I think the distinction needs to be drawn between a linear PA and a
plain old CW PA.

I have not succeeded in making a good linear. It usually either 1)
burns or 2) oscillates or 3) both oscillates AND then burns.

On the other hand, a CW PA is much easier and I am generally
successful with them, although not paying particular attention to the
efficiencies achieved.

Unfortunately I think for this project you will be wanting a linear so
you can send Multitone :-/

73 Hans G0UPL
http://www.hanssummers.com


On 8/5/06, tigger <aj-smith@...> wrote:
> I have a confession to make.  Whenever I need to make a small
> semiconductor PA I always have dreadful problems with instability, and
> achieving reasonable efficiency.  I usually blow up at least two,
> sometimes more PA devices and often become almost suicidal.  I am
> dreading tangling with the qrss tx, particularly as I have been
> proposing using the sa602, which has such a low power output that
> maybe two stages of pre-driver amplification will be required, raising
> the probability of instability further.
>
> Is there anyone out ther who finds this kind of thing easy - I am sure
> there are....
> Andy
>
>
>
>
>
>
> Yahoo! Groups Links
>
>
>
>
>
>
>
>

In my files I have found a design by Drew Diamond for a 2W 7MHz tx
line-up using a 602, 2 stages of driver, and a irf511 (similar to
510).  See diamond.jpg.  This could easily be modified for 10MHz and
seems to be an off the shelf design.  Has anyone any simpler
suggestions or comments ?

Andy

Woud anyone like to suggest a maximum power output for the TX ?  1W?

I suggest that the PA device should eb the IRF 510.  Any comments?

Andy

I have just spotted that the 7474 is positive edge-triggered, so have
ammended the spectral content doc accordingly.  I have also added a
few points to clarify the argument a little.

Andy

Hello all,

seems as if I have managed now to join this Yahoo group :-)

Will come back later after reading all the posted messages and files.


Chris, DL6JAN

I have a confession to make.  Whenever I need to make a small
semiconductor PA I always have dreadful problems with instability, and
achieving reasonable efficiency.  I usually blow up at least two,
sometimes more PA devices and often become almost suicidal.  I am
dreading tangling with the qrss tx, particularly as I have been
proposing using the sa602, which has such a low power output that
maybe two stages of pre-driver amplification will be required, raising
the probability of instability further.

Is there anyone out ther who finds this kind of thing easy - I am sure
there are....
Andy

Here is another 2-crystal filter - A ladder type this time.  This one
has 1k Ohm terminations, and has 150pF capacitors in all three
positions(shunting input & output, and in the middle).  It has an lsb
rejection of 43dB 8kHz from the passband, requiring an audio tone of
4KHz.

Look in the files for Ladder1.xls

Andy G4OEP

Andy

I don't know what level of harmonic attenuation we should be aiming
for but >50dB sounds pretty good. The only other reference I have to
make a quantitative comparison is reading somewhere in an Analog
Devices datasheet for one of the more recent DDS chips, where they
specified that the reference clock should be cleaner than 65dB (i.e.
no harmonics or spurii greater than -65dB relative to the
fundamental). It's hard to know how good it needs to be, or what
effect one would see if it was less good than optimal.

I still think it would be nice to know, using a spectrum analyser. Do
any members of the group have or have access to a SA that they could
use on a prototype of Andy's circuit? Alternatively, how about using a
receiver tuned to the fundamental to see how much of it can be
detected? I don't know exactly if that would be possible or how, but
someone might have some idea?

