OpenBCM V1.07b12 (Linux)

Packet Radio Mailbox



 Login: GUEST

G8MNY  > TECH     20.11.23 08:13z 541 Lines 28070 Bytes #12 (0) @ WW
BID : 60682_GB7CIP
Subj: Spectrum Analyser mods 88-89
Sent: 231120/0807Z @:GB7CIP.#32.GBR.EURO #:60682 [Caterham Surrey GBR]

By G8MNY                                 (Updated Sep 20)
(8 Bit ASCII graphics use code page 437 or 850, Terminal Font)

Here are 22 modifications I have done to this popular scope adaptor design that
was first published in the RSGB's Radcom Technical Topics Apr 1988.

This design used just 3 ICs, 4 regulators & 1 transistor. A MC3356 is the 1st
osc mixer & log IF (most of the 94 transistors in the IC are not used), a MC602
(NE602) 2nd osc mixer, a TL084 quad op amp to do the sweep, & 12V 1A, +5 & +6V
100mA regs. (in 1989 a similar mark 2 design & PCB kit was published, with
bells & whistles, very complex with loads of 741s & calibrated switches).

Mains  Ŀ  Mains    Box         My one was very neatly UGLY
Transf-    PSU                         constructed (not by me) with
ormer  _______ inner box             double sided PCB made box &
 ¿             aluminium folded U top cover.
   Ŀ r2 Ŀ  Op Ŀ    
   M    () Amps      r1=78L06   Then again with the RF bits
   C    ()          r2=78L05   in a 2nd bolted in PCB
   L2  () Ĵ             box with individual screen
  L          Ŀ r1               partitions & RF feed through
 Ŀ IF1     ()2nd             caps for lines in & out,
 L1    IFamp  IF2  Mix               an RF tight fitting fingered
              aluminium lid.
 ________    ____    Sweep     
  Attn    Pot    Ŀ  Ŀ 
  RF       X      Y    on/off

   Input   LPF  Protect  HF  Trim   1st  1st IF  2nd  2nd IF   IF   Filter
   Atten   0-90   Clip   EQ  Bias  Mixer  145   Mixer  10.7    Amp   10.7
   Ŀ  Ŀ  Ŀ  Ŀ  V  Ŀ  Ŀ  Ŀ  Ŀ  Ŀ  Ŀ
o)Ĵ  Ĵ~~\Ĵ ^vĴ-'Ĵ< >Ĵ__Ĵ< >Ĵ__Ĵ> Ĵ__Ŀ
     Ŀ                       Ŀ  1.5MHz Ŀ   50kHz        50kHz 
/oĴ>              145-235OSC         OSC terminated for shape 
     Markers             MHz                                
  Vernier            Ŀ          Sweep     134.3MHz                    
Frequency<>Ĵ> >                                     \/
                   Ŀ  Corrected                                     
    Ramp      Ĵ_./Sweep                                         
     Sweep<       NFB                                      Filter  
       Pot                                   Log Amp & Detector   10.7    
X o)  Ŀ  Ŀ        Ŀ             Ŀ   Ŀ   
 S  Ĵ <Ĵ < Ĵ> Ĵ<  Sync     <<<<<Ĵ__
 C       Ramp<50Hz     clamp      5 Detectors     
 O    Buffer  Osc ĴĿ   Flyback              300kHz
 P                    Amp              10kHz   sum       Spurious
 E     Ŀ1kHz                      Ŀ   Ŀ            Wall Filter
Y o)Ĵ~~\<Ĵ_/~Ĵ~~\Ĵ
   o\  LP                              Y Cal
 LPF Video                    Detector S   Video  
 Switch   Filter                   Correction   Filter

After modifications here is the upgraded Specification:-

Frequency Range 200kHz-90MHz
Level flatness 200kHz-70MHz -3dB, 80 -6dB, 90-16dB

BNC 50R input @ up to +20dBm (0.5W Max with all atten in),
70dB of input Attenuators 10, 20, & 40dB.
IF bandwidth 50kHz @ -20dB
Video Bandwidth 10kHz or 1kHz @ -3dB
Sensitivity +20dBuV (10uV) for 10dB/Noise
1st IC protected by clipper (not attenuator)

70dB of vertical Y log scale. (2dB)
60dB mixer dynamic range before onset of mixer overload
BNC Y Scope output calibrated to 100mV/10dB, with negative flyback syncs.

