|
Upgrade
Nokia wideband FM PA3GCO
|
| ... |
In
part
1 the Nokia MD59LS has been converted to a 9K6 packet transceiver.
Now it's possible to increase both rx and tx bandwidth, for wideband use.
A bandwidth of 300 kHz is easy to achieve. At this moment these wideband
Nokia's are used for 76800 Baud packet radio. Tests haven been done with
115200 Baud as well. After describing how to upgrade the transceiver, something
is told about software, modem, etc. There will be serveral references to
part
1.
... |
... |
|
Explaining
wideband modification |
|
|
|
|
|
|
|
|
|
|
|
|
|
| ... |
Current
situation: the MD59LS has been converted to a 9600 Baud packet transceiver.
Audio for modulation is applied to the 21.4 MHz crystal oscillator. A pll
locks the vco, around 452 MHz. The same vco signal is used as LO for the
receiver. First mf of the rx is also 21.4 MHz. With a 20.945 MHz crystal
this is mixed to the second mf of 455 kHz. Via the 455 kHz mf filter the
signal reaches the demodulator TDA1576, tuned at 455 kHz.
... |
... |

| ... |
In
the new wideband situation an mf of 31.4 MHz is used, using the original
31.4 MHz crystal on the synthesizer board. For wideband you don't modulate
this crystal oscillator, you modulate the vco. So audio is injected elsewhere.
Supposing the trx is used for simplex communication, the rx mf should also
be 31.4 MHz. The 21.4 MHz mf filter on the receiver board is shortcutted
and the 20.945 MHz crystal is replaced with 20.7 MHz. This 20.7 MHz oscillator
mixes the 31.4 MHz mf to 10.7 MHz. The 455 kHz mf filter is replaced by
a 10.7 MHz filter. The TDA1576 is retuned to 10.7 MHz, using a new 10.7
MHz coil.
... |
... |

| ... |
The
wideband MD59LS is certainly not suitable for moonbouncing. It appears
to work fine for 76800 Baud packet over more than 25 km, not using high
positioned antennas.
... |
... |
| ... |
In
part 1 the square adjustable inductor [06]
was shortcutted to get the low cost 21.4 MHz crystal working properly.
Remove this shortcut, make it original again. Also place back the original
31.4 MHz crystal. In part 1 a 2p2 smd condensator [18]
was soldered on top of the original 1p8 smd. Remove the 2p2 smd condensator,
so restore the original situation. |
... |
... |
 |
| ... |
Where
to inject wideband audio? Beside the variable condensator is the test point
where the pll control voltage is measured [04].
At this same test point, there's a 22 nF plastic condensator to ground.
Remove this 22 nF condensator. Now 2 holes in the pcb become available.
On of the holes is connected to the test point, the other to ground. Solder
a 12 kOhm resistor in the ground hole, place it vertically [22].
In the other hole, connected to the test point, you solder a 82kOhm resistor
[23], also vertically.
The free ends of both resistors are connected. At this junction a 6 holes
rf choke (rfc) is soldered [24].
The other side of this rfc is the new audio input. In part 1 the lf input
was at pin 1 of the 7 pin brown connector. Make this pin 1 free at the
bottom side of the pcb by cutting the track near pin 1, and solder a wire
from pin 1 to the rfc. Now pin 1 of the brown 7 pin connector is again
the audio input.
Schematics
new modulation input in PDF
|
... |
... |
... |
| ... |
In
part 1 the narrowband mf was 21.4 MHz. For wideband it is 31.4 MHz. So
the 21.4 MHz mf filter should be shortcutted, in the following way: C65
is a 10 nF smd condensator Remove this C65 [35]
which results in 2 loose contacts at the bottom side of the pcb. Solder
a wire to that contact which is closest to the edge of the pcb. C25 is
a 3.3 nF smd condensator. Cut the track between the 21.4 MHz filter and
C25 [36]. Now C25 is
not connected at one side. Connect this free end of C25 to the wire, mentioned
earlier in this chapter. Now the 21.4 MHz mf filter has been shortcutted. |
... |
... |
 |
 |
... |
... |
Replace
the 20.945 MHz crystal: It should be 20.7 MHz [25].
XF2
is the black 455 kHz mf filter. Remove this filter. Now 5 holes become
available. Three of them are connected to ground, the other two are input
and output. Place a 10.7 MHz ceramic mf filter (3 pins) in the free holes.
The center pin should be connected to ground, the outer pins are input
or output [26]. Input
and output can be switched: The filter is symmetrical. It can't go wrong.
|
... |
| ... |
The
TDA1576 is originally tuned to 455 kHz. For wideband use it should be tuned
at 10.7 MHz. Between pins 3 and 4 of the TDA1576 is a 680 pF smd condensator.
Replace it with 33 pF smd [28].
The same for pins 6 and 7: Remove 680 pF smd and place 33 pF smd back [29].
Between pins 4 and 6 of the TDA1576 there are 4 parts:
-
R72, a
2K7 smd resistor
-
C45, a
4n7 smd condensator
-
C46, a
560 pF ceramic condensator
-
L15, a
455 kHz coil
Remove
R72, C45, C46 en L15. Mark the holes where C46 and L15 used to be!
Between
pins 8 and 9 is C48, a 1 nF smd condensator. Remove C48. R43 is a potentiometer
near pin 8 of the TDA1576. Remove R43.
