OpenBCM V1.13 (Linux)

Packet Radio Mailbox

DB0FHN

[JN59NK Nuernberg]

 Login: GUEST





  
G8MNY  > TECH     07.05.08 23:12l 294 Lines 14548 Bytes #999 (0) @ WW
BID : 55280_GB7CIP
Read: GUEST OE7FMI
Subj: 198kHz Off Air Standard
Path: DB0FHN<DB0FOR<DB0MRW<DK0WUE<DB0RES<ON0AR<GB7CIP
Sent: 080507/2118Z @:GB7CIP.#32.GBR.EU #:55280 [Caterham] $:55280_GB7CIP
From: G8MNY@GB7CIP.#32.GBR.EU
To  : TECH@WW

By G8MNY                                    (Updated Apr 08)
(8 Bit ASCII graphics use code page 437 or 850)
I currently use an analogue TV lock reference system (see my buls on "Off Air
lock for Ref Osc") for calibrating RF gear, but as analogue TV will end in the
not too distant future in London, I took the opportunity to buy someone's old
homebrew veroboard construction version of the BBC 198kHz LW off air reference
project. (from a Practical Wireless article Dec 1995 by G8JVE, to go with the
Robin counter & follow up Oct 1998). This type of design might be usable in
other parts of the world if there are suitable accurate LW or MW stations.

     旼컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴커
  �袴킙xternal Active Ferrite Rod Aerial�
  �  읕컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴켸
旼懃컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴�
�     Manual    旼컴컴컴컴�   Ref Freq    �
�  o    .-.     �  Meter  �     .-.       �
쿛ower (   )    읕컴컴컴컴�    (   )      �
�  o    '-'      CZ =o  Sig     '- '      �
�  �          Not o     o Lock        (�) 쿚utput
읕컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴�

WHY WAS IT CHEAP?
Well it sort of worked, but there was far too much Phase Shift Keying on the
reference output following the low speed PSK DATA on the 198kHz broadcast. (PSK
used for U.K. power meter control?) As well as that it was quite insensitive
despite an extra RF buffer stage over the original article. It also had a
rearranged divide chain (same ICs) to give more useful reference frequencies.

旼컴컴컴커 旼컴컴커旼컴컴왯컴컴컴왯컴커旼컴컴커旼컴컴컴�
쿐xternal� �  RF  납     납 2kHz 납 IF납  IF  납   IF  � Rx  旼컴컴컴컴커 Lock
� Active �>퀯uffer척Mixer척 LC IF척AMP척Tuned 척Limiter쳐�>컴�          쳐LED
쿑errite � �      납     납Filter납x10납Filter납 x100  �2kHz �  Phase   �
쿝od Ant � 읕컴컴켸읕컫컴牡컴컴컴牡컴켸읕컴컴쬔읕컴컴컴�     쿎omparator쳐>�
읕컴컴컴켸            �200kHz             S meter      Local �          �  �
                      읕컴컴<컴컴�                  旼컴컴>컴�          �  �
旼컴컴컴커旼컴컴커 旼커 旼커 旼커넬컴� 旼커旼커 旼커넬컴�2kHz읕컴컴컴컴켸  �
쿣ariable납      � �  � �  � �  납�  � �  납  � �  납�  �           旼컴컴컨커
� 10MHz  척Buffer쳇덕2쳇덕5쳇덕5쳅덕2쳇덕2척�5쳇덕5쳅덕2첼          쿘ultiple�
쿦tal Osc납      납�  납�  납�  � �  납�  납  납�  � �  납          �  Slow  �
읕컴쩡컴켸읕컴컴켸냅컴冒읕켸냅컴� 읕켸냅컴牡컴冒읕켸 읕켸�          �  Loop  �
    �            10MHz 5MHz 1MHz    100kHz    10kHz     1kHz        � Filter �
    쿌FC             Frequency reference/marker outputs             읕컴컴컫켸
    읕컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴<컴컴컴컴컴컴�

FAULTING
On testing I found the mixer was hard clipping very early, & carefully
following the circuit I found it was incorrectly biased, due to uncut track!
Fixing that showed me there was room for improvement here. I noticed there was
2kHz IF beats on several of the Mixer bias points, I changed decoupling caps
from u1 to u47 & reduced the input symptoms. The RF buffer was rebuilt to a
simpler less lossy circuit. (see below)

The problem with the PLL was initially crudely solved with a 100uF & series 2k2
across the AFC line, then it gave a steady 10MHz note.

IMPROVEMENTS
1/ PLL Filter
Looking into the PLL jitter problem, there was a very high Z circuit using a 40
Meg ohm (4x 10m) as the R for a 40 second time constant. This obviously was not
working correctly on this layout, possibly due to leaky old PCB or varicap
diodes. (also discussed in the follow up article)

Anyway I redesigned that bit using electrolytics, something the original
designer avoided. But by using 2 identical electrolytics in series across the
5V power rail, I solved the most of the leakage problem & gained instant half
rail AFC voltage on power up, for fast lock up time.

            +5V컴쩡컴�
                +�100u      4x 1N4148
                ===      旼�<쳐컫캑<쳐커          Scope
Varicap          �       �      �      �            o    IC CD4046
 AFC  <컴쩡컴컴컴)컴컴컴컵캑>쳐컵캑>쳐컵컴컫컴�20k컴좔�< Phase Comparator
         쳐�2k2컴�       �      �      �   �             Pin 13
         �       �       읕�1M컴좔220k켸   � 1u
        ===     +�100u                    ===
       u1�      ===                        �
     컴컴좔컴컴컴좔컴컴컴컴컴컴컴컴컴컴컴컴좔컴컴컴�

I enhanced the variable CR system with 2 more diodes & another resistor, so out
of lock you now have 20k to 200uF (4S), at �1V from lock it adds a 220k (50S),
& then at less than �0.5V adds a further 1M (200S).

The 20k & 1u (20mS) is needed to remove the 5V 2kHz IF pulses & reduce the
residual 12.5Hz phase shift data to below 0.5V ripple. This however is a loop
time constant that would oscillate, so the 2k2 is added in series with the
larger C to damp this oscillation, but the R is not to big as to let through to
the AFC line, any significant amount of the PSK data. The u1 is used close to
the varicap diodes & keeps down any stray hum etc.

At lock, the scope point using a �10 probe or on AC, & locked to a fraction of
50Hz mains (e.g. 12.5Hz) you see.
        _
     +5V 납납납납� 2kHz phase
         납납납납� correction
         납납납납� needle pulses   \/ PSK Data
       ..좋좋좋좋좔-...__________..   Ripple Under
          Mean 2.3V/   납납납납�   /\ 0.5V Swing.
                       납납납납�
      0V_              납납납납�

           SLOW  P.S.K.  DATA
        /~~~~~~~~~~~\____________/

The PSK data pulls on the stored AFC voltage several times a seconds. A steady
2.3V is obtained with Xtal trimmer preset. With an SSB Rx you can hear the
10MHz osc settle down to a stable note after several diminishing 4 second
cycles.
   _
Off ~\      .-.     _._                    Power
Lock  �    /   \   /   \__/~~퀭..饗�--       Up �.__.-�-�..�-饗---
       \__/     '-'                             �
   ----------- 60 seconds ------------           --- 30 seconds ---

2/ There was an ugly hole in the box & a S meter/centre zero meter & switch
fitted, so I developed a circuit for these, a manual Centre Zero mode with a
pot to fill the hole. (I did not have the follow up article then!) 

               Varicap
                 AFC
     pot   旼27k컴�                                      o컴�< S meter circuit
+5V on/off �     \�/         Loop    Green        ______/
 픬\___o\__쳐컴컴47k컴6k8컫�<Filter  LED         �       o컴컴�22k컴�>CD4046
     �           Pot      �  AFC     �           �       CZ mode      Pin 2
     �                  �/         �/          /~~~\
     쳐컴컴컴�3k9컴컴컴캑 T1     旼� T2       쿘eter�
     �                  �\e      � �\e         \___/
     �                   _�_     �  _�_         _�_
     �                           �
     읕컴컴컴�3k9컴컴컴컴컴컴컴컴�

The switched log pot I used had a double make, & I needed a changeover. Leaving
the pot in circuit & using T1 to short incoming AFC line I achieved the same
function, & T2 shorts out the Green in lock LED to indicate it is in manual off
lock mode.

As the pot was a log volume one the varicap angle-frequency action was not
quite linear (needed square law?) so 27k was put from the max end to centre &
6k8 to the switched ground, this gave a really even feel to the offset (�5Hz @
10MHz) with a 2.3V centre.

The Centre Zero meter action was take from pulses directly from the unused IC
pin 2 via a 22k to set the meter sensitivity. If you scope that pin see.

 旼컴� 旼왯컴커
 �   � � 납   �  The average changes a little depending on the two 2kHz phases.
 �   읕� 잎

3/ I found the LC filter after the mixer was heavily loaded by the OpAmp 10k
ohm circuit negating the high Q possible, this may have been partly due to
different component values. Anyway I rewired the OpAmp for high Z & reduced the
gain from 10 to 3.9, now the Q is much greater (narrow bandwidth equals much
less AM on the 2kHz carrier!)

     +12V 쩡컴�
         47
          쳐커
      4u7 �   )||        旼컴컴컴390k커
         ===  )||L       �   -�\      �
MC1496   _�_  )||        쳐컴캑6 \    �
Mixer        �      u1   �    �   7>컴좔>Wien filter OpAmp
Pin 12 >컴컫컨컴쩡컴늘컴�)컴쩡�5 /
           �2kHz�      100k �+�/       OpAmp common
          ===  ===       �  읕컴�1M컴컴<half rail
           � Cs �     u1===
      컴컴컨컴컴좔컴컴컴컨컴컴컴컴

After experimenting with the 2 Cs values for exact resonance of the L, the 1M
OpAmp bias R does not load the tuned circuit & the much filtered IF now has
little AM left on it. So the following Wien filter has less to do to extract
the clean carrier. But I did look at that too..

              BANDWIDTHS
      10kHz                �40Hz        �20Hz         (�10Hz Jitter) <0.005Hz
旼컴컴컴커 旼컴컴커旼컴컴왯컴컴컴왯컴커旼컴컴커旼컴컴컴왯컴컴컴컴컴왯컴컴컴�
쿐xternal� �  RF  납     납 2kHz 납 IF납  IF  납   IF  납  Phase   납  PLL �
� Active �>퀯uffer척Mixer척 LC IF척AMP척Tuned 척Limiter척Comparator척Filter혽FC
쿑errite � �      납     납Filter납x3 납Filter납 x100  납          납      �
쿝od Ant � 읕컴컴켸읕컫컴牡컴컴컴牡컴켸읕쩡컴켸읕컴컴컴牡컴컫컴컴컴牡컴컴컴�
읕컴컫컴켸          200kHz            AGC�                2kHz
     읕컴컴컴컴컴컴컴컴컴컴�<컴컴컴컴컴컴�
                                 DC-20Hz

4/ The Wien filter OpAmp has a gain of half, so it can be driven with a clipped
signal from the earlier OpAmp before its output to the level detector clips. By
changing the feedback R from a 22k to 2x 22k the gain is then 1 & the output to
the signal detector is true until the Wien OpAmp clips. The Wien filter Q is
also doubled, bringing the total IF bandwidth down to about �20Hz. The
frequency setting preset needs to be about 350R, so a suitable R across the
preset to provide this value at the centre of the preset is ideal.

                       �\      旼컴> Level Detector
    Common half rail>컴�3 \    �
                       �   1>컴좔컫컴>IF Clipper
x3 amp >컴�22k컫컴늘컫캑2 /       � 
               � u22 � �/         �
               �     읕컴�22k�22k캑
               � u22              �
               쳐캑쳐컴컴컴컴컴컴켸
             Freq
             500R
     컴컴컴컴컴좔컴컴컴컴컴컴컴컴�

5/ I added an AGC, as I found the mixer & other stages could overload &
possibly add phase modulation from the AM signal before the signal was tightly
filtered. I did not want to remove RF buffer's gain, as this is useful when in
buildings with weak signals. So I used an additional NPN to short out the
external FET preamp's supply, this gave a very good AGC, with no distortion
from AGC action & gave the S meter with usable log scale.