73 Hans G0UPL
http://www.hanssummers.com


On 8/5/06, tigger <aj-smith@...> wrote:
> How about this as an estimate of the 10.14MHz content of the doubler -
>
> My present qrss TX uses a single-transistor doubler to multiply the
> 5MHz output of my home-brew dds up to 10.14MHz.  If I look carefuly at
> the doubler output on the scope I can see that alternate peaks of the
> 10MHz waveform are slightly larger or smaller.  On a full screen of
> 8cm, the wobble is about 1.5mm.  I therefore estimate the sub-harmonic
> content to be -34dB (i.e. 20 log(1.5/80)).
>
> On the 20MHz output of the push-push doubler I cannnot see any trace
> of wobble even if I expand the trace so that it is much greater than
> the screen size.  Thus if I adjust the  Y gain so that the 8cm screen
> is full at a setting of 100mV/cm, then increase the gain to 25mV/m,
> shift the display so that I can see the top of teh waveform, there is
> still no visible alternate-peak wobble.  I therefore estimate the
> 10MHz content to be below 20 log(1/320)= -50dB.
>
> Precisely what level of 10MHz attenuation do we require ?
>
> Andy
>
>
>
>
>
>
>
> Yahoo! Groups Links
>
>
>
>
>
>
>

How about this as an estimate of the 10.14MHz content of the doubler -

My present qrss TX uses a single-transistor doubler to multiply the
5MHz output of my home-brew dds up to 10.14MHz.  If I look carefuly at
the doubler output on the scope I can see that alternate peaks of the
10MHz waveform are slightly larger or smaller.  On a full screen of
8cm, the wobble is about 1.5mm.  I therefore estimate the sub-harmonic
content to be -34dB (i.e. 20 log(1.5/80)).

On the 20MHz output of the push-push doubler I cannnot see any trace
of wobble even if I expand the trace so that it is much greater than
the screen size.  Thus if I adjust the  Y gain so that the 8cm screen
is full at a setting of 100mV/cm, then increase the gain to 25mV/m,
shift the display so that I can see the top of teh waveform, there is
still no visible alternate-peak wobble.  I therefore estimate the
10MHz content to be below 20 log(1/320)= -50dB.

Precisely what level of 10MHz attenuation do we require ?

Andy

> I have posted a rigorous analysis of this question in the files area.
>  The upshot is that only the fundamental breakthrough of the doubler
> need cause concern.

Nice analysis. I agree with your conclusions (I think!). I agree that
harmonics (even and odd) are Ok, but I think that they would be Ok
only so long as their magnitude relative to the fundamental is not too
large. You could imagine a situation where if the harmonic is large
enough, the flip flop would be clocked at the harmonic frequency
rather than the desired 20.28MHz. However this would be immediately
apparent so I think it is not a significant concern.

I think that however much theory tells us there will be no fundamental
component, I bet there still will! It will get through there
somewhere. Call me paranoid, but I think in any analogue circuit there
is always a good chance that if there's anything that CAN go wrong,
then it probably will! A simple lowpass filter might therefore not be
a bad plan!

73 Hans

I have posted a rigorous analysis of this question in the files area.
  The upshot is that only the fundamental breakthrough of the doubler
need cause concern.

In my doubler design I recommend bifilar winding of the secondaries of
the input transformer to equalise the drive to the doubler
transistors.  This leaves the gains of the transistors as a source of
unbalance.  If one is greater than the other the effect will be that
the two phases of the output will be unequal, but this will not
contribute a fundamental component at the output.

Here is a challenge - can anyone demonstrate a way in which this form
of doubler can produce a fundamental component at the output ? I am
sure one could be found, but bear in mind that the system is driven by
the input transformer with two fundamental signals of opposite phase
and equal amplitude.....

Andy G4OEP

From Andy-
>>I feel that the
>> push-push doubler is unlikely to have a significant 10.14MHz output,
>> but I am not certain about how sensitive the mixer might be to this.
>> I have not thought very deeply about it, but a fundamental component
>> in the 20MHz lo sig might influence the switching points of the 7474.
>>

From Hans-
>A useful analogy is phase noise. When you convert an analogue (sine)
>signal which contains amplitude or phase noise into a squarewave, you
>turn those noise sources into jitter on the digital signal. When fed
>into the switching mixer it will have the same effect as phase noise
>would have done in a different mixer type.

>I think the same is true of a fundamental or harmonic component.

We are thinking along the same lines Hans.  I will put some thought
into this - draw timing diagrams, etc, with different
harmonically-related components and work out exactly how the output of
the 7474 might be influenced.   That way what we do will be directed
by sound principles.

Andy

Hi Andy

> Hi Hans - this point had not escaped me !  I was asking about the
> 20.28MHz signal.  We discussed at one point adding a filter to reduce
> any fundamental component which might be present.  I feel that the
> push-push doubler is unlikely to have a significant 10.14MHz output,
> but I am not certain about how sensitive the mixer might be to this.
> I have not thought very deeply about it, but a fundamental component
> in the 20MHz lo sig might influence the switching points of the 7474.
>
> I do not have a spectrum analyser.  For a high-spec design I would
> instinctively filter the lo output (I am speaking now of the 20MHz
> signal in this case) in order to reduce spurious receiver responses
> allowed by harmonic content in the lo signal.  This seems irrelevant
> in the present case since it is inconceivable that the 7474 could
> switch at a harmonically related frequency at the same time as at the
> 20Meg frequency. The system is not comparable with a linear mixer in
> that respect.  It was this idea that I was inviting comments on.
> Sorry I did not make it clear.

I am more than ready to bow to the superior expertise in this area of
yourself or anyone else reading this: but instinctively I feel that it
is unwise to dismiss the contribution of fundamental and harmonic
components.

A useful analogy is phase noise. When you convert an analogue (sine)
signal which contains amplitude or phase noise into a squarewave, you
turn those noise sources into jitter on the digital signal. When fed
into the switching mixer it will have the same effect as phase noise
would have done in a different mixer type.

I think the same is true of a fundamental or harmonic component. I
think the resultant squarewave should be as squeaky clean as possible.
For me, this means removal of as much fundamental/harmonic content
from the doubler output as possible, so that the leading edges of the
squarewave occur at precisely regular intervals. I think that if
harmonic content is present on the sinewave, this will be manifested
as a jitter in the edges of the squarewave, which will give the
switching mixer spurious responses.

So although I'm not sure if I'm right, nor am I confident I have
explained it well, I think I would be in favour of a good double-tuned
bandpass filter at the doubler output.

73 Hans G0UPL

--- In qrsstxrx@yahoogroups.com, "Hans Summers" <hans.g0upl@...> wrote:
>
> Andy,
>
> Do you have a spectrum analyser? It would be very interesting to know
> the harmonic content of the resultant doubled oscillator output.
>
> The divided down output from the '74 should be UNFILTERED! The H-mode
> mixer requires a squarewave drive!

Hi Hans - this point had not escaped me !  I was asking about the
20.28MHz signal.  We discussed at one point adding a filter to reduce
any fundamental component which might be present.  I feel that the
push-push doubler is unlikely to have a significant 10.14MHz output,
but I am not certain about how sensitive the mixer might be to this.
I have not thought very deeply about it, but a fundamental component
in the 20MHz lo sig might influence the switching points of the 7474.

I do not have a spectrum analyser.  For a high-spec design I would
instinctively filter the lo output (I am speaking now of the 20MHz
signal in this case) in order to reduce spurious receiver responses
allowed by harmonic content in the lo signal.  This seems irrelevant
in the present case since it is inconceivable that the 7474 could
switch at a harmonically related frequency at the same time as at the
20Meg frequency. The system is not comparable with a linear mixer in
that respect.  It was this idea that I was inviting comments on.
Sorry I did not make it clear.