Sweep rate 50Hz flyback locked to supply (1Hz)
BNC X scope Ramp Output 20V P-P
5MHz RF Markers up to 90MHz
Vernier Dial Frequency readout, Accuracy  1MHz
No display of lower image sideband (below 0Hz)
210-254V 50-60Hz Mains Operation 10W.

1/ There are 3 attenuators, the 10 & 20dB are accurately achievable with 1
double pole changeover switches separated by PCB screens. But the 40dB one is
not so easy, & needs some attention to detail, & an additional screening plate
between the connections to achieve it over the frequency range.
Note the 2K4 & 51R are E24 series, but 2K2, & pairs of 100R (* or 3x 150) work
quite well & may give higher dissipation if 500mW is put in. All Rs small types
& NOT WIRE WOUND! Very short leads are used on the 3 small switches.
         Ŀ      Ŀ     Ŀ   __ 50R Coax
BNC  o)o/   |   \oo/          \oo/         \o)__ to filter
Input     Ŀ |        270R     68R   
 50R   * 51R | 51R      68R    68R    100R   100R  
             40dB            20dB            10dB

I found that the losses reduced by the odd dB at higher frequencies, this is
due to.. switch crosstalk capacitance, series R capacitance or low R inductance
earth inductance etc.

2/ At the input to the low pass filter, add a 5MHz marker clock oscillator IC
mounted right beside the filter L. The output is loosely coupled by stray
capacitance with its short 1 cm lead for accurate frequency markers.
(The mk 2 SA has a marker already). The L1 is 5 turns wide spaced 5mm dia.
DC is via a push button.
Markers          14Ŀ9   stray pick up   protection
 o\470RĴ 5MHz L1    clipper        HF lift    Pin 20
+12V             OSC   from >(((()Ĵ(((()> MC3356
    ===    __ ===   IC  atten  ĴĴ  __ __   10n  8turns /\  250
   u1   5V/_\' u1       === 15p === \_/ /_\ 62R      6mm  ===   I/P Z
                 7 1     56p     56p                     / 20p
                  CLOCK IC           LPF    2x1N4148  Term

3/ Change the low pass filter termination to 62R as it is in parallel with 250R
IC input Z. My filter is 3dB down @ 90MHz with about 30dB rejection above that.

4/ Add RF clipping diodes across termination R to protect IC1. These do not
conduct at all, for on screen signal levels. But do stop you blowing up the IC
with silly signals (e.g. from a handheld), but the input attenuator is

5/ The conditioned signals from the attenuator & VHF wall filter is fed through
an L & C trimmer for best HF level @ 70MHz into the 250R input Z of the mixer
IC1 (1st half of MC3356 IC). The IC is not designed for the Local Osc @ so high
frequency, so there is some drop off of sensitivity @ 80MHz without this tweak.

1st MIXER in MC3356.
A 2:1 step up ferrite transformer & no termination resistor will give 6dB more
gain, but no transformer is flat over 0.5 - 90MHz range so this is not used.
The mixer has a gain of about 5dB.

The osc provides 2 out of phase outputs (one with no RF!) that are buffered to
drive the Gilbert mixer cell (unbalanced), which has one RF input & 1 output.

     Tuned cct        OSC      BUFFERS      MIXER  CELL
+12V 4oĿ        +12V
   ===    ) L2               /                             4.3uH )
10n__     3o)10KĴ                     IF          L3 )  1n
      Ĵ          \      \e /       ))o5 Ĵ>
     ===             10K)Ĵ                              to 145MHz
     /_\ === 2p2    e/          \e /     \   /     \   50K     IF Filter
Sweep     2oĴ              Ĵ       Ĵ       Ŀ  
>10KĴ                           \e   e/  \e   e/   
        === 5p                ))            
  1n ===                         ))  
            RF in                   /           \       
From>)) 20o)))Ĵ             10KĴ
LPF                               \e         e/      
          1M     50K         50K 50K   330R      330R ===   \_/
          Preset                                  20p __
 0V   1o

L2 is a spread out coil 5 turns 5mm dia. The 12V is well decoupled with 10n @
L3. The original Varicap was a MV209.

