If you had to pick just a handful of capacitors..

[SkrillBill]

...to always have in stock, what would they be?
I have an assorted box, has most of the values i'd need in 5-pack baggies. There are probably values i should have more than 5 of, since you could use 5 of certain values in one project if youre doing a lot of filtering.

What values should i just always have stocked?

[paulca]

pF: 10, 20, 40, 80, 160, 320, 640
nF: 10, 20, 40, 80, 160, 320, 640
uF: 10, 20, 40, 80, 160, 320, 640

More seriously... em...

100nF definitely.  They are decoupling caps everywhere.
1uF
22pF
100uF

[SkrillBill]

pF: 10, 20, 40, 80, 160, 320, 640
nF: 10, 20, 40, 80, 160, 320, 640
uF: 10, 20, 40, 80, 160, 320, 640

More seriously... em...

100nF definitely.  They are decoupling caps everywhere.
1uF
22pF
100uF

are any of those specific to electrolytic / ceramic?

[schmitt trigger]

The definitive answer is.........drum roll..........is:.................it depends.
It depends on the electronics area you are interested in.

For instance, if your area of interest is mostly microcontrollers and other logic, definitively stock 0.1 uF MLCC ceramics, but also a few 10uf, 10 volt tantalums. Sprinkle them across your board like salt and pepper. The ceramics will also be used in the following scenarios.

On the other hand, if you are interested in analog filter design, you will have to stock at least 4 decades from 100 pf to 1 uF in 1, 2, 5 fashion. (i.e., 100 pf, 200 pf. 500 pf, 1000 pf). These will have to be tight tolerance plastic poly- film types. Voltage of poly-film usually exceeds the highest voltage your circuit will likely provide, so use the lowest voltage rating available, most likely 100 or 200 volts.

Of course, if you are into linear power supplies, electrolytic caps ranging from 100 to 10,000 uF. Voltage rating will depend on your supply, but common voltages are 12, 25, 50 and 63.

[SkrillBill]

The definitive answer is.........drum roll..........is:.................it depends.
It depends on the electronics area you are interested in.

For instance, if your area of interest is mostly microcontrollers and other logic, definitively stock 0.1 uF MLCC ceramics, but also a few 10uf, 10 volt tantalums. Sprinkle them across your board like salt and pepper. The ceramics will also be used in the following scenarios.

On the other hand, if you are interested in analog filter design, you will have to stock at least 4 decades from 100 pf to 1 uF in 1, 2, 5 fashion. (i.e., 100 pf, 200 pf. 500 pf, 1000 pf). These will have to be tight tolerance plastic poly- film types. Voltage of poly-film usually exceeds the highest voltage your circuit will likely provide, so use the lowest voltage rating available, most likely 100 or 200 volts.

Of course, if you are into linear power supplies, electrolytic caps ranging from 100 to 10,000 uF. Voltage rating will depend on your supply, but common voltages are 12, 25, 50 and 63.

Good point, i didn't specify the specific area. 

Micros and logic are where i'm focusing right now. The farthest i'm going into linear power supply is LM7805/7812/317s. Pretty sure 100uF caps are all i need.. that sounds wrong. maybe 0.1uF across the fixed voltage reg.

[lordvader88]

For average ceramics where I think they are +/-20%, hopefully including cheaply made ones, I just order 'assorted' packs for 5/10$ for 500 or 1000 , and don't have to worry for ages since I don't solder up hardly any of them, just breadboarding. Seriously for 5/10$ or so on ebay u can get a package full of all the common sizes.


For electolyics tho, I just spent about $30 on no name cheapo caps for breadboarding. Hope I don't get duds and stuff tho that go off like fire crackers....I have 100s on the way, can't wait.

I ordered by voltage size, so lots of 25V, 50/63V, 100V, and some 200V/450/and higher since I have vacuum tube stuff to play with. And just common 1, 2.2, 3.3, 4.7,10,22, 33,47,68, , I don't remember exactly.


I bought about 1000 assorted green drop style caps, google could not tell me what they are for sure, very common tho, I hate google. For high voltage low capacitance stuff, so mainly 1n,10,22,33,47,68n up to 1000n. And they are like 63V/400?/630V

I really need an ESR meter.

