|
I've been wanting to get "into RF" for a while, and have wanted a spectrum analyser. I picked up a Tek 492AP cheaply because, while basically functional after recapping, the owner was clear that he didn't have the means to test it properly. Now I have time to look at it, and before I poke around too much, I'd like opinions on what might be causing "spurs" and "breakthrough" (note the quotes, indicating that they are probably incorrect terminology).
Hopefully people can direct me towards which subsystems are likely to be faulty. I'm guessing RF parts (especially the "unprotected" first mixer) are the culprit, but hoping it is something simpler or cheaper.
Available RF equipment:
- Tek 492AP, s/n 8010373, firmware 4.6/1.1, no option 001 preselecter/limiter.
- Cheapo AMD4351 signal generator, 34-4400MHz, but frequencies below 2200MHz are especially rich in harmonics since they are derived by dividing down digitally
- uncalibrated SDR, up to 1500MHz
The sig gen is coupled to the SA using an external 20dB pad and DC block, and the SA is set with 20dB internal attenuation; I believe that precludes overloading the input.
The artek manual states there are two different LOs depending on the band in use: 2072MHz and 829MHz, and a common IF of 110MHz.
So what do I notice? (I'll try to keep this readable, but the "text only" forum format may get in the way.) All measurements are made with the SA itself.
The "breakthrough" symptom
==========================
Low band, 2072MHz LO, sig gen -20dBm
- 2075MHz, display as as expected
- 2072MHz, -16dBm pure white noise, i.e. a horizontal straight line
- 834MHz, display as expected
- 829MHz, display as expected
Band 1, 1700-5500MHz, 829MHz LO, sig gen -20dBm
- 2075MHz, display as as expected
- 2072MHz, display as as expected
- 834MHz, display as expected
- 829MHz, -16dBm pure white noise, i.e. a horizontal straight line
The "spurs" symptoms
====================
Low Band (0-4200MHz)
========
Sig gen input 2200MHz -20dBm, spurs:
- 2456MHz,-43dBm
- 64MHz,-36dBm
- 192MHz,-56dBm
- 818MHz,-57dBm
Sig gen input 2210MHz (i.e. +10MHz), powers are unchanged
- 2486MHz (i.e. +30MHz)
- 69MHz (i.e. +5MHz)
- 206.8MHz (i.e. +15MHz)
- 828.6MHz (i.e. +10MHz)
What can be inferred from those spurious frequencies
and frequency changes?
Band 2 (1700-5500MHz)
======
Artek manual: 1st IF 829MHz, 2nd LO 719MHz
Sig gen input 2200MHz, -16.6dBm (not -20dBm)
- 4942MHz,-41dBm
- 2885MHz,-54dBm
- 2056MHz,-56dbm
Sig gen input 2200MHz, (i.e. +10MHz), powers are unchanged
- 4972 (i.e. +30MHz)
- 2900 (i.e. +15MHz)
- 2071 (i.e. +15MHz)
Band 3 (3000-7100MHz)
======
Artek manual: 1st IF 829MHz, 2nd LO 719MHz
Sig gen 2200MHz (i.e nominally out of band)
-3858MHz,-15dbm
Sig gen 2200MHz (i.e nominally out of band)
-3868MHz (i.e. +10MHz)
Now, what's causing/generating those frequencies?
|
a bit OT, but have a look here:
|
I'm not sure exactly what you're looking for, but I can provide some info on the 492 frequency plan and operation, which I believe is the same as in my 494. I studied it extensively when first acquired, but that was quite a few years ago.
I believe that in band 1, 0-1800 MHz, there is a low-pass filter that cuts at around 1.8 or 2 GHz, so very little higher frequency content should get through. Since this unit apparently does not have a preselector, the higher bands can have all kinds of spurs in the display that may cause some confusion, but can be figured out. First, don't worry too much about the spurs - they must come from various combinations of all the frequencies and their harmonics involved in the system, so everything has to add up.
One particular issue in your "breakthrough" listing is that on the higher bands, the IF is 829 MHz, so if that's present in the input signal, it can get through the mixer and right into the IF with little attenuation. That's one of the benefits of having a preselector - it eliminates IF and other out of band feedthrough. The corresponding issue on the low band with 2072 MHz input should not happen, unless there's no LPF - you may want to confirm whether it has this.
Concentrate on band 1 first, and confirm proper operation with the built-in 100 MHz reference. Getting the right frequencies (n x 100 MHz) displayed, with nearly correct fundamental level, will show that it mostly is OK. If you need to investigate the LOs, some things can be readily figured out.
Early on I had no microwave equipment, and tried to figure out as much as possible using only the SA itself, and a 561 sampling system. One thing I discovered is that the the second LO 2182 MHz is always on, while the 719 MHz one is only activated in the appropriate bands.
In band 1, the SA cannot see its own 2LO, but in the upper bands it should be possible to see it (as I recall, the bottom of band 2 is 1.7 GHz?) or its harmonics, since it's on all the time. These would be at 2f=4364 MHz, 3f=6546 MHz, and so on. They may be pretty far down, since the fundamental level starts out at -20 dBm or so. So, if you put 2LO into the SA input, you should be able to confirm these in band 2 or higher - maybe down in the noise, so you have to zoom in on it. If you can find these, that means that the upper band and its 719 MHz LO are OK.
One caveat though, is that at narrow spans, the 2LO is swept, while 1LO is phase-locked at a larger frequency increment. If you zoom in too far, the spur may disappear due to the sweeping action. I don't recall the details, but I think you should be able to see it down to a 5 or 2 MHz span. The manual should say what happens where. Likewise, in the upper bands, you may be able to see the 3f of the 719 MHz at 2157 MHz, etc, subject to the same limitations on span.
These tests should confirm basic operation to the correct frequency plan, and then you can experiment and figure out where all the other stuff comes from. If you're going to pursue microwave, the next logical move is to acquire a microwave counter - then you can easily look at these signals for a second opinion.
An aside: If you get a counter, a natural thing to do is look at 2LO with it. In the low band, it should show 2182 MHz, and all's well. If you go up to the next band, the 2LO includes the 719 MHz. What will the counter read? Since both frequencies are present, it may get all confused, or read one or the other. That's what happened when I viewed this on the sampling scope, until I finally figured out what was going on. The 2182 can be confirmed easily in the low band, when it's all by itself, but to clearly see the 719 in the other bands, you may need to do a little filtering. You can borrow the LPF from the 492 (if it has one), or just use a sufficiently crappy coax cable to connect the counter - it will attenuate the higher frequency so the counter can properly see the lower. Normally, bad cables are viewed negatively here, but sometimes they're your friend, and can act as a filter or lossy element to tweak your measurement conditions.
Ed
|
Thanks for a detailed reply, much appreciated. The TL;DR is that I'm still learning how SAs function, and it may be that my 492A is functioning correctly. Understanding the frequency plan is indeed top of my agenda, one step at a time. The other item is understanding single-balanced mixers, their operation, their failure modes, and the effects of bias. I made a significant error in my first message, where I referred to 829MHz and 2072MHz LO frequencies. Actually they are the IF frequencies, which I believe is significant since it means that if the RF is at the IF frequency, then my "breakthrough" symptom (display shows a flat "white noise" line) appears. That sounds like RF->IF breakthrough, which ISTR is a characteristic of single-balanced mixers. My apologies for not explaining /my/ 492A's input stage. My 492A does not have the limiter and preselector (option 1), so its lowest band is 0-4.2GHz. This also means that the input is RF -> attenuator -> cable -> first mixer -> directional filter -> second mixers -> IF. If installed, the limiter/preselector would replace the cable. Comment inline... On 07/04/16 19:57, edbreya@... [TekScopes] wrote:
I'm not sure exactly what you're looking for, but I can provide some info on the 492 frequency plan and operation, which I believe is the same as in my 494. I studied it extensively when first acquired, but that was quite a few years ago.
I believe that in band 1, 0-1800 MHz, there is a low-pass filter that cuts at around 1.8 or 2 GHz, so very little higher frequency content should get through. Since this unit apparently does not have a preselector, the higher bands can have all kinds of spurs in the display that may cause some confusion, but can be figured out. First, don't worry too much about the spurs - they must come from various combinations of all the frequencies and their harmonics involved in the system, so everything has to add up.
One particular issue in your "breakthrough" listing is that on the higher bands, the IF is 829 MHz, so if that's present in the input signal, it can get through the mixer and right into the IF with little attenuation. That's one of the benefits of having a preselector - it eliminates IF and other out of band feedthrough. The corresponding issue on the low band with 2072 MHz input should not happen, unless there's no LPF - you may want to confirm whether it has this.
That is, I now believe, a key observation. Concentrate on band 1 first, and confirm proper operation with the built-in 100 MHz reference. Getting the right frequencies (n x 100 MHz) displayed, with nearly correct fundamental level, will show that it mostly is OK. If you need to investigate the LOs, some things can be readily figured out.
The 100MHz response matches what I would expect; certainly the fundamental level is right (-22dBm) and harmonics are present up to 4GHz (-104dBm) on the 0-4.2GHz (2072MHz IF) and 1.7-5.5GHz (829MHz IF) bands. Mind you, on a 350MHz scope it barely resembles a square wave :) Early on I had no microwave equipment, and tried to figure out as much as possible using only the SA itself, and a 561 sampling system. One thing I discovered is that the the second LO 2182 MHz is always on, while the 719 MHz one is only activated in the appropriate bands.
