An Urban Legend Disproved

Discussion in 'Ham Radio Discussions' started by N2EY, Jan 1, 2015.

Thread Status:
Not open for further replies.
ad: L-HROutlet
ad: l-rl
ad: HRDLLC-2
ad: Left-2
ad: Left-3
ad: L-Geochron
ad: L-MFJ
ad: abrind-2
ad: MessiPaoloni-1
  1. K5FH

    K5FH Premium Subscriber QRZ Page

    Early Amateur SSB experimenters built filters from WWII-surplus FT-241 and FT-243 crystals in the 5 MHz range because they were cheap and widely available. This necessitated mixing with a VFO/HFO in the 9 MHz range to access 75M and 20M and, as you noted above, led to the (in)famous sideband inversion scheme.

    A sideband filter with decent skirt attenuation requires that the individual crystals that make up the filter be matched very closely. FT-241 and FT-243 crystals had the advantage that their internal crystal elements could be removed from the holders and reground if necessary to move them closer to a specific frequency and thus create closely-matched sets of crystals. This is much easier to do with lower-frequency crystals simply because the quartz plates are thicker and easier to work with. There is less chance of overshooting the target frequency by grinding away too much. Plus, surplus crystals were cheap as dirt back then so if you screwed up a few, no big deal.

    It was also possible to order closely-matched crystals from commercial sources but, hams being who and what we are, this was pretty much cost-prohibitive for the time.

    Eventually, when commercial 9 MHz crystal filters with decent form factors became available at reasonable prices (e.g., McCoy, KVG), the mixing scheme was reversed: a 9 MHz filter with a 5 MHz VFO. This had the added advantage of increasing stability since it's much easier to build a stable 5 MHz L/C VFO than a 9 MHz one. As receivers were steadily improving (e.g., product detectors and sharp IF filters) and increasing in frequency stability the stability of transmitters became more important.

    As an aside, anyone who ever used a Drake TR-4 series rig will remember the weird mixing scheme that rig used (a heterodyne arrangement using four crystals) resulted in a sideband inversion on 20M, requiring a special reversed 20M scale on the VFO dial. The sideband selector switch had a "norm" and "opp" position. Why Drake did this, I have no idea, but it worked.

    The simple solution nowadays would be to simply do everything in USB and be done with it. But, like so many things from bygone eras, the tradition survives.

    If you can obtain a copy of the ARRL's Single Sideband for the Radio Amateur (an old red-and-black volume) it will contain several articles about homebrew crystal filters. Mine is from 1965.

    Also, if you can find one, the New Sideband Handbook by Don Stoner, W6TNS, is a great resource. My copy is from 1958.

    Great insight into the early days of Amateur SSB.
  2. W7UUU

    W7UUU Principal Moderator Lifetime Member 133 Administrator Volunteer Moderator Platinum Subscriber Life Member QRZ Page

    You posted all of this back a while ago actually in one of MY threads.... and I learned a LOT.

    It's a great "article' (despite being just a post and all) and a great read.

    Thanks Jim & Happy New Year

  3. W1GUH

    W1GUH Ham Member QRZ Page

    Speakin of Drakes and filter skirts (both were mentioned).

    In TR-3 and, presumablly in the TR-4 the filters were asymmetrical.
    The skirt towards the carrier was sharper for carrier suppression, while
    the skirt away from the carrier had a less steep slope. One reason why
    those rigs were so darn GOOD sounding.

    The Technical Material Corp. sideband adapters used L-C filters at 17 kc (!).
    The reason was to reduce the distortion as much as possible. That's why
    sound so good.

    Thanks for the post, Jim. It was of great benefit to me personally.
  4. N2EY

    N2EY XML Subscriber QRZ Page

    You have the general picture, Fred, but in the interest of accuracy there are some corrections to be made. The following is derived from various amateur publications of the day (GE Ham News, QST, ARRL books, Cowan publications, etc.) plus manuals of manufactured rigs, personal experience and the recollections of OTs I have known.


    Amateurs began using SSB in the early 1930s, when W6DEI and a few others built SSB transmitters. They generally copied commercial practice of the time. These rigs were all of the filter type. They generated the SSB at around 20 kHz (just above the audio range) and then required at least two frequency conversions to get to HF. The SSB generator was at such a low frequency because the filters used were LC filters, not crystal or mechanical filters. Due to the Great Depression, the complexity and expense, only a handful of such rigs were built. One design appeared in the magazine R/9, in the early 1930s.

    Immediately after WW2, the same low-frequency-filter approach was used. One design by W2KUJ in QST about 1948 generated the SSB at a low frequency with LC filter, then converted twice to get to about 5.2 MHz. From there, a third conversion with crystals or VFO in the 9 MHz range would give 75 or 20 - and the sideband inversion.

