Author Topic: Video Teardown, Analysis and Repair of an Agilent E4407B ESA-E Spectrum Analyzer  (Read 23763 times)

0 Members and 1 Guest are viewing this topic.

Offline HugoneusTopic starter

  • Frequent Contributor
  • **
  • Posts: 962
  • Country: us
    • The Signal Path Video Blog
In this episode Shahriar takes a detailed look at an Agilent (Keysight) E4407B ESA-E Spectrum Analyzer. The instruments reports an “Unlock LO” error message which prevents internal alignments from being performed. Before the repair process can begin, the firmware of the instrument must be updated. The firmware upgrade requires the installation of additional flash memory ICs which is presented in the video. The main processor board with its various components are also described in the video.

The complete block diagram of the spectrum analyzer is presented with a detailed look at the principle of operation of the instrument. All the main functional blocks such as the YIG Oscillator, Phase Detector, LO Path, LO Amplifier, Charge Pump, Pre-tune segments, digital circuits, IF and LO paths are analyzed both in schematic and on the PCB. Several potential faults are investigated and measured.

Watch here: [2 Hours]
http://youtu.be/gb1QMJtwumQ

More videos at The Signal Path:
http://www.TheSignalPath.com

Offline kg4arn

  • Supporter
  • ****
  • Posts: 271
  • Country: us
Really enjoyed the extra detail and explanation.  Thanks
 

Offline HugoneusTopic starter

  • Frequent Contributor
  • **
  • Posts: 962
  • Country: us
    • The Signal Path Video Blog
Really enjoyed the extra detail and explanation.  Thanks

Great. Glad to hear it.

Offline Wytnucls

  • Super Contributor
  • ***
  • Posts: 3045
  • Country: be
Shahriar, the Steven Spielberg of electronics videos...  ;)

Fantastic, keep them coming.
 

Offline poida_pie

  • Regular Contributor
  • *
  • Posts: 119
  • Country: au
Yet another excellent video. I liked the sigma-delta A-D one as well.
I particularly enjoy the repair type videos you do.
 

Offline HugoneusTopic starter

  • Frequent Contributor
  • **
  • Posts: 962
  • Country: us
    • The Signal Path Video Blog
Shahriar, the Steven Spielberg of electronics videos...  ;)
Fantastic, keep them coming.

That is awesome! Thank you.

Offline vaualbus

  • Frequent Contributor
  • **
  • Posts: 381
  • Country: it
I really love the video. The schematics seem very hard to understand at first look.i just have a question. This scope go from 9khz up to 2ghz. I see the new agilent/keysight sa (psa and pxa).
For the psa the top model (e4448) go from 3hz up to 50ghz. How it can achive this resolution? What could be different from this sa?
Also I saw on eb a couple options for your sa . For example there is the high stability crystal oscillator for less tha 100 bucks and the fast adc board a cople of houndred of dollars.
By the wave keep on doing this amazing teardown!

Also any idea how the divider can be get wrong?
« Last Edit: October 03, 2014, 05:28:32 am by vaualbus »
 

Offline miguelvp

  • Super Contributor
  • ***
  • Posts: 5550
  • Country: us
I did doze off at the beginning because I just ate dinner but then I rewind it and watch the whole thing.

I think your video was plagued with the absence of Daves #666 that's why your divider was dividing by 6 instead of by 4 :)

Awesome repair even if I don't understand why the divider was affecting the voltage reference to jump all over.
 

Offline Nermash

  • Frequent Contributor
  • **
  • Posts: 257
Watched it all the way through, great content, great presentation :)

Please keep doing this kind of elaborate videos.

One thing does puzzles me: why the NI input on the U93 opamp was jumping? With stable positive and negative opamp supply, stable reference voltage, I can only guess that inverting input was being slammed beyond the CMIR, but I sure could use 2 minutes explanation why and how the fracN divider wrong division ratio influenced charge pump and voltage integrator:)
 

Offline HugoneusTopic starter

  • Frequent Contributor
  • **
  • Posts: 962
  • Country: us
    • The Signal Path Video Blog
Thank you all.

To answer a few questions,

The Op-Amp which I replaced is a fully bipolar implementation with input protection diodes between the terminals. Saturating the input/output of such an Op-Amp can cause current to flow between the terminals which is what we were seeing.

With the main Divider/4 block non-functional, the phase detector will continuously attempt to push the YIG frequency to extreme ends of its range which causes the Op-Amp (the main integrator) to saturate.

