Capacitor type confusion -what kind (ceramic, tantalum, polyester..) to get?

[analogix]

Apart from the obvious polarized electrolytic capacitors, when looking at a schematic -how do you know what kind of capacitor to use?
There's ceramic, film, tantalum, polyester, polypropylene, polycarbonate and probably a few I've left out, so do they all do the same thing, or are they used for different applications? And have certain types been replaced with newer and better technology so I can skip certain types altogether?
Searching the web for info mostly explains how they work and gets very technical, so they really don't answer my question.

I'm in the midst of ordering parts for the power supply below and there's no mention of what kind of capacitors the 100nF types should be (upper and lower left of the schematic):


I come across schematics/parts lists like this now and then, so is there a general rule you can go by in order to know what kind you should use, or do you really need some insight as to how the circuit works first?

[Audioguru]

The manufacturer of the LM317 and LM337 was National Semi but it was recently bought by Texas Instruments. The datasheets tell you exactly which types of capacitors to use.

[w2aew]

Interesting - I am in the final stages of preparing my materials for a video on this topic - describing many of the basic types of capacitors and their characteristics and typically application areas.

In your case here - the polarized caps are most likely aluminum electrolytic caps, but could be replaced by tantalum caps in most cases if desired (less leakage, lower ESR but more expensive).  The 100nF caps are likely ceramic caps.  Given the choice, choose ceramic caps with X7R dielectric vs. Y5V or Z5U dielectric (which have terrible stability vs. temperature).  Class 2 ceramic caps like these also have some voltage dependence too.  You can use film caps here if you like (they'll be phyiscally larger and more expensive, but much more stable with temperature and voltage).

[james_s]

It mostly comes down to knowing the general characteristics of various capacitor types so that you know which types are most appropriate for a specific application. For most hobby projects you can use aluminum electrolytic for high values and ceramic for low values unless mentioned otherwise.

[Atheus]

Just to note - voltage regulators have come on a long way too. The LM317 has terrible specs. It drops about 2V over it and it gets VERY hot if you push it near it's 1A (I think... without a heatsink) limit. You'll find that for a few pennies more you an get a nice new design which wires up just the same way. You wouldn't have to change the circuit even.

These are usually referred to as "LDO"s which means low dropout (maybe 10% of the 2V the LM317 does) and it can dissipate 3A+ without a heatsink. Get them in TO220 packages and screw a fat heatsink on and I'd put 5A through them any day. I've prototyped something with cheap LM317s (and blew one up!) and then when I got it working I swapped out the part for a nice new LD1085 (just an example) LDO with a heatsink. All that happened is performance improved. Do check your datasheet of course, but for my one, the 3 pins are the same as LM317 and it was a direct swap.

~Atheus

[mariush]

I agree with Atheus, there are quite a lot of linear regulators that are just as cheap but have better specifications compared to LM317, and have the same pinout. Just the adjustment resistors may have different values.

Speaking of those, I'm not 100% sure but I think the 220 ohm resistors may be a bit too high value in that circuit, the circuit would probably more happy with 100 or 120 ohm resistors there.
In the datasheet , you can find there at page 5 "Minimum load current to maintain regulation" with 3.5mA typical value, max 10mA ... so with these linear regulators you should aim to make them consume always at least 10mA to be sure the output is always regulated.
Note that the led also uses some power, but due to the 1k resistor in series, it's only going to use about 1-2mA.

Some of the current will be used by that feedback, and that's where lower resistor like 100-120 ohm helps  but for extra safety you can also simply connect a resistor between output voltage and ground of each regulator which would dissipate around 10mA as heat ... you have the formula V = I x R so for let's say minimum 4v output and 10mA , you'd place a R = V / R = 4/0.010 = 400 ohm across the output and ground (make it 470 ohm for a standard value). Considering the led will also use 1-2mA and the feedback resistors will also take some current, a larger resistor will probably be Ok to use... something like 1000 ohm would be just fine.

You'd have to be careful to use a properly rated resistor, to make sure it would handle the dissipated power.. you have the formula P = I^2 x R .. so for 10mA and 1000 ohm resistor, the power dissipated would be 0.01 x 0.01 x 1000 = 0.1 watts, so you should use at least a 0.25w or 0.5w resistor.

Also note that you have there 1.5kohm resistor in series with a potentiometer with a 2kohm value.  That would make the minimum resistance value possible in your circuit to be 1.5 kohm and the maximum 3.5 kohm.