N.B. never accidentally put earth on pin 2 or 3, that will destroy the OSC NPN!

6/ To get the last drop of balance out the 1st mixer, I found a 1M preset from
RF input to ground could give a slight improvement in balance & reduce a 2nd
harmonic of a pure RF signal by 2dB.

 dB  0Hz
+60        .Fo                    If the balance is the other
+50   ^                          way, try the pot to +12V.
+40   |       Adjust 1M
+30 ~60dB      for Min           Use a well filtered Osc (2-30MHz)
+20   |          \|/             so the 2nd harmonic > -60dBc
+10   v          2Fo    Noise
  0 Floor

7/ The varicap tuned first local VHF oscillator's range has extended to tune
from 145MHz to 235MHz by adding further UHF Varicaps across the initial one &
stretching/reducing the osc L2. This may depend on the varicap used & stray C.

8/ The local oscillator coil is adjusted to give 0Hz line (e.g. osc = 145MHz)
when the FREQUENCY control vernier is set to 0 (mechanically near the -12V end
of the pot, so that tuning sweep voltage after the amp is near to +12V). This
also stops the unwanted image side of the 0Hz from being displayed & causing
confusion. The 22uF bipolar cap removes any scratchyness in the freq pot.

            FREQ<>Tuning                0Hz
           10K POT             DC                  
                       ===22u                No         5MHz
           80MHz CAL     Bipolar           Inverted    Markers
           5K PRESET    __                 Display  |    
-12VĴ<                               \/        
     Thermal                                ______________
    compensation. (See 21/ & 22/)

9/ Mine used a standard 50R 2M three pole TOKO 144-146MHz filter, (the original
article said to build your own). The TOKO one can be modified from a bandwidth
of about 3MHz (-10dB) with 3 peaks, to a single peak 1.5MHz wide, by adding 2
small metal shielding strips (6mm x 8mm) to cover the 2 apertures between the 3
coil sections, BUT this is fiddly to do!

Underside view                   After mod
 Ŀ                  Ŀ
 ( ) ( ) ( )                  ( )( )( )

To see the 1st IF response on its own on the display, to tune it up, temporally
remove connections to the 2nd IF filter & bypass it with a bridging 1nF cap &
feed a carrier in (marker). 
        _   _   _                        ..
      / \./ \./ \                     /  \
     |           |                   |    |
   |               |               |        |
_./       3MHz      \._      ____./  1.5MHz  \.__

Once the 1st IF's bandwidth is reduced then the number of signals reaching the
2nd mixer is greatly reduced & hence this reduces the amount of unwanted
distortions & close in mixing products being displayed.

2nd MIXER.
This uses a NE602 osc & a balanced Gilbert cell mixer (used unbalanced) with
similar internal circuit to that of the MC3356 RF part, it runs on its own +6V
regulator & RF decoupled. Mixer gain is about 17dB.

+6VĿ   * see /11 for value
   ===   8t ,/\              *    === 10n
 1n__  5mm (|         8     430R   __
           /(|   6Ŀ   
    134.3MHzĴ         4         to 10.7MHz
       3p9 ===   7    NE   >ceramic
145MHz      Ĵ   602              filter
IF >Ĵ)Ĵ                             N.B. as with 1st osc
      1n        1                    an accidental earth
           ===     2  3  5                    on pin 6 or 7 will
        6p8    1n===    ===1n                  destroy the osc NPN.

10/ The VHF oscillator in the 602 should be run on a lower frequency (core "in"
position) to the 1st IF, this is to reduce spurious images. e.g. 10.7MHz 2nd
IF needs 134.3MHz, (or a 6MHz IF needs 139MHz). Due to the ferrite core this
osc is susceptible to changes in magnetic fields, so mains transformer flux can
be a problem for "zoomed in" stability!

