMP, Joe Galeski boxed up the original and sent it to him.  FB Joe. 

In December 1961 Joe Galeski published a QST article describing a transistorized version of the IMP -- this rig ran on 15 meters.  K5BZH wrote that Joe later published an article about a small, solid-state transceiver,  appropriately called "The Shrimp." 

W2EWL's "Cheap and Easy SSB" Rig -- And The LSB/USB Convention Myth

In March 1956 Tony Vitale published in QST an article about a "Cheap and Easy" SSB transmitter that he had built around the VFO in an ARC-5 Command Set transmitter.  Vitale added a 9 MHz crystal-controlled oscillator,  and around this built a simple phasing generator that produced SSB at 9 MHz.  He then made excellent use of the ARC-5's stable 5 - 5.5 MHz VFO.  His rig covered both 75 meters and 20 meters.  Here is the article:

http://nebula.wsimg.com/2b13ac174f7f2710ca2460f8cf7d6b8b?AccessKeyId=D18ED10DA019A4588B7B&disposition=0&alloworigin=1

Because it used the 9 and 5 frequency scheme, over the years many, many hams have come to think that Vitale's rig is the source of the current "LSB below 10 MHz, USB above 10 MHz." This is  wrong.   An example of this error popped up on YouTube just this week (the video is otherwise excellent): 

First, Vitale's rig had a phasing SSB generator. All you would need to switch from USB to LSB was a simple switch.  And indeed Vitale's rig had such a switch. Pictures of other Cheap and Easy transmitters all show an SSB selection switch. So with a flip of the switch you could have been on either USB or LSB on both 75 and 20.  With this rig, you didn't even need sideband inversion to get you to 75 LSB and 20 USB. 

Second, even if hams somehow became so frugal that they wanted to save the expense of the switch, leaving the switch out (as suggested above) would NOT yield the desired "75 LSB 20 USB" that the urban legend claims that W2EWL.   As we have been pointing out, a 9 MHz SSB generator paired with a 5 MHz VFO (as in the Vitale rig) will NOT -- through sideband inversion -- yield LSB on one band but USB on the other.   

W2EWL's rig could not have been the source of the LSB/USB convention.  I still don't know where the convention came from. I am still looking for the source. 

But leaving the LSB/USB convention issue aside, Tony Vitale's rig is an excellent example of early SSB homebrewing, and of a very clever use of war surplus material.  In the January 1992 issue of Electric Radio magazine, Jim Musgrove K5BZH writes of his conversations with Vitale about the Cheap and Easy SSB.  Tony told Jim that this rig came about because the Central Electronics exciters required an external VFO -- they recommended a modified BC458.   B&W had recently come out with a phase shift network. Vitale went ahead and built the whole rig inside a BC458 box.  FB Tony! 

In the December 1991 Electric Radio, Jim K5BZH reports that Tony was recruited into the ranks of SSBers when he watched a demonstration of SSB by Bob Ehrlich W2NJR in November 1950. Tony very quickly started churning out SSB rigs.  His daughter Trish Taglairino recounted that when her father had "done something great again" there would be a parade of hams to the basement shack.  About 30 guys showed up when Tony put his first SSB rig on the air -- they sent out for beer.  

Thanks to Jim for preserving so much SSB history. 

QST Recognized Error on Sideband Inversion, But Continued to Make the Same Mistake

 

I don't really know if this is good news or bad news.  It's good that in November 1985 they recognized the error, but then they allowed the same error to be repeated by the same author in the 1989 article "A Four-Stage 75-Meter SSB Superhet," and again in 1990 in W1FB's Design Notebook.   It also made it into the 2002 ARRL Handbook.  

Thanks to Chuck WB9KZY for alerting us to this Feedback piece. 

QST Repeatedly Got Sideband Inversion Wrong

It kind of pains me to do this. These articles are from a long time ago, and the author is an esteemed  Silent Key,  but the myth about the origins of the USB/LSB convention is still out there, and as a homebrewer of SSB gear I feel obligated to point out these examples of the error that that myth is based on.   