Re ladder filters- This configuration will make a good filter for our
requirements with 2 crystals.  I have not explored it so thoughly as
the 1/2 lattice.  Its main advantage is that you can easily control
the coupling between the two crystal resonators by the choice of
capacitors (3 in total) which form the 'steps' of the ladder.  The
system works like a textbook 'coupled resonator' system.  As you
reduce the coupling a double-humped responce arises, the humps
separating as you reduce the coupling further (greater capacitance).
This has the important advantage in our application that it enables a
last-ditch, sledgehammer technique for placing the pass-band (or at
least one of the humps) where you want it.  I was concerned at one
stage that it would be difficult to place the pass-band peak in the
required place with off-the-shelf crystals of a predetermined marked
frequency.  But the half-lattice has turned out ok in this respect, so
I lost interest in the ladder.  I think it might not be possible to
get quite such good lsb attenuation with the ladder type, since the
deep null in the stopband is a characteristic of the half-lattice and
phaser types.

The latest filter seems to be so good that I feel it is unlikely that
I would be able to do better with just 2 crystals whose frequency I
cannot choose, whatever the configuration.  After struggling with this
I feel I know a good one when I see it - and this one strikes me as a
miracle !  There are so many things to get right all at once -
passband freq, lsb attenuation, insertion loss.  Usually when you get
one right the others go wrong !

Des - is your filter anything like mine ?  You have not given us any
details yet...

It seems to me that the design of the rx is now complete (pending
comments from others, and minor details like adding more af gain, and
trimming the af frequency respone - I plan another op-amp with some
1st-order filtering).  So the tx comes next....

Andy

Des,

I am very excited by your results with the signal frequency crystal
filter. I think it is a very clear demonstration of how much
difference it makes, and should be a great encouragement to
development of this dedicated QRSS receiver.

Can you post details of your circuit arrangement? It might make a good
short SPRAT article in the next issue, as a follow up to my QRSS
article in the current SPRAT.

73 Hans G0UPL


On 8/3/06, Des <d29602960@...> wrote:
> Hi everyone,
>
>   I have posted the complete set of articles from Rad-Com's Technical
> Topicics which describes a method of improving front end selectivity
> with the use of crystals.
>
> Here, we have had some modest success with a crystal filter ahead of
> my SA602 based DC-RX (built along the same lines as Hans DC-RX) even
> though no optimisation had been done the prototype filter instantly
> removed all signs of AM breakthrough. Rough estimate of insertion loss
> is about 12 to 15 dB (ouch!) but because band noise is so high it does
> not seem to have made a whole lot of difference! Andy (G4OPE) has
> allready done some work on filters with a much more scientific
> approach, my efforts are intended to solve a short term problem that I
> and other SA602 based DC-RX users suffer, AM breakthrough!
>
> Hans, you were right, the signals outside of the passband did effect
> Argos AGC badly, with the narrowed BW of the xtal filter Argo behaves
> much better and both the contrast of screen captures and general
> performance are much improved. The filter also makes improvement with
> only one xtal fitted too. QRP cost Hi.
>
> Early days yet, still excited about the first results but I have much
> more to try. Equipment here is very limited so I rely partly on
> guesswork and "instinct", not very scientific I know but I get there
> in the end... slowly Hi. I will knock together some notes and
> schematics for the files area when I have it all figured out. None of
> my experiments will push back the frontiers of crystal filter design
> but they might be useful to the SA602 based DC-RX users.
>
> 73,s to all
>
> Des (M0AYF)
>
>
>
>
>
>
>
>
> Yahoo! Groups Links
>
>
>
>
>
>
>

Andy,

Do you have a spectrum analyser? It would be very interesting to know
the harmonic content of the resultant doubled oscillator output.

The divided down output from the '74 should be UNFILTERED! The H-mode
mixer requires a squarewave drive! Otherwise there is no point at all
to having done the doubling and divide-by-2 process. If you feed the
H-mode mixer (FST3125) with a sinewave, it will switch at some point
in the cycle and create effectively a non-50% duty cycle. It is
absolutely essential to leave the '74 putput exactly as it is: nice
squarewave, 50% duty cycle, as perfect as can get. If you can get a
74AC74 that would be optimum.

73 Hans G0UPL
http://www.hanssummers.com


On 8/4/06, tigger <aj-smith@...> wrote:
> I have just completed a breadboard push-push doubler.  This type of
> doubler has a low fundamental breakthrough, and I think it will be ok
> without additional filtering (just a tuned cct in the collectors).  It
> is ackling strongly - 8dBm output into 50 Ohms !
>
> What are people's views about filtering the lo output ?  The 7474 is
> going to produce more odd harmonic output than you can shake a stick at !
> Andy
>
>
>
>
>
>
>
> Yahoo! Groups Links
>
>
>
>
>
>
>
>

Looks great Andy.

I wonder also about ladder filters using 10140 crystals, and how they
would compare in this application?

73 Hans G0UPL


On 8/4/06, tigger <aj-smith@...> wrote:
> I have had some difficulty getting reproducible results in my filter
> tests, but feel I am now getting it together.  I have just uploaded my
> latest attempt to the files area as halflattice2.xls.  It has an
> asymmetric response, with insertion loss of about 4db and lsb
> rejection of -46db.  The lobe-null spacing is 1.7kHz, giving an audio
> tone @ 850Hz.
>
> This filter uses 2 HC18 xtals marked 10.140 in a half-lattice
> configuration without a shunting capacitor.  There are tuned matching
> transformaers at input and output with 50 Ohm terminations on the
> primary (input) and secondary (output).
>
> I cannot explain why this filter is asymmetric while the previous one
> using 8.275MHz HC18U xtals was symmetrical.  It might be that the new
> xtls are not as well matched as the previous ones.  If it turns out
> that crystal variations are significant with different pairs, it is
> possible that adding a capacitor across one crystal will allow the
> filter response to be adjusted to the required shape.  I have 2 more
> crystals on order and try them when they arrive.
>
> The filter seems to be ideal for our apllication, and I do not propose
> to look further for a suitable design.
>
> It now seems that the rx design is finalised (at least in my mind).
> The line up is -
> xtal filter,
> H mode mixer in adapted form,
> doubler, filter, bistable LO system,
> max4107 opamp.
>
> The lo needs to be worked out in detail, but this strikes me as being
> routine stuff...
>
>
> Andy G4OEP
>
>
>
>
>
>
>
>
> Yahoo! Groups Links
>
>
>
>
>
>
>
>

I have just posted a circuit diagram of a 10.14MHz VXO and doubler O
system to the files area

Andy G4OEP

I have just completed a breadboard push-push doubler.  This type of
doubler has a low fundamental breakthrough, and I think it will be ok
without additional filtering (just a tuned cct in the collectors).  It
is ackling strongly - 8dBm output into 50 Ohms !

What are people's views about filtering the lo output ?  The 7474 is
going to produce more odd harmonic output than you can shake a stick at !
Andy

I am following with interest your tests.
The idea to put a cristal filter in front of my DC RX seems good.
I have here some 10140 xtals , I just tested a serial xtal to
a 50 ohm resistor (rx side)  .
The LSB rejection should be at a first sight ar. 15-20 db.

To have the best LSB and power line interferences rejection
I suggest with DC RX
to set the VFO around 10 137 000 HZ (beat 3000 hz) .

I will try  if this simply xtail net will makes some difference.I
have the maximum Broadcast rejection using a 12v car  battery
for all the DCRX , and sometimes I used an LC preselector


with 2 sections  variable air capacitor.

My RX is under the roof near the antenna , away from the Home PC.
I have multiwire line for the remote control functions + analog audio
out.


                                          73 de Paolo I1DFS.



--- In qrsstxrx@yahoogroups.com, "tigger" <aj-smith@...> wrote:
>
> I have had some difficulty getting reproducible results in my filter
> tests, but feel I am now getting it together.  I have just uploaded
my
> latest attempt to the files area as halflattice2.xls.  It has an
> asymmetric response, with insertion loss of about 4db and lsb
> rejection of -46db.  The lobe-null spacing is 1.7kHz, giving an
audio
> tone @ 850Hz.
>
> This filter uses 2 HC18 xtals marked 10.140 in a half-lattice
> configuration without a shunting capacitor.  There are tuned
matching
> transformaers at input and output with 50 Ohm terminations on the
> primary (input) and secondary (output).
>
> I cannot explain why this filter is asymmetric while the previous
one
> using 8.275MHz HC18U xtals was symmetrical.  It might be that the
new
> xtls are not as well matched as the previous ones.  If it turns out
> that crystal variations are significant with different pairs, it is
> possible that adding a capacitor across one crystal will allow the
> filter response to be adjusted to the required shape.  I have 2 more
> crystals on order and try them when they arrive.
>
> The filter seems to be ideal for our apllication, and I do not
propose
> to look further for a suitable design.
>
> It now seems that the rx design is finalised (at least in my mind).
> The line up is -
> xtal filter,
> H mode mixer in adapted form,
> doubler, filter, bistable LO system,
> max4107 opamp.
>
> The lo needs to be worked out in detail, but this strikes me as
being
> routine stuff...
>
>
> Andy G4OEP
>

I have had some difficulty getting reproducible results in my filter
tests, but feel I am now getting it together.  I have just uploaded my
latest attempt to the files area as halflattice2.xls.  It has an
asymmetric response, with insertion loss of about 4db and lsb
rejection of -46db.  The lobe-null spacing is 1.7kHz, giving an audio
tone @ 850Hz.

This filter uses 2 HC18 xtals marked 10.140 in a half-lattice
configuration without a shunting capacitor.  There are tuned matching
transformaers at input and output with 50 Ohm terminations on the
primary (input) and secondary (output).

I cannot explain why this filter is asymmetric while the previous one
using 8.275MHz HC18U xtals was symmetrical.  It might be that the new
xtls are not as well matched as the previous ones.  If it turns out
that crystal variations are significant with different pairs, it is
possible that adding a capacitor across one crystal will allow the
filter response to be adjusted to the required shape.  I have 2 more
crystals on order and try them when they arrive.

The filter seems to be ideal for our apllication, and I do not propose
to look further for a suitable design.

It now seems that the rx design is finalised (at least in my mind).
The line up is -
xtal filter,
H mode mixer in adapted form,
doubler, filter, bistable LO system,
max4107 opamp.

The lo needs to be worked out in detail, but this strikes me as being
routine stuff...


Andy G4OEP

The Story so Far

The question of the design for a simple qrss transceiver originated on
the qrss net.  http://mail.cnts.be/mailman/listinfo/knightsqrss_cnts.be
An ongoing interest there is the design of distinctive identification
patterns which can be displayed via Argo.  One way of creating these
is to drive a varactor controlled vxo from a dac.  This technique can
be used to create ramps, sawteeth, triangular waves, etc, but is very
limited since fm does not lend itself to the simple generation of
more complex patterns.  Helschreiber-like patterns require a complex
yet easily defined spectrum of rf signals (so-called 'multitone'), and
this in turn requires a ssb transmitter.  The discussion therefore
focussed on the design of a simple ssb transmitter.

As would be expected, the rival merits of phasing-type direct
conversion (pdc) and superhet schemes were discussed.  Hans (G0UPL)
pointed out that the extremely narrow bandwidth (100Hz) of the qrss
sub-band greatly simplified the design of the phasing systems required
by pdc, and made it an attractive option.  Andy (G4OEP) replied that
the narrow band width requirement could also simplify the design of
the filter in a superhet, since extremely simple 1 or 2-crystal
filters such as the crystal phasing filter (1 crystal) or the
half-lattice filter (2 crystals) would give satisfactory performance.
  He believed that this gave the superhet an edge in terms of simplicity.

Des, M0AYF, who uses a dsb dc receiver, and is occasionally troubled
by broadcast breakthrough, introduced the idea of a signal-frequency
crystal filter ahead of a dc receiver, and distributed some copies of
Technical Topics (Radcom) material describing the application of
signal-frequency filtering using a variety of crystal filter
configurations.  Andy recognised that a single-frequency ssb receiver
with a crystal filter in the front end was the logical solution to the
narrow-band qrss requirement.  The principle of using the superhet
technique to apply the bandwidth-defining properties of a fixed if
filter to a range of rf signals at different frequencies just is not
required.  Simple crystal filters such as the phasing filter or the
half-lattice have deep nulls in the stop-band, and a narrowly peaked
passband;  if the carrier insertion oscillator (CIO) is placed between
these, good lsb rejection can be achieved very simply, albeit over a
narrow range of frequencies.  Thus was born the concept of a TRF ssb
receiver with a signal-frequency crystal filter. Discussion of other
types (pdc and superhets) stopped at this point, and there seems very
little point in re-opening this area of discussion unless some
previously missed point can be contributed.

At present the main topics of discussion are 1) filter design, and 2)
mixer design.

Filter
A crucial concern is the availability of suitable crystals.
Fortunately a supply of 10.140,000MHz crystals is planned by a member
of the qrss net, and the question then becomes – can a suitable filter
be designed using these crystals ?  It is required above all else that
low passband attenuation be located in the region 10,140,000 to
10,140,100Hz.  The attenuation slope of the filter is not of such
importance, since qrss software (e.g. Spectrum Lab) can deal with
audio tone up to 15kHz or more: the pass- and stop-bands could
therefore be separated by up to 7.5kHz.  Ultimate rejection is also
not crucial, since it could be assisted with additional LC filtering
if necessary. This is a narrow ssb filter; it should have good lsb
rejection, and everything else is of less importance.

Three configurations (possibly more) should be considered. – 1)
single-crystal phaser, 2) 2-crystal half-lattice, 3) ladder filters
with 2 or more crystals.  Andy performed some initial trials using
crystals marked 8.275MHz from his junk box before he acquired any
10.14MHz crystals. He could not find a way of making a phasing filter
with the passband in the required position relative to the nominal
frequency of the crystal.  The half-lattice seemed to have more
potential in this respect: a plot of a typical response curve has been
placed in the 'Files' section.  Alternatively, a two-crystal ladder
filter could be made in such a way that, increasing the coupling
capacitor to reduce coupling, a classic two-humped response can be
made, one hump of which can be tuned on to the required frequency
range. The other hump is at a lower frequency,  between the passband
and the stopband.  Getting good sideband separation by this technique
is not easy.  Later experiments (currently progressing) with 10.14MHz
crystals, confirms that a half-lattice filter with acceptable
performance (>40dB sideband rejection, acceptable bandwidth correctly
placed, and low insertion loss) is possible.  It remains to be seen
whether other configurations can be better or comparable.  Experiments
in this area are very relevant to current interests.

Mixer
Two options have been discussed – the H-mode design, and diode-ring
mixers such as the SBL1.  Mixers such as the SA602 having low
third-order intercept are of no interest in the receiver, but could be
used in the transmitter.

The H-mode mixer has outstanding imd performance and low noise, and is
of great interest.
http://www.warc.org.uk/cdg2000/The%20CDG2000%20Transceiver.htm
  It requires fairly complex LO drive circuitry (e.g. frequency
doubler, filter, bistable waveform  squarer) of which several variants
have been published, and requires adaptation for use as a ssb detector
with audio output.  Both of these issues have been discussed.  Hans is
strongly in favour of a no-compromise full design using all of the
techniques recommended for achieving high intercept point.  Andy is
more inclined to compromise to achieve simplicity.  But overall,
agreement is settling in favour of a rigorous, if more complex design,
even if this results in imd performance far in excess of what is
strictly required.  Andy has given some initial thought to adaptation
for af output, and has developed a solution which replaces transformer
primaries with differential op-amp circuitry.  A proposal for this
modification has been placed in the 'Files' section.

Diode Ring.  This type of mixer has a lower intercept point, and
requires a high level of LO injection, but is a generally robust type
of mixer, and could undoubtedly give good performance in a design with
narrow front-end selectivity.  It remains a possibility, but interest
in the H-mode is strong.  For optimal performance, all ports of this
mixer should be properly terminated; this requires a buffer of some
kind (probably an amplifier) between the filter and mixer's rf port.
This adds complexity to the design , and involves the possibility of
degrading the imd performance by raising the signal level, and might
introduce non-linearity in the amplifier.  A passive attenuator pad
could be used, but would degrade noise performance.  A suitable
amplifier buffer with 10dB gain is published in the 1991 ARRL Handbook
p 30-11.

Gain.
It is hoped that adequate noise performance can be achieved by
providing all of the gain in the receiver at af.   Sprat 121 (Winter
2004/5) p8 ff contains a comparison of the noise performance of
several op-amps, and quotes a noise figure of 0.75nV/rtHz for the
MAX4106.  This is 13dB below the NE5534, and holds out great promise
for achieving a low-noise design.  Using this op-amp instead of a
NE5534 is equivalent to providing 13dB of low-noise rf gain.

The Transmitter.
Little thought has been given to this at present.  It is assumed that
a simple mixer such as the SA602 could be used.  Detailed design of
the vxo, crystal oven, PA and output filter are all required, but seem
to be routine issues raising no particular areas requiring debate:
nevertheless contributions are invited. A suitable thermo-controller
for the crystal oven can be found here -
http://g4oep.atspace.com/qrss/qrss.htm.

I hope that these notes will stimulate discussions in areas of
interest.  I shall be posting supporting literature on this site in
the near future – keep an eye on the 'Files'.

Hi everyone,

    I have posted the complete set of articles from Rad-Com's Technical
Topicics which describes a method of improving front end selectivity
with the use of crystals.

Here, we have had some modest success with a crystal filter ahead of
my SA602 based DC-RX (built along the same lines as Hans DC-RX) even
though no optimisation had been done the prototype filter instantly
removed all signs of AM breakthrough. Rough estimate of insertion loss
is about 12 to 15 dB (ouch!) but because band noise is so high it does
not seem to have made a whole lot of difference! Andy (G4OPE) has
allready done some work on filters with a much more scientific
approach, my efforts are intended to solve a short term problem that I
and other SA602 based DC-RX users suffer, AM breakthrough!

Hans, you were right, the signals outside of the passband did effect
Argos AGC badly, with the narrowed BW of the xtal filter Argo behaves
much better and both the contrast of screen captures and general
performance are much improved. The filter also makes improvement with
only one xtal fitted too. QRP cost Hi.

Early days yet, still excited about the first results but I have much
more to try. Equipment here is very limited so I rely partly on
guesswork and "instinct", not very scientific I know but I get there
in the end... slowly Hi. I will knock together some notes and
schematics for the files area when I have it all figured out. None of
my experiments will push back the frontiers of crystal filter design
but they might be useful to the SA602 based DC-RX users.

73,s to all

Des (M0AYF)

Explore the 'Files' section - I have placed some material there which
might be of interest.

Andy G4OEP

This site contains original material by the inventer of the h-mode mixer.

http://www.warc.org.uk/cdg2000/The%20CDG2000%20Transceiver.htm

Andy G4OEP

Hallo to the group: I0SKK is just came too...:-) Best 73 and glad to
join to you!
73

Alex I0SKK

I have just received 5 Max4107 low noise opamps.  smd - ooh-er

Andy G4OEP

Discussions for the qrss transceiver are based on an initial proposal
which can be found here -

http://g4oep.atspace.com/qrss/qrss.htm#Proposals for a Dedicated
Home-Brew QRSS Transceiver

Comments directed towards realising a design along these or other
lines are invited.  The aim is to develop a design which has high
performance, yet simple enough to be attractive for home construction.


Andy G4OEP