    In the late 1940s, SSB generation got simpler in two ways:

    1) WW2 surplus FT-241 crystals became available - and they were cheap. FT-241s were not HF crystals; they were in the range of about 370 to 1000 kHz. They aren't easily reground or reetched, but they were of such precision that they were usable as-is. They were used to make lattice filters in the 400-500 kHz range, for both transmitters and receivers. With such a filter, one could reach 75 meters with a single conversion, and 20 meters with two conversions.

    2) The phasing system of SSB generation had been known in principle for many years, but nobody used it because nobody knew how to make a simple, practical circuit that would give the required 90 degree phase shift over the audio range required for voice transmission (typically 300 to 3000 Hz). Then, in the magazine Electronics for December 1946, an article by R.B. Dome showed how to design and build wide-band audio phase shift networks, using only resistors and capacitors. The "Dome network" made the phasing system practical almost overnight - and divided amateur use between phasing and filter methods, both for receiving and transmitting.

    The Dome audio phase shift networks were simple circuits but required high precision resistors and capacitors if significant rejection of the unwanted sideband was to be obtained. The phasing method got a big boost when manufacturers began packaging the audio phase shift networks as a pre-adjusted unit, at reasonable cost. The most popular example was the B&W 2Q4/350 phase shift network, packaged in a metal tube shell the size of a 6J5.

    The phasing method had the advantage that it could be done at the output frequency - no conversions at all! The problem was a lot of adjustments when a significant frequency change was made. So frequency conversion phasing rigs became common.
    The use of 9 MHz SSB generation and 5 MHz VFO came mostly from the phasing folks. By the early 1950s it was being used in the CE SSB exciters (10A, 10B, 20A) and others, as well as homebrew rigs (W2EWL's "Cheap And Easy SSB"). With the phasing method, all you have to do to change sidebands is to reverse the phase of one audio channel, which can be done easily with a DPDT switch, so conforming to the "correct" sideband was easy.

    The use of FT-243 crystals to make SSB filters became common in the mid-1950s, once the design and adjustment techniques became well known. In this same time period, mechanical filters made by Collins became available to amateurs and were used in their equipment. Mechanical filters worked well, but were expensive and only practical up to about 500 kHz, requiring two conversions for the higher amateur bands.

    Packaged HF crystal filters for SSB first began to appear in the mid-1950s, made by companies such as Hycon Eastern. (see QST article "What's Wrong With Our Present Receivers?" in QST for January 1957). Such filters were intially very expensive (the Hycon filters cost about $44 each in the mid-1950s!) but over time prices came down. By the early 1960s, the McCoy "Golden Guardian" and "Silver Sentinel" filters were widely available.

    The "Silver Sentinel" price was $32.95 in 1963, and included two matched carrier crystals. Although relatively expensive, it was actually cheaper than ordering six new, precision crystals at $5 or more each, plus all the other parts, and making your own filter.

    In the interest of accuracy, a few words about the Drake TR-4 and TR-3:

    The heterodyne scheme used in those rigs is ingenious, and permits coverage of all the pre-WARC HF bands from 80-10 pretty easily. But it does have a few compromises.

    The design is single conversion with 9 MHz IF. It uses a premixer system (heterodyning the VFO signal) to cover 40, 15 and 10. The system uses 5 heterodyne crystals to cover 7 ranges of 600 kHz each.

    Here's how it works:

    On 80/75, the VFO tunes 5.5 to 4.9 MHz and is used directly to convert 9 MHz LSB to 3.5 to 4.1 MHz. Since the VFO is not the highest frequency, the sideband does not invert. Note that the VFO is tuning "backwards" compared to the operating frequency.

    On 40, the VFO tunes 5.5 to 4.9 MHz and is pre-mixed subtractively with 21.5 MHz from a crystal oscillator. This gives premixer output on 16.0 to 16.6 MHz, which is then used to convert 9 MHz USB to 7.0 to 7.6 MHz. Since the premixer output is the highest frequency, the sideband inverts. Again, the VFO tunes "backwards".

    On 20, the VFO tunes 4.9 to 5.5 MHz and is used directly to convert 9 MHz USB to 13.9 to 14.5 MHz. Since the VFO is not the highest frequency, the sideband does not invert. Note that the VFO is tuning "frontwards" compared to the operating frequency - which only happens on 20.

    On 15, the VFO tunes 5.5 to 4.9 MHz and is pre-mixed subtractively with 35.5 MHz from a crystal oscillator. This gives premixer output on 30.0 to 30.6 MHz, which is then used to convert 9 MHz LSB to 21.0 to 21.6 MHz. Since the premixer output is the highest frequency, the sideband inverts. Again, the VFO tunes "backwards".

    On the first 10 meter range, the VFO tunes 5.5 to 4.9 MHz and is pre-mixed subtractively with 42.5 MHz from a crystal oscillator. This gives premixer output on 37.0 to 37.6 MHz, which is then used to convert 9 MHz LSB to 28.0 to 28.6 MHz. Since the premixer output is the highest frequency, the sideband inverts. Again, the VFO tunes "backwards".

    On the second 10 meter range, the VFO tunes 5.5 to 4.9 MHz and is pre-mixed subtractively with 43.0 MHz from a crystal oscillator. This gives premixer output on 37.5 to 38.1 MHz, which is then used to convert 9 MHz LSB to 28.5 to 29.1 MHz. Since the premixer output is the highest frequency, the sideband inverts. Again, the VFO tunes "backwards".

    43.0 MHz is the second harmonic of 21.5 MHz. The 21.5 MHz crystal used on 40 meters could theoretically have been reused on the second 10 meter range by using the second harmonic, saving a heterodyne crystal, but Drake chose not to do that.

    On the third 10 meter range, the VFO tunes 5.5 to 4.9 MHz and is pre-mixed subtractively with 43.6 MHz from a crystal oscillator. This gives premixer output on 38.1 to 38.7 MHz, which is then used to convert 9 MHz LSB to 29.1 to 29.7 MHz. Since the premixer output is the highest frequency, the sideband inverts. Again, the VFO tunes "backwards".

    The end result is that on some bands the sideband inverts during the heterodyne process, but in others it does not. The tuning direction is reversed on 20 relative to the other bands, too.

    But Drake had one more trick up their sleeve!

    The 9 MHz filter in the TR-4 is actually two filters, one for LSB, one for USB, both designed to use a carrier crystal at exactly 9 MHz. As mentioned by W1GUH, the filters responses are slightly asymmetric, with the steeper slope on the carrier side, maximizing unwanted-sideband rejection. This design eliminates the need to use two dial cursors or a shifter circuit to make the dial read accurately on either sideband.

    When you operate the sideband knob on the TR-4, you're actually selecting which filter is used. The "upper" and "lower" indicator lights on the front panel are operated by a circuit that includes a section of the bandswitch and the sideband selector switch so that the lights indicate correctly for each band.

    The "X" on the sideband selector is for CW and AM operation and tuneup; it does not indicate "normal" or "opposite" sideband, and isn't marked that way. You just look at the indicators to know which sideband you're on. Or, just listen in the receiver.


    Note that sideband-inversion and tuning-direction-reversing in receivers and transmitters are independent of each other. Sometimes a heterodyne scheme will reverse one and not the other, sometimes both are reversed, sometimes neither. It all depends on the math - simple addition and subtraction.

    The tradition was so engrained by 1960 or so that many SSB rigs were made which could not work the "wrong" sideband at all. Others could, but only if you bought an optional extra carrier crystal.

    I have one. Practically everything in it is from QST articles. Note the dates of the articles - tells what the trends were.


    The GE SSB Handbook is online, free for the download. Has a collection of good articles from GE Ham News.

    If you want to see what may be the ultimate in homebrew SSB in the hollow-state era, google "LWM-3".
  5. K7RQ

    K7RQ Subscriber QRZ Page

    There is some good historical info here. When SSB was a new concept and no commercial gear was yet available, ARRL came out with a book called "Single Sideband for the Radio Amateur." It described the home brew setup described above, with a 5 Mhz. command set VFO mixed with a 9 Mhz. filter. You could buy those WW2 surplus command sets for peanuts and they were an amazingly stable VFO. This was the cheapest way by far to get on sideband, and by selecting sum or difference mixing it worked on either 75 or 20 meters. It was so popular with home brewers that McCoy and no doubt others began marketing 9 Mhz. filters. And this rig's output was USB on 20 and LSB on 75, and that is how the tradition started.
    I attended a convention in those early days when a rep from Central Electronics, one of the first companies out with some excellent SSB ham exciters, stood at a blackboard and using simple math, showed that on SSB, using an 807 as an output stage, you could get more talk power at the receiving end than you got with a kilowatt of AM. I was sorry to see them go out of business, their gear was right up there with Collins, qualitywise.
  6. W9JEF

    W9JEF Platinum Subscriber Platinum Subscriber QRZ Page

    The power of each sideband of 1kW AM is 250 watts.
    Of course on a selective receiver, only one sidband is detected.
    But isn't 250 watts pushing it a bit, for a single 807? :eek:

    Jim's debunking of the myth corrects some admittedly
    sloppy thinking on my part, and I thank him for it. :)

    I have an old CE 20-A exciter that Terry, W9DIA/SK gave me
    (when I worked for him at Amateur Electronics in Milwaukee).
    It's missing its audio phasing network; on my bucket list is
    to make another, with a 5 MHz ARC-5 Command xntr as a VFO.
    Those WWII aircraft radios were well-built, and rock-stable
    (for their time). Gear driven, so a logging scale will be on its knob.

