EMRFD Message Archive 172

Message Date From Subject
172 2006-10-18 17:07:36 Wes Hayward The ARRL Challenge--SSB Vs DSB
Hi all,

I've received several notes from folks interested in the ARRL "$50
Challenge." Its really encouraging to see that so many people are
thinking about the problem. W1ZR at the League is to be
congratulated, for the program is already a success.

One of the common questions that I hear is "SSB or DSB?" There is
no simple answer. Indeed, one of the things that this competition
might do is to generate experimental data about that question. I
have commented that DSB may be plagued with greater on the air
distortion and that is the thing I want to address in this note:

Consider a simple example: Assume that we have a two tone audio
generator with outputs at 1200 and 1700 Hz. We will first inject
this into a filter type SSB transmitter and then into a DSB rig.
Assume the carrier is at 1 MHz.

In the SSB transmitter, the two tones are present in the audio at
1200 and 1700, and there could be IMD within the audio stages.
The separation of the tones is 500 Hz, so we will have the potential
for audio IMD 500 Hz above the high tone and 500 Hz below the low
tone at 700 and 2200 Hz. This is easily investigated with
experiments with a base band spectrum analyzer operating with a
sound card in a computer. Typically the audio stages would use
heavy feedback, probably with op amps, and there will be minimal
distortion. The same can be said for a well designed balanced
modulator (with proper drive levels) and an IF amplifier before the
crystal filter. It is all a matter of design, but it can be done.

The transmit mixer will not be a really bad problem, again if
handled with care. This mixer could offer a more interesting
design challenge.

We now arrive at the RF power chain. The input to the chain is a
single sideband (assume USB) consisting of two tones 1200 and 1700
Hz above 1 MHz. The two will now intermodulate in the power chain
to create a spectrum with 3rd order IMD products at 1,000,700 and
1,002,200 Hz, along with the desired outputs at 1,001,200 and
1,001,700 Hz. Consider only the 3rd order distortion products,
and forget about the higher order stuff. Notice in our example
that the 3rd order IMD products are still within the spectrum that
would contain the desired voice stuff from perhaps 1,000,300 to
1,003,000 Hz. There is no "splatter" in this ideal SSB case.

We now apply the same two tone audio generator to a DSB rig.
Assume that we continue to do a good job with the audio design and
the balanced modulator. We also assume that in both transmitters,
SSB and DSB, the balanced modulator is so good that there is no
carrier in the output.

We now have four desired tones in the RF spectrum appearing at 1200
and 1700 Hz above and below the 1 MHz suppressed carrier: 998,300;
998,800; 1,001,200; and 1,001,700. These are the
desired, "designed" outputs. They would be received in a SSB
receiver, so only half would contribute to a meaningful voice
output. The RF power chain can now generate two third order IMD
products for each combination of two tones. Let's label the tones
as A, B, C, and D, respectively. The combinations are then AB, AC,
AD, BC, BD, and CD. (AB is the same as BA, etc.) There are six
combinations, each generating two IMD tones, so we have 12
distortion products where we had just 2 in the original SSB case.
(We have confined our thinking to simple distortion between two
tones. But a real world amplifier subjected to four equal tones
may well have more complicated distortion. The usual two tone
analysis would no longer apply. This case could get really

The extreme frequency spread case, AD, is interesting. The
separation between these tones is 3400 Hz, so the related distortion
products will appear 3400 Hz below low tone A and 3400 Hz above tone
D, at 994,900 and 1,005,100 Hz respectively. This represents 10
kHz of occupied spectrum when the distortion products are included.

The example audio tones that I chose were picked for simple
arithmetic and to illustrate that there could be distortion in all
parts of the transmitter chain, SSB or DSB. What if we considered
the most extreme audio input cases of 300 and 3000 Hz instead of the
original ones. Or how about 2900 and 3000 Hz. You can "do the
math" for both modes and see that it is possible to generate a lot
of distortion (splatter) with DSB and that it can extend for quite a
distance, even considering only the third order IMD.

This problem with DSB was pointed out to me by Rick, KK7B, in a
discussion about six months or a year ago. I had a new VHF
DSB/CW transmitter that was nearly finished at that time. It has
remained in that partially finished state for months. I'll
eventually finish it, but the priority has dropped by about 20
dB. The discussion with Rick, however, continues.

There are other factors to consider for "The Challenge." The
popular system that we hear folks talking about, for both SSB and
DSB, is an IRF510 or 511 PA at 5 watts output with a 12 volt power
supply. We reported IMD measurements in EMRFD for a PA of this
sort. The short story is that these FETs are cheap and easy to use,
but are not very good for IMD. They get much better when you run
the voltage up to 24 to 28 volts, but they are poor at 12. A
better choice might be a bipolar such as the 2SC1969, which is still
affordable from RF Parts (although higher than the IRF510.) The
super cheap 2sc5739 may or may not be a good performer in linear
service. Only measurements will tell.

One recent question I heard regarded the use of bipolar transistors
Vs FETs for power amplifiers. Both can offer low distortion.
Data suggests that FETs are capable of lower high order (5th and
7th) distortion than bipolar transistors.