The output voltage is set with the formula  V out = Vref x ( 1 + R2/R1) where Vref for LM317 and LM337 is 1.25v ,  R1 is that 220 ohm (which I recommend to lower to 100-120 ohm)  and R2 is the other resistor (or resistor+potentiometer in series)
So your circuit will have the minimum voltage  Vmin = 1.25 x ( 1 + 1500/220) = 1.25x7.8 = 9.75v and the maximum would be Vmax = 1.25x(1+3500 / 220) = 1.25*16.9 = 21v
If you want larger adjustment range, you should use lower minimum resistor. For example, I'd recommend r1 = 100 ohm and maybe a 220 ohm resistor in series with a 1k potentiometer (easier to find)
this will give you a V min = 1.25 x (1 + 220 / 100) = 1.25 x (1+2.2) = 4v  and Vmax = 1.25 x (1 + 1220/100)  = 1.25 x (1+12.2) = 16.5v ... so your output range would be 4v .. 16.5v

If you want more than maximum 16.5v, you'd use either a larger potentiometer or a bigger resistor in series with the potentiometer (but that will set the minimum configurable voltage higher)


You also used 1n4004 diodes in that circuit. They're good choice, but they may be harder to find compared to 1n4007 for example. The difference between 1n4001 to 1n4007 is just the maximum voltage the diodes can handle... 1n4001 is rated for maximum 50v , 1n4004 is rated for maximum 400v and 1n4007 is rated for maximum 1000v (and slightly lower reverse voltages, 35v for the smallest 1n4001).

Your circuit works with up to around 25v DC so even 1n4001 could be used in your circuit, but the diodes in this series are so cheap and easy to manufacture you're most likely going to find 1n4007 in bigger stocks and at the same price (or lower price) compared to the other models in the series.

Here's a datasheet for the 1n400x series diodes.

Most linear regulators aren't very picky about how much capacitance is the output, so you don't have to use exactly 22 uF which may be harder to find , or may be more convenient other values. For example, you may want to use 10uF , same as you use in other parts of the circuit because it may be cheaper to buy 5-10 capacitors of same value instead of 2-3 pieces of two different capacitance values.
Or, you may find 100uF or 220-470uF capacitors easier than you find 22uF capacitors... just be careful to buy capacitors with voltage rating higher than the maximum voltage you'd want your power supply to output.

SOME linear regulators require capacitors at the output with some particular specifications, like having a minimum ESR value between some range, like 0.1 ohm to 1 ohm. In such cases, datasheets suggest using tantalum capacitors or electrolytic capacitors (because ceramic capacitors typically have ESR values below 0.1 ohm and small capacitance electrolytic capacitors can have ESR values above 1 ohm so since most tantalum capacitors have ESR values within a narrow range of let's say 0.1ohm to 0.6 ohm they're safe and easy to recommend in a datasheet). One such linear regulator is LM1117 (or any with 1117 in the name).

But the thing is some designs for such linear regulators were invented 10-15 years ago, when there wasn't such a wide selection of electrolytic capacitors and the chemical formulas used to produce the substance in electrolytic capacitors weren't as modern as today. Today you can actually buy electrolytic capacitors that would fit those requirements (for example use a 100uF 25v capacitor with ESR value below 1 ohm, which was very hard to find when those datasheets were written years ago).

[Housedad]

Everything I read around here reminds me how much things have changed since I last studied electronics and how much I have to learn.   I did not know about the LDO's  available so easily.  It was 78xx  and LM317 or nothing back then. (very early 80's)

Thanks for the info. 

Now back to the OP's thread....

[james_s]

Just to note - voltage regulators have come on a long way too. The LM317 has terrible specs. It drops about 2V over it and it gets VERY hot if you push it near it's 1A (I think... without a heatsink) limit. You'll find that for a few pennies more you an get a nice new design which wires up just the same way. You wouldn't have to change the circuit even.

These are usually referred to as "LDO"s which means low dropout (maybe 10% of the 2V the LM317 does) and it can dissipate 3A+ without a heatsink. Get them in TO220 packages and screw a fat heatsink on and I'd put 5A through them any day. I've prototyped something with cheap LM317s (and blew one up!) and then when I got it working I swapped out the part for a nice new LD1085 (just an example) LDO with a heatsink. All that happened is performance improved. Do check your datasheet of course, but for my one, the 3 pins are the same as LM317 and it was a direct swap.

~Atheus

Are you thinking of switching regulators perhaps? Low dropout regulators are typically linear, provided the input voltage is not too much higher than the output voltage you can get away with less heatsink, but certainly not *no* heatsink if pulling 3A through it. The LD1085 you mention has a typical dropout of 1.3V, that's better than the LM317 but not *that* much better. At 3A you're still going to be dissipating a minimum of 3.9W, that's going to require a heatsink.

If you blew up a LM317 it was probably a knockoff, the real reputable brand LM317s are pretty bulletproof, they shut down if they overheat or are overloaded.

[mikerj]

Just to note - voltage regulators have come on a long way too. The LM317 has terrible specs. It drops about 2V over it and it gets VERY hot if you push it near it's 1A (I think... without a heatsink) limit. You'll find that for a few pennies more you an get a nice new design which wires up just the same way. You wouldn't have to change the circuit even.

The power dissipation of an LM317T is almost exactly the same as any other linear regulator operated under the same conditions, irrespective of how new the design is.  The only slight difference would be down to quiescent current, which is pretty low in the 317 anyway.

[mariush]

No, he meant LDOs which are a "special" type of linear regulators, with very low dropout voltage.

A lot of the high current linear regulators have a dropout voltage in the 1.3-2v range because they use darlington transistors or similar circuits on the output, so you always have 2 x 0.5-1v voltage drops.

LDOs are regulators which use single npn trasistors or a mosfet at the output which makes it possible to have very low dropout voltages, there are regulators that need as little as 20-50 mV above the output voltage to be stable.
See regulators like TI LP2951 (i think it uses npn transistor) , microchip mcp1824 (pmos)    , LDFPVR from ST (some mosfet)

the lower voltage drop makes it possible in some situations to lower the input capacitor values (so cheaper, less space used), or allows for less power wasted in the regulators - you could have some designs for example where you need 3.3v or 3.6v from a lithium battery which could be between 3.7v and 4.2v

More modern linear regulators / LDOs also use less current on feedback, some may have compensation stuff in them to make them work with very low esr capacitors on output, have low quiescent current, some have enable/standby pins , power good pins etc etc

[rs20]

Neglecting quiescent current considerations, replacing a high dropout regulator with a low dropout regulator, while leaving the rest of the circuit unchanged (including input voltage), will not reduce dissipation. Hence, saying that you can just replace one with the other and remove the need for a heatsink without further comment is "leaving out some important details" to say the very least.

Yes indeed, using a low dropout regulator allows you to reduce the input voltage (e.g. by modifying the input transformer too), and if you take that extra step, that reduces the actual voltage drop across the regulator, and accordingly the power dissipated and possibly the requirement for a heatsink.

No need for an argument as long as you clearly articulate what's going on.

[Zero999]

Yes, a low drop-out regulator won't save any power, if V_IN-V_OUT is the same.


Many low drop-out regulators will also oscillate when, low ESR, high value capacitors are connected to the output. Don't look at the LDO in isolation. Consider the design as a whole.

[mariush]

It's still tricky with low frequency (50-60hz) transformers.

If the transformer is a low VA one (low power, small size etc) .. the lower VA it is the higher the output variation at low power consumption will be.

So for example, a 12v AC transformer rated for let's say 25-50VA could output as much as 10-15% more at low power consumption - let's say when your device idles and uses 20-50mA. But it could drop down to the nominal 12v AC when you're in the 300+ mA range.

Also, the forward voltage of each diode in a bridge rectifier will vary depending on current flowing through it and the temperature... so at 20-50mA each diode may have only 0.5v..0.6v voltage drop, while at 300+mA the voltage drop could go up to 0.7-1v .. not much of a difference but it can add up.

Then you may have variation in the mains voltage ... for example you have a 230v to 12v AC transformer but you have to keep in mind your mains could be anything between 210v and 240v - for example if you want to sell your product in EU, some countries or even some regions inside some countries have nominal 220v and some have nominal 230v. So basically with the 230v to 12v ratio of  19.16:1 your transformer could output as little as 210v/19.16 = 10.9v AC  or as much as 240v/19.6 = 12.56v , and you add on top of that those 10-15% of variation at low currents...
So the peak voltage could be very high ... 1.414 (sqrt[2]) x vac (at max mains voltage) + 10-15% margin  - 2 x Vdrop in diode will give you the peak voltage which tells you the rating of the "reservoir" capacitors .. but the capacitance will depend on how much maximum current you plan your circuit to be able to output continuously and how much over the output voltage you can afford to be.

For example, let's say i want to make a 3v to 12v adjustable linear power supply, at up to 1A.. so I'm going to use that LDFPVR linear regulator that has a typical dropout voltage of 200mV and up to 500mV at 1A.  For rectification, I can use 4 x 1n5817 which have about 0.5v drop at 1A.
With the worst case scenario of 0.5v dropout voltage, 11v AC mains if there's low mains voltage, I'd want at least 12.5v DC at any time.
So in worst case scenario  Vdc peak = 1.414 x 11v ac  - 2 x 0.5v (bridge rectifier) = 14.55v
so to keep the dc voltage above 12.5v i'd have to use a capacitor  c = 1a / [ 2 x 60 hz x [14.5 - 12.5v ] ) = 1 / 240 = 0.004166 Farads or 4166uF , with a rating of 25v

The alternative would be to use a potentially more expensive 15v AC transformer , which means you'd get a higher peak dc voltage so you won't need as much capacitance but you'd need capacitors rated for 35v instead of 25v which may also be more expensive or uncommon compared to 25v and not to mention that at lower currents, your capacitor would still be configured for enough charge to provide 1A of current,  so at low loads like 100mA the voltage will be quite high,so your linear regulator will produce a lot of heat, so you'd potentially also need slightly larger heatsink (or more heatsink than just a big copper pad on a pcb).

Basically with the 12v AC trasformer you could have 12.5v ..16.5v at the input and get 12v out , while with a 15v ac transformer you'd have 12.5v..20v dc .. so with 20v 12v out at something like 0.5a you'd dissipate 4 watts of heat.
 


note: my example would be bad, because at 240v mains, the peak dc voltage would be about 16.7v which is above the recommended maximum voltage (the absolute maximum for that one is 20v) for the chosen linear regulator.. before even including those extra 10-15% at low loads.

