Message-ID: <4E307D02.7080905@iup.uni-heidelberg.de>
Date: Wed, 27 Jul 2011 23:02:58 +0200
From: =?ISO-8859-1?Q?Stefan_Sch=E4fer?= <Stefan.Schaefer@iup.uni-heidelberg.de>
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 6.1; de; rv:1.9.1.8) Gecko/20100227 Thunderbird/3.0.3
MIME-Version: 1.0
To: rsgb_lf_group@blacksheep.org
References: <4E2410BC.7070405@iup.uni-heidelberg.de> <CAA8k23TQMhV_Ew-WMA8i1P5H94ty5bLaksWqhxsjwDxZKsEbEQ@mail.gmail.com> <4E2452BC.5050408@iup.uni-heidelberg.de> <D81113E5E794498285F29E6EEED8C2DC@JimPC> <4E255A9B.8090502@iup.uni-heidelberg.de> <6F17BBDDB2624CB4B81DF1B446E928F2@JimPC> <4E287230.40708@iup.uni-heidelberg.de>
In-Reply-To: <4E287230.40708@iup.uni-heidelberg.de>
Subject: Re: LF: TX converter for 137 kHz
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: quoted-printable
Sender: owner-rsgb_lf_group@blacksheep.org
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org
X-AOL-SCOLL-URL_COUNT: 0  
x-aol-sid: 3039ac1d60c94e307e0a596a
X-AOL-IP: 195.171.43.25
X-AOL-SPF: domain : blacksheep.org SPF : none


LF,

I going on with my monologise  ;-) ...about my LF TX converter (maybe=20
someone silent is interested) that directly convertes 12.7 kHz=20
(generated by SpecLab) to 137.7 kHz in an IQ mixer. It is a modification=20
of G4JNT's design...

The converter is quasi finished and works well!
The last steps have been:
I measured the output impedance of the transformer, it is 100 Ohm. Then=20
i added a PI low pass filter that transforms to 10 kOhm. This filtered=20
signal looks almost like a perfect sine wave, just very small peaks are=20
modulating the signal. I checked the narrow band spectrum when applying=20
a single tone at 12 kHz and when applying a white noise, limited to=20
11...12.8 kHz.
Single tone:=20
http://dl.dropbox.com/u/19882028/LF/Sidebands%20with%20single%20tone%20at=
%2012%20kHz%20analog%20output.jpg
White noise:=20
http://dl.dropbox.com/u/19882028/LF/Sidebands%20with%20white%20noise%2C%2=
0analog%20output.jpg
Harmonics on the analog output (12 kHz tone):=20
http://dl.dropbox.com/u/19882028/LF/Harmonics%20on%20the%20analog%20outpu=
t.jpg

After filtering the analog signal is fed to a LM393 comparator running=20
at 12VDC. First i used a LT1028 but the comparator has a much higher=20
voltage gain. Rise time is about 50 ns while it took 800 ns on the=20
LT1028. Furthermore the obtained duty cycle is 50+-1 % when using the=20
comparator but it was totally apart fraom that when using the LT1028.=20
After the LM393 i added a ICL7667 to invert the signal and reduce the=20
output impedance of the mixer.

Actually it is possible to vary the soundcards audio level without=20
losses in performance of the digital output! I.e. i can vary the=20
soundcard level from 0 dB (full range) to -20 dB without any changes on=20
the output. If i reduce the level even more, strong jitter occurs on the=20
output. Maybe this could be even improved by further filtering the=20
analog signal. But a range of 20 dB is sufficient. It allows to QRO/QRP=20
say from 10W...1000W remote controlled :-) I hope i will get some kind=20
of linear mode PA by this "AM stage" (future project) that steers the=20
SMPS of the PA...

The digital output of the mixer has strong harmonics of course. There=20
are many peaks in the spectrum that made my worry if this is a suitable=20
signal! But the spectrum of my DDS VFO didn't look much better. So i got=20
the idea to directly measure the spectrum of the TX antennas signal. I=20
put the scope probe out of the window, about 3m distant to the TX=20
antenna while running about 100W by the DDS and the class E PA. The=20
spectrum looks impressing good to me so i lost all my worries about=20
making QRM. See the harmonics at=20
http://dl.dropbox.com/u/19882028/LF/LF%20TX%20antenna%20spectrum%20with%2=
0120W%20class%20E%20mode%20PA%20driven%20by%20DDS%20VFO.jpg=20
and a narrow band view at=20
http://dl.dropbox.com/u/19882028/LF/narrow%20band%20spectrum%20of%20my%20=
TX%20signal%20with%20class%20E%20PA.jpg

These curves can be compared to the new H bridge PA that i will run soon.

