ฝัง
- เผยแพร่เมื่อ 10 ก.ค. 2024
- In this episode Shahriar repairs an Agilent PSA Series Spectrum Analyzer. The instrument generates many errors during self-alignment and produces no measurements below 3.2GHz. The block diagram of the unit is thoroughly presented and various possible failure points are considered. Based on the observation of the noise floor, the most likely cause is the second LO module. The measurement of the LO power indicates that the second LO power is fall below nominal.
The second LO signal (at 3.6GHz) is generated by a DRO PLL module which is locked to 600MHz. The PLL is functional, however the output power is below -20dBm. Teardown of the module reveals a simple design with a doubler and filter. The filter (which is discolored) shows a huge loss likely caused by RF losses due to trace degradation. Thinning the traces provides some improvement and the remaining loss is compensated with a two stage RF amplifier. With this modifications, the instrument’s alignment errors are partially resolved.
The next problem is with the input attenuator at the 20dB range. This problem is simply resolved by providing lubrication on the attenuator solenoids after disassembly. The instrument is then used for various measurements to verify its correct operation.
The Signal Path
www.TheSignalPath.com
/ thesignalpath
www.Patreon.com/TheSignalPath - วิทยาศาสตร์และเทคโนโลยี
![[Full Episode] MasterChef Junior Thailand มาสเตอร์เชฟ จูเนียร์ ประเทศไทย Season 3 Episode 5](http://i.ytimg.com/vi/g5wpgeAIzMs/mqdefault.jpg)
![CAMPปลิ้น | EP.77[2/2] "รับทราบ" แก๊งเพื่อนรัก 10 ปี ที่ไปไหนก็มีตำนาน!](http://i.ytimg.com/vi/ifc0ofD51B0/mqdefault.jpg)
Signalpath is by far the most knowledgeable and productive video series about use and repair of complex rf instruments. Thank you for doing great job.
Simply the best electronics channel.
The most expensive too.
That's some high price gear...
- Eddy
Great microwave and millimeter wave channel for sure.
Replacing and adding the amplifiers in that module is a seriously impressive piece of repair work. Makes you wonder how much otherwise good equipment gets trashed that could have been repaired by a knowledgeable and dedicated engineer.
Fantastic work as always, keep the repairs coming..
Absolutely fantastic repair! Great work!
Awsomeness repair. Thanks Shahriar!
Congratulations for great repair and thank You for a wonderful video!
The most enjoyable serious electronics channel in YT, without ant kind of doubt, IMHO. And for this particular job, as ever, clean and elegant. My most sincere congratulations 👏👏👏👏👏👏👏👏👏👏👏👏👏👏👏👏👏👏👏👏👏👏👏👏👏👏
Thanks as always! Your videos are the best in the business!
As usual, great video. I am encouraged to see that these specialised parts are not always so fragile/sensitive that they can't be hacked into submission. Thanks!
Fabulous repair. Thank you
What an insane repair!!
This was such an awesome repair!
Great fix. Thanks for doing these kind of repairs. I just repaired an e3632a off ebay for about 15% of list. I would never have had the guts to do this without these videos.
Another amazing journey with you Sir. I have watched this 5 times already. I can follow, but your reasoning skills and troubleshooting are a testament to your knowledge and education and that is a talent only a very small percentage of engineers have. For someone with a PHD, you are refreshingly down to earth and approachable.
This is what complete mastery of electrical engineering looks like
Fantastic video !
Amazing repair , you are amazing guy and I really appreciate your videos
A really inspiring video!
Man that was one of your best..
Great video! I think that these videos showing a way to approach a problem are great. As per the pace of the video I don't think there is any problem with going faster through stuff already done.
Amazing work. Thanks again for these videos (even if I don't understand a lot of it)
Thank's for not quizzing us, because there is no way a beginner like me is going to know what a notch filter does.
You can slow the video down in the settings if it's too fast.
FYI, a notch filter eliminates a certain frequency from a signal, but not those frequencies above or below. A common use is to eliminate a strong harmonic from some generated signal: when you make a given frequency signal you'll get extra signals at multiples (and/or fractions) of that frequency. You don't want those extra signals (harmonics), so you can filter them out in various ways.
You may have noticed that there are two filters in the section where the notch filter is. The other one is a band-pass filter. It eliminates frequencies above and below the signal of interest. The band pass filter operates first, but clearly there was a harmonic that it wouldn't eliminate. Making it stronger would probably distort the signal of interest, so they added a notch filter to just eliminate the harmonic.
Great video, I really enjoyed it
Yep, the 2nd LO on these units are known for filter drift: literature.cdn.keysight.com/litweb/pdf/E4440A-20.pdf
The standard real time BW of the PSA is 10 MHz. Basic mode gives you digital demodulation, time domain and power statistics for signals up to 3 GHz (similar to the E4406A VSA).
Nice phase noise option :)
Wow good find.
Indeed the stripline filter drifts, probably due to the substrate DK drift (moisture adsorbtion?). I have repaired several of these DRO-PLLs by disconnecting the filter, attaching it with semirigid cables to VNA and "live" retuning by shortening each resonator slightly from each end with x acto knife until recentering. This works as a charm and no additional changes are necessary.
I've watched this one several times...one of my favorites besides the 2+ hour long Agilent Spec An you did which is my favorite. I really like the fact that you used common sense & experience to note the oxidation of the filter traces & realized the skin effect depression on the signal especially at 3.6GHz that could have.....excellent deductive reasoning sir & I applaud it. You & W2AEW's channels are tops in this area due to you both having great insight, background & experience in the field...don't stop for sure. I also think your style of NOT making use watch you take out 234985767 screws etc makes it even more valuable as an aid to our learning....one suggestion, consider some new video production software to allow you to do split screens....so you can show where your probing on 1 screen & the result on SA or DVM etc on another especially when the angles or spaces between don't lend themselves to a single camera shot... my 3 cents....but great job Shahriar & I'll become a Patreon supporter of your channel this year (so you can do those split screens haha)
Thanks...even Rigol couldn't specify the probe jack they used on their own test equipment...my applications engineer friend at Rigol went all the way to the Chinese support folks to discover they didn't even know. Oh well, but thanks for the tip and keep up the great work, always look forward to learning a lot from your repair journeys! :-)
Nice catch...
I have no experience with HF equipment. I learn from these videos.
Thank you for the post
This was a very beautiful repair. After oiling the plunger none of them should stick. Thinking about the oil no dirt will interfere because the unit is screwed together so tight. I would love to have this instrument and think it is cured. I really admire your logical thinking on error detection. It is very admirable. Good job on this very useful instrument. The frequencies this unit observes is within the frequencies i am studying with ultrasonic underwater harmonics. Harmonics of splitting water molecules for more rapid desalination.
I have repaired a higher frequency variant in this same series. Same issue with the Elcom DRO in the low band module. Interestingly, the replacement part I ordered from HP (Keysight Never!) was an entirely new design, apparently done in house, as it looks similar to their newer modules in more recent test gear.
They appear to have gone through a number of subs before getting fed up and designing an entirely new part. or possibly they had a duplicate function that they needed in something newer and actually paid attention to lessons learned when designing it. I have the old Elcom carrier board sitting around, might take a stab at it if I ever get bored.
Nice catch on the lubrication on the attenuator. I brought an HP8494b back to life with a little machine oil in the right places.
All that and it still somehow operates using a time portal to 16 years into the future. I guess you need to wait for EEVBlogs 121GW multimeter to fix that little problem ;)
Excellent repair. Using the spare PCB space to chain lower gain amplifiers in that 3.6 oscillator was genius as was plating over the PCB tracks on the filter. Though I would have liked to have seen you doing the dirty work!
I finished all of the video and just WOW. I'm not surprised that the oxidation happened but rather that it was made to happen in the first place. That part was made subpar for its application. I'm sorry to say that but really it is. To be included in an Agilent instrument I'm shocked even more. I hope they avoided buying parts like this from Elcom or Elcom stops with these designs for future instruments but I suspect they are the cheaper bid...
Look at this vs say the inside of an Avantek or a HP assembly from the 90s (I made a recent preview video on this). Even the attenuator at the end you can see what I mean. You'll note 2 things. Most important one is GOLD. Then there is a part somewhere usually made of ceramic. The gold is used for plating copper traces to prevent oxidation. Even connections are made by wire bonding than soldering. Ceramic (ex. Alumina, AlN, and BeO) is used for thermal reasons (like thermal conductivity) as the substrate. Using just PCB is a great way to kill components in a metal can. Acts like a thermal insulator.
Since this unit is not hermetically sealed, it is no surprise that oxidation would appear on the traces from leaked air and a heated environment made worse by materials used. For a more reliable long term fix would be to try gold plating a newly made filter or creating a good seal with a "servicing screw" and periodically purging the unit with nitrogen. I have to make a PVS-14 vid about nitrogen purging if one doesn't exist already. This gives an idea about what I mean by nitrogen purging an enclosed space.
The only reason why gold and ceramic wasn't used was to cut costs which shocks me because you are still paying the same fortune or more on these instruments if they were bought new just like if something from the 90s was bought new. Worse off is that our military or high risk cilivian organizations (like NASA) uses these instruments for critical debugging of their own equipment. Sorry about my long rant but considering how pricey this is and you end up with an instrument with subpar parts boggles the mind...
In any case great repair vid. I loved the reasoning and the macgyvered solution towards the problem. Looking forward to your next repair. I'm gonna guess that other unit in the beginning is the Anritsu MS9710B from earlier.
Bravo!! You are very sharp and clever. For sake of precision: the pll is based on a coaxial resonator vco working on a pure TEM mode. Basically, a DRO uses a high constant dielectric puck working on a TE01 delta mode. The field configurations are different. For frequencies below, say 2-3 GHz, the puck would be too big and the coaxial resonator is preferred. You are right. It is possible to tune the coax ceramic resonator with the tip of your iron!!! Many years ago I used to be a designer of both type of oscillators...
Brilliant!
You are the number one !!
nice repair 👍
What I think could be intresting too is a repair video using only the most basic tools necessary and some "hacks" to measure things. Imagine you are a student with only a rigol DS1000Z series scope, a few multimeters, PSUs etc. but nothing fancy, and score a dumpster find of a similar spectrum analyzer and want to repair it now....
This is a good point. He did extensively use a working spectrum anaylser to fix the broken one. Most people don't have a spare hanging about. Prehaps just use a 100MHz scope, a frequency counter and some mixers?
Great video, thanks. You mentioned possibly going too fast, I would have been interested in a little more detail on the mods you made to the elcom unit. Maybe take a minute with a diagram of the circuit and describe the mods a little more.
Amazing
Dear Shahriar,
As of your question about the attenuator: it is the 26.5GHz version. They used a similar approach at the CXA (and likely other X series analyzers too), that they used a single type. Even our 3GHz version had the 26.5GHz capable attenuator inside.
two levels to reach: 1) being able to buy the equipment 2) being able to confidently say what you measure is real and not some garbage introduced by incorrect usage.
Will I ever reach these levels? :(
It depends. The primary thing is that you need good attenuation to detail.
Can you explain more with the DSO on how the filter (zig zag traces) works?
21:10 "I can also tap it to see if the power changes..."
Very nice video, quite the complicated repair in this one! PS- the 'mystery' instrument on your bench is clearly an OSA, probably an Anritsu judging by the location/style of the optical input and floppy disk location, possibly your MS9710B?
I am trying to think what it will hurt putting two amplifiers in. Instead of tinning the filter I think one could use petroleum jelly to remove the oxidation with a q-tip. It did raise the signal 6db by tinning the filter. Good work.
Another great video!
Just curious about that crud found on the snakey bandpass filter (inside the Elcom SPDRO).
What would cause that stuff to accumulate (some form of electroplating effect, etc.)?
It looks like typical silver oxidation. Silver probably wasn't the best choice of material.
👍👍
Just rewatched this older video, still very interesting! Did you find the source for the 520MHz yet?
Very nice work!
But even after this video, I still avoid to touch RF filters, attenuations, and mixers...
For Me, is still black magic.
Another great video, thank you !
73 N8AUM
Great repair ! May be you could explain little more in detail this modification. How to calculate proper resistors for biasing and so. Thx
Can you comment on what's involved in designing a spectrum analyzer that goes all the way down to 3 Hz? Surely there's more to it than just removing the AC coupling capacitor at the input. What other challenges are involved?
I'd say most mixers (e.g. double balanced ones) can't go down to < 1 kHz. Also I assume having a LO in the
In the sat comms industry seen a lot of the phase lock block units going down exactly the same as your unit low output !, looks to be very common problem with these units low output.
Epic fix. Botching in the 2 additional amplifiers was great. Surely they introduce more noise that the old amplifier, can you measure the difference?
The instrument weighs 30kg!? Thats really impressive!
Very informative presentation! Where can I find lecture how to repair Power Supply module for E4440A Agilent 0950-4447 Rev 2 . It is very difficult without schematic. Yellow LED doesn't lit on the power ON switch, +15V is not present, but I have +/- 162 Vdc on the big electrolytic caps and at the first switching MOSFets.
Put solder over those traces doesn’t affect the behavior of the filter? Impedance or so
Hi. I have the same model which has problem with the power supply model. Do you know how to bypass the on/off switch so I can troubleshot the PS alone?
Is it true a spectrum analyser us a bit like my software defined radio for listening to ameteur radio?
Can someone explain why did they assemble the attenuator unit in this order (10, 6, 20, 30). Why not 6,10 etc.
Great video as usual!
Do you sell these after you fix them?
He does sell some of the things he fixes. If you are clever you can figure out his eBay account, but I won't post it here.
Depends on the instrument. If I don't have something similar int he lab, I keep them.
The track between the cavity and the case wall looks weird, like if it is burned or something like that.
Could not find the mmic you used, what was it?
Blah, i wished tge problem was in on of the more complicated board ahaha. Anyways good videos there is another unit selling on ebay at an incredible price. Good find!
I had to lubricate piston on my marconi 2955a because these relays were no case protected ad yours... So dust and dirt will get into piston, i dont understand why they design relay without case shielding against dirt...
I would love to see a RYTHM "Routing YIG Tuned Harmonic Mixer" teardown if you have a spare broken one on hand. That part in and of itself alone is interesting. Its a preselector filter, PIN switch, mixer all in one. It has 4 YIG balls in it too. I've seen a broken 5086-7884 (for the 8563E/EC models) on ebay going for $200 - $400. If you don't have one on hand and can't do ebay perhaps we could work something out? I'm also able to loan one to you for review/educational purposes. I have a dead one on hand currently. For information on these there is patent US 5553319 A and a QSL site by Luis Cupido (under his wire bonding repair link.)
I don't like what you did to the filter, was this just a do or die action? I mean by putting tin on the tracks they now all have a different coupling to the next stage... Another thing - why does it have an optional audio input, is this because it goes down to 3 GHz so is it an extra signal input?
Did you ever find a source for the front panel "LEMO" style connectors? Apparently it's some sort of industry standard...would love to power my homebrew active probe from the Rigol DSA1030 instead of a separate power supply.
I ended up buying a probe power supply from eBay for about $50. It had two of them on the front panel. That is the only source I found. :(
Old Metrosonics noise dosimeters used LEMO connectors to attach the mic cable. Maybe you could salvage some?
WOW a blast from the past, I worked for METROSONICS when they were still in the basement of the owner's house. Digi-Key and Mouser sell LEMO connectors.
Hi, could someone please guide me to figure out what is wrong in my SA E4440A. Its reading less 30dBm power over 4GHz. I check different portions. Cannot find 321.4MHz from RYTHM and LO as well going to A13 board
AFC=automatic frequency calibration?
Good video, Asking for advice fix N9320A , Measure Amplitude Loss -3dB Thanks!.
So a non-Keysight part caused the Keysight instrument to fail.
You should grab a bunch of those floppy drive to USB or SD card adapters to replace the floppy drive on these old devices
Also that other thing on your bench is CLEARLY an Anritsu MS9710B, duh, everyone knows that ;P
[RF Noob here]
Something I never understood is why there is no solder mask on RF PCBs. Wouldn't some kind of varnish have protected that filter agains corrosion ? I'd say it has to be about controlling the impedance or whatnot, where the soldermask cannot be properly controlled.
My guess is that the solder mask is basically some form of paint and that that might interfere with the properties of the PCB. Also, that PCB on which the filter is made is no ordinary PCB material. For RF usually special materials are used to have better controlled dielectric properties. I suspect that a solder mask might harm these properties or at least make them less predictable.
It would certainly protect it from corrosion, but it would influence the RF properties (e.g. impedance matching ; filter Q) due to skin effect etc. So the RF performance would depend on the solder mask being simulated properly during design of the filter, but in pratice there will be manufacturing differences which would create mismatches (e.g. thinkness).
The black magic pixies that do all the RF stuff don't like the smell of solder mask. Interferes with their otherworldly incantations.
I can say what the others have been saying is true. Solder mask would degrade the signal. Its like if I were to add water inside a wave guide. The metal housing itself is acts as an important component as seen here when Shahriar opens and closes the assembly with its cover. There is a way to help against corrosion and still keep it free of any painted coating. The solution is to gold plate copper and to use wire bonding instead of solder for connections (flux residue and cleaners can change characteristics as well).
PCB is a terrible subtrate to use in RF as well but not just because of electrical properties. You can design around electrical problems. The problem is heat. PCB acts as an insulator and it comes as no surprise that the component he replaced died. The subtrate should be a ceramic material so that heat can conduct itself out to the metal housing which is screwed to the case that has airflow from the fans cooling it.
See another dirty connection :) :)
Mmm así se siente hacer el primer comentario... interesante
991Mhz is a bit off. Is it normal?
I think I probably would have given up after 3,999 screws.
LOL 35:35 - IS BROKEN
This instrument is super expensive! never seen second hand equipment this expensive.
Where did you get the money to buy those instrument? Those equipment in that lab worth probably about a million dollars.
May 18th 2033 would be somewhat logical to reset to, but July not so much
I sell two devices of this type if you are interested
Adding the amplifier adds noise… ie… the performance/dynamic range was changed.
While this did get the level up so that the alignment could be run… it did not fix the “REAL PROBLEM”, as in, the real issue will cause this 2nd Lo level to drop in power once again in the short term…. So now you will have a broken unit plus the higher noise…
I have seen many peoples attempt to fix this issue… and like this one… it works for a limited time…
Even Agilent/Keysight does not “Fix” this… They changed the design to use a completely different Lo Board…
I know you can see listings on ebay from a bunch of places calming to have working or refurbed Boards, but there is only one Company that I know of that is actually fixing this the right way… as in, it will last for many years… and does not change the dynamic range… That Fix is being done by a company by the name of Alltest or Expresstest, and they are doing it for the same price as the rest of the hacked up attempts that others are selling…
I am afraid your concerns are not valid. The original unit used an amplifier and an attenuator. I used two low-noise MMICs and no attenuator. The second amplifier operates in large signal. The combined amplifiers impact on the phase-noise of the signal LO is below the phase-noise of the DRO itself and does not impact the system performance. I verified the DANL as well as the phase noise of the unit and it meets all specifications.
Furthermore, the whole point of operating the second amplifier in compression was to build margin into the unit. The filter can degrade by another 10dB without loss of output power. This fix will outlast the age of the unit itself.