[Atheus]

Just to note - voltage regulators have come on a long way too. The LM317 has terrible specs. It drops about 2V over it and it gets VERY hot if you push it near it's 1A (I think... without a heatsink) limit. You'll find that for a few pennies more you an get a nice new design which wires up just the same way. You wouldn't have to change the circuit even.

The power dissipation of an LM317T is almost exactly the same as any other linear regulator operated under the same conditions, irrespective of how new the design is.  The only slight difference would be down to quiescent current, which is pretty low in the 317 anyway.

Okay, without changing ANYTHING else, you're right. But if I had a target output of 30VDC or something and a 317, I would need to get ~32V from the transformer/rectifier/etc, whereas I might only need 30.2V for the LDO to achieve 30V variable. This makes it more efficient and therefore less wasted energy. Am I right? Serious question - I'm a total noob (almost - I'm a software dev) to this and might WELL be talking complete shit. Please do call me out if I give wrong information or I misunderstand something... please tell me exactly WHAT I misunderstnd too please

Cheers!

~Atheus

[Housedad]

Don't forget to pad it for variances in the input voltage to that transformer.  In the US it can be 110v up to 125V coming from the wall and that is under ideal conditions.

[james_s]

The problem is that you can't typically adjust the input voltage with such fine granularity, you normally will be using an off the shelf transformer which come in a selection of specific output voltages. Then you have to account for the voltage sag that occurs under load, and varying line voltage, so you don't want to cut it too close. You need to make sure that the voltage under worst case conditions, low line voltage and max load is still above the dropout voltage of the regulator. LDOs are most useful in battery powered applications where you want to be able to operate all the way down to the depleted cell voltage without having to add an extra cell and the associated waste in the regulator.

The dissipation (thus heat production) of a linear regulator is calculated by the voltage drop across the regulator and the current passing through the regulator. The voltage drop across the regulator can be slightly lower *if* you are able to lower the input voltage but even under ideal conditions the savings are not exactly revolutionary compared to a 1970s LM317.

Now switchmode regulators, those are another story. With those you can get efficiencies approaching or better than 90% and indeed it is possible to deliver >1A with a considerable input to output voltage delta with little or no heatsink.

[analogix]

It mostly comes down to knowing the general characteristics of various capacitor types so that you know which types are most appropriate for a specific application. For most hobby projects you can use aluminum electrolytic for high values and ceramic for low values unless mentioned otherwise.

Thanks! That's basically what I wanted to know 

I've heard a lot about "recapping" gear, and have gotten to understand that electrolytic capacitors can go bad after several years and need to be replaced now and then (preferrably with high quality replacements from name brands such as Panasonic and Nichicon) which is something I'm looking into as I'm servicing some electronic music equipment from back in the 80s/90s, so I've started replacing their large power supply capacitors for starters.
I just came across this interesting document which discusses how cheap, low quality Chinese capacitors are being used to cut costs, but in the long run prematurely results in failure: how to choose replacement capacitors and resistors.

[james_s]

Yes I've replaced a LOT of bad capacitors. I like to use Nichicon, Rubycon and Vishay capacitors when I can. 105C or higher, longest life that will physically fit. Longer life capacitors tend to be physical larger but it helps that overall caps are smaller physically than in the 80s.

[langwadt]

Interesting - I am in the final stages of preparing my materials for a video on this topic - describing many of the basic types of capacitors and their characteristics and typically application areas.

In your case here - the polarized caps are most likely aluminum electrolytic caps, but could be replaced by tantalum caps in most cases if desired (less leakage, lower ESR but more expensive).  The 100nF caps are likely ceramic caps.  Given the choice, choose ceramic caps with X7R dielectric vs. Y5V or Z5U dielectric (which have terrible stability vs. temperature).  Class 2 ceramic caps like these also have some voltage dependence too.  You can use film caps here if you like (they'll be phyiscally larger and more expensive, but much more stable with temperature and voltage).

don't forget that some tantalum caps have a nasty habit of shorting out and catching fire if they see voltage or current spikes

[analogix]

On the subject of "recapping"; is there an FAQ or beginner's guide on this?
I'm planning on replacing most of the big PSU caps of my synthesizers which are 20-30 years old, just to be on the safe side. Apart from the obvious need of changing a capacitor (bulging top), do you measure them with an ESR meter (I don't have one and never tried it) or do you just take an educated guess (like I'm doing with the mentioned music gear) based on the age of it and replace them one by one in any case?

About the PSU schematic and capacitors: I'll see if I can find some X7R type ceramic capacitors for the 100nF values. For the rest I've found some Panasonic 105C rated electrolytics which according to the "How to choose replacement capacitors and resistors" is the way to go for power supplies.
I understand that there are shortcomings with that PSU schematic but most of the stuff is over my head. Actually the PSU I'm about to build is on a PCB which a project-partner of mine designed as a slightly modified version of it, but I believe the basic idea is still teh same. I'm attaching it here.

So if I understand correctly I should replace the LM317 and LM337 regulator ICs with something else that is pin-pin compatible and get better performance and less heat?