[paulca]

One of my first component purchases were these:
https://www.bitsbox.co.uk/index.php?main_page=product_info&cPath=65_286&products_id=1991
https://www.bitsbox.co.uk/index.php?main_page=product_info&cPath=65_286&products_id=1990

I then bolstered the nano range with 50 100nF ceramics and for fun, a few bags of 10x 100nF polys.

[David Hess]

For high frequency decoupling, I would start with film (better) or X7R ceramic capacitors of 0.01uF and 0.1uF.  For medium frequency analog work like frequency compensation, I would extend this to values in the 1, 2.2, 4.7 sequence from 0.001uF to 1uF.  Extend to the 1, 1.5, 2.2, 2.7, 3.3, 4.7, 5.6, 6.8, 8.2 series if doing a lot of this work.

For bulk decoupling, I would want electrolytic capacitors in the 1, 2.2, 4.7 series from 1uF to 100uF.  These could be aluminum electrolytic or solid tantalum with the later being higher performance.  I would not bother with polyester aluminum or tantalum electrolytic capacitors unless I was building switching regulators or some other specific application that calls for these.

For higher frequency work, I would want C0G/NP0 ceramic capacitors in the 1, 2.2, 4.7 series from 10pF to 1000pF.  Extend to the 1, 1.5, 2.2, 2.7, 3.3, 4.7, 5.6, 6.8, 8.2 series if doing a lot of this work.

[mrpackethead]

100nF definitely.  They are decoupling caps everywhere.
1uF
22pF
100uF
[/quote]

and 1nF

[SkrillBill]

I just wanna say thanks to everyone in this post.  I have a pretty good idea of the caps im gunna keep on hand. My original plan was to go craycray on DigiKek. Fortunately, a friend of mine is selling off a bunch of stuff, including a kit of capacitors. In the meantime, i'm getting these:

https://www.ebay.com/itm/182341004929

If the link is broken or the item is gone, its a 1000pc pack of 50v/0.1uF ceramic disc capacitors.

 

[Zero999]

You'll probably notice that people have already suggested capacitor values like 10, 15, 22, 33, 47 and 68, in nF or µF.

There are standard values, which resistor and capacitors are mostly made in. These are known as the E-series of preferred numbers. The ones mentioned above are the E6, because there are 6 values per decade. The tolerance is of the E6 values is 20%, meaning the nearest value to an exact number, will always be within 20% accuracy. There are also E3 (50%), E12 (10%), E24 (5%), E48 (2%), E96 (1%) and E192 (0.5%) numbers, but most capacitors are generally only available in E6 values (some being limited to E3), with even the most precision units being made in E24 values.

https://en.wikipedia.org/wiki/E-series_of_preferred_numbers

A generic set of E3 values from 10pF, to 10 000µF will do for most applications. If you do some filter or oscillator design, which require a bit more precision, then E6 values between  100pF and 10nF is also handy.

[SkrillBill]

You'll probably notice that people have already suggested capacitor values like 10, 15, 22, 47 and 68, in nF or µF.

There are standard values, which resistor and capacitors are mostly made in. These are known as the E-series of preferred numbers. The ones mentioned above are the E6, because there are 6 values per decade. The tolerance is of the E6 values is 20%, meaning the nearest value to an exact number, will always be within 20% accuracy. There are also E3 (50%), E12 (10%), E24 (5%), E48 (2%), E96 (1%) and E192 (0.5%) numbers, but most capacitors are generally only available in E6 values (some being limited to E3), with even the most precision units being made in E24 values.

https://en.wikipedia.org/wiki/E-series_of_preferred_numbers

A generic set of E3 values from 10pF, to 10 000µF will do for most applications. If you do some filter or oscillator design, which require a bit more precision, then E6 values between  100pF and 10nF is also handy.

That's one thing i was hoping to find out, the 'standard values', which I now know about

I think the kit my friend is giving me is E6, although im not 100% sure. I want to see what all is in that kit first before i buy anything else other than the 0.1uF caps. I would imagine any kit is going to have the standard values to some degree(unless theyve been used up).

[JacquesBBB]

As it is said above,  for capacitor, you can  often restrict yourself to  the E3 series

1, 2.2, 4.7

But the main problem comes with the voltage rating.
Ceramic capacitors come often  rated at 50V, and this  is enough in lot of situations,

But for electrolytics,  the situation is more complicated.
You can decide  to overate your capacitors, and stock only 50V, or 100V,  but this is only vlid for low values.
For larger values ( > 100uF, the size of the cap and its price will change significantly  with the voltage rating )

My advise would be to stock only small (physically)  caps  that are cheap, and get cheaper by quantities (100),
and buy large caps only when needed,  by the piece.

