Shelf life of Electrolytic Capacitors

[SeanB]

After a few hours it now has 12uA of current flowing in, so will leave it running overnight at 6V3 to finish it's reforming then see how it has changed from sitting unused in a store for a few decades.

[The Electrician]

If we're talking old caps, just wanna mention that high voltage caps are sometimes rated with a minimum voltage, below which the capacitance really goes down.

So don't rely on low-voltage capacitance test as a measure of goodness for these caps. Just check for shorts and toss into circuit.

Are you referring to aluminum electrolytics? How high a voltage is "high voltage"?  What manufacturer's caps have you seen where the rating provided for lower capacitance at lower voltage?

[SeanB]

After an overnight reform it dropped to 4uA of current, so disconnected it and left it standing with 6.3V on it for the day. Came home, and it is now at 5.3V, so will leave it for another day and see further, then do the tests again to see if anything has changed on this 40 year old capacitor.

[sam coniglio]

Thank you all for the help gentleman!

I like the idea of reforming a group of random samples. It seems that turning on the power of old piece of equipment and let just sit and burn in would help to restore all components.

I have a few old dinosaurs like DOS 3.0 PC, PC /XT and finally a Windows 95 AT. They all boot fine. The components such as Electrolytic capacitors, carbon based resistors, and  Epsoms ect have  not failed the system.

However in the name of progress, faster CPU speeds, and then don't forget the thrill of getting to tell  your next door neighbor that my PC is bigger, faster, more expensive than his unit and just plain the best thing out there.

Of coarse Intel is grinning in the background, fully knowing the next ego trip Microprocessor is all ready to hit the market in a few months. WOW a PC that can do 5 billion calculations per second. This is much faster than my old TI 59 pocket calculator that I used in school.   I was so proud of it that I kept it safe in my "pocket protector".

So frankly hardware and software doesn't ever get a chance to fail due to age.

But what about hard drives and solid state mass storage?

How long does a typical hard drive last. The new buzz word today is  "Cloud Drive". Okay, but the hard drives on the cloud fail every day, just as our drives fail. This is because the cloud uses the same exact hardware that we do.

I can remember one of the first projects that I constructed at 4 yrs old. It was a Chrystal set that used a Germanium diode, and a wire wound Oatmeal Box. Can our man-made memory and other mass storage compare to the human brain with regard to instant access of data that has been stored more than 5 decades.

Anyway the next question:


How safe is the data on our storage systems? I have heard that with regard to hard drives , the conservative estimate is four years, but no-one knows for sure. Do we phase out our Hard Drives  every  four years? Do we put them in the same dumpster as the Electrolytic Capacitors?

Best regards,

Sam Coniglio

[Kjelt]

Harddrives lifetime depend on the design and the usage.
AFAIK, the number of full power on/off cycles with parking the heads is usually the limiting factor but parameters like heat and vibration are very important to.
 
To give an example, with the arrival of affordable NAS equipment from Synology or QNAP and others , WD Green Harddrives that are designed for desktop usage were used in the NAS were the NAS does a spin down (time parameter). The WD Greens failed a lot in that usage.
Professional server companies use very expensive Raid or Server Edition pro harddisk that cost twice as much per GB and have their own vibration algorithms and do not park unless power is shut down. But these are made for 24/7 operation and rather stay on for 5 years at a row than turned off three times a day.
The WD Red harddrives also do no longer park their heads unless total power down.

[mzzj]

Took that 20mF capacitor and measured it with the LCR meter. Overrange..... so used the $20 component tester and it came back, 32.82mF, 0.07R ESR and loss factor of 4.3%. Now busy reforming on a power supply at 6V3, via a 820R resistor ( nearest to hand) and I will check it again after a few hours. Takes a long time for the RC circuit to charge up, it was at 4V4 after a minute.......

Sometimes old electrolytics show increased capacitance because the insulating oxide layer has become thinner. After reforming the capacitance will drop a bit.

Mostly I have had good experiences with old electrolytics,  only real problems have been the caps from batches of the infamous capacitor plaque couple of years ago like Nichicon HM-series that rotted on my self before I actually used them for anything!
Also have bucketload of ~20 years old Jamicon's that have borderline acceptable leakage current even with reforming.

Because sometimes my "stock" sits in self for long perioids I mostly try to avoid water-based electrolytics when i order new stuff. (sometimes tricky to tell, avoid ultra-ultra-low esr types and look for a -55Cel temperature rating)

[SaabFAN]

Some of my HDDs are over 10 years old. One has even been manufactured in the 90s and is running occasionally when I decide it is time to play some old games that require DOS or Windows 98.
I had only one HDD fail due to old age: A 13GB HDD made by Seagate that I salvaged from an old Aldi-PC made by Medion in 1998. It failed because it became stuck and couldn't spin up anymore. If I had kept it running, it would probably still work. Had no sector-errors ever.

It seems like this for HDDs: They either fail within 6 months, or run for a at least a decade.

About EPROMS: Is there a way to get their data without desoldering them if they're not socketed? I would like to make Backups of the ROMs of old equipment to eliminate the possibility of them failing permanently when the ROMs lose their data.

[Kjelt]

About EPROMS: Is there a way to get their data without desoldering them if they're not socketed?

In theory yes if you can make sure all other parts on the same databus are disabled and in tristate (deselected) you should be able to read it out in circuit.
But if the rest of the circuit is powered you could have a conflict with the microcontroller for instance driving the same CS lines so be carefull and examine the schematic first to identify possible conflicts on pins. If you are able to isolate the Vcc of the eprom and power only the eprom that would make it much easier.

What I usually do with old equipment is desolder the eprom and put sockets in, read it out multiple times (also preferably with different voltages) and see if all reads are binary equal. Then I rewrite the eprom (not erase) just write over it and the contents are refreshed for the next years to come.

[SeanB]

After 24 hours of reforming, and after another day it now has 4.3v across it. When it reaches 2V I will check it again for capacitance on the $20 component tester, seeing as the LCR meter is a little out of range with 20mF.