Products > Test Equipment

When to replace your capacitors?

(1/3) > >>

I recently acquired an Agilent 34410A with an unknown work history. I meticulously examined it, cleaned it, spot-checked items, and conducted some side-by-side measurements.

My guess is that this unit was built around 2006-2007. 

I've seen wisdom on this forum for replacing CAPs at around 10 years, but I know there are manufacturer specifications and so on, too. The truth is, without knowing if the unit sat on a bench and ran 24/7 for the last 15 or more years, it makes me think I should replace the capacitors and be "Safer than Sorry."

I've searched this forum, but maybe I missed the capacitor replacement wisdom thread. Please share your thoughts. I'm leaning toward just swapping them, but that means I'll have to have it calibrated.

Kim Christensen:
I rarely replace components on spec. If it's working to spec, I leave it alone. An exception would be something like old RIFA capacitors with cracks in them.
Otherwise, why do unnecessary rework?

If it ain't broke or known to trash the whole thing don't fix it.


The lifetime of the capacitor depends on:

The quality of the capacitor and this can even change if a manufacturer change a little in his production line.

The type of circuit where it is used.

How often it is used.

The housing and temperature where it is used.

The place where it is used. (hot/cold humidity, vibration and so on)

So what did you expect?


Depends on duty and life history.  Commodity equipment running near ratings and operated frequently, or continuously, might run out capacitors in some years (or even 91 days, if you get my drift), and regular replacement might well be considered.  But then, keep in mind what this is: commodity equipment should either be bought to last ("prosumer" or better quality?), or replaced with equivalents as needed (since, that's the point, it's a commodity, funge it all you want, right?).

Professional equipment, test equipment, items operated infrequently (but regularly, so as not to fall into capacitor reforming issues), can last very long indeed, approaching the shelf life of the capacitors themselves (or other things expire in the mean time).  30-50 year old equipment can still have good capacitors in them.

Consider the failure statistics of a piece of equipment.  If one capacitor is failing, maybe that's neither here nor there; maybe it's the most stressed one (e.g. PSU main filter), maybe everything else is fine -- or more precisely, to guess they have a few decades left in them.  Then, replacing the one is fine.  Maybe several are failing, including cooler, lighter-duty parts: they may all be reaching EOL even given the use of higher-quality materials and low operating temperatures: replacing all is likely a good idea, else the repair won't last long before the next failure and so on.  Maybe you test the ESR of everything, and find most are adequate at room temperature, but several are significantly temperature-sensitive and so should be replaced; maybe all are temperature-sensitive, and still adequate for now, but they'll only get worse in the upcoming years and replacement is very much desirable.

Not that ESR or ESR(temp) is all that great of an indicator of aging -- it is at least correlated, but not highly I think.

Also, keep in mind there are very few bad brands.  Mostly, types are poorly chosen for a given application, and any given brand is subject to scorn due to confirmation bias, spurious correlation -- obviously, the cheapest and most popular types are also the most likely to fail, there's more of them out there!  But also, keep in mind the sense of "poor" here.  We as end users or repairers are disappointed by short lifetime parts, but good engineering is concerned with optimal overall design, which can include compromises between many positive and negative variables.  And manufacturing is primarily concerned with cost.

Often a manufacturer skimps on ratings, or chooses a cheaper / general-er-purpose type where an SMPS type would otherwise be preferred, etc., to eke out that tiniest of profit margin, or perhaps more controversially, for planned obsolescence.  But whatever the case, it's largely a matter of industrial process control and design intent or budget.  The main exception of note being the mid-2000s capacitor plague, a result of formulation errors from chemical suppliers (I think it was?).  It's perfectly normal to have e.g. Capxons last for decades (I have a 2008 laptop that's still running fine with Capxons in its power supply, very light duty mind, continuous operation but nearly idle), and also to run out in months (short-lifetime design).

My guess is, test equipment from the 2000s, if not subject to plague (and it's probably failing already if that's the case?), will be fine for some decades hence.

Point of reference, from my own experience: my '90s TDS460 (not -A) suffered from bad SMT capacitor seals, corroding the boards in places; I recapped it to prevent further damage (and there were some electrical malfunctions associated with degrading capacitors).  I also have a Tek 475 that I'd used for quite a long time (roughly early 2000s to 2010ish?), which the capacitors dried out in, and I had replaced them somewhere along the way; vintage, 1973 I think?, so they were well in need.  You'll see exceptions from both sides of the bell curve, of course, but these seem very typical cases: some things are prone to failure for various reasons (poor component quality, poor soldering, etc.), some just get lucky, or survive the dice-roll of life (some components are randomly made with lower impurity levels, and last longer than their brethren in the same product line, or batch even).



[0] Message Index

[#] Next page

There was an error while thanking
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod