-
I am no longer doing 'tear-downs' but thought some readers may be interested in a repair that I am about to undertake on a very nice portable U4941 Spectrum Analyser from the Advantest Test Equipment Stable. I used a similar model some years ago and it was very useful for portable applications as it can run on a battery pack or external 12V supply.
These analyser's normally sell for over £1000 in the UK, but this one was poorly so I risked £200 on it. I says 'risked' because there is no affordable support for these units in the UK. R&S offer a fixed (very high) price repair service. If something unusual has failed within, it becomes an expensive door stop. I have loads of spares for my R4131D SA's but none for this U series model.
The U4941 arrived and was found to be in excellent condition. It has been well cared for in its life and it came with its official Khaki protective soft case and the User Manual.
There is no service manual openly available for this model but the U3661 is similar and the service manual proves what I already expected. These units are module replacement only with no component level repair supported. No schematics are provided in the service manual. It has just basic faulty module identification routines. Not a whole lot of help really.
I love a challenge so the lack of a set of schematics is of no concern to me. I just reverse engineer what I need to better understand. All part of the fun for me.
Upon receipt of the Advantest Spectrum analyser, I carried out the usual initial checks before opening the casing. With power and an external monitor connected I was able to establish that the boot system self check showed Error 900 which is a lock problem in the Phase Locked Loop. The microprocessor board appeared to be working OK as the front panel controls all worked. This was good news for me as I dislike microprocessor related faults.
During the brief testing phase I detected the smell that I associate with degraded and leaking electrolytic capacitors. It was coming from the main units extraction fan rather than the external power supply block. The smell is not forgotten once experienced. It is a rank sweet fishy smell that can be quite overpowering in extreme cases of electrolytic capacitor leakage. The leaking of capacitors in equipment built in the 1990'sis well known and named the 'Capacitor Plague' as it is so common an can be very destructive. There is a dedicated web site detailing this nasty and all too common failure in modern electronic equipment. It is most prevalent in equipment power supplies, especially IT related equipment where such capacitors are rated close to their operating voltages. The leaking electrolyte is very conductive and corrosive and can cause all manner of havoc on a computer motherboard.
The Wiki page on Capacitor plague is here:
https://en.wikipedia.org/wiki/Capacitor_plague
The web page dedicated to the problem is here:
http://www.badcaps.net/index.php
As soon as I had smelt the well known failed capacitor odour I knew I would need to inspect all the capacitors on the various boards within the SA, starting with the Power supply related module. This Advantest SA is designed to be field portable so it has the ability to run from a 12V battery pack or external 12V supply. As such the unit contains a DC to DC converter board that provides all the regulated power supply rails for the unit. This SA requires multiple +12V, -12V, + 8V, -8V and +5V for analogue circuits, and +5V for the digital electronics needs. The Analogue and Digital supplies are kept separate to avoid digital noise cross contamination issues.
On the DC-DC module board there are nine separate isolating TDK DEL series DC-DC converter modules providing the requires power rails. There are three +12V, two +5V, two -12V, a +8V and lastly, a -8V. The use of individual DC-DC converters is a little unusual as normally all the required rails are provided by a multi output single DC-DC converter module. The good news is that a faulty module is easily identified and replaced or repaired. This is why the SA still appears to partially work. The digital power rail is being correctly generate and so the microprocessor is running fine.
Further investigation of the DC-DC module proved that it was the sole source of the awful leaking capacitor smell. Every electrolytic capacitor on all of the TDK DEL mini DC-Dc converter modules was leaking ! They really reeked when brought close to the nose. Some of the larger Brown coloured electrolytic's on the modules PCB also appeared to have leaked electrolyte onto the board. At that point I knew I was in for a complete replacement of all electrolytic capacitors in the DC-DC module. I will also need to clean the PCB's and repair any corrosion that has occurred. There was visible overheating damage on the -8V TDK DEL module.
A test of the complete DC-DC module was possible by supplying it with a 12V lab power supply and activating all of the TDK DC-DC modules via the remote control on/off line. Tests proved that two TDK modules were not producing an output. One +12V and the -8V rails were missing. Both modules have been removed for investigation and repair.
The faulty TDK modules were visually inspected and the -8V module is in a bad way. Its power Mosfet has seriously overheated, partially desoldering itself from the PCB, and cooking the PCB in the process. Exactly what caused this overheating is not known as the TDK module contains over-current detection and automatic shut-down to protect itself. Something has gone wrong n that little PCB. It could be just the leakage of conductive capacitor electrolyte that was soaking the PCB's surface. we shall see when it is cleaned and new capacitors and Mosfet are fitted. The +12V module has no damage evident except electrolyte leakage. It may be in over-current fold-back mode.
I ordered all new capacitors and the required power Mosfets from Radio Spares to repair the two modules. In the mean time I quickly removed the failed capacitors and faulty Mosfet. I also reverse engineered the TDK DC-DC modules so that I better understand their circuit topology. They are actually pretty simple devices but they use an obsolete switching controller so I hope neither have failed on the faulty modules. If so, I will fit an alternative DC-DC module of suitable spec. All capacitors on all TDK modules and the supporting PCB will be replaced with quality parts from Panasonic to ensure high performance and long life.
I will treat the whole DC-DC module PCB and associated TDK DC-DC modules to a good bath in my specialist Electronics compatible Ultrasonic Cleaning Bath. It is designed for delicate work and uses a swept Ultrasonic frequency and 'electronics safe' cleaning solution. It will remove all the nasty remaining leaked electrolyte that I could not get to with IPA and swabs. This will ensure that corrosion of the PCB's does not continue and ensures the long life of the module.
I need to restore the DC-DC power module to full working order before i do any further diagnosis on the SA as the correct DC powers rails are a fundamental requirement for correct operation, as you will appreciate. I can then check for over current situations on the power rails causing fold-back of the associated TDK module.
I inspected the microprocessor module and some of the electrolytic capacitors on that modules PCB have some corrosion around the pins. That is enough for me to replace them as a precaution, even if the corrosion is not leakage related. I have ordered replacements for all those capacitors as well. I shall be busy soldering new capacitors onto the PCB's, that is for sure. Prevention is better than cure however.
Fraser
-
The DC to DC module etc
-
More of the DC-DC module etc
-
More of the DC-DC module etc.
-
Leaking Nichicon 100uF 16V electrolytic capacitors. All have failed in this module and show between 6 and 40 Ohms ESR. No significant corrosion is visible at this time. Also, pictures of the dead Power Mosfet that was cooked to death
-
The module and removed failed DC-DC converters are now awaiting the fitting of the new parts. The microprocessor module will have all new electrolytic capacitors fitted as well.
The total cost of the parts from Radio Spares was less than £20 so not bad really. There are a lot of capacitors to be fitted though !
I chose good quality 105C rated Panasonic capacitors with low ESR and decent life specs so they should last a while even if the module gets hot when in operation.I did not buy the capacitors from eBay as quality is not always assured and I do not want to repeat this capacitor replacement any time soon. Fake Branded capacitors are also common on eBay making a bit like Russian Roulette when they are needed for a serious application like this.
The full set of parts arrived from RS today via their usual excellent free next day service
Fraser
-
Excellent write-up so far, Fraser. I look forward to reading subsequent episodes as you progress.
I like you calling RS Radio Spares, I'm old enough to remember when they were called that!
-
This repair could end up being pretty straight forwards, or it could turn out to be a rabbit hole to hell, depending upon what else I find wrong with the unit.
The original symptoms were a white LCD screen and no signal base line on the display when n external composite video monitor is connected.
With the unit completely missing one of its +12V rails and the -8V rail, I m not surprised it was malfunctioning. The seller said that it worked before going into storage so I may be lucky. I have noted that the RF module screws have been moved since manufacture though so there may be issues yet to be discovered. Once i have the DC-DC module running correctly I will be able to determine whether all is well with the other modules.