[analogix]

don't forget that some tantalum caps have a nasty habit of shorting out and catching fire if they see voltage or current spikes

It sounds like I should stay away from these despite what has been mentioned about less leakage etc.

So to conclude, if a project hasn't specifically stated the type I should go for electrolytic capacitors (high values and polarity) or ceramic capacitors as this will do fine. And if it won't the type/technology will be clearly stated?

[rs20]

It sounds like I should stay away from these despite what has been mentioned about less leakage etc.

So to conclude, if a project hasn't specifically stated the type I should go for electrolytic capacitors (high values and polarity) or ceramic capacitors as this will do fine. And if it won't the type/technology will be clearly stated?

Sounds like a good summary of my feelings on the matter. However, the more you learn, the more nuanced your decisions will become. For instance, you should, at same point, make yourself aware of the DC bias dependency of certain types of ceramic capacitor (), and the piezoelectric effect which is the topic of many recent EEVBlog videos. You shouldn't just be silly and avoid ceramic caps entirely because of these issues, that'd be throwing the baby out with the bathwater. But it does lead to important improvements to your design, whether it be increasing the nominal value of power supply decoupling caps to compensate for the DC bias effect, or avoiding ceramic caps for the one or two caps along your audio path.

And don't worry about this until you work on mains electricity, but when you do come around to working with mains electricity, you should study X and Y safety capacitors too.

[Zero999]

You shouldn't just be silly and avoid ceramic caps entirely because of these issues, that'd be throwing the baby out with the bathwater.

I'd say the same about tantalum. Yes, they can blow up if used improperly but so can aluminium electrolytics. I've never had a problem with solid tantalum capacitors. They're often more reliable than aluminium capacitors, especially at higher temperatures and in near-vacuum applications.

[james_s]

Those are linear power supplies, there's nothing you can replace those regulators with that will be any better. All else the same, all linear regulators will dissipate exactly the same amount of power as heat. Linear power supplies are also rather easy on capacitors, unless they're leaking or test high ESR I don't even bother to replace them normally. If the equipment has a lot of hours on it or the capacitors are known to fail then yeah replace them but it's not like a switching power supply where the caps have to deal with very high ripple current and high frequencies. Don't start trying to modify things unless you know there is a design deficiency and you know what you're doing. There are a lot of people, especially with audio equipment that do stupid modifications that do nothing to improve the function, people with just enough electronics knowledge to be dangerous start passing around advice.

Ceramic caps rarely fail and don't need to be replaced unless they are bad. Yes tantalums can short out but used properly they're pretty reliable and they don't contain any liquid to leak.

[rs20]

In case the OP is feeling overwhelmed; I should point out one thing -- while professionals might (maybe) agree one the one correct type of capacitor for a given usecase, there might be many perfectly adequate choices which work just fine, with things that you might not care about (size, cost, vacuum tolerance) being the only differences.

For example, suppose a circuit calls for a 10uF, 25V capacitor for supply decoupling. I'd choose MLCCs like these,  which are sized at 3.2 by 1.6mm, and cost around AU$0.30 in one-off quantites. However, you could also choose to use film capacitors like these, which are like 5x the area at 5.7mm x 5.0mm, and cost a whopping AU$13.45 each. If you used the film capacitors, I think your circuit would work perfectly well too, it'd just cost more.

So these aren't always massively important choices where the circuit only works if you make the one right choice.

[tggzzz]

For example, suppose a circuit calls for a 10uF, 25V capacitor for supply decoupling. I'd choose MLCCs like these,  which are sized at 3.2 by 1.6mm, and cost around AU$0.30 in one-off quantites.

... and realise that the capacitance is specified with 0V DC. When running at the rated voltage, the capacitance can fall by 80%! Also, don't forget to look at the capacitance degradation over time.

[Housedad]

I'm starting out again in electronics and just got my bench together.   I bought a whole pile of Electrolytic's, Polyester, ceramic, and tantalum capacitors in huge assortments of values and range of voltages (and other components)  off of Ebay for cheap, just so I could learn about them while I fiddle around experimenting and prototyping  circuits.  Now when I will make something I want to keep, I will buy brand name components from Digikey or Mouser just for that project.   I expect these cheap ones to go bad easily or have crappy specs.   I found the only way to find out is to read a ton first, then have cheap parts laying around to play with not care if I burn them up while I find out if what i read was true or not.

[bitseeker]

On the subject of "recapping"; is there an FAQ or beginner's guide on this?
I'm planning on replacing most of the big PSU caps of my synthesizers which are 20-30 years old, just to be on the safe side. Apart from the obvious need of changing a capacitor (bulging top), do you measure them with an ESR meter (I don't have one and never tried it) or do you just take an educated guess (like I'm doing with the mentioned music gear) based on the age of it and replace them one by one in any case?