As an example,  for good brand (Nichicon  100uF, 25V ULD1E101MED1TD), you have at Mouser

by      price

1      $0.201
10     $0.137
25     $0.112
100    $0.083
250    $0.074
500    $0.063
1,000  $0.052

So the price drops significantly up to 100 :  P(1)/P(1000) = 3.86   ;  P(1)/P(100) = 2.42

On the opposite,  if you consider   (Nichicon  220uF, 400V, UVY2V221MRD)
the price sequence is different

by      price
1      $4.06
10     $3.66
25     $3.35
100    $2.77
250    $2.64
500    $2.28
1,000  $2.11

So you can buy by the unit.  P(1)/P(1000)= 1.92   ;     P(1)/P(100) = 1.46

[Zero999]

It's possible to make any E12 value from two E3 values or any E24 value from two E6 values.

The tables below are for resistors, but they can be used for capacitors, if they're connected in series, where it says parallel and vice versa, e.g. to make 15nF, you need 22nF and 47nF in series, rather than parallel, as would be the case for resistors. Note that some capacitors come in very wide tolerance ranges, so there'd be no point in making an E12 value, if the capacitors you have are 20% tolerance.

[SkrillBill]

As it is said above,  for capacitor, you can  often restrict yourself to  the E3 series

1, 2.2, 4.7

But the main problem comes with the voltage rating.
Ceramic capacitors come often  rated at 50V, and this  is enough in lot of situations,

But for electrolytics,  the situation is more complicated.
You can decide  to overate your capacitors, and stock only 50V, or 100V,  but this is only vlid for low values.
For larger values ( > 100uF, the size of the cap and its price will change significantly  with the voltage rating )

My advise would be to stock only small (physically)  caps  that are cheap, and get cheaper by quantities (100),
and buy large caps only when needed,  by the piece.

As an example,  for good brand (Nichicon  100uF, 25V ULD1E101MED1TD), you have at Mouser

by      price

1      $0.201
10     $0.137
25     $0.112
100    $0.083
250    $0.074
500    $0.063
1,000  $0.052

So the price drops significantly up to 100 :  P(1)/P(1000) = 3.86   ;  P(1)/P(100) = 2.42

On the opposite,  if you consider   (Nichicon  220uF, 400V, UVY2V221MRD)
the price sequence is different

by      price
1      $4.06
10     $3.66
25     $3.35
100    $2.77
250    $2.64
500    $2.28
1,000  $2.11

So you can buy by the unit.  P(1)/P(1000)= 1.92   ;     P(1)/P(100) = 1.46

I have noticed the price difference myself as well. I was kind of already thinking that i should only need stock of the 'low vlaues' of caps, and as you said stock up on larger ones when they are needed for a project. Most of the caps i've been looking at have ranged from 50v to 200v. I don't actually need that much for project i'm working on now. But at least i have the rating if a project that needs it comes up.

It's possible to make any E12 value from two E3 values or any E24 value from two E6 values.

The tables below are for resistors, but they can be used for capacitors, if they're connected in series, where it says parallel and vice versa, e.g. to make 15nF, you need 22nF and 47nF in series, rather than parallel, as would be the case for resistors. Note that some capacitors come in very wide tolerance ranges, so there'd be no point in making an E12 value, if the capacitors you have are 20% tolerance.

Great info. I didn't know you could do that with capacitors as well. Although something confuses me when you said:
"...to make 15nF, you need 22nF and 47nF"

Did you mean something else here?

[Zero999]

Great info. I didn't know you could do that with capacitors as well. Although something confuses me when you said:
"...to make 15nF, you need 22nF and 47nF"

Did you mean something else here?

Yes, you need 22nF and 47nF in series to make a 15nF capacitor. The table is for resistors. To use it for it for capacitors, the series/parallel column needs to be reversed, so it where it says 22 and 47 P (meaning parallel for resistors), the capacitors need to be connected in series, because the formula for resistors in parallel, is the same as the formula for capacitors in series and vice versa.