With regard to Radio Spares, yep I remember them as that . The problem is that if I say 'RS', some of our American friends might confuse that with Radio Shack, a very different company !
I will update this thread as an when there is progress to report.
Fraser -
Thank you for sharing your work.
Can you let us also know a little more about your reverse engineering methods you use in your work?
Thank you.
-
Hi Jane,
I have discussed some of my reverse engineering work, and approach, in this thread:
https://www.eevblog.com/forum/eda/pcb-reverse-engineering/
I can recommend the Book detailed in the thread for those new to the 'art'.
Fraser -
Yes, I read the info about the book but I was afraid of Visio software . For me it can be a time consuming task to draw reverse results . What I need is to find only a connection between parts on PCB and use a hand-writing only. Reverse Visio schematic can be nice but not needed for repair workers/ hobby people like me.
-
Jane,
The book is not just about using Visio to draw schematics. That is just the documentation part. The principles and methods of reverse engineering a PCB are fully covered. I still use a noteboook, A3 drawing board and A3 paper to draw the schematic as I trace the routing of the various connections between components. I often do a rough sketch of the schematic before drawing it neatly on the drawing board, if needed. It is true that I a converting onto Visio now, but that is by no means essential.
Basic points on reverse engineering a complex PCB.......
1. Try to get a decent image of both sides of the PCB using a flat bed scanner or camera. Print the pictures and use these to draw PCB routings and points of interest. I often create a node map as well to trace where a particular signal or power rail appears on the PCB. There is a problem of course.... if a PCB is faulty with a broken track or feed through, you will not find continuity where it should be ! That will, of course come to light when a particular chip or component appears unconnected on a pin, or one end, that should go somewhere.
2. Obtain the data sheets for ALL of the IC's on the PCB and study how they are normally used in a circuit. That can give vital clues to the PCB's topology and functionality.
3. I use the continuity function on my FLUKE 87 III as it is very fast to react if you run the moving probe along a row of pins etc. I also use the Wavetek continuity finder brush and probe that was detailed in the thread I referred to. I have five of those Wavetek/Meterman/Fluke units as they are just too useful to risk not having one in the future. Such a shame they are now discontinued.
4. Another trick is to take your time and split a PCB up into specific areas that will be reverse engineered to schematic level. That makes the task far less daunting on a complex PCB. The chip datasheets will have provided an idea of what chips are responsible for what functions and these chips are often grouped together with their associated support components to keep PCB tracks as short as possible. This is not always the case however and you can have a nightmare PCB that does not obey any common sense or logic in its layout.
5. Multi layer PCB's and those with several FPGA's present special challenges and should be considered a bit advanced for someone starting out in PCB reverse engineering.
6. PCB reverse engineering is not black magic, it just takes time, patience, a methodical approach and logical thinking. It is true to say it is a bit of an art that is learnt as experience removes much of the fear from the challenge and also helps to better predict how the designer likely designed the circuit. This helps me to take the datasheets and effectively 'guess' how everything is connected together before I prove such on the PCB. That does take quite a bit of experience to do well however. I specialise in RF circuits and thermal imaging cameras so have built up a decent knowledge to assist me with such schematics. Equipment that contains a single board computer has to obey the basic topology of a computer, be it in separate elements, or a single chip computer.
7. It could be said that reverse engineering an embedded computer PCB is simpler than working on an analogue electronics PCB. With a computer you know that the micro-processor needs suitable power supplies, a power on reset circuit, RAM, ROM and I/O. Its is a bit like joining a dot-to-dot as the major chips are connected together, either directly or via buffers and glue logic. The address and data lines are very easy to trace out between the various chips that need such. Control lines can be a little bit more challenging however as they can conect to all manner of support components depending upon intended function.
That is enough from me for now......... it is possible to write a book on this topic
Fraser -
All the capacitors are off the PCB's and the whole board assembly has been given its first bath in IPA to remove flux and other contaminants.