Offline German_EE

  • Super Contributor
  • ***
  • Posts: 2399
  • Country: de
Two hours of RF goodness whilst the girlfriend is at the gym. Life is good  :)
Should you find yourself in a chronically leaking boat, energy devoted to changing vessels is likely to be more productive than energy devoted to patching leaks.

Warren Buffett
 

Offline HugoneusTopic starter

  • Frequent Contributor
  • **
  • Posts: 962
  • Country: us
    • The Signal Path Video Blog
Two hours of RF goodness whilst the girlfriend is at the gym. Life is good  :)

Haha, lucky you. My girlfriend was sitting across the room making faces at me while I recorded.

Offline HighVoltage

  • Super Contributor
  • ***
  • Posts: 5560
  • Country: de
I just watched your video completely through and would like to say thank you for these repair videos. They are just perfect and definitely not too long. I have never seen an OpAmp behave like this, so that was something new. And so many other great points you made. Just make more repair videos on Agilent test equipment, I really deeply enjoy them.
There are 3 kinds of people in this world, those who can count and those who can not.
 

Offline Lukas

  • Frequent Contributor
  • **
  • Posts: 412
  • Country: de
    • carrotIndustries.net
I watched the video from the beginning to the end, was really informative. Keep it up!
You asked why the LO sweeps at varying speeds, my take on it: (Select text to see it) On the first ramp, it uses the fundamental of the LO, on the second the first harmonic and so on. Since the sweep speed is constant the LO sweep speed has to be halved when using the first harmonic to maintain the sweep speed.
 

Offline saraht

  • Newbie
  • Posts: 8
I watched it last night. I'm not a short-attention-span person, I love longer format videos like yours. I'd hate to miss the detail.

Great work, please keep it up! :-)
 

Offline HugoneusTopic starter

  • Frequent Contributor
  • **
  • Posts: 962
  • Country: us
    • The Signal Path Video Blog
I watched the video from the beginning to the end, was really informative. Keep it up!
You asked why the LO sweeps at varying speeds, my take on it: (Select text to see it) On the first ramp, it uses the fundamental of the LO, on the second the first harmonic and so on. Since the sweep speed is constant the LO sweep speed has to be halved when using the first harmonic to maintain the sweep speed.

Yes, you are right! Thanks for posting the answer!  :-+

Offline G0HZU

  • Super Contributor
  • ***
  • Posts: 3238
  • Country: gb
It's interesting to see inside one of these analysers. We have several of these at work for doing simple system work etc.

I only skimmed the video to look at the schematics and the PCB layout bits but a few things in the video confused me a lot. eg there were several references to an IF at 312.4MHz

I can't see how the IF can be 312.4MHz because most oldschool HP uW analysers have an IF at 321.4MHz. This allows easy generation of the 300MHz LO (from 10MHz) to mix 321.4MHz down to the common 21.4MHz IF.

The 5.5GHz filter looks like a simple LPF using radial stubs and I can't initially see how this can provide a BRF notch at 5.5GHz whilst passing about 3.92GHz with low loss. This is despite the block diagram showing a 5.5GHz BRF symbol. Very confusing. The filter is too simple to allow this (unless some tradeoff against highish insertion loss at 3.92GHz is tolerated?). At a guess those radial stubs will give notches up around 6.5GHz or higher. Normally an analyser will have a steep LPF here in order to provide some isolation between the first two mixers across as much of the LO1 range as possible. My guess is that they have cascaded two LPFs here in order to try and get some steep rolloff to get LO isolation to minimise spurious terms in the two mixers. eg ABS(nLO1 - mLO2) = 321.4MHz. There's lots of smaller radial stubs littered along the IF path and I think these will be there for similar reasons. i.e. to notch out harmonics of the LOs to provide mixer to mixer isolation at many, many GHz. But that is just a guess.

At 1:16:58 the prescaler output is to the left rather than on the right as the bit pointed to by the finger looks like a control line used to select different division ratios in the device?
« Last Edit: October 04, 2014, 01:03:24 am by G0HZU »
 

Offline apelly

  • Supporter
  • ****
  • Posts: 1067
  • Country: nz
  • Probe
Thanks for the video. I could watch quality content like that for hours.

If anything, I thought the end felt a bit rushed. I'm sure you were getting a bit sick of it by then though! I certainly wasn't.

You make great videos, and I appreciate the effort you put in. Thanks again.
 

Offline Noise Floor

  • Regular Contributor
  • *
  • Posts: 117
  • Country: us
Woot woot.  Thanks for doing that.
 

Offline HugoneusTopic starter

  • Frequent Contributor
  • **
  • Posts: 962
  • Country: us
    • The Signal Path Video Blog
Thanks everyone for your contribution.

