OpenBCM V1.07b12 (Linux)

Packet Radio Mailbox



 Login: GUEST

G8MNY  > TECH     23.04.22 08:23z 325 Lines 17368 Bytes #7 (0) @ WW
BID : 644_GB7CIP
Subj: Analyser Takeda Riken TR4122B
Sent: 220423/0813Z @:GB7CIP.#32.GBR.EURO #:644 [Caterham Surrey GBR] $:644_GB7C

By G8MNY                              (Updated Nov 21)
(8 Bit ASCII graphics use code page 437 or 850, Terminal Font)
Being a sucker for test gear, I bought this rather large & very heavy vintage
Japanese Spectrum Analyser, also badged as Adventest, for a "song" at a club
junk sale. It has a tracking generator & 2 digital stores. It was not known how
well it worked, but I had seen it switched on at a club's test evening, so I
knew it was not quite all OK.

                          Vernier Readout
        旼컴컴컴컴컴컴컴컴컴컴컴컴컴컴컴풣req Tune
        쿽旼컴컴컴컴컴컴  ~~~~~~( O )
   CRT  쿽              坪컴컴컴     쿗HS/Center
Controls쿽              . _  -      
        .   DISPLAY    Sweep        쿞can &
        .              ( O )   ( O ) 쿍andwidth
  On/Off쿚              & Spot      _
         읕컴컴컴컴컴컴 Atten         쿣id Filter
        . . . 旼컴컴컴  & Gain        
 Presets. . . 쿎ounter  ( O )   (0)  0쿎al BNC
              읕컴컴컴   .   __      
Tracking() ()       o . __   ()쿙 RF In
 Gen Out읓컴컴컴컴컴컴컴컴컴컴컴컴컴컴컫
           _ O  [] []     [][][][]   
          Phones   Digital store unit

It came with no information or front cover, but here is my spec for it so far..

Freq   : 100kHz-1500MHz (Display & Mechanical Vernier counter to 1750MHz)
Input  : 50 N,+20dBm(100mW)Max, 0-50dB atten, in 10dB steps, 1dB comp @-10dBm
IF Gain: 0 to +50dB, in 10dB steps (Gain/loss scale) & a Calibrated +12dB Pot
IF B/W : 3MHz, 1MHz, 100kHz, 10kHz, 1kHz, & 500Hz, all sweep friendly shape
Video  : Video filter, 1/30 of IF BW. For noise averaging, sweep speed adjusts
Cal out: 50 BNC with 200MHz @ -30dBm, harmonics check to 1400MHz sweep cal.
Sweep  : 2kHz-100MHz/Div (B/W linkable). Sweep stop for envelope & Phones use
Phones : AM 3.5mm jack, and AM/FM switched 6mm jack on memory unit
Marker : Movable spot stops sweep momentarily, for 8 digit Freq Counter readout
Modes  : Trace (U sweep marker spot) / Manual select sweep time / Auto (best)
Display: Linear / 2dB/Div / 10dB/Div. 10 x 10 Div etched tube graticule
Gen out: Tracking Generator, optional filters. 0 to -50dBm in 10dBs, 50 N
Store  : 2 Digital memories, one can be subtracted to normalise display
Outputs: IF @ 10.7MHz, X, Y, & Z, for a 2nd display. Data/printer socket out
Power  : 122VA selectable AC Volts internal / External DC (Special Batt pack)
Weight : 24kg!

Well it was very cheap, so I expected the odd problem, I soon saw one. The
precision input attenuator was broken. The first 10 20 & 30dB worked, but 40 &
50dB fell off the screen. It had a DC isolator to the input N socket, so I had
to open it up & unbolt bits to test further.

I located the attenuator, to find it was a SMA connected relay controlled
remote module after the N input. On testing I found all the relay control lines
went on & off OK, so it nt a control switch or relay drive logic fault.
Removed attenuator, took off the RF cover. Inside I saw 6x 6mm dia tiny uWave
relays & 3 RF attenuators & silver track line on the RF low loss substrate. It
was fairly symmetrical, & I was able to identify an open circuit metal
deposited R only 2mm square, on the input side of the offending 20dB attenuator
section. So it had been "burnt out", luckily only this attenuator & not the RF

Searching Jessop G6JP's RADIO DATA book, I found the theoretical value for the
series arms of a 50 20dB T atten should be 41R. I initially solder tinned the
silver strip lines either side of the faulty R, & tried a small lead resistor
of 47R, that proved OK, so this realy was the fault!

                                          2 double pole change over relays 1
       1a     o-50컫50훟     2b          & 2 are used, this ensures high
         o___  1b     2a ___o            isolation with 4 contacts bypassing
 SMA o)_/    \o_________o/    \__Next     the through path when attenuating.
Input   o                   o   Atten
 50     읕41R컴컫컫컴41R컴            The unused paths are terminated.
            /   20R 20R    trimR
         Fault                         The fixed trimR used in calibration?

So for a proper job, the 41R had to be accurate, small & non inductive, if it
was to work well to 1500MHz. After consulting an old Basic Hamhelp Program for
multiple Rs, I used 3 surface mount Rs in parallel 2x100R & a 220R to give the
41R. Once stacked up & soldering together the SM Rs were not too difficult to
solder to the pre-tinned stripline, although it looked a bit messy in the end.

Success, re-assemble & tested, display now showed nice even steps of the 40 &
50dB atten & all exactly 10.0dB & quite flat to 1.5GHz.

Later after many days of use & taking it to club fixit nights, it packed up!
The fault was a humming transformer & I found -15V was being shorted out!
Disconnecting the PSU load wire brought the volts back. The wire went to the
motherboard, so I pulled out the 5 cards & short went on my meter. The Log
detector PCB had the short, no high power components, just 10 x 10dB gain amps/
detectors. With the -rail the track was easily identified with its electrolytic
caps +ve to earth. I removed 3 of them, but they were not faulty, eventually I
came to the end of the double sided weaving track, to see a mechanical 3mm long
soldered lead bent over & shorting to adjacent earth. Fixed OK.

After a lot of soak testing, to null some of the frequency drift, I noticed
ripples appearing on the digital store trace line. This was hum was on the +5V
in the difficult to access add on lower memory unit. The reservoir cap value
before the regulator, was low at only 1000uF, & the +15V PSU had 2 of them, so
I swapped 1 of them over to the 5V PSU, & halved the pre reg ripple, so all OK.
On closer examination of piggyback 2nd memory PCB, I noted 2 Tant caps on the
5V rail had changed colour. Testing them 1 was as low 1K. It seems that this
added load to the mis wired PSU C gave the fault. After replacing them, &
destroying 2 more, I DC tested to realised the silk screen legend was actually
wrong, with + on the -ve lead! So an original maker's PCB fault I was not

 Log                                    The Long persistent White/Blue P7
10dB  -5        0        +5  Lin  2dB   Phosphor is viewed through a clear
/div 0旼쩡쩡쩡쩡쩡쩡쩡쩡쩡100% 0 /div  thick blue glass gives a clear display.
    20쳐탠탠탠탠탠탠탠탠탠80%          Usual Brill, Focus, & a bottom blanking
    30쳐탠탠탠탠탠탠탠탠탠70%          controls, Astig & Geom are presets.
    50쳐탠탠탠탠탠탠탠탠탠50% -10      10 x 10 div graticule with 0.2div
    60쳐탠탠탠탠탠탠탠탠탠40%          marks, are etched inside the tube
    70쳐탠탠탠탠탠탠탠탠탠30%          for accurate no parallax readings.
    90쳐탠탠탠탠탠탠탠탠탠10%          The Tune Freq calibration can be
   100읕좔좔좔좔좔좔좔좔좔0%  -20      switched to LHS or Display centre.
      -5        0        +5


X: With centre freq mode, Zooming in with sweep range & on a signal (Zero Line)
   in the middle & then zooming out, sets the X screen position middle.
   Frequency cal lines (& harmonics) set the X Gain to get scale right.

Y: Linear scale mode, no signal defines the lower Y graticule. Y Gain is set in
   10dB/Div mode, over 90dB of atten/IF gain, to get the scale right.

Memory: Repeat for the 4 preset Shift & Gains on the side of the memory unit,
so there is no difference with memory on/off. N.B. Position & Gains interact &
repeat adjustments may be needed.

The underside has the RF modules, which are well sign written, hence I was able
to work out much of a schematic for it..