11/ The 2nd mixer's output has in internal pull up of 1K5 to so get 330R source
impedance for the filter a 430R on pin 4 determines the 2nd IF filter source
impedance. (This is not applicable to the Mark 2 with narrow filter option.)

     .---.                  .--.          If the filters are deliberately
                           /    \         mismatched then they can give
     COMMS                |      |        a better analyser friendly
   |  Filter |             |        |       "rounded peak response" rather
_./   50kHz   \._       _./          \._    than the flat topped ringy
      edges of communication filters.
 Correctly terminated      Mis Terminated
    RINGY FILTER        sweep friendly filter

To find the best values for your filters use small 1k presets to source &
terminate the filters to see this effect on the display of a carrier, find the
optimum value for best IF shape & then replace the presets with nearest fixed
                   === u1    IF    
          +6V      __       AMP  470R
2nd  Pin 8<Ĵ           220K   Ĵ
Mixer      430R             /     
602  Pin 4>Ŀ   Ĵ T1      Ŀ   >Pin 7  To
     Pin     __ __    BFX90\e   __ __  __ __ 330R            MC3356
      3      Ŀ              Ŀ  Ŀ  >Pin 9  Log
       50kHz         100R      Ĵ>Pin 8  Detector
       10.7         Preset         ===
    >Pin 19
            1st filter             2nd     Wide filter

Adjust the 100R gain preset for the optimum noise floor that can just be seen.

12/ Adding an inter filter buffer stage using a single transistor T1 adds some
preset gain for setting the overall noise floor of the analyser, as well as
matching into a 2nd filter. This filter is terminated by the input load on the
detector 330R. Two 50kHz 10.7MHz ceramic filters provide a reasonable
compromise of selectivity for sweeping 0-80MHz @ 50Hz without too much ringing
distorting & loss of the peaks levels (up to 10dB) while still looking good in
close "zoomed in" sweeps. However a 3rd filter was put in tandem with the 2nd
filter to clean up poor filter skirt rejection of my particular narrow filters
@ 7MHz!

13/ The detector (S meter output) uses 5 IF amps & detectors (with limiters) to
obtain log response & it is quite accurate for over 40dB range. But ignoring
the slight overload in the mixers this range can be extended on the display, by
increasing the gain calibration preset (1K preset now 2K2), & then adding a non
linear correction attenuator with diodes D1 (Schotky/Ge) & D2 (Si) to give 30%
stretch @ the highest & lowest levels where the detector has lower sensitivity.

                      Ge                  1.0 Output from        _.-'
MC3356  pin     Ĵ>Ŀ          Y to      .9 Detector        .-'
LOG     14>Ĵ     > scope    .8               .'
DETECTOR       10KĴ                   .7             .'
         2n2  680R        T2            .6           .'    S
     Video ===      22K    \  Timebase   .5         .'  Correction
     Filter   2K2   __     <Sync/      .4     _.-'
              Y CAL \_/Si e/  Blanking   .3 _.-'
Pin 11>Ĵ      .2 
Lim out  1n       __                      .1                         Display
is RF                                      0v >output
grounded                                        0  10 20 30 40 50 60 70 dB

With this correction you can get good display linearity to 70dB. Easily tested
with input attenuator & a signal generator to see equal height 10dB steps.

14/ The sensitivity is set by the Y Calibration preset to give 100mV/10dB. A
2.2nF capacitor limits the Y video bandwidth to about 10kHz (50%), but having
hardly any degradation of pulse height at the widest sweep range.

15/ For some applications lower video bandwidth is needed to reduce noise, I
added a 33nF switched across the Y output incorporated with the above mod, to
give about 1kHz Y bandwidth (50%).

10k>Y to
 __ 33nF

It needs to be switchable as it causes the output to lie about the fine detail
with wide sweeps.