The formula for resistors in series is:
R_T = R1 + R2

And for resistors in parallel:
R_T = (R1*R2)/(R1+R2)

For capacitors in series:
C_T = (C1*C2)/(C1+C2)

For capacitors in parallel:
C_T = C1+C2

[SkrillBill]

Great info. I didn't know you could do that with capacitors as well. Although something confuses me when you said:
"...to make 15nF, you need 22nF and 47nF"

Did you mean something else here?

Yes, you need 22nF and 47nF in series to make a 15nF capacitor. The table is for resistors. To use it for it for capacitors, the series/parallel column needs to be reversed, so it where it says 22 and 47 P (meaning parallel for resistors), the capacitors need to be connected in series, because the formula for resistors in parallel, is the same as the formula for capacitors in series and vice versa.

The formula for resistors in series is:
R_T = R1 + R2

And for resistors in parallel:
R_T = (R1*R2)/(R1+R2)

For capacitors in series:
C_T = (C1*C2)/(C1+C2)

For capacitors in parallel:
C_T = C1+C2

Ahh, seeing the formula makes a lot more sense. Thanks for clarifying that!

Now, this might be too long of topic for this post so if you'd rather link me to something, please do. But..

When connected in series, what changes about the ( physics? capacitance?) that the 22nF and 42nF combined equal a lesser end capacitance?

[JacquesBBB]

For a simple capacitor made of two plates,

The capacitance is proportional to the area of the plates (A) and inverse proportional to the distance between the plates (d)

C =  k   A/d

When  you put two caps in series,  you double d, so  you divide the capacitance by two

When you put two caps in parallel, you double A, so you  double the capacitance

See https://en.wikipedia.org/wiki/Capacitance for more.

[Zero999]

Another thing to note is, in a set of capacitors connected in series, the smallest capacitor always sees the highest voltage. In fact, the same formula used for a resistive potential divider, can be used for two capacitors in series, if reciprocal of each capacitor is taken first.


C1 = 47nF
C2 = 22nF
Since both capacitors are in nF and it's only the ratio between them that matters here, we can ignore the nano part, which stops us having to deal with big or small numbers.

Take the reciprocal of each capacitor:
Y1 = 1/C1 = 1/47 = 0.02128
Y2 = 1/C2 = 1/22 = 0.04545

Now use the usual formula for a resistive potential divider:
V(Output) = Y2*V1/(Y1+Y2)

V(Output) = 0.04545*12/(0.02128+0.04545) = 8.173V
https://en.wikipedia.org/wiki/Voltage_divider

The only trouble with the above calculation, is it assumes each capacitor is perfect and doesn't leak any current. In real life capacitors have a leakage current, which be unpredictable and will start to dominate, once the capacitors have charged up. In the previous example, if C2 has a higher leakage current (lower equivalent parallel resistance) than C1, the voltage across C2 will settle to a lower value, than C1 after awhile. If there's no DC present, for example the two capacitors are across a purely AC voltage source, then the above formula will work. If the voltage rating of both of the capacitors is higher, than the supply voltage, and it's not being used as a voltage divider, then the unequal leakage current can be ignored..

If the voltage ratings of the capacitors is below the total supply voltage, then the uneven leakage current could be an issue. The solution is to add voltage sharing resistors across the capacitors. If the ratio of the resistors is the same as the reciprocal of capacitor values, then the steady state and initial voltages will match. In the above example, if a 2.2M resistor is connected across C1 and a 4M7 resistor across C2, then the steady state voltage, will be similar to the initial voltage when the capacitors are rapidly charged.

One short cut which is useful is, if the values of both of the capacitors are the same, then the total capacitance halved and any voltage sharing resistors used can be the same value.

[SkrillBill]

Wow, i'm learning a lot more than i thought i was going to. This is awesome.

For a simple capacitor made of two plates,

The capacitance is proportional to the area of the plates and inverse proportional to the distance between the plates

C =  k   A/d

When  you put two caps in series,  you double d, so  you divide the capacitance by two

When you put two caps in parallel, you double A, so you  double the capacitance

See https://en.wikipedia.org/wiki/Capacitance for more.

Another thing to note is, in a set of capacitors connected in series, the smallest capacitor always sees the highest voltage. In fact, the same formula used for a resistive potential divider, can be used for two capacitors in series, if reciprocal of each capacitor is taken first.


C1 = 47nF
C2 = 22nF
Since both capacitors are in nF and it's only the ratio between them that matters here, we can ignore the nano part, which stops us having to deal with big or small numbers.

