30 January 2011

The Leaking Electrolytic Capacitor Plague and my Tek TDS-460

There are many good things in life, but bad ones tend to take more of humanity's time. There is compassion, love, science, understanding, but people spend most of their life - if not all of it - in greedy bellicose hunger, hating, believing in ghosts (holy or otherwise), and missing the - equally valid - other guy's perspective. The news every day provide ample proof.
Having put that off my chest, I had better now focus  my "elegant tapestry of quotations, musings, aphorisms, and autobiographical reflections" to the more mundane matter of  The Capacitor Plague, which, although completely unknown to the Medical Science, has caused many a frustrated consumer's tear to flow. In short, electrolyte leakage from bad (bad, bad) electrolytic capacitors destroys thousands of electronic devices every day, by corroding copper traces and creating parasitic conductivity on printed circuit boards. The plague doesn't discriminate and victims appear in every class of electronic devices. I have seen my car's ECU, personal computers, power supplies, test instruments, transceivers (and my friend's Stavros Sony ICF-80) die a gruesome death by the infamous Capacitor Plague.
In all godlike modesty, I have been able to resurrect most of the victims, save a few that were truly beyond redemption, as the time and cost to fix them was more than getting a new one. The latest unfortunate victim in my troubled experience was my beloved Tek TDS-460 digitising oscilloscope, a true work of four-channel art, which suddenly (and scaringly) started failing the self-test and showing erratic triggering (quite blasphemous for a prestigious Tek stallion). As the original SMD electrolytic capacitors, with a tarnished history of leaking, had been very wisely replaced with new ones before I acquired the instrument, I was almost certain that the malfunction couldn't be attributed to the electrolytics - although the symptoms - in an eerie way - pointed straight to that direction. Turns out that the Plague is like a time-bomb, and a very delayed one sometimes.

After thoroughly cleaning the (amazingly beautiful) acquisition board with isopropylic alcohol around the electrolytics, the symptoms went away (to my great joy), but only to come back to haunt me a couple of days later. I immediately repeated the cleaning and drying - same story. I replaced some of the caps in despair - nada. So I was clearly missing something. My trusty 4X Russian magnifying glass in hand, I started examining the PCB in detail around the electrolytics - and I finally spotted what you see in the pictures (click to enlarge), near the pins of U82 and U140. The discoloration was well concealed UNDER the conformal coating of the PCB and very hard to see, due to the light reflecting off the epoxy coating and the glare obscuring the surface under.
As there were no signs of electrolyte leakage from the new capacitors, the damage must have started with the old, leaky caps, and it took years to finally manifest itself as a malfunction of the instrument. The electrolyte had crawled under the conformal coating and gradually compromised the insulation, eventually disturbing trigger control potentials and making the self-test fail (it logged the error message: "trigComparatorTest, TRIGA status after trigger: exp(ected) = 1, act(ual) = 0).  The dried-up murderous electrolyte residue shows best in the second photo, of U82 (the first is of U140).
After scraping off the coating and affected spots under it with the point of a scalpel, the symptoms vanished, and this time I think it will be for good. I couldn't find any more such discolorations elsewhere on the board, but there's always the inaccessible area under the SMD  integrated circuits, so I shouldn't be unduly surprised if...!

Well, there's a nice thought to keep a test instrument lover twisting and turning in his bed at night...

27 January 2011

The Withering Filters: A microscopic view

In a previous post I discussed the deleterious effects of electromigration and corrosion on ceramic intermediate frequency filters, when DC potentials are applied to their pins, especially the output pin. I have collected some more data that may be of interest to those following the subject. Take a look at the photos, click on them to see an enlarged version.

Life on the edge. The first photo (25X magnification) shows the corroded edges of a filter element (the shunt output element, which due to its small thickness gives the most trouble). Observe the damage to the plating at the corner and along the upper edge, also the metal / oxide deposits that have short-circuited the element, rendering the filter inoperative. Measuring with an ohm-meter, the resistance was about 30 ohms between the plated surfaces (where it should have been a very large value, in the tens of megohms or more). However, after cleaning the element's edges in the way I have described, the filter (amazingly) returned to normal! I have repaired numerous such filters this way, and, after the addition of DC blocking capacitors in the circuit, they seem to stay healthy and happy.
 A rough corner. The corner of the element in the previous photo (now in 200X magnification). Electromigration and corrosion have admittedly done a great job of stripping away the metal plating and short-circuiting the element. Check out the dendritic growths at the edge of the remaining plating. The small magnification factor doesn't do them justice. Anyone with a spare electron scanning microscope? I could accept a hand-me-down, you know, I am not that snubbish...
Filter-pox. The effects of moisture inside the filter, on the other elements. Although rather spooky-looking, the elements actually checked out all right. The leftmost thick element is the one at the input. Owing to its thickness, the input element is a lot more tolerant of the DC bias abuse. In fact, I have yet to find a troublesome input element, the trouble is always at the thin (~0.35mm) output element (seen to be missing at the far right). The elements between the input and output elements of the filter don't "see" the DC bias, so they don't suffer the dire consequenses. They get their spots, nevertheless. Moral: It doesn't pay being in the middle of any mess.

An inside job. The spring plate compressing the elements in the filter case is oxidised, too. Yes, those shiny sparkles are indeed tiny droplets of water. Some of the filters I've examined contained a surprising quantity of water. The damage was roughly proportional to the quantity of water, which suggests that the manufacturing could be improved, so as to prevent water from entering the filter's case. The problem is, we're talking about filters that cost a couple of euros retail, and it's always true you get what you pay for. I have yet to see a good crystal filter deteriorate due to moisture ingress, they're truly "hermetically sealed".

A very green face. The end plate at the input side, also oxidised. Although certainly a disturbing view, the oxidation there doesn't do much harm to the functionality of the filter (well, up to a point, I guess!)

It's aliiiive!! To test a little theory of mine, I experimented by applying about 60V DC through a 33KΩ resistor to the input and output pins of a new TOKO ceramic filter. The high voltage was intended to speed things up (things = the deterioration of the filter). This went on for two months, after which the output element was short-circuited (as expected), but the thicker input element was mostly unaffected. So, in conclusion, one may take a calculated risk and add a DC blocking capacitor only at the output pin of the ceramic filters of this type, as the input seems a lot more forgiving to abuse (and most transceivers apply only about 8V to the filter). The output pin is the one closer and opposite to the arrangement of the three grounding pins (in the TOKO brand filters I have found in Kenwood transceivers). Check the manufacturer's data sheets to find out about the other brands and types.

Radio amateurs (and every sensitive person) should establish a movement for the rights of those wonderful, innocent, so unjustly suffering components, the ceramic filters. Don't just sit there! You could write to your manufacturer of ham radio transceivers and ask for the continuing torment of those poor beasts to end at last, or something like that!...

***ADDENDUM: For those who want to learn more about the phenomenon that I propose that causes DC-biased filter failure, please take a look here: http://www.ami.ac.uk/courses/topics/0158_emgr/index.html
There you can admire two great photos of the results of electromigration across tracks and solder resist on printed circuit boards, plus lots of interesting relevant information. Clearly, humidity and voltage gradients at small distances are a bad combination!!