    My first SSB radio was a used Swan 175, for which I paid $50.
    (At first I tapped into the power supply of my old hombrew AM rig.)
    It only covered the 'phone portion of 75 (3800-4000 at the time).
    Fortunately, it's heterodyning scheme allowed me to also go on 20,
    (using 3 bandswitches). It's lower (inverted) sideband on 75,
    became USB on 20, since it was adding, not subtracting.

    Last edited: Jan 2, 2015
  7. K7RQ

    K7RQ Subscriber QRZ Page

    Probably so, but he was there to sell us on SSB!
  8. N2EY

    N2EY XML Subscriber QRZ Page

    Yes there is....but also some inaccuracies.

    No, it isn't. Urban legend with no basis in fact. Sorry, but the tradition did NOT come from a filter rig using 9 MHz SSB generator and 5 MHz VFO.

    Here's proof:

    1) Yes, ARRL published a book called "Single Sideband For The Radio Amateur". It was a collection of QST articles about SSB, reprinted in convenient book form. The first edition came out in 1954, and SSB was advancing so fast back then that the second edition came out on 1958, and by 1970 they were on the fifth edition.

    But by 1954, when the book first appeared, the LSB/USB tradition was already in place. There were also several SSB rigs on the market back then, too - the Central Electronics 10A and 10B, for example.

    2) The construction of homebrew high-frequency crystal filters happened after the first edition of "Single Sideband For The Radio Amateur". Many different frequencies were tried.

    3) The use of 5 MHz VFO and 9 MHz SSB generation was popularized by phasing rigs of the early 1950s, not filter rigs.

    4) The McCoy filters appeared in 1960 - and came with both LSB and USB carrier crystals.

    In short, the use of a 9 MHz filter-type SSB generator and 5 MHz VFO is not the source of the tradition at all, because such a rig does NOT invert the sideband on either 75 or 20. The math proves it, and so does the history.

    He may have showed that, but unless the 807 is run at several times maximum ratings, it just isn't so.

    Here's proof:

    In those days, amateurs measured DC power input, not RF out. "A kilowatt of AM" meant 1000 watts DC in, and maybe 750 watts carrier out.

    With 100% modulated AM and no "supermodulation" and other tricks, the power in the sidebands is about half that in the carrier. So with a "kilowatt of AM", assuming a high-efficiency Class-C plate-modulated stage, you get 750 watts of carrier and 375 watts of sidebands.

    A single 807 is can produce maybe 40 watts of RF, at max ratings, Class AB2 SSB, under ideal conditions. That's less than 1/9 of the sideband power of a "kilowatt of AM".

    However, there's a little more to the story. With SSB, the receiving station can use a much narrower receiver passband - less than half the bandwidth of that required to receive AM and get all the sideband power. That narrower receiver means better signal-to-noise ratio - about equal to a doubling of power. Maybe even a tripling.

    But that's not enough to make a single 807 on SSB have more "talk power" than a kilowatt of AM. Not even close. The AM kW will still be about an S-unit stronger.

    However, if you look at a somewhat bigger tube - say, the 811A - things are a little different. A single 811A can produce an honest 150 watts in SSB service, and that, combined with the noise reduction of a narrower SSB receiver, can put the single-811A SSB rig in the same ballpark as a kW of AM.

    What happened to CE is that they got bought by Zenith, and the parent company decided to get out of the amateur radio market.

    The story is doubly sad because CE did some really advanced stuff.

    Their 100V, 200V transmitters, and the 600L amplifier, had patented bandpass output couplers and NO tuneup controls! No peaking the grid, no dipping the plate, no loading. Just like a modern SS rig - but more than 55 years ago, with tubes!

    CE had developed an extraordinary receiver called the 100R, too - but only one prototype was built, and then CE got out of the amateur radio business.
  9. N2EY

    N2EY XML Subscriber QRZ Page

    I have the info on what's inside a B&W 2Q4/350 audio phasing network, if you need it.

    Also have the info on converting an ARC-5 to be a multiband VFO for the 10A/10B/20A. It involves using one of the 1625s as a multiplier on bands other than 20 and 75.
  10. G4ALA

    G4ALA Ham Member QRZ Page

    I do not know whether of not US amateurs are required to use USB/LSB as appropriate. Here in UK bandplans are merely adivisory rather than manatory and any frequeqency (within a band) and mode can be used. Nearly everyone sticks to the band plan. Some switch the sideband used from the recommended sideband. This seems to help with adjacent channel interference. That is easy with modern rigs.

    That was a nice long posting at the head of string. I had read many times that the set of sidebands selected on each band was a solution to a component count mimimization problem founded in manufacturing/engineering. Could both answers be correct?


    John G4ALA
Thread Status:
Not open for further replies.

Share This Page