In band 1, the SA cannot see its own 2LO, but in the upper bands it should be possible to see it (as I recall, the bottom of band 2 is 1.7 GHz?) or its harmonics, since it's on all the time. These would be at 2f=4364 MHz, 3f=6546 MHz, and so on. They may be pretty far down, since the fundamental level starts out at -20 dBm or so. So, if you put 2LO into the SA input, you should be able to confirm these in band 2 or higher - maybe down in the noise, so you have to zoom in on it. If you can find these, that means that the upper band and its 719 MHz LO are OK.
One caveat though, is that at narrow spans, the 2LO is swept, while 1LO is phase-locked at a larger frequency increment. If you zoom in too far, the spur may disappear due to the sweeping action. I don't recall the details, but I think you should be able to see it down to a 5 or 2 MHz span. The manual should say what happens where. Likewise, in the upper bands, you may be able to see the 3f of the 719 MHz at 2157 MHz, etc, subject to the same limitations on span.
I'll check all those things later, partly to confirm my understanding of the frequency plan. These tests should confirm basic operation to the correct frequency plan, and then you can experiment and figure out where all the other stuff comes from. If you're going to pursue microwave, the next logical move is to acquire a microwave counter - then you can easily look at these signals for a second opinion.
I was planning on using the SA for that! My attitude, which may well need re-adjustment, is that with synthesised sources it is easy to ensure the correct frequency, give or take phase noise. What am I missing? An aside: If you get a counter, a natural thing to do is look at 2LO with it. In the low band, it should show 2182 MHz, and all's well. If you go up to the next band, the 2LO includes the 719 MHz. What will the counter read? Since both frequencies are present, it may get all confused, or read one or the other. That's what happened when I viewed this on the sampling scope, until I finally figured out what was going on. The 2182 can be confirmed easily in the low band, when it's all by itself, but to clearly see the 719 in the other bands, you may need to do a little filtering. You can borrow the LPF from the 492 (if it has one), or just use a sufficiently crappy coax cable to connect the counter - it will attenuate the higher frequency so the counter can properly see the lower. Normally, bad cables are viewed negatively here, but sometimes they're your friend, and can act as a filter or lossy element to tweak your measurement conditions.
I like the devious imagination :)
|
That is quite a bit different from the "old" 492, I think. Does this model have digital storage? If so, it could be that the 1IF/2LO are supposed to automatically switch between 2072/2182 MHz and 829/719 MHz, depending on the center frequency, even during a wide sweep. If not, I don't think there's a straightforward way (except maybe with precision notch filters or some DSP applied) to avoid the IF feedthrough frequencies noted earlier - the manual should explain.
In the older models, the 1IF and 2LO are fixed, depending on the band selected, but they don't necessarily have to be, if the brain has enough control over the process. With digital storage, it's possible for the SA to do all sorts of switching of that stuff behind the scene, and reconstruct and present the correct info on the display. Is there a manual available? I'm curious now about the architecture.
BTW the YIG-tuned preselectors typically only work at 1-2 GHz or so and up, so are used for the upper bands. The lowest bands typically use a LPF to limit the highest input frequency applied.
Ed
|
On 08/04/16 17:50, edbreya@... [TekScopes] wrote:
That is quite a bit different from the "old" 492, I think. Does this model
have digital storage?
It does have digital storage, but I'm not sure why that might /enable/ automatic switching mentioned below. If so, it could be that the 1IF/2LO are supposed to automatically switch
between 2072/2182 MHz and 829/719 MHz, depending on the center frequency, even
during a wide sweep. If not, I don't think there's a straightforward way
(except maybe with precision notch filters or some DSP applied) to avoid the
IF feedthrough frequencies noted earlier - the manual should explain.
The notch filters have occurred to me too. So far everything I've read in the manual assumes option 01 is installed. I believe option 01 means 1.8GHz LPF plus limiter on band 1, and the YIG preselecter on all other bands - which corresponds with your statements. In the older models, the 1IF and 2LO are fixed, depending on the band
selected, but they don't necessarily have to be, if the brain has enough
control over the process. With digital storage, it's possible for the SA to do
all sorts of switching of that stuff behind the scene, and reconstruct and
present the correct info on the display. Is there a manual available? I'm
curious now about the architecture.
I'm reading the artec manuals for the 492A and 492B. I haven't yet spotted any non-correspondence with the hardware. BTW the YIG-tuned preselectors typically only work at 1-2 GHz or so and up, so
are used for the upper bands. The lowest bands typically use a LPF to limit
the highest input frequency applied.
|
By digital storage, I mean that the brain has full control over what's displayed, versus older types that have to present in real time, and can't store the information. Without storage, the repetition rate and activities of the SA have to be fast enough and transparent enough to present a usable display. With storage, the SA can do whatever it has to do, and take a fair amount of time doing it, but then only present the correct data on-screen. I assume the 492 and 492AP can do this.
So in my hypothetical SA process, let's say the 1IF and 2LO can be switched on the fly between 2072/2182 and 829/719 MHz. Let's also say there's a PIN-switched notch filter (stubs) in the input structure that can greatly reject either 2072 or 829 MHz, also switched on the fly.
Now suppose you want a full 0-4.2 GHz sweep. The first LO would start at 2072 MHz, and sweep to 2072 plus something less than 2072, say 1800 MHz to use round numbers. So, 1LO goes from 2072 to 3872 MHz, while 1IF is 2072, and 2LO is 2182, and the 2072 MHz notch is active. This would gather the correct amplitude info for 0-1800 MHz, while anything around 2072 MHz present in the signal would be rejected.
Then let's say the SA changes 1IF and the notch filter to 829 MHz, and 2LO to 719 MHz, then sweeps 1LO from 2529 to 5029 MHz. This would gather the correct amplitude info for 1700-4200 MHz, while rejecting signal content around 829 MHz.
Then let's say the SA takes the info from these two overlapping band results, and presents 0-1750 MHz from one, and 1751-4200 MHz from the other, in a continuous 0-4200 MHz display on screen, without telling you all the stuff it had to do.
Of course, without the benefits of LPFs or preselectors, there can still be various extra spurs, but at least the IF problem can be greatly diminished.
This may all appear confusing at first, but if you sketch it out it should make some sense. I made up these numbers on the fly while writing - hope I didn't make any logic or math mistakes that could confuse it more.
Ed
|
What does your 492A state when its starting up? it should say "492A FW n.nn" or somthing similar. The firmware number, and what version of the 490 series your analyzer thinks it is, is important. Also what options are your instrument? The thing I find particularly interesting is if its covering 0-4.2GHz in a single band. The regular 492, 492A and 492B should cover 0-1.8GHz in band 1. Could your analyzer be option 8? this limits the max frequency of the analyzer to 8GHz. The 2072MHz IF should be used at frequencys below 1.8GHz where the analyzer are upconverting, and the 829MHz IF for bands above 1.8GHz. The firmware for the later instruments (279x series I beleve) were able to utilize this to do multiband sweeps, but as far as I know not the 492 series. There are actual, slow coax switces that needs to switch to do the change between the IF's. Digital storage is option 2, and would be required to do multiband sweeps. With a lack of option 1 (preselector) you just use the identify (ident) function to determine if its a real or false product you are seeing on the screen for bands above 1.8GHz. The DIP switches on the modules are extremly fragile, thus your analyzer may think it has a different option setup than it has, and this leads to some quite interesting problems that can be hard to track down. Also notice that the switch setup changes with different firmware versions. There is a map for the 9.7 version at ko4bb.com. BR. Thomas. 2016-04-08 17:24 GMT+02:00 Tom Gardner tggzzz@... [TekScopes] < TekScopes@...>:
Thanks for a detailed reply, much appreciated. The TL;DR is that I'm still
learning how SAs function, and it may be that my 492A is functioning
correctly.
Understanding the frequency plan is indeed top of my agenda, one step at a
time.
The other item is understanding single-balanced mixers, their operation,
their
failure modes, and the effects of bias.
I made a significant error in my first message, where I referred to 829MHz
and
2072MHz LO frequencies. Actually they are the IF frequencies, which I
believe is
significant since it means that if the RF is at the IF frequency, then my
"breakthrough" symptom (display shows a flat "white noise" line) appears.
That
sounds like RF->IF breakthrough, which ISTR is a characteristic of
single-balanced mixers.
My apologies for not explaining /my/ 492A's input stage. My 492A does not
have
the limiter and preselector (option 1), so its lowest band is 0-4.2GHz.
This
also means that the input is RF -> attenuator -> cable -> first mixer ->
directional filter -> second mixers -> IF. If installed, the
limiter/preselector would replace the cable.
Comment inline...
On 07/04/16 19:57, edbreya@... [TekScopes] wrote:
I'm not sure exactly what you're looking for, but I can provide some
info on
the 492 frequency plan and operation, which I believe is the same as in my
494. I studied it extensively when first acquired, but that was quite a few
years ago.
I believe that in band 1, 0-1800 MHz, there is a low-pass filter that cuts at
around 1.8 or 2 GHz, so very little higher frequency content should get
through. Since this unit apparently does not have a preselector, the higher
bands can have all kinds of spurs in the display that may cause some
confusion, but can be figured out. First, don't worry too much about the spurs
- they must come from various combinations of all the frequencies and their
harmonics involved in the system, so everything has to add up.