Last Friday, Pete WB9FLW and I were talking about homebrewing SSB rigs.  I recommended a series of QST articles by Doug DeMaw.  "Beginner's Bench:  The Principles and Building of SSB Gear" started in QST in September 1985. There were at least five parts -- it continued until January 1986. (Links to the series appear below.) I hadn't looked at these articles in years, but when I did, a big mistake jumped right out at me:  In the first installment, on page 19, Doug  makes the same mistake that he made in his Design Notebook:

"Now comes the conversion section of our SSB generator.  We must move (heterodyne) the 9-MHz SSB signal to 3.75-4.0 MHz. Our balanced mixer works just as it does in a receiver. That is, we inject the mixer with two frequencies (9 MHz and 5 MHz) to produce a sum or a difference output frequency (9 - 5 = 4 MHz, or 9 +5 = 14 MHz) If we are to generate 75 meter SSB energy, we must chose the difference frequency. We could build an 20-meter SSB transmitter by selecting the sum of the mixer frequencies. The RF amplifiers and filter (FL2) that follow would then have to be designed for 14-MHz operation.  In fact, many early two-band homemade SSB transmitters were built for for 75 and 20 meters in order to use this convenient frequency arrangement.  The use of upper sideband on 20 meters and lower sideband on 75 meters may be the result of this frequency arrangement (the sidebands become inverted when switching from the difference to the sum frequency.) " 

Those last two sentences are incorrect.  They repeat the "Myth," or the "Urban Legend" about the origins of the LSB/USB convention.  Contrary to what many hams now believe, with 9 MHz filter and a 5.2 MHz BFO it takes more than just switching from sum frequency to difference frequency to invert one of the sidebands. 

There are two conditions needed for sideband inversion to take place:  

1) You have to be taking the difference product (DeMaw got that right) 

2) The unmodulated (VFO or LO) signal must be larger than the modulated signal. (DeMaw and the ARRL obviously missed that part.  Repeatedly.) 

This is another way of stating the simple, accurate and useful Hallas Rule:  Sideband inversion only occurs when you are subtracting the signal with modulation FROM the signal without modulation. 

For DeMaw's claim to be correct, one of the SSB signals going into the balanced mixer would have to invert, and the other would have to not invert.  Let's see if that happens: He has the sideband signal being generated at 9 MHz and the VFO running around 5 MHz. 

9 - 5 = 4  But we are not subtracting the modulated signal FROM the unmodulated signal.  SO NO INVERSION

9 + 5 = 14   We are not subtracting at all.  SO NO INVERSION.  

Doug's convenient frequency scheme WOULD work if he'd just switch the frequencies of the filter and the VFO.  With a sideband generator on 5.2 MHz and a VFO around 9 MHz you do get the happy 75  LSB, 20 USB arrangement without the need to switch the carrier oscillator/BFO frequency.   That is what happened in the Swan 240, and that is what I have in my Mythbuster rig. I am listening to both 75 LSB and 20 USB without changing the carrier oscillator/BFO frequency.  My filter/BFO/product detector is set up for USB.   With this arrangement the 75 meter LSB signals DO invert, and the 20 meter USB meter signals do not, so both are able to make use of my USB BFO/product detector without shifting the BFO frequency. 

This error shows up again in DeMaw's the May 1989 QST article "A Four Stage 75-meter SSB Superhet" (reprinted in the ARRL's QRP Classics book).  Here he writes: 

"Should you want to cover both the 75- and 20-meter bands you can build a 20-meter version of FL-1 and band switch the two filters. As with the 75 meter only version, an IF of 9.0 MHz (Y1) is required. With this arrangement the 20 meter band will tune backwards from the 75 meter band, but upper- and lower-sideband reception will occur, as required, without changing the BFO frequency (Y2). This two band scheme with a 5-MHz VFO is an old one!"   NOTE: FL1 is the bandpass filter, not the IF filter.  