                           旼컴컴컴컫컴컫컴�+12V
                           �        �  2k2                   旼컴>S meter
            d             22k      1M   � 4n7              100K
         g 쳐컴컫�15k커    �        �   쳐늘�>Mixer        preset
||旼컫컴쩡>�    �     �    �   4n7  � �/                     �   u1      IF
||(  �  �\ 쳐쩡�)컴늘컨�>>컵컴컴늘컴좔큃F Buff    旼컴쩡100k컨쩡늘컴*컴<OpAmp
||( === �/\ s�  � 10n       \�        �\e         �   �+     _�_    No  Pin 1
||(  � ===   � ===        AGC쳐컫컴�    �         �  ===     /_\  Series
||(  � /�   4k7 �10n       e/� === �    �         �   �2u2    �     R
  읕컨컴좔컴컨컴좔컴컴컴>>컨컴컴좔�)컴컴좔컴�     � 컴좔컴컴컴좔컴�
External Active Aerial  BNC     u1 �          AGC �
                                   읕�100k컴컴컴컴�

The two 100k Rs set the AGC gain & with the 2u2 set the AGC time constant. It
is just fast enough to remove some of the LF AM components below 20Hz, but
still give a stable AGC, leaving a fairly unmodulated carrier after the narrow
IF filters have removed the higher frequency sidebands.

6/ Local Osc
The narrow 1:5 200kHz pulses fed to the mixer has been ramped a bit with an
additional 4n7 & 1k (was 10k), this yields 3dB more mixer gain (ideally a 100mV
square wave is best).
                               旼�>Mixer Pin10
                           4n7 �     �.    �.
                        0V캑쳐캑     � \_  � \
                               �10n  �   ~-�
                              ===
       74LS132        74HC390  �         74HC390
   +5V컴컴컫컴컴컴컴컴컴컴쩡컴�)컴컴컴컴컴컴쩡컴�
           �14          16�   1k          16�    2kHz
10MHz   旼컨커  10MHz  旼컨컴� 홵200     旼컨컴�  o
Buffer>쨈1 11쳐컴컴컴컴�1  15쳐� kHz    渼3  15쳇컨컴>to Phase
       윱2  8첼        �    9쳐�      50윱4   9쵬     Comparator
                            悶    悶
                            납____납
                              5:1
7/ Output Frequencies
As this design was not quite as published, here is the arrangement.

       74LS132        74HC390  �         74HC390
   +5V컴컴컫컴컴컴컴컴컴컴쩡컴�)컴컴컴컴컴컴쩡컴�
           �14          16�   1k          16�    2kHz
10MHz   旼컨커  10MHz  旼컨컴� 홵200     旼컨컴�  o
Buffer>쨈1 11쳐컴컴컴컴�1  15쳐� kHz    渼3  15쳇컨컴>to Phase
       윱2  8첼        �    9쳐�      50윱4   9쵬     Comparator
       渼3 12척     旼캑3    �        kHz�     �
       척4 13쵬     쳐캑4  13쳐컴컴컫컴컴�1  13쳐컴컴�
       윱5  6첼     �  �     �      �    �     �     �
        �   9척     � 渼14 12첼     �   渼14 12첼    �
        �  10척     � 척2   7척     �   척2   7척    �
        읕컫켸�     � 냅컴쩡켸�     �   냅컴쩡켸�    �
          7�  �     � �  8�   �     �   �  8�   �    �
     컴컴컴좔�)컴컴�)컨컴컨컴�)컴컴�)컴컨컴컨컴�)컴컴)컴
              o     o         o     o           o    o
            10MHz   5MHz     1MHz  100KHz     10kHz  1kHz

8/ Output Protection
By adding a pair of 1N4148 diodes as a clipped attenuator the TTL chips are
protected from external static damage & small amounts of accidental RF (10W?).

              o     o         o     o           o    o
            10MHz   5MHz     1MHz  100KHz     10kHz  1kHz
             /�\  +
              읕컴캑쳐470컴쩡컴쩡�(o Output
                   1u     _�_ _�_ �  1V p-p
                          \_/ /_\ �
                          컨컴컨컴�
RESULT
I now have a very accurate marker for HF (VHF/UHF on harmonics with a steady
pure tone), & I can calibrate Frequency counters or lock them up to this
source, as well as lock up my 100Hz-1GHz PLL signal generator.

For VHF & UHF, Xtal oscillators age & can't be relied on to maintain high
accuracy over several years. This accurate source (better than 1 in 10^7)
enables checking of standards. See my buls "Simple Crystal Oven" & "Crystal
Drift Compensation" for more information.


Why Don't U send an interesting bul?

73 De John, G8MNY @ GB7CIP


Read previous mail | Read next mail


 31.10.2026 18:40:16lGo back Go up