Will be getting 200x 22 nF capacitors soon

[Howard]

For the heck of it, I will measure them all and record the data. They are +/- 5% tolerance.

I think I might be able to check the accuracy of my LCR meter this way too. What can I do with the data that might help me with that?

EDIT: 200, not 150

 

[PottedMeat]

What kind of meter is this?

I built a meter like this ( same measurement technique, different PIC ) http://electronics-diy.com/lc_meter.php .

I had a box of caps and inductors and the values seemed to be pretty close to the printed values, but I never really tried to increase the accuracy.

 

[Howard]

Eh, some cheapie off eBay.

 

[Colt45]

I'd wager the accuracy on the meter will be worse than the actual tolerance of the caps.

This is my antique... 1949! 😉
http://www.simpson260.com/303/hickok_209a.jpg

 

[Born2bwire]

Originally posted by: Colt45
I'd wager the accuracy on the meter will be worse than the actual tolerance of the caps.

This is my antique... 1949! 😉
http://www.simpson260.com/303/hickok_209a.jpg

Their logo is awesome.

 

[AlricLantra]

I have to ask . . . . why are you getting 200 22nF caps?

 

[Fox5]

Originally posted by: Colt45
I'd wager the accuracy on the meter will be worse than the actual tolerance of the caps.

This is my antique... 1949! 😉
http://www.simpson260.com/303/hickok_209a.jpg

Looks like something out of fallout.

 

[fornax]

Assuming that the deviations from the true capacitance are random and are normally-distributed (which is by no means guaranteed), this is what you can do:

Plot a histogram of the values that you measured. Then take two measurements:

(1) the central value of the distribution
(2) the width of the distribution

If the average (middle) is not 22nF, then either your LCR meter is systematically off, or the production line was off

If the width is more than 5%, then either your meter has large random errors, or the batch you got was defective.

 

[bobsmith1492]

Originally posted by: fornax
Assuming that the deviations from the true capacitance are random and are normally-distributed (which is by no means guaranteed), this is what you can do:

Plot a histogram of the values that you measured. Then take two measurements:

(1) the central value of the distribution
(2) the width of the distribution

If the average (middle) is not 22nF, then either your LCR meter is systematically off, or the production line was off

If the width is more than 5%, then either your meter has large random errors, or the batch you got was defective.

Often, capacitors will be manufactured toward the high-end of their tolerance range since their capacitance diminishes with age and use. That's probably not the case, though, with a 5% tolerance part...

 

[Howard]

Originally posted by: AlricLantra
I have to ask . . . . why are you getting 200 22nF caps?

Building an active crossover for dipole speakers.

Originally posted by: bobsmith1492

Originally posted by: fornax
Assuming that the deviations from the true capacitance are random and are normally-distributed (which is by no means guaranteed), this is what you can do:

Plot a histogram of the values that you measured. Then take two measurements:

(1) the central value of the distribution
(2) the width of the distribution

If the average (middle) is not 22nF, then either your LCR meter is systematically off, or the production line was off

If the width is more than 5%, then either your meter has large random errors, or the batch you got was defective.

Often, capacitors will be manufactured toward the high-end of their tolerance range since their capacitance diminishes with age and use. That's probably not the case, though, with a 5% tolerance part...

Isn't that just with electrolytics?

 

[Howard]

21.9 2
22.0 3
22.1 3
22.2 5
22.3 10
22.4 7
22.5 14
22.6 13
22.7 15
22.8 13
22.9 13
23.0 20
23.1 20
23.2 18
23.3 10
23.4 12
23.5 12
23.6 6
23.7 1
23.8 3
23.9 0
24.0 0

mean = 22.9115

 

[Born2bwire]

Okay, now calculate the median, mode, standard deviation, variation and fit the best curve to the data.

 

[NeoPTLD]

0.022uF isn't that small, but when you get to <1,000pF range, the capacitance of leads contribute significantly to the error.