SMD Electrolytic Capacitor: Decoding Capacitance & Voltage Rating

[JDW1]

I have an old accelerator board used in the PDS slot of an Apple Macintosh SE/30. The board was manufactured around 1994. I want to replace the electrolytic capacitors but I'm not sure how to decode the numbers printed on the caps. Here's a photo:



As you can see the numbers are:

22
6 A
7C6

I could take guess and assume that 22 means 22uF, but what about the voltage spec? Would 6A mean 6.3V? Do any of you know? There's no logo mark on the cap that I can see to indicate manufacturer.

Capacitor dimensions are:

Diameter = 4.0mm
Height = 5.9mm (including plastic base)

Thank you.

 

[JonSea]

Google is your friend

 

[JDW1]

I already Googled prior to posting here and came up with those same diagrams, which as you can see do not match up with the printing on my capacitor.

It's highly unlikely the 1st row of text on my cap is a "Lot No." (as indicated by your diagram at left. More likely, your diagram at right is correct about my "22" being 22uF. However, your diagram at right indicates the 2nd row of text is Voltage and my cap says 6A, which isn't listed on that Voltage table. Again, this is why I posted here to see if there might be yet another explanation for what the voltage rating of my cap may be.

If you have further thoughts that go beyond Google, I would appreciate hearing them.

Thanks!

 

[rjenkinsgb]

I'd say it is this variation, using a date code that includes letters:
**broken link removed**

Worst case, the voltage may be confused - 6.3 if it's the 6, 10V if it's the A from the table.

Use a high ripple current rated 22uF 10V and you should be safe either way.

 

[JDW1]

Thanks. I'm actually thinking about using 10V Niobium Oxide 22uF replacements since they will last longer. They are less expensive than tantalums, won't burn, and aren't too much more expensive (at Mouser) than electrolytics in the small quantities I need. ESR is only 700m-ohm too, which is nice. The only caveat is the body size is very tiny, so I need to remove an existing cap and see if the pads beneath are close enough.

 

[rjenkinsgb]

Last longer? The failure rate is given as 0.5% per thousand hours in the data sheet for those...
That seem rather poor to me.

You can get conventional electrolytics rated 5000 hours & higher.

Go up a bit on the voltage rating and the ripple current gets better than the 10V ones and the life at low voltage should be even better, eg. something like these:
https://uk.rs-online.com/web/p/aluminium-capacitors/7270818/

 

[JDW1]

Thank you for alerting me to the failure rate and for the replacement suggestion. Unfortunately, that particular cap you suggest is too large since the the distance from the outer edges of my PCB pads is only 6.5mm. That cap would overshoot the pads by 2.5mm.

The main reason I considered tantalum is because they won't leak over time like the electrolytics. It might take 15 years, but the electrolytics will at some point leak. I was hoping Niobium Oxide caps might be the superior solution as compared with tantalums.

 

[kubeek]

Last longer? The failure rate is given as 0.5% per thousand hours in the data sheet for those...

Doesn´t that mean that the lifetime should be about 100000 hours (to get 50% failure rate)?

 

[rjenkinsgb]

Doesn´t that mean that the lifetime should be about 100000 hours (to get 50% failure rate)?

IF you pick one single component in a product and ignore everything else.

You need to think of the failure rate of an overall system. If you had 100 components like that in a device, it's a 50% chance of failure after 1000 hours, which is roughly 40 days at 24/7

That's a quick route to bankruptcy!

For serious gear that needs to be reliable, the predicted failure rate of any single component should be infinitesimal.

 

[JDW1]

So are we now saying that physically tiny aluminum electrolytic capacitors in 22uF sizes and rated at voltages between 6.3 Vand 10V and in quantities of 5pcs on the same PCB (my application) are more reliable overall and in the long term than other capacitor technologies like tantalum or niobium oxide?

In my experience with vintage Macintosh hardware, the capacitors that leak the worst tend to be the physically small aluminum electrolytic capacitors. The electrolytic capacitors that are physically large, say 3cm in diameter, tend to still work well, without leakage, even to present day, even though they may have been manufactured 30 years ago.

So when considering failure rates, I also need to consider failure modes. I don’t want leaked electrolyte eating through my traces, and I don’t want a capacitor catching fire, and I don’t want the capacitor to fail shorted which may fry components.

Using a 16V rated 22µF SMD solid tantalum capacitor would probably be safe in my application because I see other vintage boards that use 16V rated solid tantalums in these 5V circuits and even after all these years they are still fine. But again, a 22µF 16V tantalum capacitor is getting rather pricey and it’s a bit physically too large for my existing pads, which is why I’ve been pondering alternative technologies like Niobium Oxide caps.

Your further thoughts in light of this would be greatly appreciated. Thank you.

UPDATE-1: According to AVX, their "low ESR NOS devices" have a failure rate of "0.2%/1000 hours" which they say is "more reliable than tantalum capacitors."

UPDATE-2: Page 109 of this AVX Tantalum and Niobium Oxide Capacitors document does in-depth into the failure rate and its calculation. Example calculation is given at the top-right of page 110. All said, we ought not to so easily write-off Niobium Oxide caps as being failure prone.

 

[rjenkinsgb]

It looks like the "0.5% per thousand hours " is their base figure for use in other calculations - but it is just based on running the capacitors at their full ratings continuously.

All the rest of the info is about how much you must de-rate and under-run them in practical designs to obtain a decent service life.
eg. no more than 35-40% rated voltage for high reliability (page 109 fig. 2b) which in turn gives hundreds of times longer life (fig. 2a).

Likewise for current and temperature.


I suppose it's no different to a mechanical part, if you subject it to it's maximum load capability all the time it's not going to be reliable; use something ten times stronger than the load and it should last just about forever.



Without knowing the exact conditions those caps are running at in the circuit, it's guesswork. I alway try to fit the highest rated part I can find (lifetime & ripple current rating etc.) when replacing failed caps.

Looking at the photo, is there room to use a larger wire leaded cap layed down, with the leads formed to fit to the surface mount pads?
That's a fairly easy work-around if there is space. I suspect you could fit tants like that and a 22uF 25V tant bead is around £1, not that bad.


The niobium or tantalum ones should be fine as long as the ratings are appropriate - but those are guesswork...

ps. Your datasheet link did not work for me, I found a direct one:
https://pdf.datasheet.live/9b236ebb/avx.com/TPSD336K035Y0300.pdf

 

[JDW1]

Thanks for sharing your thoughts.

I just finished removing the stock cap shown in my photo and taking precise measurements:

Outer Edge to Outer Edge, Pads + Gap: 6.32mm
1 pad Width: 2.6mm
Pad height: 1.6mm
Gap between pads: 1.12mm

Electrolytic cap dimensions:
H: 5.87mm (including plastic base)
D: 4.04mm
W: 4.27mm plastic base
L: 4.3mm plastic base
Leg-to-leg (outer edges): 4.8mm


Some replacement caps I'm pondering...

Niobium Oxide:
NOSB226M008R1800
NOJB226M010RWB

Aluminum Electrolytics:
EEE-HC0J220R
865230140001
6SVPS22M

Solid Tantalum:
TAJC226K016SNJ

 

[oemcar]

22uF, 6V DC-
I've had good luck substituting tantalums for small electrolytics.

 

[JDW1]

22uF, 6V DC-
I've had good luck substituting tantalums for small electrolytics.

Yes, I have seen my stock parts are indeed 6.3V rated. But you don’t replace electrolytic capacitors with solid tantalum (or even polymer tantalum) rated at the same voltage though.

I thought about polymer tantalum replacements but the prices are higher than niobium oxide and lack the no-burn and won’t-fail-shorted features of Niobium.

I like the NOS-series specs because the failure rate is less than half the NOJ-series. But I can get a 10V rated 22uF NOJ part with ESR of 0.7-ohms, whereas the NOS part is rated at 8V and 1.8-ohm ESR. Part numbers:
NOJB226M010RWB
NOSB226M008R1800

 

[oemcar]

IMO- I like electrolytics for audio pass circuits...

However, I dont subscribe to their proclivity to leak dialectric fluid to the surrounding pc board over time- causing localized circuit damage- as I've witnessed many times at my OEM audio repair shop of 22 yrs experience.

The field I'm hired to consult in requires stability of oscillator networks for GM manufacturing test equipment. Tant's have proven superior stability over time V electrolytics for this app..
Again- JMO