SCHEMATIC         __________
Cal o)컴컴컴컴컴캑200MHz Osc쳐컴컴컴컴컴컴컴컴컴컴컴컴
                 읕컴컴쩡컴켸      ___________               _____
Track  _____     旼컴컴좔컴컴컴__10.7MHz Osc         15V킘C or쳐DC
Gen o)퀮tten쳇컴캑Gen Up Mixers_ 읕컴컴컴컴컴          +5V퀾AINS
Out   읕컴컴冒   읕쩡컴컴컴컴컴  __________          +150V PSU 쳐230V
    _______  넬컴컴좔커  _______ 윱411MHz Osc              읕컴컴
    Gated  납YIG UHF쳐 Sweep  읕컴컴컫컴켸        
   쿎ounter쳐冒  Osc   쿎ontrol-----------------------------풞UTO
   읕컴쩡컴  읓컴컴컴 읓컫컫컫                             |
   旼컴좔커     Frequency   |speed                         |
   쿘arker쳐컴)컴컴컴컴컴켸  |                              |
   읕컴컴켸                  |                              |
      _____ 旼컨커旼컴컴컴   |旼컴컴왯컨컴왯컴컴커旼좔커旼컴컴좔컴컴컴컴커
In o)퀮tten척Mix1척IF Gain쳐))613IF척Mix2척210IF척Mix3척10.7IF & Filters첼
     읕컴컴牡컴컴牡컴컴컴켸   |읕컴컴牡컴컴牡컴컴켸읕컴켸읕컴컴컫컴컴컴컴켸
   旼컴컴컴                X ------<---풣ilter--<------------          
     CRT       旼컴컴컴컴컴좔 Y     旼컨컴컴컴좔커   旼컴컴컴컴컴컴컴커  
   쿏ISPLAY쳐쩟쩡퀱IGITAL STORE쳐컫컴컴큋ideo Filter쳐컴퀽in/Log Detector쳐캑
   읕컴컴컴 납 읕컴컴컴컴컴컴      읕컴컴컴컴컴켸   읕컴컴컴컴컴컴컴켸  
X o)컴컴컴컴켸납      & PSU                                                
Y o)컴컴컴컴컴冒         ______                          ____________      
Z o)컴컴컴컴컴켸   o컴컴퀮M AMP쳐켸                      쿑M Det & Amp쳐컴컴
Phones o)컴컴컴컴\o    읕컴컴켸                         읕컫컴컴컴컴켸     

Understanding the internals does help you use gear better.

This is the heart of this Spectrum Analyser; it uses a free running, UHF osc
with a temperature controlled high Q "Yttrium Iron Gamete" bead, like a ferrite
core, in the centre of the UHF osc coil. It's magnetic inductance is changed
with powerful external magnetic field at right angles to the UHF coil, & this
linearly sweeps the oscillator between 613-2113MHz. YIGs are very stable (when
temperature controlled) & like Xtal oscillators have very low noise sidebands,
unlike PLL & DDS sources, have NO sprogies/spurs, but with no PLL they drift!

On close in work I have found the YIG osc drifts a good 30MHz @ 1400MHz from
cold to hot. To try to compensate this, I used a 47k NTC thermistor with an
insulated in heat shrink & polystyrene cover, all taped to top of the warming
YIG Osc Box mounting point, 10cm below the main geared vernier Pot, & fed with
a screened lead to reduce any hum pickup.

    Earthed @   Preset1
   SWEEP CAL 0V컴100K컴컴47kTh컴커
                 - - - - - - - 엿     ~+5V DC from 
 SWEEP PCB __                       __ SWEEP PCB
    LF cal _()컴컴컴Verner10K컴컴좔()_ HF cal
   pot R86             /\              pot R83                    Comparitor
                                    __                            FET
                        읕컴컴컴컴컴()_ SWEEP PCB Mean Freq10k컫컴Gate
One end of the thermistor goes to ground (Cal pot earth point) via Preset 1.
This limits the temperature compensation action at hottest temp. The other end
is wired to the 10 turn vernier tuning pot (most +ve lead ~5V) which is the DC
feed from pot R83, & which also sets the slope (gain) of the compensation
action feed from pot R83.

Minimising drift..
1/ Set FRONT PANEL "CAL" & "FINE TUNE" controls to mid position.
2/ When warmed for only 5 mins, & repeatedly adjust R86 & R83 to give the
   correct vernier calibration @ 200MHz R66 & R83 on the 7th harmonic at
   1400MHz of the 200MHz Cal Ref. to agree with centre marker spot counter
3/ When warmed up with cover on for 1hr, set vernier to 1400 & adjust Preset1
   (cover off) for 1400MHz on the counter. Note there is some temperature /
   magnetic / hysterisis settling time on the YIG osc until it & the control
   circuit settle down & stop drifting.
4/ Let it cool down for an hour or so (cover off?), & go around the loop from
   2/ again & again, until hot & cold frequencies do not need adjusting
   (<3MHz out in 1400MHz).

N.B. This will not stop all display drift when zoomed right in & @ 500Hz B/W,
but it will make a big difference!

CALIBRATION (After warm up)
Using the internal 200MHz -30dBm Calibrator & its harmonics, with a patch lead
you can do most of the calibration:- Test Input Attenuators, IF preset gains, 
Frequency & Sweep controls, & you can check & set front panel CRT Preset
Geometry, Shifts, & Gains, etc. Under the top cover are lots of presets &
timers for..