If you'll be doing a lot of recapping, get an ESR meter unless your budget (both time and money) isn't a factor. In switching power supplies and devices where temperatures near the caps are very high (e.g., next to heatsinks, power transistors, regulators) will degrade electrolytic caps much more quickly than most other applications. So, for harsh uses/environments you might replace aging electrolytics as preventive maintenance. For other cases, blindly replacing them could just be a waste of money and time, not to mention adding in the potential for accidents such as installing an electrolytic cap backwards.

ESR meters can be expensive, but they don't have to be especially for basic checking during diagnosis and repair. The $10-20 ESR/component tester kits from China work just fine. If a cap is bad the ESR will be obviously high (i.e., the accuracy of the cheap tester won't be a factor in determining good vs. bad).

[bitseeker]

Interesting - I am in the final stages of preparing my materials for a video on this topic - describing many of the basic types of capacitors and their characteristics and typically application areas.

I'm looking forward to your vid, Alan. Hope you're all better now!

[james_s]

If you'll be doing a lot of recapping, get an ESR meter unless your budget (both time and money) isn't a factor. In switching power supplies and devices where temperatures near the caps are very high (e.g., next to heatsinks, power transistors, regulators) will degrade electrolytic caps much more quickly than most other applications. So, for harsh uses/environments you might replace aging electrolytics as preventive maintenance. For other cases, blindly replacing them could just be a waste of money and time, not to mention adding in the potential for accidents such as installing an electrolytic cap backwards.

ESR meters can be expensive, but they don't have to be especially for basic checking during diagnosis and repair. The $10-20 ESR/component tester kits from China work just fine. If a cap is bad the ESR will be obviously high (i.e., the accuracy of the cheap tester won't be a factor in determining good vs. bad).

I bought a Capacitor Wizard about 20 years ago when that was the only game in town. At $179 it was a lot of money from the crappy fast food job I had at the time but it has paid for itself many times over. Now there are circuits out there to build your own very cheaply, or those ebay China ones.

[analogix]

Those are linear power supplies, there's nothing you can replace those regulators with that will be any better.
.......................
Don't start trying to modify things unless you know there is a design deficiency and you know what you're doing. There are a lot of people, especially with audio equipment that do stupid modifications that do nothing to improve the function, people with just enough electronics knowledge to be dangerous start passing around advice.

I don't want to be one of those people, which is why I'm asking for advice here 
So I should really just leave the PSU as it is then and build it with the components/values as in the schematic, and not replace the LM317 with an LD1085 as suggested earlier?

[analogix]

ESR meters can be expensive, but they don't have to be especially for basic checking during diagnosis and repair. The $10-20 ESR/component tester kits from China work just fine. If a cap is bad the ESR will be obviously high (i.e., the accuracy of the cheap tester won't be a factor in determining good vs. bad).

And you can check a capacitor's ESR in-circuit (without desoldering them from the PCB), right?
As I understand it, the meter is used to check a capacitor's internal resistance, and if a capacitor shows less than 3 Ohms it's time to replace it. Seems easy enough without needing a Ph.D. in electronices 
Is an ESR meter only used (or needed to be used) with electrolytic polarized capacitors, or any sort of capacitor?
The recapping discussions only seem to be about electrolytics.

Yes, I just checked up on eBay, and sure enough there are ESR meters that are pretty cheap. Will probably be cheaper in any case than replacing caps at random just to be on the safe side 

[cowana]

As I understand it, the meter is used to check a capacitor's internal resistance, and if a capacitor shows less than 3 Ohms it's time to replace it. Seems easy enough without needing a Ph.D. in electronices 

The opposite - the ESR (Equivalent Series Resistance) will rise when the cap is bad.

The larger the capacitor, the lower you would expect the ESR to be - for a small (<10uF) capacitor, 3 Ohms might be a typical value. For a big reservoir cap (say 1000uF), 3 Ohms would indicate a bad capacitor.

[mariush]

And the ESR also depends on the series of capacitors and what they're aimed for.

For example, you may have a 100uF 16v 85c rated electrolytic capacitor in an audio circuit which could have 0.5-1 ohm esr and that would be perfectly fine for that series. The datasheets wouldn't even list the ESR/impedance because you're not expected to use it in a circuit that needs low ESR.
But on the other hand, you could have a Panasonic FR series capacitor of same 100uF 16v rating but 105c and the normal ESR for this capacitor would be below 0.5 ohm esr, so if you measure one at 1 ohm you'd be worried.

Here's panasonic FC datasheet, a low impedance/esr series but quite old and compared to newer low esr series, it's not that low esr : https://industrial.panasonic.com/cdbs/www-data/pdf/RDF0000/ABA0000C1209.pdf
And here's Panasonic FR datasheet, a very low esr series :  https://industrial.panasonic.com/cdbs/www-data/pdf/RDF0000/ABA0000C1022.pdf

you can see 100uF 16v = 0.35 ohm esr for FC series, and it's 0.3 ohm esr for FR series and in a smaller diameter.

Also, you can't just replace a capacitor with one that has very different specifications.
for example, you may have a capacitor on the output of a *1117 linear regulator. This particular series of regulators requires a capacitor on the output with the ESR between 0.1 ohm and 1 ohm.
So let's say you have such a regulator configured to output 5v and it may have a 470uF 10v rated capacitor from  Panasonic FC series and you replace it with a Panasonic FR... you screwed up , the FC had esr of around 0.11 ohm while the new FR series has an ESR of about 0.05 ohm so now your linear regulator may be unstable..

[james_s]

I don't want to be one of those people, which is why I'm asking for advice here 
So I should really just leave the PSU as it is then and build it with the components/values as in the schematic, and not replace the LM317 with an LD1085 as suggested earlier?

Unless you have reason to believe that the existing design is faulty, or you have a bad part and a modern substitute is cheaper or easier to obtain then yes I would suggest using the values on the original schematic. Are you building a replica, or repairing an existing unit? If building a replica then you have a lot of freedom, but I don't see anything particularly wrong with the original design. Linear power supplies with 3 terminal regulators are pretty bulletproof in general.

[bitseeker]

And you can check a capacitor's ESR in-circuit (without desoldering them from the PCB), right?

Yes, but it depends. If the tester uses a low voltage signal, it's less likely to cause active components to alter the measurement. However, you also have to take into account what else is in-circuit with the capacitor. For example, if there's another capacitor in parallel with the one you're measuring, then you have two "resistors" in parallel (each capacitor's ESR is like a resistor) which will result in an ESR reading that is of the pair, not just the one you have the probes across.

So, when in doubt, at least desolder and detach one lead of the capacitor to isolate it from the rest of the circuit.

As I understand it, the meter is used to check a capacitor's internal resistance, and if a capacitor shows less than 3 Ohms it's time to replace it.

Higher ESR is bad. Remember, this is a capacitor, not a resistor. An ideal (i.e., theoretical) capacitor has no ESR. How high is too high? It depends on the capacitor's type, model, voltage rating and capacitance value. Check the manufacturer's datasheet to be sure. There are also generic tables of maximum ESR, but beware that they can be significantly different since measurement methods weren't always standardized.

Is an ESR meter only used (or needed to be used) with electrolytic polarized capacitors, or any sort of capacitor? The recapping discussions only seem to be about electrolytics.

Any kind. Electrolytics are the most common type of cap to fail in devices, hence all the discussion about them. Contrast that with ceramics, which rarely have to be replaced, but they also have ESR.

Yes, I just checked up on eBay, and sure enough there are ESR meters that are pretty cheap. Will probably be cheaper in any case than replacing caps at random just to be on the safe side 

Exactly, especially if you'll be working on linear power supplies and large audio amps where the caps are bigger and, hence, cost more.

[analogix]

And you can check a capacitor's ESR in-circuit (without desoldering them from the PCB), right?

Yes, but it depends. If the tester uses a low voltage signal, it's less likely to cause active components to alter the measurement. However, you also have to take into account what else is in-circuit with the capacitor.

I'll have to read some more upon how to use an ESR meter -I can see how having one is useful when servicing stuff, but obviously it's not as easy as seeing an "OK" or "Fail" on its display. Also I assume the resistance in a capacitor increases gradually, making it more a question of "how bad is this capacitor, and how much useful life might it have left?".

I looked up the datasheets of some large Panasonic electrolytic capacitors I'm going to order for the power supply (I'd rather get some expensive quality components now than replace them soon anyway) but there was no mention of ESR there. Also, if servicing stuff there will probably be cases of unknown capacitor brands where datasheets are unavailable, so what do you do then when determining if a cap is "dead" or not?

I came across a nice downloadable PDF entitled How to choose an ESR meter, but they were all more expensive than the eBay ones. And there are several to choose from on eBay. Does anyone here have any experience with the cheapo ones and can recommend one (or recommend which ones to stay away from)?

[analogix]

Unless you have reason to believe that the existing design is faulty, or you have a bad part and a modern substitute is cheaper or easier to obtain then yes I would suggest using the values on the original schematic. Are you building a replica, or repairing an existing unit? If building a replica then you have a lot of freedom, but I don't see anything particularly wrong with the original design. Linear power supplies with 3 terminal regulators are pretty bulletproof in general.

My project partner (we're building this 14 band vocoder which was printed in an early 80s electronics magazine) already had several PCBs (professionally made) of the (slightly) modified PSU (the Eagle schematic I attached a few posts back). I intended to build it exactly as the schematic says regarding components and values as I'm not qualified to determine what could possibly be improved upon. In any case I'm sure it'll be miles ahead of the original 80s original PSU for the vocoder (see attachements below). And I'm ensuring I get quality name-brand parts (and not from eBay) for it as a lousy PSU could potentially damage the vocoder which is a big, expensive project.

Getting back to capacitors: in the vocoder parts list it says that some capacitors should be of the polyester and polycarbonate types, but I'm having trouble locating them on the Farnell website where I'm ordering from (I have no problems finding electrolytic, tantalum and ceramic capacitors though, which are the other types used in the project). Are these "film capacitors" or have they been replaced by some other types which are better than what was available back in 1980?