Don't forget about the Twitter account (@TheSignalPath)!

Offline electronic_eel

  • Regular Contributor
  • *
  • Posts: 201
Thanks for your videos, I really like the longer videos which go down into the details.

When you show the repaired SA with the PLL module in the last few minutes of the video with the full sweep (before zooming in on the freq of the pll), there is a large signal just at the left side of the display. What is that? Shouldn't the signal level be at the noise floor on the left side?
 

Offline HugoneusTopic starter

  • Frequent Contributor
  • **
  • Posts: 962
  • Country: us
    • The Signal Path Video Blog
Thanks for your videos, I really like the longer videos which go down into the details.

When you show the repaired SA with the PLL module in the last few minutes of the video with the full sweep (before zooming in on the freq of the pll), there is a large signal just at the left side of the display. What is that? Shouldn't the signal level be at the noise floor on the left side?

The PLL is unlocked and the instrument is trying to interpret various IF signals it receives from its several converters. I suspect that some DC offset signal from a mixer is wrongfully interpreted as a tone. Once the PLL is locked, this issue is resolved automatically.

Offline HighVoltage

  • Super Contributor
  • ***
  • Posts: 5560
  • Country: de
The PLL is unlocked and the instrument is trying to interpret various IF signals it receives from its several converters. I suspect that some DC offset signal from a mixer is wrongfully interpreted as a tone. Once the PLL is locked, this issue is resolved automatically.

Before I ask my question, I need to say, that I have not done much with RF, but find it fascination.
So, may be the answers are obvious to an RF specialist...

1)
I have an older HP Spectrum Analyzer Model 8596E.
This one also shows a peak at the very beginning of the sweep, when it starts the sweep at 9 kHz.
(Noting connected to the input)
Is it possible that this is inherent to HP spectrum analyzers?
I get a -1dBm at -100Hz
(see pictures)

2)
Why can I set my HP Spectrum Analyzer to start sweeping at a negative value?
i.e. -100 Hz

3)
And one more question, if you know the answer....
Why do most of these spectrum analyzers have a listed start frequency of 9kHz?
Why not 5kHz or 10 kHz or any other value?
Even some HP RF signal generators start at 9 kHz?
Was this an arbitrary selection by HP at the time or is there a significant technology reason behind this number?

Thanks for all your great videos.

There are 3 kinds of people in this world, those who can count and those who can not.
 

Offline HugoneusTopic starter

  • Frequent Contributor
  • **
  • Posts: 962
  • Country: us
    • The Signal Path Video Blog
The PLL is unlocked and the instrument is trying to interpret various IF signals it receives from its several converters. I suspect that some DC offset signal from a mixer is wrongfully interpreted as a tone. Once the PLL is locked, this issue is resolved automatically.

Before I ask my question, I need to say, that I have not done much with RF, but find it fascination.
So, may be the answers are obvious to an RF specialist...

1)
I have an older HP Spectrum Analyzer Model 8596E.
This one also shows a peak at the very beginning of the sweep, when it starts the sweep at 9 kHz.
(Noting connected to the input)
Is it possible that this is inherent to HP spectrum analyzers?
I get a -1dBm at -100Hz
(see pictures)

2)
Why can I set my HP Spectrum Analyzer to start sweeping at a negative value?
i.e. -100 Hz

3)
And one more question, if you know the answer....
Why do most of these spectrum analyzers have a listed start frequency of 9kHz?
Why not 5kHz or 10 kHz or any other value?
Even some HP RF signal generators start at 9 kHz?
Was this an arbitrary selection by HP at the time or is there a significant technology reason behind this number?

Thanks for all your great videos.

The signal you see below 9kHz is actually at DC. It is normal that spectrum analyzers show that signal especially the ones which are AC coupled at the input.

Spectrum analyzers (in normal sweep) will mirror the signals at 0Hz if you set the start frequency to a negative value. This features is useful when using external mixers or frequency offsets where the frequency axis can show relative negative frequencies.

The 9kHz value is the cut-off frequency of the internal DC blocking capacitor. This is a standard value. Making it less than 9kHz is very difficult while simultaneously meeting the higher bandwidth requirements. Very high performance spectrum analyzers are actually DC coupled and can measure down to 3Hz.

Offline electronic_eel

  • Regular Contributor
  • *
  • Posts: 201
Very high performance spectrum analyzers are actually DC coupled and can measure down to 3Hz.
But they usually have a big downside: you have to be very careful when measuring with such an instrument as the input stage will be destroyed if you input a dc signal.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf