Eng-Tips is the largest engineering community on the Internet

Intelligent Work Forums for Engineering Professionals

Register Log in

Electrolytic capacitors - low voltage/reduced capacitance 1

Status
Not open for further replies.
Skogsgurra

Skogsgurra

Electrical
Mar 31, 2003
11,815
0
0
SE
I am running a column in the Swedish paper Elektronik i Norden where different topics are covered. A Little like Bob Pease did in Electronic Design.

A question about elctrolytic capcitors and how capacitance changes with voltage was brought up and I ran a quick test.

The result: Yes, there is a reduction when going from near rated voltage down to one tenth of that voltage. The measured change was between 3 and 8% for capacitors in the 1 through 47 uf range with 50 V rated voltage.

Considering that tolerances for these capacitors usually is -20% and +50% I don't think that it matters much.

Question: Does anyone among the readers have any experience where this capacitance variation has been a problem? Any other Comments?

Gunnar Englund
--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.
 
Replies continue below

Recommended for you

Sort by date Sort by votes
The name 'electrolytic capacitors' might not be sufficiently specific. There seems to be a wide variety, presumably with different characteristics.

I'd guess Aluminum, Non-Solid are the common sort as typically used in power supplies.

Purely for convenience of reference, a family tree from Wiki.

320px-E-cap-family-14-02-02.jpg
 
Upvote 0 Downvote
The characteristic that you've found, the change of capacitance with applied voltage, probably has an associated time constant. I doubt that it's strictly instantaneous.

Therefore, this effect would have to be deeply entangled with other well-known 'dynamic' characteristics; such as the effect of frequency on effective capacitance, or non-linear distortion (conceptually within a single cycle).

If you could determine the effective time constant, then you might be able to deduce a physical cause; as well as ensure that the effect is correctly distinguished from other effects that may be complicating measurements.

This point here is basically that the Axis of Time offers so much information that it's worth considering more complex, and more subtle, dynamic (AC) measurements.

The AC characteristics obviously need to drift into alignment with the DC characteristics as the frequency is lowered towards DC.

Obviously this effect has to be compatible with the existing body of knowledge (e-caps' many well known non-ideal effects), somehow integrated in a self-consistent manner.

 
Upvote 0 Downvote
Does it have to do with the voltage effect on forming the insulating layer?

Bill
--------------------
"Why not the best?"
Jimmy Carter
 
Upvote 0 Downvote
I don't know, Bill.

It is the first thought I had, but the effect is instantaneous. At least what is what it looks like when taking a few measurements over a minute and also seems to be reversible. I had a "severe accident" with my little YOGA C930 that I used for the measurements (Don't laugh - cry! I was handling superglue with a certain amount of carelessness and it got sucked into the tiny start button which, of course, got stuck immediately) so I have to rig the measurement anew to find out better what it looks like.

The measurements were taken with 47.5 and 4.75 V DC and the capacitors were charged via a ten k resistor. Time to 30 and 3 V was measured using a 12 bit USB scope and zoomed for best precision.

The forming of heavy electrolytic capacitors is probably akin to the effect - in some way. But that would make capacitance smaller at higher voltages.

BTW: Anyone knows how to release that Start button? Debonder from Great Planes didn't work. I hope there is a Super-Duper or perhaps even a Miracle Debonder out there. Any tips appreciated.

Gunnar Englund
--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.
 
Upvote 0 Downvote
Does anyone among the readers have any experience where this capacitance variation has been a problem?
Click to expand...

No problems ever. No one ever uses electrolytics in applications where 10% even matters. They do the math. "Need this much." Go to the offerings and see +50%/-10% or something similar. Select from the meager standard value offerings one that the -10% (or whatever) still exceeds and call it a day. The cap goes into service and starts drifting from day one. No circuit I've ever seen cared the least. If a non-bulk capacitance matters never use an electrolytic..

But then you know this. I think you may be hanging out with the wrong crowd if someone cares.

PS VE1BLL thanks for letting me poke fun with you. I laughed again at that.

Keith Cress
kcress -
 
Upvote 0 Downvote
I am glad that you laughed at that - and not at my superglue thing.

BTW, I tried the massacre solution, first thing. No luck with that, so I ordered the debonder an waited almost a week for it to arrive.

I now use an older (Ideapad 320S) and it happened to install Firefox - don't ask why. It seems that Firefox and this 2+ yr machine work far better than the YOGA ever did with IE. Isn't that unfair?

Next step is to open the C930. But can't find a single screw. How is it done? Did they use magnets?

Gunnar Englund
--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.
 
Upvote 0 Downvote
Acetone (findernail polish remover) is a commonly used solvent for cyanoacrylate (CA, 'Superglue'). Unfortunately it's also a solvent for some plastics, so you'd need to be extremely careful to test it on an inconspicuous area.

Interesting that nitromethane is listed as the primary agent in some CA debonding solvents.
 
Upvote 0 Downvote
GE said:
...effect is instantaneous. At least what is what it looks like when taking a few measurements over a minute...
Click to expand...

When I suggested looking at the timing, I was thinking in terms of milliseconds.

The non-linearity should reveal itself as a subtle distortion (conceptually within a single cycle).

The timing, possibly measured as variability of distortion over frequency, would provide information.

 
Upvote 0 Downvote
OK, I think that it might show as a second harmonic if I run sub-100 Hz. Will definitely try that. This is getting interesting.

Gunnar Englund
--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.
 
Upvote 0 Downvote
Gunnar. Whenever you start a thread, I KNOW that it will be interesting. Thanks for sharing.

Bill
--------------------
"Why not the best?"
Jimmy Carter
 
Upvote 0 Downvote
No. There's no second harmonic in capacitor current. I went down to 10 Hz without seeing any trace of 20 Hz. Or any other harmonics. Could do 1 Hz, but the time needed isn't worth it.

Still, there is a pronounced change in capacitance vs voltage in the three capacitors I have worked with. The funny thing is that capacitance goes down when voltage goes down. I expected the opposite. And the opposite is what most references show. THAT is a mystery. Pictures:

image_e2ipl4.png


Blue is applied voltage. Red is current.


image_ro970p.png


Current spectrum has a thick noise mat. But no pronounced harmonics.



Gunnar Englund
--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.
 
Upvote 0 Downvote
GE said:
Blue is applied voltage. Red is current.
Click to expand...

From the point of view of the capacitor (in the 'ELI the ICE man' sense), I'd phrase that as follows:

Red is applied current. Blue is the (resultant) voltage.

Your original phrasing left me scratching my head for a minute! :)


 
Upvote 0 Downvote
THAT long?

I applied a voltage from a generator and that resulted in a current. No need for head-scratching. But some head-banging, perhaps?

I picked up where Smoked left. BUT - and this is important: Do not cut the blue wire!

No milliseconds effect seen. So I think that the variation I saw when measuring on a minute scale may have been larger if I measured over longer time. The Al2O3 probably takes some time to change.

Gunnar Englund
--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.
 
Upvote 0 Downvote
Status
Not open for further replies.

Similar threads

jabba99
  • Locked
  • Question
replicating an old circuit- unexpected voltage drop
2
Replies
23
Views
38
IRstuff
IRstuff
xtal01
  • Locked
  • Question
min NiMH charge voltage and extra parts
Replies
1
Views
14
EnzoAus
EnzoAus
Zabee.Khan
  • Locked
  • Question
Voltage step up
Replies
2
Views
7
LionelHutz
LionelHutz
Helepolis
  • Locked
  • Question
Low Voltage System wiring (5-24VDC), Rules of Thumb? 1
Replies
2
Views
9
danw2
danw2
SolarTrap
  • Locked
  • Question
Battery protection low voltage disconnect issue
Replies
14
Views
15
3DDave
3DDave
Back
Top