How
to place the new 10.7 MHz coil, type KACSK-586HM: This coil has 5 pins,
I call them A, B, C, D and E. Cut B, D and E, as short as possible. Now
you have only 2 pins left: A and C.
|
... |
... |
 |
|
 |
... |
Where
to put pins A and C? L15 is the removed 455 kHz coil. You marked its holes.
Find the L15 hole which is closest to pin 5 of the TDA1576. I call it hole1,
in dutch gat1. Hole1 is the first hole for the new 10.7 MHz coil. C46 is
the removed ceramic condensator. You marked its holes. Find the C46 hole
which is closest to pin 8 of the TDA1576. I call it hole2, in dutch gat2.
Hole2 is the second hole for the new 10.7 MHz coil. |
... |
| ... |
Stick
the 2 pins of the 10.7 MHz coil through the holes [27]
: Hole1 [30] and hole2
[31]. Now watch the
solder side of the pcb. It appears both pins of the new coil are connected
to pin 6 of the TDA1576, the coil is shortcutted. To solve this problem,
free the pin in hole1 by cutting a track [32].
Now make a connection [33]
from the free pin to pin 4 of the TDA1576. The new 10.7 MHz coil is now
placed between pins 4 and 6 of the TDA1576.
In
part 1 the wire for rx-audio was connected to pin 8 of the TDA1576. Switch
it to pin 9 [34] of
the TDA1576.
This
completes the wideband upgrade.
|
... |
... |
 |
| Number |
Part |
Supplier |
| 1 |
crystal
31.4 MHz |
original
from Nokia |
| 1 |
resistor
12 kOhm |
Display
51.00.12K |
| 1 |
resistor
82 kOhm |
Display
51.00.82K |
| 1 |
RFC
6-holes |
Display
02.42.201 |
| . |
. |
. |
| 1 |
crystal
20.7 MHz HC50 |
Klove |
| 2 |
33
pF smd |
Display
81.50.33P |
| 1 |
coil
10.7 MHz type KACSK-586HM |
Display
02.59.651 |
| 1 |
ceramic
filter 10.7 MHz SFE-10.7MA |
Display
02.59.676 |
| ... |
In
this example I use a working frequency of 434.0 MHz, the same frequency
my neighbour's cordless headphone is using. The vco frequency is set 31.4
MHz higher, so 465.4 MHz. Set the pll control voltage to 2 Volt. A low
voltage at this point results in minimum influence on the audio characteristics
of the Nokia. During rx adjust the 10.7 MHz coil for an optimal bit-error-rate
test.
... |
... |
| ... |
We
use standard G3RUH compatible modems, interfaced to the pc via an scc card
(Z85C230). Speed can be increased by changing some jumper posistions, a
few resistors and condensators. It is described in the modem documentation.
Use the 76K8-only-dcd because it requires no tuning. This dcd doesn't appear
to be spike free. For most software no problem, TFPCX accepts this without
problems. JNOS however requires a spike free dcd. Modify the hysteresis
of this dcd as described in several documents. See your modem documentation.
Otherwise use the noise dcd: It is spike free but it requires tuning.
From
PD1ACF and PE1MVX, two of the persons behind PI8HGL, I received a nice
document containing modem modifications. Click here
for the PDF, click here
for the PI8HGL website.
. |
... |
| ... |
Caution:
you modulate the vco, the same vco that is needed during reception. Therefore
the modem should not produce audio during rx. Use pin 22 of the tx eprom
to mute the audio: Connect it to 5 Volt during rx, this switching voltage
is available on the modem board.
With
the tx-EPROM the FIR filter constants are adjustable. Pins 2, 21, 23 and
26 are used for selection. For 76800 Baud the jumpers have to be set as
follows:
... |
... |
... |
| Pin
filter-EPROM |
Voltage |
| 2 |
5
Volt |
| 21 |
0
Volt |
| 23 |
0
Volt |
| 26 |
5
Volt |
|
| ... |
I
use a home made scc card, based on the Z85C230 chip. Besides the scc chip
only 3 other chips (7404, 7474 and 74688) are needed. Standard is a 4.9152
MHz crystal. By doubling it to 9.8304 MHz, the maximum speed goes up from
45000 Baud to 115200 Baud. Of course you need a 10 MHz scc chip: Z85C230-10.
... |
... |
| ... |
I
find standard AX25 software not suitable: no GP, SP or TSHOST. Every transmission
7 frames of 255 bytes each are sent. Very little for 76800 Baud packet
radio. If you do decide to use such software with TFPCX after all, keep
in mind that the scc clock frequency has been doubled. So if you tell TFPCX
you'd like to do 38400 Baud, your scc card does 76800 Baud. A better solution
is software that can transmit larger blocks of data every transmission,
like FTP with JNOS, a DOS program. Every time my Nokia is keyed, 32K of
data is sent in one block. This goes well on an old 386DX40 with JNOS.
Linux users should tell their software to use the internal FIFO buffers
of the Z85C230 to avoid underrun errors.
... |
... |
| ... |
Using
76800 Baud packet radio we exchange more than 7000 bytes of data every
second on a simplex channel. We use 100 kHz bandwidth. With 20 Watt output
stations can be reached more than 20 km away. A power reduction of 20 dB
has nearly no effect: it still goes very well.
... |
... |
|