I submerged the assembly in IPA and then rocked the container to create a gentle wave across the surfaces. From the colour of the IPA that I later drained off, the board had a lot of contamination with excess flux etc.
Tomorrow the assembly goes back into an IPA bath but this time it will be agitated automatically by a laboratory mixing/swirl platform. I will leave it to self clean for an hour and then rinse it off in de-ionised water before giving it a thorough clean in the Ultrasonic cleaner. This may all seem a little excessive but I want to remove as much contamination as possible before the ultrasonic bath and ensure that any electrolyte contamination has been exposed to both an IPA bath and the Ultrasonic cleaning solution. I do not want corrosion problems to pop up on these PCB's sometime in the future due to poor cleaning.
I will then fit the new capacitors and clean off any excess flux that results. The two faulty DC-DC modules will be dealt with separately and will hopefully be repaired. They can then be re-fitted on teh main board.
The PCB's were all 'swimming' in electrolyte, including the main PCB, it was bad.... but no corrosion to speak of so that is good. The smell is rank when you desolder an area that is contaminated with electrolyte. Nasty job.
Fraser
-
The DC-DC PCB's were given a thorough clean today and I thought some might be interested in my technique and the equipment that I use.
I own a Ultrawave Hygea swept frequency Ultrasonic cleaning bath and a Zinsser Analytic Desyre-Mix orbital mixing table. Both are used in the following cleaning process.
The PCB's were badly contaminated with capacitor electrolyte but thanks fully corrosion does not seem to have set in yet. I need to removal all traces of the electrolyte from every nook and cranny on the PCB's to avoid future issues caused by such. The motto here is do it properly and do it once.
The process I used was as follows:
1. All suspect capacitors were removed from the PCB's to permit easy access to the area beneath them.
2. All old solder was removed and new solder laid down on the pads to protect them.
3. The PCB's were soaked in an IPA bath for 1 hour to dissolve flux residues and loosen the electrolyte residues. Occasional manual bath agitation kept the IPA moving over the PCB's surfaces.
4. The PCB's were rinsed off using tap water and left to dry
5. The PCB's were given another bath in IPA with mild agitation to loosen dirt.
6. The PCB's were rinsed off with de-ionised water (no lime content)
7. The PCB's were placed into my Hygea Ultrtrasonic cleaning bath and the bath programmed for a 30C then 40C temperture process. The duration of each process was set to 3 minutes. The full process was carried out for each side of the PCB facing down towards the Ultrasonic transducers (of which there are six for good wave propagation). The Hygea is filled with a cleaning solution designed specifically for delicate electronics and was bought from Allendale electronics.
8. After both sides of the PCB's had been cleaned in the Hygea, the PCB's were allowed to drain off excess cleaning fluid before being placed in the rotary table mixer bath. The Desyre-Mix is a good quality lab grade mixer that effectively 'jiggles' whatever is placed onto its top table at a frequency set by the user. The plate is also capable of heating up to 140C if desired. I attach a water bath to the plate and do not use the heater. When running the PCB's are subjected to swirling water passing across the top and bottom surfaces. The action provides efficient and effective cleaning of the PCB's to remove any detergent residues. The PCB's are cleaned in this way for 30 minutes.
9. The PCB's are removed from the water bath and then placed into a bath of IPA that is agitated on the Desyre-Mix for 10 minutes.
10. The final stage in the cleaning is to remove the PCB's and place then in a bath of de-ionised water that is agitated on the Desyre-mix for 10 minutes to remove all IPA residues.
11. The PCB's were inspected and found to be spotless and ready for drying in a warm air flow, and then repair.
This was a very thorough clean due to the specific issues with the PCB contamination. A slightly shorter process is used for just removing dirt and flux residues.
Pictures of the process follow.
Fraser
-
More pictures
-
More pictures
-
@Fraser: Thank you for sharing your reverse engineering methods with all of us .
-
I forgot to mention the technique I employed to remove the SMT 100uF capacitors from the TDK DC-DC PCB's. I used to desolder them with either two soldering irons or hot tweezers but,if I am honest, they were a PITA to get off the PCB. I saw some demonstrations of alternative removal methods and was pretty horrified at what was done. I am not one to discount a method until I have tried it however so I investigated one method that employed brute force, yet never seemed to cause PCB damage. I have now mastered the grip, push and twist removal method
The techniques is as follows:
1. Grip the SMT capacitor case with a pair or pointed pliers
2. Push down on the capacitor with the pliers
3. Whilst pushing down, twist the capacitor clockwise, then counter clockwise a small amount, increasing the amount with each twist.
4. Continue to push down and twist until the capacitors leads shear away from the PCB.
5. Apply solder to the capacitors remnant leads on the PCB and wipe away the short lead remnant in the process.
6. Clean the area under the capacitor with IPA and remove excess solder from the pads with solder wick.
To me this resembles the technique used by dentists to remove Molar Teeth !
I have employed this method to good effect for a while now and have not damaged a PCB track to date. It sounds brutal, feels pretty brutal and definitely looks brutal, but it works
Fraser -
Here is a video off Youtube showing the brute force method of capacitor removal.
I personally like to push down whilst twisting and also to twist through small angles back and forth to cause a 'worked' metal fracture. This, IMHO applies less lateral force to the PCB track than a simple full twist method.
Fraser -
I tried the twist method but I prefer hot air to remove the SMT electrolytics. Safety glasses are a must though because some will literally pop off the board.
-
Excellent work as always, Fraser. Looking forward to seeing how it turns out.
-
Minor update.
I found time to fit the new capacitors and MOSFETs to the TDK DC-DC modules yesterday.
I used my new Pace ST50 and TD100 micro soldering hand piece. It only arrived yesterday and I was keen to see how it performed. In short, it was superb. Amazingly fast to reach temperature from cold and so nice to use. It made short work of the SMT capacitor and MOSFET soldering. The unit came with the three most useful tip cartridges so I am now better equipped for the more delicate soldering work. My Pace PS90 irons are great, but a little bulky for some tasks.
The rest of the capacitors will be fitted today.
Fraser
-
All of the capacitors have now been refitted to the DC-DC module PCB and the two repaired TDK DC-DC PCB's have been refitted.
Bench testing proved that all the TDK DC-DC PCB's are now operating correctly and producing the correct output voltages etc.
The DC-DC module was refitted into the Analyser and it was switched on.
My efforts were rewarded with a normal Advantest splash screen on the previously non operational LCD display. A good start
The splash screen was replaced with the fitted options screen and then the spectrum view display. No errors were displayed. The correct noise floor base line was present and correct. Things were looking very promising. A quick sniff test around the rear of the analyzer did not reveal any nasty smells or that horrible hot semiconductor smell when something is getting too hot. No problem with the -8V line then
The analyser was set to the UK VHF broadcast band and the familiar signal peaks appeared and are at reasonable levels considering the whip antenna and the fact the analyser is in low sensitivity mode with the RF pre-amplifier out of circuit.
She lives
I let the unit run for a while and then extracted the DC-DC module again to check the TDK PCB's for signs of excess current draw on any rail. All MOSFETS were just warm to the touch. Excellent news. All issues were due to the failed electrolytic capacitors and electrolyte that coated the PCB's.
I am giving the DC-DC module another clean in the Ultrasonic cleaner to remove the rework flux that I applied. After it has been washed in De-Ionised water, it will be left to dry in the drying cupboard before being refitted into the analyser.
A happy ending and hopefully an interesting little repair for readers.
Fraser
-
The Advantest start-up splash screen and spectrum display of the UK VHF broadcast band.
Finally, the DC-DC board is back in the Ultrasonic Cleaner.
Fraser
-
A happy ending and hopefully an interesting little repair for readers.
Yay! Thanks, Fraser. It's good to see more coverage of Advantest repairs.