One particular issue in your "breakthrough" listing is that on the higher
bands, the IF is 829 MHz, so if that's present in the input signal, it can get
through the mixer and right into the IF with little attenuation. That's one of
the benefits of having a preselector - it eliminates IF and other out of band
feedthrough. The corresponding issue on the low band with 2072 MHz input
should not happen, unless there's no LPF - you may want to confirm whether it
has this.
That is, I now believe, a key observation.
Concentrate on band 1 first, and confirm proper operation with the built-in
100 MHz reference. Getting the right frequencies (n x 100 MHz) displayed, with
nearly correct fundamental level, will show that it mostly is OK. If you need
to investigate the LOs, some things can be readily figured out.
The 100MHz response matches what I would expect; certainly the fundamental
level
is right (-22dBm) and harmonics are present up to 4GHz (-104dBm) on the
0-4.2GHz
(2072MHz IF) and 1.7-5.5GHz (829MHz IF) bands. Mind you, on a 350MHz scope
it
barely resembles a square wave :)
Early on I had no microwave equipment, and tried to figure out as much as
possible using only the SA itself, and a 561 sampling system. One thing I
discovered is that the the second LO 2182 MHz is always on, while the 719 MHz
one is only activated in the appropriate bands.
In band 1, the SA cannot see its own 2LO, but in the upper bands it should be
possible to see it (as I recall, the bottom of band 2 is 1.7 GHz?) or its
harmonics, since it's on all the time. These would be at 2f=4364 MHz, 3f=6546
MHz, and so on. They may be pretty far down, since the fundamental level
starts out at -20 dBm or so. So, if you put 2LO into the SA input, you should
be able to confirm these in band 2 or higher - maybe down in the noise, so you
have to zoom in on it. If you can find these, that means that the upper band
and its 719 MHz LO are OK.
One caveat though, is that at narrow spans, the 2LO is swept, while 1LO is
phase-locked at a larger frequency increment. If you zoom in too far, the spur
may disappear due to the sweeping action. I don't recall the details, but I
think you should be able to see it down to a 5 or 2 MHz span. The manual
should say what happens where. Likewise, in the upper bands, you may be able
to see the 3f of the 719 MHz at 2157 MHz, etc, subject to the same limitations
on span.
I'll check all those things later, partly to confirm my understanding of
the
frequency plan.
These tests should confirm basic operation to the correct frequency plan, and
then you can experiment and figure out where all the other stuff comes from.
If you're going to pursue microwave, the next logical move is to acquire a
microwave counter - then you can easily look at these signals for a second
opinion.
I was planning on using the SA for that! My attitude, which may well need
re-adjustment, is that with synthesised sources it is easy to ensure the
correct
frequency, give or take phase noise. What am I missing?
An aside: If you get a counter, a natural thing to do is look at 2LO with it.
In the low band, it should show 2182 MHz, and all's well. If you go up to the
next band, the 2LO includes the 719 MHz. What will the counter read? Since
both frequencies are present, it may get all confused, or read one or the
other. That's what happened when I viewed this on the sampling scope, until I
finally figured out what was going on. The 2182 can be confirmed easily in the
low band, when it's all by itself, but to clearly see the 719 in the other
bands, you may need to do a little filtering. You can borrow the LPF from the
492 (if it has one), or just use a sufficiently crappy coax cable to connect
the counter - it will attenuate the higher frequency so the counter can
properly see the lower. Normally, bad cables are viewed negatively here, but
sometimes they're your friend, and can act as a filter or lossy element to
tweak your measurement conditions.
I like the devious imagination :)
-- Please avoid sending me Word or PowerPoint attachments. See <> PDF is an better alternative and there are always LaTeX! [Non-text portions of this message have been removed]
|
On 08/04/16 21:08, 'Thomas S. Knutsen' la3pna@... [TekScopes] wrote:
What does your 492A state when its starting up? it should say "492A FW
n.nn" or somthing similar.
The firmware number, and what version of the 490 series your analyzer
thinks it is, is important.
Firmware 6.0, front panel 1.1 What's puzzling me is that in an earlier message I stated 4.6, 1.1. Either my brain is turning to mush or the processor's memory is turning to mush. Also what options are your instrument?
Good question. Is there a means of using the front panel to determine the system settings? Option 01 is definitely missing - there's no hardware on the RF deck for preselector/relays/limiter. There is the digital storage in the form of save A, view A-B, and save waveforms to memory. ISTR the high resolution mode is there, but I can't see an option for that. The thing I find particularly interesting is if its covering 0-4.2GHz in a
single band. The regular 492, 492A and 492B should cover 0-1.8GHz in band
1. Could your analyzer be option 8? this limits the max frequency of the
analyzer to 8GHz.