Doug's mistakes in this area may simply be due to the fact that he was more of a CW guy.  And this is something that is quite easy to confuse:  9 and 5 will get you to 75 and 20, but you have to make sure the VFO is at 9 if you want to make use of sideband inversion and avoid having to shift the BFO/ carrier oscillator.   I've made this mistake myself: 

https://soldersmoke.blogspot.com/2012/05/usblsb-urban-legend-debunked.html

In October 1993 I wrote to DeMaw about his Four Stage 75 meter SSB Superhet.  I think I was looking for details on how to put it on 20 meters.  As I recall, Doug wrote back telling me to just pick 20 meter values for the input bandpass filter.  Had I done so, I would have discovered that  -- for the reasons cited above -- this just wouldn't have worked on 20.  His BFO and filter were set up to receive LSB signals. That's fine for the incoming 75 meter LSB signals.  But on 20 -- contrary to DeMaw's thinking -- there would be NO sideband inversion. I'd be trying to listen to 20 meter USB signals with a receiver set up for 20 meter LSB.  

Did anyone else notice these errors.  Were there ever errata notices in QST on this?  

This is a reminder that you should take all technical articles and schematics with a grain of salt.  Many contain errors. We are all human, and this is a complicated subject with lots of details. 

https://pe2bz.philpem.me.uk/Comm/-%20ARRL/Build-Service-Design/Receivers/PrincipleOfBuildingSSBgear/8509017.pdf

https://pe2bz.philpem.me.uk/Comm/-%20ARRL/Build-Service-Design/Receivers/PrincipleOfBuildingSSBgear/8510027.pdf

https://pe2bz.philpem.me.uk/Comm/-%20ARRL/Build-Service-Design/Receivers/PrincipleOfBuildingSSBgear/8511016.pdf

https://pe2bz.philpem.me.uk/Comm/-%20ARRL/Build-Service-Design/Receivers/PrincipleOfBuildingSSBgear/8512037.pdf

https://pe2bz.philpem.me.uk/Comm/-%20ARRL/Build-Service-Design/Receivers/PrincipleOfBuildingSSBgear/8601029.pdf

The Unicorn! A 75 LSB /20 USB Receiver (That Can't Work)

 

Don't get me wrong -- I'm a huge fan of Doug DeMaw.  His books and articles are a treasure trove for ham radio homebrewers.  Also, Doug was an honest guy who admitted in the preface to his QRP book that at times he did not fully understand the circuits he was building; that kind of honesty is rare,  and is very helpful to amateurs who struggle to understand the circuits we work on.  

But everyone makes mistakes, and Doug made one in his "W1FB Design Notebook."  I present it here not as a "gotcha" effort to nitpick or sharpshoot a giant of homebrew radio, but because this error illustrates well the depth of the 75 LSB/20 USB myth, where it comes from, and how important it is to really understand sideband inversion.     Here is the mistake: 

That's just wrong.  A receiver built like this will not allow you to listen to 75 LSB and 20 USB "without changing the BFO frequency." (Am I the first one to spot this error?  Didn't anyone build this thing, only to discover that it, uh, doesn't work?)

Here's a little drawing that I think illustrates why the mythical scheme will not work: 

All confusion about sideband inversion could be avoided with the simple application of what I think we should call "The Hallas Rule": 

"Sideband reversal occurs in mixing only  if the signal with the modulation is subtracted from the signal that isn't modulated."  

Be careful here:   I think some arithmetic carelessness is responsible for much of the myth. Taking the difference frequency is not enough to produce sideband inversion. Read the Hallas Rule carefully:   For sideband inversion to occur, the signal with the modulation must be subtracted FROM the signal without the modulation.

--------------------------------------------- 

About the Swan 240's SSB generation scheme: 

I first stumbled on this problem when building my first SSB transmitters in the Azores.  I was using a VXO,  and a filter pulled out of a Swan 240 (5.173 MHz).  I started with VXO crystals at around 12.94 MHz.  The rig worked,  but I couldn't pull the VXO crystals very far.  So I switched to crystals at around 23.3 MHz (you can pull higher frequency crystals farther).  But look what happened:  My Carrier Oscillator frequency had been set up to receive USB signals on 17 Meters.  With the 12.94 MHz rocks, that worked fine:   18.150-12.977 = NO INVERSION.  But it all changed when I went to the 23 MHz VXO rocks:  23.323-18.150 = INVERSION!   This had me scratching my head a while.  I had to draw myself little spectrum pictures (like the one above) before I realized what had happened.  To get it to work -- to get it to produce USB on 17 meters -- I had to move the Carrier Oscillator to the other side of the passband. Good thing that Swan 240 came with TWO BFO crystals (5.1768 MHz and 5.1735 MHz). I just had to change the crystal. 

For 75 and 20 meters, the Swan 240 uses the correct 5.173 MHz filter with a 9 MHz VFO to get the happy situation of 75 meter LSB and 20 Meter USB WITHOUT changing the BFO/Carrier Oscillator frequency.  This is the Mythbuster scheme.  Unlike Doug's receiver, it works.  The scheme also works in the Swan 240 on 40 meters because for 40 the Swan rig has the VFO running from 12.073 MHz to 12.513 MHz. Here too, no change in the BFO/Carrier Oscillator  frequency is needed. But the Swan recommended a modification that would allow operation on 20 LSB and 75/40 USB!  It used a BFO/Carrier Oscillator crystal of  5.1765 MHz and a switch mounted on the front panel.  Luckily,  my junker Swan (acquired from HI8P in the Dominican Republic) had the second crystal -- mine was 5.1768 MHz.  It was that crystal that allowed me to get my Azorean SSB transmitter to work using the 23.9 MHz VXO rocks.    

SolderSmoke Podcast #231 -- Travel, SST, Mythbusting, Filters, TIAS, NanoVNAs, DC RX in SPRAT, Drake A Line, Spillsbury, STICKERS! Mailbag

SolderSmoke # 231 is available: 

http://soldersmoke.com/soldersmoke231.mp3

Annual Field Day Special Edition

Travelogue:   To the Dominican Republic! 

New dog -- Meet Guapo (see below). 

A great Father's Day for Pete and Bill.  I got a TinySA. 

Pete got some cool chick magnet glasses (see below).  

Watch out Newbury Park! 

Bill's Activity

SST Transceiver. Took it to Dominican Republic.

Made only one contact, but QRP-QRP. 

Not a lot of CW activity, and not a lot around 14.060. 

A lot more FT8 visible on the NA5B WebSDR.

That might be better for this kind of operation.  

I might try SST CW out today from the backyard. Field Day!

Fired up my 20 meter DSB NE602 rigs.  Made two contacts.  

Still trying to fully understand the NE602 Gilbert Cell.  Lots of mystery in there. 

Building "The MythBuster."  75/20 with sideband inversion.  

10 pole 5.2 MHz filter.  Used Dishal and AADE. 

Used NanoVNA to see the passband. 

G3UUR for crystal parameters.

Cohn Constant K topology. 

Also used NanoVNA to check input and output impedance on the TIA amps I will use around the filter. 

Pete's activity: 

Article on DC receiver in SPRAT. FB response. 50 receivers under construction.   

Work on Drake A Line.  

Jim Spillsbury. 

------------------------------

The N5JHH IBEW SolderSmoke stickers (see above and below).  How we will use them. 

Mailbag:

N2SVD

K8ITY

Tom (Junkbox receiver)

N0ZIB (Curse you, SolderSmoke!) 

DL6ID

N2NLY

VK2EMU

ZS1KE

AF7O

NG2E

VU2ESE

OK1RP

N5JHH