16/ The hot +12V regulator has been heatsinked, & the + rail input smoothing
capacitor increased from 680uF to 2m2. Transformer & rectifier pulse currents
wiring & layout have been kept away from the regulators as far as possible to
reduce supply hum ripple pickup.
                  4x 1N4001     +17V Ŀ
                Ĵ>Ĵ7812> +12V    Much larger output
 L >o/ oĿ     __   __  2m2 + +330u    200mA   caps have been
        100k )(   /_\   /_\  25v===      ===16v            used to reduce
240V        )(___  ___  ____________________\ 0V      the last remnants
        NEON )(           680u+         +    /         of hum & noise.
            )(           25v===        ===330u          This is most
 N >     )Ĵ    Ŀ  16v           important
               Ĵ<Ĵ7912> -12V    for close in
 E > 14-0-14         -19V           20mA   stability.
      > 0V    0.3A     >50Hz

The mains transformer has also been varnished to reduce acoustic hum & an outer
copper short circuit added to reduce magnetic fields that can affect the 2nd
osc stability. The other 2 low power regulators +6V for 2nd osc, & +5V for Log
amp, are placed near those circuits for best noise/voltage error rejection.

17/ A 50Hz synchronisation line is provided for ramp timebase locking. This is
important for close "zoomed in" stability of the sweep.

In the simple mark 1 design, it uses 4 operational amplifiers IC3 (e.g. TL084)
that run on the 12V. The original circuit produced a symmetrical 500Hz ramp up
& down oscillator which was far too fast for wide sweeps & half the time was
wasted during the flyback. Mod 18/ solves this.

18/ IC3a forms a 50Hz ramp oscillator with the 100K & 12K in parallel during
flyback due to the diode D3, & a 1uF timing capacitor to give close to mains
frequency. Then a small injection of 50Hz from the mains transformer alters the
flyback time (D3, 12K & 1uF) to cause lock up to mains frequency. This method
ensures constant sweep MHz rate & the mains lock ensure a stable display even
with some sweep hum present when zoomed close in.
         50Hz 12VAC >100K12KĿ           _
         from bridge       __     100k          ڿ      _ \____
MAINS                   D3 /_\  100k            
LOCKED            /Ĵ       Ĵ\    -      Ĵ>22k>Y Blanking
50Hz <<'IC3b             `>Ĵ\                Transistor
RAMP,.      n47 `\Ŀ   )Ĵ/'    + `>2k2Ĵ         T2
  ,/      ĴĴ              IC3a  Ĵ/'          
,/   ,/   10k 1k2 ===1u    10k         IC3d     u1===
          Buffer x9   ,/',/'   50Hz Ramp Osc     Sweep Settle Delay

19/ IC3d buffers &
inverts the banking          dB  0Hz
pulse & it is               +70          |        /|\   + 0.9V
lengthened with CR          +60        Sweep       |
& a diode before it         +50        Centre     100mV
drives blanking             +40          |        /10dB
transistor T2. This         +30                    |
eliminates any sweep        +20                    |
folding due to VHF osc      +10     Noise Floor   \|/
sweep settling delay          0 Ŀ     + 0.2V
with RF sweep filtering     SYNCS     SWEEP       ^ 
to be masked. And the                          |  - 0V
banking also gives          Scope^               Scope
the scope a 0V sync        Trigger              Flyback
pulse to lock to.                < - - - - 20mS - - - - ->

20/ IC3c is the sweep correction amplifier, this amplifies the selected sweep
width together with the centre frequency DC, then corrects for VHF oscillator
varicap frequency control non linearity by pre-distorting the ramp waveform
with 5 gain changes using 4 diodes, D4-D6 & ZD2 zener.

            -12V>3K32K2<+12V    MHz Tune
10 Turn                390K              240Ramp                          _
 Tune<68K             R1               230Input                 _..--''~
  Pot     27KĴ<Ĵ                220     Five       _.--'~    R4
                27K   __               210    Slopes  _.-'  R3
Ramp       Ĵ\      /_\    To VHF     200         .-'
           `>)>Osc       190      .-' R2
Sweep     )Ĵ/'IC3c        Varicap    180    .'
  Pot<68K             2.7V           170  .' R1
  10k                  Ĵ>Ĵ<Ŀ   160 ;
                  R2 47K   15K   4K7  150:    Varicap Volts (ref to +12V)
                              R3    R4     +12  9   6   3   0  -3  -6  -9  -12

The R1-4 values used for the gain corrections are set up using the marker to
give even spacings on the display.

