[Date Prev][Date Next][Thread Prev][Thread Next] - [Date Index][Thread Index][Author Index]

NOTES on AIDC 3733

Here are some notes on the Transystems downconverter:

The chip marked 510 is a Fujitsu MB510 prescaler. This chip can divide and
incomming signal (up to 2.7 GHz) by 128,144,256, or 272 times. This divided
signal in the 8 or 16 MHz range (with a 2.3 GHz signal in) is compared to
the VCO and locked by the HC4046. What does this mean? It means that you
have a choice of four crystals to achive the desired LO frequency. Here is a

For a 2256 MHz Local Oscillator:

128x  17.625 MHz
144x  15.66667 MHz
256x  8.8125 MHz
272x  8.294118 MHz

The desired division ratio is selected by putting either ground or Vcc on
pins 3 and 6. More information on the MB510 can be found here:

Look around in the junkbox and see if any of the crystals that you have
might work better with a different divide ratio. Some of the components in
the 8 MHz oscillator might need to be changed when using the other divide
ratios. The loop components might also need tweaking if going to the 16 MHz
ones. Keep in mind that most crystals higher that 20 MHz or so might be
overtone crystals. A crystal marked 28.4 MHz might oscillate close to 8.8
MHz. I have not tried the CB crystals sold at Radio Shark.

Another interesting point is that the phase noise can be improved by several
dB's by changing the multiplication factor. The calculated degradation of a
PLL of this type is 20 x log(multiplication ratio). This value is
theoretical and you actually get more degradation. This means that the
degradation from the phase noise of the reference will be:

x128    = 42.1 dB
x144    = 43.2 dB
x256    = 48.2 dB
x272    = 48.7 dB

>From these numbers it is possible to improve the phase noise by 6 dB by
going to a 17.625 MHz crystal. Improvements will be less due to slight
degradation of the Q of the crystal going from 8.8 to 17 MHz.

Another very important note is that the bandwidth of the bandpass filters
have passband at both the desired frequency (LO+IF) and the image (LO-IF).
So when using a 2256 LO and receiving a signal at 2401, you are also
receiving at 2546 MHz. This undesired signal or noise power passes through
the bandpass and ends up as noise at 145 MHz.  This will create a 3dB error
in the noise measurements. LO leakage can also degrade these measurements.

The chip marked R24 is the VCO transistor. The wavy trace on the lower left
pin is the 2278 MHz resonator inductor . It is bypassed to ground and has a
220 ohm bias resistor. The 1PZ chip is the varactor . The varactor is fed by
a 10K resistor marked 103. The round test pad to the right of the resistor
is used to test the lock voltage. Idealy you want the voltage on this pad to
be the same before and after the conversion of the LO frequency. I have not
measured mine but it needs to be in the middle of the voltage rails so that
the PLL has range to compensate for the temperature and other drifts. On a
locked PLL the voltage can be adjusted by changing any of the resonator
components (varicap, series cap, etched inductor).

The signal out of the VCO goes to an Agilent (HP) AT-41511 transistor. The
datasheet is here:

Most of the resistors around it are for biasing. The three resistors after
it are a small attenuator pad going into the mixer. This pad can be removed
to make a stronger signal source. The mixer diodes are inside the chip
marked SH2. There are two diodes inside that package.The 2.4 GHz signal is
feed on the upper left of the mixer and the IF comes out below the etched
"U". This "U" looks like a LO notch or something.

Receiver stages:
The small coax at the input of the second filter is a notch filter. It can
be adjusted to notch the image (around 2546) by cutting it. How much? You
need test equipment. I would just remove the thing.  The part marked 414 is
an Agilent AT-414xx transistor for additional gain. The LNA board works on
12 volts and requires no negative supply. It can be used for a much better

Things I don't like or would do different:

1. I would remove the input filter. This would buy me better noise figure.
Overload should not be a problem if aiming the antenna  away from powerful
transmitters.The second filter would be modified to filter the image
frequency and the LO. For weak signal work it is very important to filter
the LO at the receiving frequency (2401). The LO specially on these types of
converters has a lot of undesirable noise at 2401. This degrades your noise

2. The LO is a very poor performer. This design is just not suitable for SSB
work! I would totally get rid of the LO and replace it by a 2256 design
based on a 94 MHz multiplied crystal in an oven.

The mixer should be ok. I plan on taking the converter to work this weekend
to see if I can come up with reproducible modifications to improve this
converter. I personally think that it can be modified to really perform well
for mode S reception. I will keep everyone updated.

Pieter N4IP

Via the amsat-bb mailing list at AMSAT.ORG courtesy of AMSAT-NA.
To unsubscribe, send "unsubscribe amsat-bb" to Majordomo@amsat.org