I have used DSB in simple VHF rigs for years with success. DSB has
provided me with a simple rig that I can use to chat with the local
guys. Almost always, the situation I encounter here in the
Pacific Northwest is that of sparse band occupancy. I may find
a group of guys chatting on or near the 144.200 MHz 2 meter calling
frequency without much else going on. I may also find a net up at
144.240, but that is so far away that one could be using DSB at
144.2 and not bother those folks, so long as the DSB rig is not over
driven. Similar arguments apply to the 6 meter band. Band
occupancy will be much different in other parts of the country.

A VHF contest is a different situation, and this happens a few times
per year. Then you can find a proliferation of signals, often all
grouped very close to the calling frequencies of 50.125 and
144.200. An opening on 6M during a contest is the extreme
case. My attitude is that DSB should rarely be used in such a
situation, unless the power level is well below that possible with
full drive to the PA. This is the time to switch to CW. The DSB
transmitter that could generate a lot of junk on phone turns into a
pristine, clean source of RF when operated on CW. The same goes
for a SSB/CW rig. Of course, switching to CW on a VHF band will
immediately reduce the number of folks that can copy you. This is
rarely a major problem, for the interesting "DX" stations are
usually not CW challenged.

The situation is slightly different on 40 M, the designated band
for "The Challenge." A weekend with a lot of activity, even during
the day, may create a situation where DSB might not be advised. A
transceiver using a DSB transmitter and simple DSB direct conversion
receiver would work well. The argument here is that if you don't
hear a lot of stuff on the sides with your direct conversion
receiver, there is no need to worry about distortion products from
the transmitter. An old friend, Jeff--WA7MLH, has built numerous
rigs of this sort for 75 and/or 40 M over the years and has used
them with great success on backpacking junkets. A major downside
of this is that two of these rigs are not suitable for phone
communications with each other. Jeff uses homebrew SSB rigs when
at home.

The bottom line is a reflection of personal goals and values. In
spite of the ongoing debate with Rick, I think that a DSB/CW rig is
practical, and especially appealing for its simplicity. But it must
be designed and used with an understanding of the extended potential
for close in interference. My own design philosophy is to design
ham gear is limited by what is possible within reasonable practical
constraints for a given application rather than to design things
that barely meet an existing set of loose specifications, such as
those imposed by the FCC. Most other ham designers will disagree
with me on this.

I think that experimentation and careful thought will be required
of "The ARRL Challenge" winner. A successful entry will require
an expenditure on the part of the designer/experimenter (DE) that
exceeds the $50 duplication goal. The DE may have to invest money
or effort in instrumentation well beyond the multimeter limit
allowed for the final builder. I also suspect that a successful
design will be one in a large number that the DE will have built.
Alternatively, he or she may just built just one bench "prototype",
but it will undergo numerous changes during the design process.

Remember that no matter what happens so far as winning The League
Challenge, all who participate are sure to learn a lot from the
process. All participants will win.

73, Wes, w7zoi
173 2006-10-18 19:50:49 jr_dakota Re: The ARRL Challenge--SSB Vs DSB
Do AM rigs (assuming they are not overmodulated) suffer from the same
sideband splatter as DSB or does the (usually) high level modulation
scheme filter this out somewhat?

In case anyone is interested Dieter, W8DIZ over at Parts and Kits (
www.partsandkits.com ) sells the 2SC1969 2 for $6 making it quite
cost effective using 1 in the Challenge ... he also has ADE-1 2 for $4
which Jim Kortge, K8IQY has used these with great success in a couple
of his 2N2 line of CW transceivers

174 2006-10-18 21:10:39 Mark Re: The ARRL Challenge--SSB Vs DSB
Wes, you suggest using the bipolar 2SC1969 instead of the MOSFET
IRF510 or IRF511. Is the 2SC1307 a good substituti
175 2006-10-19 16:43:27 Wes Hayward Re: The ARRL Challenge--SSB Vs DSB
Hi Mark, et al,

I am not familiar with the 2sc1307 so I can't really comment.
Clearly experiments are in order here. This is the sort of thing
that will probably be required when designing for "The Challenge."
I suggested the 2sc1969 as a good bipolar PA in this power class.
The IMD specifications are impressive. The IRF510 and 511 HEXFETs
are easy to use and are cheap. However, the IMD is not outstanding
at 5 watts out. It is an individual value judgement as to what you
want to use. IMD from a 511 may not be a major issue with a 5W QRP
rig. The IMD was excellent in a rig I built long ago that used an
IRF511 at 1.5 watts output.

I would purchase a 2sc1969 and a NTE advertised equivalent and
measure the performance of both before publishing anything using the
NTE part. No matter what you use, I would urge you to tell the
world exactly what you used and not assume any accuracy about claims
of equivalence.

Good luck.

73, Wes, w7zoi
176 2006-10-20 00:03:25 Mark Re: The ARRL Challenge--SSB Vs DSB
Thanks for your reply. Unfortunately, I don't have a scope or a
spectrum analyzer. I do homebrewing and experimenting, but I don't
have the equipment to run these kind of performance tests.

73 de Mark