Take the reciprocal of each capacitor:
Y1 = 1/C1 = 1/47 = 0.02128
Y2 = 1/C2 = 1/22 = 0.04545

Now use the usual formula for a resistive potential divider:
V(Output) = Y2*V1/(Y1+Y2)

V(Output) = 0.04545*12/(0.02128+0.04545) = 8.173V
https://en.wikipedia.org/wiki/Voltage_divider

The only trouble with the above calculation, is it assumes each capacitor is perfect and doesn't leak any current. In real life capacitors have a leakage current, which be unpredictable and will start to dominate, once the capacitors have charged up. In the previous example, if C2 has a higher leakage current (lower equivalent parallel resistance) than C1, the voltage across C2 will settle to a lower value, than C1 after awhile. If there's no DC present, for example the two capacitors are across a purely AC voltage source, then the above formula will work. If the voltage rating of both of the capacitors is higher, than the supply voltage, and it's not being used as a voltage divider, then the unequal leakage current can be ignored..

If the voltage ratings of the capacitors is below the total supply voltage, then the uneven leakage current could be an issue. The solution is to add voltage sharing resistors across the capacitors. If the ratio of the resistors is the same as the reciprocal of capacitor values, then the steady state and initial voltages will match. In the above example, if a 2.2M resistor is connected across C1 and a 4M7 resistor across C2, then the steady state voltage, will be similar to the initial voltage when the capacitors are rapidly charged.

One short cut which is useful is, if the values of both of the capacitors are the same, then the total capacitance halved and any voltage sharing resistors used can be the same value.

i'm making notes of all this information guys. It's the same as me reading it but having it explained this way seems to make it easier for me to understand because i do.

[Zero999]

I hope you don't mind being a bit off-topic, but the calculator tools linked below are helpful for selecting standard capacitor and resistor values to make a certain value. Again the parallel and series need to be swapped for capacitors, as these tools are only for resistors.
http://sim.okawa-denshi.jp/en/teikokeisan.htm
http://jansson.us/resistors.html
http://www.qsl.net/in3otd/parallr.html

[james_s]

I find I use quite a few 470, 860 and 1000uF capacitors repairing things, occasionally a 330, 100 or 47, mostly 16 or 25V.

For building my own stuff I go through a lot of 100nF decoupling caps, most other values I buy as needed. I also have one of those binders with a zillion different values of 0603 capacitors I can use in a pinch. I almost always forget to order at least one value I need when I go to build something.

[SkrillBill]

I hope you don't mind being a bit off-topic, but the calculator tools linked below are helpful for selecting standard capacitor and resistor values to make a certain value. Again the parallel and series need to be swapped for capacitors, as these tools are only for resistors.
http://sim.okawa-denshi.jp/en/teikokeisan.htm
http://jansson.us/resistors.html
http://www.qsl.net/in3otd/parallr.html

Don't mind at all! A good resource is still a good resource.

I find I use quite a few 470, 860 and 1000uF capacitors repairing things, occasionally a 330, 100 or 47, mostly 16 or 25V.

For building my own stuff I go through a lot of 100nF decoupling caps, most other values I buy as needed. I also have one of those binders with a zillion different values of 0603 capacitors I can use in a pinch. I almost always forget to order at least one value I need when I go to build something.

I've pretty much come to the conclusion that i need to have 0.1uF decoupling caps on hand at all times, a few other values then stock the rest as needed.

[paulca]

I've pretty much come to the conclusion that i need to have 0.1uF decoupling caps on hand at all times, a few other values then stock the rest as needed.

Yep.  You'll always find a circuit somewhere that someone has used a calculation to determine the EXACT capacitor value and specifies something really weird and rare.  Usually you can replace it with a more common value with little to no effect.

I've been at this for only a few months, but the capacitors I have actually used would be:

22pF  Ceramic
100nF Ceramic or poly - almost every circuit to decouple signal stuff
1uF Elec - almost every circuit to smooth/decouple power rails.
100uF Elec
1000uF Elec - power supply circuits to smooth/decouple the output.

I think there was a 47pF once.  The rest have stayed in their bags more or less.

I do need to pay more attention to my voltage ratings on the elecs though, some are 30V, some 16V.  I don't think I've built a circuit that requires more than 15V though.

[mikeselectricstuff]

I could mostly live with the following - very rarely use anything else

100n and 100u through-hole
100n, 1u and 10u ceramic SMD