Another thing: I actually started building the vocoder years ago, but never finished it due to lousy home-made PCBs. I'm now awaiting new, professionally made PCBs and intend to desolder many of the components from the old boards instead of buying everything once again. Should I replace all the capacitors even though they haven't/or only barely been used (as in power applied, apart from a few short test sessions)? I'd say the components might be 15-20 years old.
I really look forward to the desoldering process  Fortunately they're all single-sided PCBs and I don't have to worry about not damaging the circuit traces.

[mariush]

Switch mode power supplies usually run at frequencies above the hearing range, which would be above 30kHz.  Most these days are around 60-100 kHz. At such high frequencies, the ESR value of a capacitor is pretty much the same as the impedance listed in datasheets, so you can basically look at the column Impedance / 100 kHz / 20c (or some other positive voltage) 

ESR is not equal to impedance but at high frequencies it's pretty much very close to that value. For repairs and stuff like that, it's good enough.

Also, the cheap ESR meters you find on eBay don't actually measure the "ESR", they calculate something which - again - is super close to the ESR value or the impedance of a capacito... to determine if a capacitor is good or bad, a reported value a few percentages less or more than the actual value won't make a difference. Bad capacitors usually have significantly higher ESR than what's in datasheets.

If you want accuracy, you would buy a LCR meter with ESR measurement function at 100kHz (or at least 10kHz if you're on a budget), but such LCR meters usually cost from 90-100$ and up.   

[james_s]

If it's a big expensive project, one improvement you may consider is adding crowbar protection circuits to the output. An example of these uses a zener diode to trigger an SCR across the supply rails which clamps them and causes the fuse to blow in case the regulator fails shorted and causes the full unregulated voltage on the output, not likely but it can happen.

Polyester and polycarbonate are both examples of film capacitors. Whether you need to use precisely those types depends on the characteristics of the circuit in which they're used.

Personally I would not bother replacing all of the capacitors unless they are leaking or test bad. I have lots of 30+ year old devices that still work fine with the original capacitors. If it's not a switchmode application, electrolytic capacitors will usually last a long time. If it had a lot of hours on it and showed any signs of performance degradation then I would replace all the electrolytics but since they're essentially unused I personally wouldn't, that's up to you though.

[bitseeker]

I'll have to read some more upon how to use an ESR meter -I can see how having one is useful when servicing stuff, but obviously it's not as easy as seeing an "OK" or "Fail" on its display. Also I assume the resistance in a capacitor increases gradually, making it more a question of "how bad is this capacitor, and how much useful life might it have left?".

How fast a capacitor degrades depends on many factors. Heat is a major contributor.

Also, if servicing stuff there will probably be cases of unknown capacitor brands where datasheets are unavailable, so what do you do then when determining if a cap is "dead" or not?

You can either look at a datasheet for a typical capacitor with similar specs (type, voltage, capacitance), refer to one of the many generic tables of ESR values, or if you have access to a bunch of known-good capacitors, measure them with your ESR meter and make a table. The latter is a good way to go since it'll be relative to your equipment.

A bad electrolytic will often have an ESR that is many times higher than the maximum that's considered to be "good". Thus, it'll usually be obvious that it's too high.

And don't forget that if the equipment was made during the Capacitor Plague, you'll want to change the electrolytics to avoid corrosive leaks.

[Mr. Scram]

In your case here - the polarized caps are most likely aluminum electrolytic caps, but could be replaced by tantalum caps in most cases if desired (less leakage, lower ESR but more expensive). 

Don't tantalum capacitors have the disadvantage of causing a dead short when they are overloaded?

[bitseeker]

Tantalums often short when they die and blow up when hit with too much current. I've seen a few pop several feet into the air.

[james_s]

They do tend to fail shorted when they fail, but so do most semiconductors. Overall tantalum capacitors are quite reliable when not abused. There is no perfect capacitor, they all have advantages and disadvantages.

[bitseeker]

Yep, gotta pick the right one for the job.

[Harb]

I think a good video needs to be done to cover this, as a lot of correct answers here are being lost in the noise of techno babble.....

Like a lot of Beginners I think what the OP is asking is what Cap for what App

What do you use for Power supplies.......what cap for RF .........what cap for etc etc etc..........and then the naming convention, given their are so many different types on offer that are called different names etc......ie Electro's as apposed to Aluminum Caps .......red drops / green caps etc etc and the list goes on......I can see why its so confusing to a new player to figure out the mess !!

And add to that what Caps are now considered obsolete and should be just forgotten about, moving onto better technology now.

[nugglix]

Waiting for Alans video on the topic...   

In the mean time one might have a look at:



Cheers

[w2aew]

Here's my video on the subject - very similar to Mr. Carlson's Lab's excellent video:

[Harb]

Here's my video on the subject - very similar to Mr. Carlson's Lab's excellent video:

And a great Video it is too........cheers

[bitseeker]

Great job, Alan. Looking forward to the Class 2 ceramic caps video.

[w2aew]

Great job, Alan. Looking forward to the Class 2 ceramic caps video.

It's on my long list, so don't hold your breath too long....

[bitseeker]

No worries. Just putting my vote in.