    .    Even spaced 5MHz markers
     .    |   |   .   .   .
         |  |  |  |  |  | 
     5  15  25  35  45  55  65  75  85   MHz

              FREQ<>Tuning    The multiturn vernier
            10K POT              DC     centre frequency control
                         === 22uF       has a 22uF bipolar
           80MHz CAL       Bipolar!     capacitor to ground
           5K PRESET      __            (or elect to +12V near the ICs)
-12VĴ<         ////            to remove any pot scratchiness.

22/ A multiturn preset pot on the positive rail of the control is added to
calibrate 80MHz position on the vernier scale. A diode in series gives some
temperature drift compensation. Together with the correction circuit of IC3c
fairly accurate frequency readouts are possible on the vernier scale. 0-90MHz.

I N  U S E
Other than the 0 Hz line, there is only one unwanted image @ 10.7MHz, it is at
a low level & its appearance depends on the IF gain setting. It is due to the
second IF detector being in the same IC as the RF input section!

 dB  0Hz
+60                                  VHF images are all well
+50                                  down due to the VHF LPF
+40                                  & the chip sensitivity
+30                                  cutting off as well as
+20                                  the input filter & double
+10                          Noise   screened box.
  0^ Floor

These can be seen as higher levels of harmonics increasing at a greater rate
than the fundamental. e.g. a 10dB increase in level, causes the fundamental to
increase by 10dB (1 division), but the 2nd harmonic increases by 15 to 30dB!

dB   0Hz                            dB  0Hz
+70                               +70      Fo
+60      Fo                       +60   ^  
+50  /|\                         +50   |       Mixer
+40   |                          +40   |     Generated
+30 <60dB                        +30 >60dB   Harmonics
+20   |                          +20   |       2Fo
+10   v      2Fo   3Fo  Noise    +10   v               3Fo
  0 Floor      0
     Filtered Osc test                     Filtered Osc test
                                          OVER LOADING 1st MIXER

They can also be detected as unwanted sidebands around the markers too.

 dB                                    dB
+70     Signal                       +70
+60                                 +60
+50                                 +50
+40        Marker                   +40
+30                                +30  Base Noise Floor
+20 SM          S+M               +20     Raised     __
+10 Mix          Mix               +10      __..--""~~
  0              0""~~

Other signs of overload is a raised noise floor.

These are much the same as above, but occur when the 2nd mixer sees 2 large
signals passing through the 1st IF filter. So if the narrowing of that filter
has been done, strong signals will need to be closer than 1 MHz to suffer this
problem. (e.g. using the analyser closer than 1 MHz from 0 Hz reduces dynamic
range due to the increased noise floor from its' own 2 oscillators)

 dB                                     With large carriers are looked
+70                                    at close in, you will see noise
+60              /~\                   sidebands (phase noise) added
+50                                  to the filter response, this is
+40                                  normal for this sort of analyser.
+20 Sideband         Sideband      Lower frequency Y display filtering
+10 Noise  _       _  Noise      can mask this, but at the cost of
  0ͼ  <50kHz>     peak pulse height accuracy.

Some of this phase noise can be noisy sweep amps, as the S/N needed on the VHF
osc will be >120dB, e.g. 24V max sweep & < 24uV of noise! I have used active
sweep filtering on some analysers to overcome this failing where the sweep rate
is low & a CR filter after the last opamp does reduce the HF noise sent to the

         Ĵ>Ŀ                The diode direction (depends on circuit)
    \                        ensures the flyback charges up the cap
 OP  `>R>VHF OSC   voltage quickly, so there is little cramping
amp /'        === C            at the start of the sweep.
                           e.g. Xc = R (-3dB) @ 10x sweep freq.

See my tech buls "Spectrum Harmonic Demo circuit", "RF Directional Coupler", 
"Analyser SWIRES RESEARCH SA87" & "Analyser Takeda Riken TR4122B".
Why don't U send an interesting bul?

73 de John G8MNY @ GB7CIP

Read previous mail | Read next mail

 09.12.2023 01:27:32zGo back Go up