I can select up to 21GHz. The 2072MHz IF should be used at frequencys below 1.8GHz where the analyzer
are upconverting, and the 829MHz IF for bands above 1.8GHz. The firmware
for the later instruments (279x series I beleve) were able to utilize this
to do multiband sweeps, but as far as I know not the 492 series. There are
actual, slow coax switces that needs to switch to do the change between the
IF's.
Digital storage is option 2, and would be required to do multiband sweeps.
With a lack of option 1 (preselector) you just use the identify (ident)
function to determine if its a real or false product you are seeing on the
screen for bands above 1.8GHz.
I'm going to have to find out what the identify function is. The DIP switches on the modules are extremly fragile, thus your analyzer
may think it has a different option setup than it has, and this leads to
some quite interesting problems that can be hard to track down. Also notice
that the switch setup changes with different firmware versions. There is a
map for the 9.7 version at ko4bb.com.
Oh joy.
BR.
Thomas.
2016-04-08 17:24 GMT+02:00 Tom Gardner tggzzz@... [TekScopes] <
TekScopes@...>:
Thanks for a detailed reply, much appreciated. The TL;DR is that I'm still
learning how SAs function, and it may be that my 492A is functioning
correctly.
Understanding the frequency plan is indeed top of my agenda, one step at a
time.
The other item is understanding single-balanced mixers, their operation,
their
failure modes, and the effects of bias.
I made a significant error in my first message, where I referred to 829MHz
and
2072MHz LO frequencies. Actually they are the IF frequencies, which I
believe is
significant since it means that if the RF is at the IF frequency, then my
"breakthrough" symptom (display shows a flat "white noise" line) appears.
That
sounds like RF->IF breakthrough, which ISTR is a characteristic of
single-balanced mixers.
My apologies for not explaining /my/ 492A's input stage. My 492A does not
have
the limiter and preselector (option 1), so its lowest band is 0-4.2GHz.
This
also means that the input is RF -> attenuator -> cable -> first mixer ->
directional filter -> second mixers -> IF. If installed, the
limiter/preselector would replace the cable.
Comment inline...
On 07/04/16 19:57, edbreya@... [TekScopes] wrote:
I'm not sure exactly what you're looking for, but I can provide some
info on
the 492 frequency plan and operation, which I believe is the same as in my
494. I studied it extensively when first acquired, but that was quite a few
years ago.
I believe that in band 1, 0-1800 MHz, there is a low-pass filter that cuts at
around 1.8 or 2 GHz, so very little higher frequency content should get
through. Since this unit apparently does not have a preselector, the higher
bands can have all kinds of spurs in the display that may cause some
confusion, but can be figured out. First, don't worry too much about the spurs
- they must come from various combinations of all the frequencies and their
harmonics involved in the system, so everything has to add up.
One particular issue in your "breakthrough" listing is that on the higher
bands, the IF is 829 MHz, so if that's present in the input signal, it can get
through the mixer and right into the IF with little attenuation. That's one of
the benefits of having a preselector - it eliminates IF and other out of band
feedthrough. The corresponding issue on the low band with 2072 MHz input
should not happen, unless there's no LPF - you may want to confirm whether it
has this.
That is, I now believe, a key observation.
Concentrate on band 1 first, and confirm proper operation with the built-in
100 MHz reference. Getting the right frequencies (n x 100 MHz) displayed, with
nearly correct fundamental level, will show that it mostly is OK. If you need
to investigate the LOs, some things can be readily figured out.
The 100MHz response matches what I would expect; certainly the fundamental
level
is right (-22dBm) and harmonics are present up to 4GHz (-104dBm) on the
0-4.2GHz
(2072MHz IF) and 1.7-5.5GHz (829MHz IF) bands. Mind you, on a 350MHz scope
it
barely resembles a square wave :)
Early on I had no microwave equipment, and tried to figure out as much as
possible using only the SA itself, and a 561 sampling system. One thing I
discovered is that the the second LO 2182 MHz is always on, while the 719 MHz
one is only activated in the appropriate bands.
In band 1, the SA cannot see its own 2LO, but in the upper bands it should be
possible to see it (as I recall, the bottom of band 2 is 1.7 GHz?) or its
harmonics, since it's on all the time. These would be at 2f=4364 MHz, 3f=6546
MHz, and so on. They may be pretty far down, since the fundamental level
starts out at -20 dBm or so. So, if you put 2LO into the SA input, you should
be able to confirm these in band 2 or higher - maybe down in the noise, so you
have to zoom in on it. If you can find these, that means that the upper band
and its 719 MHz LO are OK.
