Need help with Tektronix 2710 spectrum analyzer

[Esa Heikkinen]

Hi

I need help with Tektronix 2710 spectrum analyzer...

It's having some frequency domain fault above 1 GHz. Infact there are no
exact point where the fault begins: sometimes it will work even to 1.3
GHz but there's a days when even 800 MHz is too much for it... However
with lower frequencies (below 700 MHz) it always works.

Firmware version is 01.15.90 and installed options are 300 Hz RBW
filter, phase lock, tracking generator, centronics and frequency counter.

When the fault kicks in there's something going wrong with center
frequency control. Frequency domain of the spectrum is shifted about 26
MHz lower, for example is center frequency of 1.2 GHz is set and that
signal is fed to the analyzer the peak will be found at 1174 MHz. Some
times it may "jump" between correct and wrong frequencies when the
center frequency knob is turned. Amplitude is measured correctly.

When changing to narrow spans (20 kHz/div or less) and the phase lock
option is turned on it will sometimes even lock to the wrong frequency!
When this happen the frequency counter option (which can be used to
count any displayed peak's exact frequency) will also show about 26 MHz
wrong.

However it's quite common situation that when the frequency domain fails
the phase lock and frequency counter options won't work anymore. Error
message "FREQ NORM SUGGESTED (1ST LO)" or "FREQ NORM SUGGESTED (FIND
SIDE)" is displayed instead. Of course running that normalization
process does not help and the normalization process will be passed
without any errors.

This unit has YIG local oscillator. I have service manual and I'm
already done the reference normalizations, span attenuator and center
frequency accuracy adjustments, LFVCO and HFVCO adjustments and counter
level adjustments to fix the fault - no help. Service manual won't help
with those error messages for example what's the "FIND SIDE" is meaning?

The center frequency control system is quite complex (it's 80's
technology) and infact there's 26 MHz VCO on center frequency board,
which controls the YIG's coil currents/center frequency. This could
somehow explain the 26 MHz frequency error but how - and why this
problem only happens on higher frequencies?

 

[Joerg]

Hi

I need help with Tektronix 2710 spectrum analyzer...

It's having some frequency domain fault above 1 GHz. Infact there are no
exact point where the fault begins: sometimes it will work even to 1.3
GHz but there's a days when even 800 MHz is too much for it... However
with lower frequencies (below 700 MHz) it always works.

Firmware version is 01.15.90 and installed options are 300 Hz RBW
filter, phase lock, tracking generator, centronics and frequency counter.

When the fault kicks in there's something going wrong with center
frequency control. Frequency domain of the spectrum is shifted about 26
MHz lower, for example is center frequency of 1.2 GHz is set and that
signal is fed to the analyzer the peak will be found at 1174 MHz. Some
times it may "jump" between correct and wrong frequencies when the
center frequency knob is turned. Amplitude is measured correctly.

When changing to narrow spans (20 kHz/div or less) and the phase lock
option is turned on it will sometimes even lock to the wrong frequency!
When this happen the frequency counter option (which can be used to
count any displayed peak's exact frequency) will also show about 26 MHz
wrong.

However it's quite common situation that when the frequency domain fails
the phase lock and frequency counter options won't work anymore. Error
message "FREQ NORM SUGGESTED (1ST LO)" or "FREQ NORM SUGGESTED (FIND
SIDE)" is displayed instead. Of course running that normalization
process does not help and the normalization process will be passed
without any errors.

This unit has YIG local oscillator. I have service manual and I'm
already done the reference normalizations, span attenuator and center
frequency accuracy adjustments, LFVCO and HFVCO adjustments and counter
level adjustments to fix the fault - no help. Service manual won't help
with those error messages for example what's the "FIND SIDE" is meaning?

The center frequency control system is quite complex (it's 80's
technology) and infact there's 26 MHz VCO on center frequency board,
which controls the YIG's coil currents/center frequency. This could
somehow explain the 26 MHz frequency error but how - and why this
problem only happens on higher frequencies?

You could check the power supply and bias amp for the YIG. But if the
YIG itself is toast I think most are sealed assemblies that can't really
be field-repaired.

 

[Max65]

Hi Esa.
First of all, let me tell you that I never seen the 2710 schematics,
the things I write below are just suggestions based on intuition
(maybe you had already do that).

The 26Mhz VCO you find on the center frequecy board, should enter into
a mixer. Look for the LO intput level of the mixer, maybe it is too
low or near the level at which the mixer starts to work (a passive
mixer usually need a LO signal of 4-10dbm to work).
Of course, maybe the 26MHz VCO frequency or the other frequency
incoming into the mixer the LO signal, so check both the signals for
the right level.
Maybe it's just a "cold" solder joint.

Hope the above will help.

Good luck.

Have a great day.

Massimo

 

[Esa Heikkinen]

Hi Max.

First of all, let me tell you that I never seen the 2710 schematics,
the things I write below are just suggestions based on intuition
(maybe you had already do that).

I did some scans from service manual schematic and circuit description
pages. Maybe these will help to get further:

http://www.amigazone.fi/files/tek2710/schema1.png
http://www.amigazone.fi/files/tek2710/schema2.png
http://www.amigazone.fi/files/tek2710/schema3.png
http://www.amigazone.fi/files/tek2710/schema4.png
http://www.amigazone.fi/files/tek2710/schema5.png
http://www.amigazone.fi/files/tek2710/description1.png
http://www.amigazone.fi/files/tek2710/description2.png
http://www.amigazone.fi/files/tek2710/description3.png
http://www.amigazone.fi/files/tek2710/description4.png
http://www.amigazone.fi/files/tek2710/description5.png
http://www.amigazone.fi/files/tek2710/description6.png
http://www.amigazone.fi/files/tek2710/figure1.png
http://www.amigazone.fi/files/tek2710/figure2.png
http://www.amigazone.fi/files/tek2710/figure3.png

I do how the frequency synthesizers work but still can't get the figure
of this! For example there are no N divider at all and the description
claims that the harmonics of signal called "strobe" is used instead? And
this "strobe" signal is infact that ~26 MHz signal mentioned before! But
there are 11 MHz VCO also... And sometimes it counts "beat" and
sometimes phase offset like normal PLL. I don't really know why this is
so complex - may be there wasn't dividers fast enough to count 2.11 -
3.91 GHz in the end of 1980's?

I guess that it locks to wrong harmonic of the strobe, because the
frequency is always ~26 MHz off when the fault kicks in. But why this is
happening only with higher frequencies - just no idea. Also, the service
manual claims that with wide spans the strobe signal is not even turned
on and the YIG main coil is driven with suitable current controlled by
CPU. So how the frequency can be same amount off without "strobe"? Only
explanation that I could guess is that the fault was on when frequency
normalization was run and because of that there are wrong DAC values in
some kind of "main coil current table" in the CPU.

Clearly I'm too novice with those 1980's RF circuits...

 

[Max65]

If I were you, I checked the incoming signals at the mixer HY450. If
I'm right, its output (BEAT NOTE) should be either the PLL feedback
signal when used in small span scans and the center frequency tuning
signal in the other cases.
Does Q460 work correctly?

Massimo

 

[Esa Heikkinen]

If I were you, I checked the incoming signals at the mixer HY450. If

I'm right, its output (BEAT NOTE) should be either the PLL feedback
signal when used in small span scans and the center frequency tuning
signal in the other cases.
Does Q460 work correctly?

Yes I already did some measurements from it's output from (from J430)
with oscilloscope. Signal at this point is weak, infact it can't be
clearly measured with digital oscilloscope at all, execpt when it's
frequency is low (couple tens of kHz) when it's amplitude raises. The
frequency here changes when spectrum center frequency is tuned. The
strobe signal in J460 is strong.

Infact this strobe signal is not always turned on when instument is in
normal mode. It turns only momentary on when frequency correction is
done between sweeps, occassionally. I guess the reason for that is
keeping noise floor good (-143 dBm in this unit). But of course it's
possible to turn the strobe signal continuously on with instument
diagnostics menu.

When J430 is disconnedted there will be an error message "CANNOT COUNT
BEAT FREQUENCY" in the screen and sweeping also stops, instrument goes
useless. There are also frequency counter tests in diagnostic menus.
With this function it's possible to monitor the VCO and beat frequencies
real time. If J430 is not in place / no signal these counter tests will
also fail. So there must be some signal on J430 because there are no
error messages and instrument works perfectly below 700-1 GHz.

 

[Max65]

Infact this strobe signal is not always turned on when instument is in
normal mode. It turns only momentary on when frequency correction is
done between sweeps, occassionally. I guess the reason for that is
keeping noise floor good (-143 dBm in this unit). But of course it's
possible to turn the strobe signal continuously on with instument
diagnostics menu.

It should be continuously on when you use narrow spans too (below
10kHz),
the reason it's not continuous on moderate spans is because it is
used
as reference for a while, then the correction error is sampled and
hold
to keep the YIG sawtooth signal centered on the desired frequency.

If J430 is not in place / no signal these counter tests will also fail.

Do you know what's the frequency test?
I mean, does the test run below or above 700MHz?

So there must be some signal on J430 because there are no
error messages and instrument works perfectly below 700-1 GHz.

I don't agree your statement above, maybe that (in some
circumstances)
the mixer is underdriven by Q460 or the YIG signal is too low, in
those
situations the mixer may have some conversion losses above 700MHz.
The fact that the defect is frequency dependant, leads me think that
the
problem should be into the YIG interface (limited to the HY450 mixer
circuitry) or the YIG oscillator.
Check the YIG level on J100 too. The YIG oscillator is ovenized.
For that reason its circuitry is thermal stressed every instrument's
power on/off cycle, and it's not rare to have cold solder for that.

Have a nice day.

Massimo

 

[Esa Heikkinen]

Max65 kirjoitti:

The fact that the defect is frequency dependant, leads me think that
the problem should be into the YIG interface (limited to the
HY450 mixer circuitry) or the YIG oscillator.

That could be bad. I'm quite sure that there are no spare parts
available for those. Expecially HY460 which internal circuit is unknown.

Check the YIG level on J100 too. The YIG oscillator is ovenized.

Need another spectrum analyzer for that because I don't have any power
meter for those frequencies... It seems that there are no adjustments
for YIG level. The RF level adjustment in the local buffer seems to
control only the 1st mixer injection but not HY460 input. The YIG has
some internal trimmers but those are glued. I have no idea what those do
and there's nothing about those in service manual. Service manual says
that if YIG is defective it will be changed as a compete installation.

For that reason its circuitry is thermal stressed every instrument's
power on/off cycle, and it's not rare to have cold solder for that.

Is it possible to fix soldering INSIDE YIG? There is a PCB on top of YIG
can, it's the 1st lo interface. I have already check solderings in it.
But the YIG can itself is closed with screws. I didn't open it because
I'm unsure if it will damaged if opened?? However schematics show that
there is circuit inside YIG also and cold solder could be there.

 

[Max65]

Hi Esa,

 > HY450 mixer circuitry) or the YIG oscillator.

That could be bad. I'm quite sure that there are no spare parts
available for those. Expecially HY460 which internal circuit is unknown.

I don't want let you down for this, my own was just an hypothesis.
The HY460 should be just a mixer and I don't believe it could be
defective
because it's placed in a safe area (I mean, it's not like the input
mixer that
could be damaged by the unknown incoming signal of the analyzer test
port).

Need another spectrum analyzer for that because I don't have any power
meter for those frequencies... It seems that there are no adjustments
for YIG level. The RF level adjustment in the local buffer seems to
control only the 1st mixer injection but not HY460 input. The YIG has
some internal trimmers but those are glued. I have no idea what those do
and there's nothing about those in service manual. Service manual says
that if YIG is defective it will be changed as a compete installation.

I can't figure out what trimmer you see inside the YIG, since the
schematic
doesn't show any there.
Anyways here in Italy we say: "don't cry over spilt milk, go ahead".
If it is really defective, you could try to open it and see what to
do.
But before it, you should establish whether it is faulty or not.

You should look the Beat Note at J430 with your oscilloscope when the
analyzer shows the wrong behaviour and see if the signal is similar to
the
one you see at lower frequencies where it works right. Since the
signal is the
result of the beat of the YIG and the VCO signals, you should change
the
center frequency slowly and see what happen for at least 26-30MHz and
compare the two behaviours.
If you see differences, the faulty component/solder should be before
J430,
otherwise it should be after J430. This could be a good start point by
my
point of view, don't you?

Is it possible to fix soldering INSIDE YIG? There is a PCB on top of YIG
can, it's the 1st lo interface. I have already check solderings in it.

I remember that a good friend of mine who worked for Rohde & Schwarz
for
a long time repaired one of that one day, I see it. The real problem
was to
solder the components on the alumina boards which are highly thermal
dissipative, so you need an high power soldering iron and limits the
time.
(one other issue was don't lose the small yttrium iron garnet, I
remember
it was not greater than 3mm in diameter).

But the YIG can itself is closed with screws. I didn't open it because
I'm unsure if it will damaged if opened?? However schematics show that
there is circuit inside YIG also and cold solder could be there.

Yes, unluckily often the cold solders are there because of the
thermal
stresses I told you before.

Check for the behaviour on J430 and let me know, nothing is lose for
the
moment.

Massimo

 

[Max65]

Hoops I feell an idiot, in the previous message I wrote: "I can't
figure out what trimmer you see inside the YIG, since the schematic
doesn't show any there." Better looking to the schematic I realized
that you refer to R330, R340 and R130 on the YIG bias board called
A19A1 1st LO INTERFACE.

More, did you adjust the RF OUT LEVEL (R288 on the 1st LO BUFFER
board?
If you had to do it because the signal on J380 was to low it clould be
a symptom of bad behaviour of the YIG.
Anyways you could measure the voltage level on the emitter of Q190.
The higher is the voltage the lower is the YIG signal incoming on the
board. You should measure it below 700MHz and above it and loot for
any difference.
Massimo

 

[Esa Heikkinen]

Hi Max,

I don't want let you down for this, my own was just an hypothesis.
The HY460 should be just a mixer and I don't believe it could be
defective because it's placed in a safe area (I mean, it's not

Here's a picture of HY460:
http://www.amigazone.fi/files/tek2710/HY460.jpg

I checked that with microscope and no any visual defects. But of course
that doesn't prove anything about it's condition...

You should look the Beat Note at J430 with your oscilloscope when the
analyzer shows the wrong behaviour and see if the signal is similar to
the one you see at lower frequencies where it works right.

That's a good idea and I planned to do that now in weekend. But like I
told before the signal here is so weak that it can't be seen with my
digital oscilloscope and my analog oscilloscope is too slow. So I plan
to build some quick-and-easy preamp for oscilloscope - with some OP AMP!
However after digging all my "treasure shelfs" it seems that I haven't
got any opamps with enough bandwith. Maybe I will try one with 5 MHz
bandwith (-3 dB point) it's possible that it will give some gain also
with higher frequencies - or not... :-/ However the signal on the
digital oscilloscope is so weak that i would need a gain of 10 or 100 or
more. Maybe some video amplifier taken from scrap VGA monitor could do it...

center frequency slowly and see what happen for at least 26-30MHz and
compare the two behaviours.

Last time I tried to do that the signal was visible only to few hundred
kHz and then the amplitude decreases rapidly. But maybe with preamp...

Check for the behaviour on J430 and let me know, nothing is lose for
the moment.

Yes I'll back when there's someting new.. Have to find some solution for
the oscilloscope preamp..

 

[Esa Heikkinen]

More, did you adjust the RF OUT LEVEL (R288 on the 1st LO BUFFER

board?

For now I haven't touch it at all. If I understood correctly it will
only adjust the 1st mixer injection and has no effect on HY460 input?

Also I'm not sure if I lose the amplitude measurement calibration if I
adjust that? Of course it's possible to recalibrate the analyzer after
that, but may be it's wise to do some reference measurements before
adjustment...

Anyways you could measure the voltage level on the emitter of Q190.

Have to try that also when 1st lo buffer is open - it's almost hour of
disassembling, tens of screws etc. and some semi rigid cables have to
replaced with coax until it's possible to test run 1st lo buffer on the
table... I hope that these kind of set-up or running it without it's
shielding will not change it's performance too much. You can see the
part of 1st lo buffer assembly on HY460 picture.

 

[Max65]

Here's a picture of HY460:
Uhmm... It looks like a signle balanced mixer, but the outcoming
signal from T450 should have a great shape, it must have harmonics
over the GHz limit to work with that couplers.
Do you have tried to see the signal on R450 with the scope?

Last time I tried to do that the signal was visible only to few hundred
kHz and then the amplitude decreases rapidly. But maybe with preamp...

Never use x10 probes (if you did) since they highly reduce the
bandwidth of the scope. Anyways the T450 transformer output seems to
be insulated from ground and have very low impedance, so your x1 probe
should be used without any problem (doens't matter where you put the
ground terminal).
If your scope has at least 40MHz bandwidth you should see a discrete
square wave there (I guess).
I wonder reading that you read a so low signal at J430 because looking
to the schematic of the following stage, you can see just two opamp
having low gains (about 10) connected there.

For now I haven't touch it at all. If I understood correctly it will
only adjust the 1st mixer injection and has no effect on HY460 input?

Yes, it is. I asked just to know if the YIG has fall down of some dB
there.

Also I'm not sure if I lose the amplitude measurement calibration if I
adjust that? Of course it's possible to recalibrate the analyzer after
that, but may be it's wise to do some reference measurements before
adjustment...

Do not adjust any trimmer if you don't know the procedure for do it, I
asked just to know if the service manual said something about it.
Have a nice weekend.

Massimo