PSU, has 5 pots for setting left to right +15, -15, +5, & +150V.

EHT Box, next to the CRT has 2 presets for "EHT" & "Focus" etc.

Sweep board PCB, is next to the CRT & at back "Pincushion", & "Trapezium"
geometry pots, then "unknown pot", then "Y shift", "2dB mode gain", & "2dB base
line" pots.

The Log Amp Screened PCB adjacent, has no accessible adjustment presets.

IF PCB, has gain equalising pots between filter bandwidths "100kHz", a "shape
trimmer", "under 100kHz". Then pots for the "+30dB", +20dB", "+10dB" control, &
gain control "+12dB", "0dB gain control" cal pots, then an overall gain preset
pot used with front panel Cal.

IF Filter screened PCB, has 1 preset for gain equalising "wider than 100kHz".

Timebase PCB, has 3 interacting sweep speed pots that control the complex sweep
stop & measure spot frequency & carry on. Scope test point & adjust time with
spot off screen for best time accuracy in manual time mode. Nearest pot is
another for "trigger", used in mains Line lock & Video modes.

Sweep PCB, has control of YIG osc magnetic field current & freq limits. Middle
of PCB is "X gain", an "unknown", "R83 1400MHz", "R86 200MHz".

On the underside there are several metal boxes..

The RF unit has 200MHz Calibrator "RF Level" pot, & "Freq" trimmer. With a
patch cable in, the 613 & 230MHz IFs can all be carefully peaked. Several of
the RF unit controls feed ref osc to the tracking generator box, & are adjusted
with a Generator patch cable in place. Use wide sweep to see a flattest line to

The tracking generator box, has harmonic filters in filter mode, & a rippled
line is displayed. Adjust all, for best flattest line. Adjust O/P match
flattest near 1500MHz & Output level for correct level.

The Counter can be calibrated with an external ref osc @ 100MHz patched in
after the 1/10 pre-scalar box, to read 1000MHz.

Underneath & difficult to access, is the add on memory unit. It has presets for
"Y top" gain clip, "Playback sharpness" trimmer, "Y in" Bottom clip, (test on
2dB range), "X position" trigger, "X in" gain. On the 2nd memory Piggy back
card is a "spot" location pot to set identically to the non memory position.

Compared to simple home made & commercial Spectrum Analyser I have used, this
"Proper one" is not as sensitive & takes some getting use to. With the multiple
bandwidths available from 3MHz for fast sweeps right down to 500Hz close in
work. The inter-lockable Sweep & Bandwidth control & the Auto sweep speed,
stops you getting a faulty spectrum display (warning LED). But as this is not a
modern PLL system, slow drift of the swept YIG UHF oscillator is noticeable on
close in work. but it does reduce after 1 hr warm up or so. One shot sweeps
with memory easily solves most of this drawback though.

The Zero Hz line can be seen when the YIG Osc is on the 1st IF freq, as well as
about 300MHz of the spectrum mirror image.

With slow sweep, narrow bandwidth, video filter in & the storage memory, the
noise floor can be seen as low as 98dB down on a just overloading 0dBm input.

Harmonics & images are well down, if the signals are >3MHz apart so they are
not in the 1st IF together.

The tacking generator with sweep stopped can be used as an RF generator, it is
also very useful on filters & pre-amps etc. But the first thing you notice is
all your bad cables & connectors! Of course it can also be used just a sweep
gen too, triggering a scope from Z out etc.

Best accuracy for adjustments is in Linear mode or the expanded 2dB/Div.

The 2 memories can be used in subtraction mode, so a change in display can be
seen, or an un-flat response of test leads normalised before measurement. And
the Peak store mode lets you capture a rouge signal too.

For power RF inputs, I have a N connected 50W 40dB attenuator DC-5GHz to put in
front, so hopefully I will not be burning out the input!

I have only found 3 pages of the user manual on line so far, without buying the
full price set of 5 manuals, so I wonder if any readers have used this S.A. or
have any information on it?

Also see my Tech buls on "Spectrum Harmonic Demo circuit", "A Versatile Pulse
Tester", "Clip on QRM Probe". "Analyser SWIRES RESEARCH SA87", "Spectrum
Analyser mods 88-89", "Power Line Telecomm QRM", "SSB Demo circuit", "Marconi
2019A Sig Gen", "FeelTeck Dual DDS Osc FY6600-60M", & "RF Directional Coupler".

Why don't U send an interesting bul?

73 de John G8MNY @ GB7CIP

Read previous mail | Read next mail

 16.05.2022 09:39:49zGo back Go up