One caveat though, is that at narrow spans, the 2LO is swept, while 1LO is
phase-locked at a larger frequency increment. If you zoom in too far, the spur
may disappear due to the sweeping action. I don't recall the details, but I
think you should be able to see it down to a 5 or 2 MHz span. The manual
should say what happens where. Likewise, in the upper bands, you may be able
to see the 3f of the 719 MHz at 2157 MHz, etc, subject to the same limitations
on span.
I'll check all those things later, partly to confirm my understanding of
the
frequency plan.
These tests should confirm basic operation to the correct frequency plan, and
then you can experiment and figure out where all the other stuff comes from.
If you're going to pursue microwave, the next logical move is to acquire a
microwave counter - then you can easily look at these signals for a second
opinion.
I was planning on using the SA for that! My attitude, which may well need
re-adjustment, is that with synthesised sources it is easy to ensure the
correct
frequency, give or take phase noise. What am I missing?
An aside: If you get a counter, a natural thing to do is look at 2LO with it.
In the low band, it should show 2182 MHz, and all's well. If you go up to the
next band, the 2LO includes the 719 MHz. What will the counter read? Since
both frequencies are present, it may get all confused, or read one or the
other. That's what happened when I viewed this on the sampling scope, until I
finally figured out what was going on. The 2182 can be confirmed easily in the
low band, when it's all by itself, but to clearly see the 719 in the other
bands, you may need to do a little filtering. You can borrow the LPF from the
492 (if it has one), or just use a sufficiently crappy coax cable to connect
the counter - it will attenuate the higher frequency so the counter can
properly see the lower. Normally, bad cables are viewed negatively here, but
sometimes they're your friend, and can act as a filter or lossy element to
tweak your measurement conditions.
I like the devious imagination :)
|
On 08/04/16 21:08, 'Thomas S. Knutsen' la3pna@... [TekScopes] wrote:
With a lack of option 1 (preselector) you just use the identify (ident)
function to determine if its a real or false product you are seeing on the
screen for bands above 1.8GHz.
I now understand what that is and how it works. Unfortunately it can only be invoked at <=50kHz/div, i.e. 500kHz across the display. That makes some sense for narrowband signals, but I was hoping to use the SA for wideband and/or ultra-wideband signals.
|
I now understand what's happening much better, although still imperfectly. Thanks to all those that have contributed their thoughts, and now I'd be grateful for any insight into the question the last paragraph.
The key issue is that all textbooks describing SAs show a block diagram including a preselector or low pass filter depending on whether a high IF or low IF is used. My 492 has neither, since those essential components are an "optional extra" (option 01).
Given what I paid for the SA, I'm not complaining. But quite frankly I expected more from "old school Tek", and I can only assume that the sales department had to advertise something (anything) under an arbitrary price limit ("from only $4999"). All other options are, reasonably, to add non-essentials or delete essentials.
It seems I have three ways forward:
1) use it "as is", which is less useful than I was expecting, albeit still useful
2) add the essential components, probably $500, problems with calibration and/or getting everything working with the firmware
3) ensure the first mixer isn't subtly damaged, and replace if necessary
Given that the first mixer is mixing, is it working as well as can be expected or has it been subtly borked?
What level of spurious responses would be expected in a working/damaged first mixer?
|
2) add the essential components, probably $500, problems with calibration
and/or
getting everything working with the firmware
I think you can find the things you need to add option 1 to the analyzer on ebay. Adding option 1 is just getting the YIG filter and driver, then is just flipping the correct DIP switches to get it to recognize it.
3) ensure the first mixer isn't subtly damaged, and replace if necessary
Given that the first mixer is mixing, is it working as well as can be
expected
or has it been subtly borked?
What level of spurious responses would be expected in a working/damaged
first mixer?
Page 15: That should cover most anything about the 1. mixer. Br. Thomas.
__
|
On 11/04/16 16:59, 'Thomas S. Knutsen' la3pna@... [TekScopes] wrote:
2) add the essential components, probably $500, problems with calibration
and/or
getting everything working with the firmware
I think you can find the things you need to add option 1 to the analyzer on
ebay.
Yes; I would have to check if there is any significant difference between driver board 670-5552-03 and -05. Adding option 1 is just getting the YIG filter and driver, then is just
flipping the correct DIP switches to get it to recognize it.
I haven't found any indication of the relevant switches in any of the documents I've seen. From that I have inferred, correctly or not, that the driver's presence is auto detected. Without that I would be reduced to randomly flipping switches, and there are too many combinations for me to feel comfortable with that. My firmware is v6.0/1.1 (not 4.6, doh). 3) ensure the first mixer isn't subtly damaged, and replace if necessary
Given that the first mixer is mixing, is it working as well as can be
expected
or has it been subtly borked?