Some smaller works have to be done until i can replace the DDS by the=20
new converter and then transmitting with SpecLab :-) Apart from QSY this=20
will finally allow me to use a higher DFCW shift than 1 Hz when=20
operating at DFCW-3...

I thought about trying CW to UK. Probably this can only be done when=20
leaving the city. A few km apart i could steer the TX by a 70cm FM link=20
and got the idea that i can even QSY when feeding the 70cm TX with a=20
tone in the passband (e.g. 1...3 kHz). The AF signal on the RX could be=20
put to the soundcard (a 2nd one) and is then converted by SpecLab to=20
11...13 kHz.... LF seems to be a never ending story regarding upcoming=20
ideas and projects :-)

73s Stefan/DK7FC



Am 21.07.2011 20:38, schrieb Stefan Sch=E4fer:
> Hello Jim, LF,
>
> Am 19.07.2011 23:11, schrieb James Moritz:
>>>> Scale the values of either R or C proportional to the inverse of=20
>>>> the centre frequency (or both - it is 1/RC time constant you need=20
>>>> to scale). Yo might find faster op-amps neccessary at high audio=20
>>>> frequencies.
>>> OK, i did that for a center frequency of 12 kHz but the result was=20
>>> poor. Then is systematically played with the values and found two=20
>>> pairs of 1n/5k8 and 15n/2k27 which showed at least 47 dB image=20
>>> rejection in the LF band, peaking at -76 dB at 12.55 kHz, which will=20
>>> be a theoretical value of course. I'm going to try these values in a=20
>>> practical circuit and see what comes out.
>>
>> Well, it just depends what you are trying to optimise - do you want a=20
>> wide frequency range with moderate unwanted sideband supression (this=20
>> is what Andy's initial values on the spreadsheet seem to be aimed=20
>> at), or a higher supression over a narrow frequency range - you might=20
>> like to try C1 1n, R1 5470, C2 4.7n, R2 6790. These are feasible=20
>> values - but to get the response shown on the spreadsheet, you might=20
>> need a screwdriver with reduction gears for adjusting the preset=20
>> resistors ;-)
>
> I tried 1n / 5800 Ohm (5k1 and a 1k Pot) and 14.7n / 2300 Ohm (1k8 and=20
> a 1k Pot) and the 74HCT4052. After struggling a bit due a hardware=20
> mistake i found that the output waveform looked like nonsene. But this=20
> was due to massive overloading the input of the OPA which runs at 5=20
> VDC (i tool the LM358 which seem to work reasonable). When setting=20
> SpecLab to -5dB of the full audio level, everything seems to work well=20
> for the first attempt.
>
> This is the board:=20
> http://dl.dropbox.com/u/19882028/LF/TX%20converter%20board.JPG
>
> I recorded the output voltage of the 2 busses. It looks not really=20
> like a sine wave ;-). But i think this is normal since the RF has a=20
> much lower frequency than the LO and a 90 deg shift. So the curve will=20
> look a bit like a staircase. Or am i wrong? This output of the busses:=20
> http://dl.dropbox.com/u/19882028/LF/Pin3and13of74HC4052.BMP
>
> When feeding this differential signal to a transformer ( i took a=20
> FT50-77 and 12 turns primary, 10 secondary, so far) and adding a 100=20
> Ohm Pot, the output spectrum looks OK for me, for the first attempt,=20
> see:=20
> http://dl.dropbox.com/u/19882028/LF/Output%20spectrum%20without%20LPF%2=
0and%20without%20optimising.BMP=20
>
>
> This is the spectrum without optimising the two Pots. It shows above=20
> 30 dB image rejection and the LO (=3D125 kHz)  is not visible at all :-=
)
>
> Next week i will add a white noise to the RF input and will optimise=20
> in the region of 11...12.8 kHz. Then i will a simple and uncritical=20
> LPF and will take a look at the spectrum. Then another amp stage will=20
> follow. Then i'll come on air on 137.7000000 kHz (both, the LO and SL=20
> is GPS "locked")...
>
> Any comments please? Do the waveforms look OK for you or is there a=20
> big mistake?
>
> Vy 73, Stefan/DK7FC
>
>
>
>