Checking Mouser about 30µF 450V Polypro film caps, I just discover they are not more expensive than 63V caps. I am amazed, it is even the other way around; 450V can be cheaper and there is more choices over 30µF. Size does go up with voltage, as expected, but cost doesn't !!!!!Maybe the designer got them extremely cheap, cheaper than the low-voltage ones?
So, now, I see small size makes them rare and costly.
There can very well be good deals in large over voltage capacitors.
I don't think there's any math for this. It is an empirical phenomenon reported by Douglas Self, Cyril Bateman, others. The lower the ac across the electrolytic, the smaller the distortion ratio.
Jan
DC and amplitude can both have something to do with it: non-linearity in elements is caused by some kind of curvature (= non-linearity).
When you take a small enough portion of the curvature, it ends up looking linear, and that remains valid whether the distortion either increases with signal level (compressive distortion, the most common) or decreases (expansive distortion).
This means that when you work in a restricted enough domain of the characteristic, it will reduce the apparent distortion. This could include AC or DC aspects: offsetting the quiescent point of a class B stage could reduce the apparent Xover distortion, as could reducing the absolute AC voltage seen by a stage.
The rule is valid for well-behaved elements: some may have non-motonic transfer functions or hysteresis, etc., and they will be exceptions, but non-rogue, simple elements like capacitors tend to obey the rule
When you take a small enough portion of the curvature, it ends up looking linear, and that remains valid whether the distortion either increases with signal level (compressive distortion, the most common) or decreases (expansive distortion).
This means that when you work in a restricted enough domain of the characteristic, it will reduce the apparent distortion. This could include AC or DC aspects: offsetting the quiescent point of a class B stage could reduce the apparent Xover distortion, as could reducing the absolute AC voltage seen by a stage.
The rule is valid for well-behaved elements: some may have non-motonic transfer functions or hysteresis, etc., and they will be exceptions, but non-rogue, simple elements like capacitors tend to obey the rule
Yes it does. Its the combination of the DC bias and the actual ac level across it that determines the non-linearity. The result is that for coupling capacitors, the lower in frequency you go the higher the distortion (because the lower the F the higher the AC level across the C).
That is why Douglas Self recommends selecting a cap value that is at least 10 x as large as the value needed for the LF roll off - to keep the distortion low at low frequencies.
Jan
We cannot exclude any option firsthand.
You have the choice: it could be one of the reasons you mentionned, or none in fact.
Why do you assume the designer made a rational choice?
It can happen, but generally designers make a choice based on previous experience (not necessarily relevant), directives or assumptions.
You should be able to question the rationality of that choice, and it is probaby questionnable.
A good test would be to substitute a more normal capacitor, and compare the result with the original part.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.