What level of spurious responses would be expected in a working/damaged
first mixer?
Page 15:
That should cover most anything about the 1. mixer.
Unfortunately it doesn't indicate the acceptable/unacceptable level of spurs, and I don't have sufficient experience to understand them. If mixers are either OK or dead, then I don't have a problem.
|
I think you will need a few more pieces if you want to go back to a "normal" 492-ish SA. You should study the manual and circuits for that model and the 494 for comparison - the RF block diagram should tell most of the story. If this one is like the 494, there should be a couple of coaxial relays to route the input to the mixer through the LPF, or the preselector, and maybe an extra 3 dB attenuator, as I recall. And the associated coaxial cables, driver circuitry and wiring, etc. The low band of course is then only up to 1.8 GHz, but the IF feedthrough problem will be gone.
If you mostly need to use the low end of the low band, you can put an external LPF up front. I think the original one for that architecture cuts at 1.8 GHz, and is pretty far down (80-100 dB?) by 2072 MHz, for IF rejection. Even a 2 GHz one may be pretty good for this. The 1.8 and 2 GHz LPFs are probably quite common on the used market - likely since they were used in so many SAs for exactly this purpose.
If you really only need that low end, it would be best to put the LPF inside behind the step attenuator, where it normally would go, for better isolation. Then if you ever do need the higher bands, you can rehook it or add the optional parts.
|
On 11/04/16 18:50, edbreya@... [TekScopes] wrote:
I think you will need a few more pieces if you want to go back to a "normal" 492-ish SA. You should study the manual and circuits for that model and the 494 for comparison - the RF block diagram should tell most of the story. If this one is like the 494, there should be a couple of coaxial relays to route the input to the mixer through the LPF, or the preselector, and maybe an extra 3 dB attenuator, as I recall. And the associated coaxial cables, driver circuitry and wiring, etc. The low band of course is then only up to 1.8 GHz, but the IF feedthrough problem will be gone.
If you mostly need to use the low end of the low band, you can put an external LPF up front. I think the original one for that architecture cuts at 1.8 GHz, and is pretty far down (80-100 dB?) by 2072 MHz, for IF rejection. Even a 2 GHz one may be pretty good for this. The 1.8 and 2 GHz LPFs are probably quite common on the used market - likely since they were used in so many SAs for exactly this purpose.
If you really only need that low end, it would be best to put the LPF inside behind the step attenuator, where it normally would go, for better isolation. Then if you ever do need the higher bands, you can rehook it or add the optional parts.
Understood and accepted. I'd included most of those components in my $500 guesstimate, different hardline coax excepted. I had a quick look at the Minicircuits LPFs but they aren't as good as you indicate the Tek part is. I suspect the Tek is a rod-and-bead filter, unlike the Minicircuits filter.
|
Yes, you want a true multi-section coaxial LPF there for best performance. There should be some available at the usual places. The common ones are about 4-6" long by 1/2" diameter, with SMA connectors. The most common brands I recall from that era are K&L and RLC. Whatever you get will have to fit the limited space on the RF deck, so choose wisely, and consider the connectors etc. You may be able to find an actual one from a Tek SA, since they're fairly common too.
Another thing to consider is to maybe look for a junker one of the 49X series that happens to have the stuff you need - it may be cheaper than buying all of it separately.
Ed
|
Just noticed there's a whole 492 "not working" on ebay for $799 BIN. It does light up but the display is all messed up. From the pictures it appears to say options 1 2 3 on a tag. This may be a good possibility to make one SA out of two, and have plenty of spare parts and extra stuff to fool around with. May be worth a look, and contact seller for more info.
Ed
|
On 11/04/16 20:06, edbreya@... [TekScopes] wrote:
Just noticed there's a whole 492 "not working" on ebay for $799 BIN. It does light up but the display is all messed up. From the pictures it appears to say options 1 2 3 on a tag. This may be a good possibility to make one SA out of two, and have plenty of spare parts and extra stuff to fool around with. May be worth a look, and contact seller for more info.
I can't see that on ebay.com or ebay.co.uk, but I can see a listing (121934026947) that ended on 4th April due to a "spelling error". Do you still have the ebay number or URL?
|
I think this URL should work. If not, the auction number appears to be 141929945400
Good luck
Tekt
Tektronix 492 Spectrum Analyzer Power On Tested ... US $799.00 For parts or not working in Business & Industrial, Electrical & Test Equipment, Test Equipment
View on www.ebay.com
Preview by Yahoo
ronix 492 Spectrum Analyzer Power On Tested Free Standard Shipping!
|
The URL works even from my old POS computer. I don't know why yahoo makes all those multiple copies - I only pasted it into the message once. BTW that item also says free shipping, which saves some dollars and grief.
Ed
|
