Why are good capacitors so expensive, and why don't they make these any more?

[MathWizard]

It's time for me to stock up on some parts, and I'll get a bunch of Polyester/Mylar caps, for audio-RF type projects. Why don't they make them down into the pF level, I mainly see 1nF as the smallest?

And I see polypropylene capacitors are better, and maybe 10x more expensive (at least where I'm shopping). And I read polycarbonate caps are better again, so are polystyrene and polytetrafluorethylene caps. But already with polypropylene caps, there's less values commonly available, and hardly anything under 1nF (maybe the big retailers, but I bet it's expensive).

So for quality through hole caps under 1nF, for low RF/radio work, what do people use these days (if they aren't rich)?

But some of these hardly anyone makes anymore, so why not? and what do they use now then? Is it since everything is SMD these days and they use something else for all that ?

And why is this stuff so expensive anyways ? Is it just lack of competition or is it that hard and expensive to make the chemicals, or make the capacitors ? My memory of organic chemistry would say it's easy and cheap to make these chemicals now a-days.

[tom66]

Depends on the application. For RF work ceramic is plenty fine - and you can get down to sub-pF with ceramics.  And it's fine for many audio applications if you're careful with how you use them.  You generally only need other types of capacitors when you have high stability requirements (limited drift in capacitance with temperature/voltage), reliability requirements (ceramic can be vulnerable to shock often failing short), or specific anti-microphonic requirements.

I suspect that since ceramics can fulfill almost every niche now besides a few very specific ones, there is much less commercial incentive to produce other types of capacitor.  The market moves with technology.  For the few cases where specialist caps are needed the cost will obviously be higher, because there's much higher overheads to produce lower quantities of these things.

[Kleinstein]

For the small values COG ceramic capacitors are working very well. The average C0G cap is more stable than the film capacitors and the losses are about on par with PP or PS. Good ones can even get lower loss - on par with exotic PTFE capacitors.  If the larger TC of the film caps is wanted, there are special ceramis ones with positive and negative TC.
The C0G ceramic is essentially not microphonic - possibly less than film types.
MLCC C0G caps up to some 10 nF are still affordadble and hardly a need for small film capacitors anymore. Very low leakage (sup pA) may be an exotic exception, where one may still want PS capacitors.
For the RF use the size matters and SMD is big plus and for the high frequencies THT is a no go - even in the 1970s they used a kind of SMD (or in plane mount), even with germanium transistors (e.g. AF379) .

The PP and PS film materials are not compatible with the high temperatures of SMD soldering. SMD film capacitors are mainly limited to PPS.
Polycarbonate was OK with a relatively low TC, but otherwise not especially good. AFAIK the essentially olny manufacturer of the foils stopped quite some time ago and hardly any such caps after that.

[T3sl4co1l]

Don't use polyester at RF, the Q is poor.  They're fine as bypass/coupling caps, but mind the current ratings, which are relatively low for the same reason.

That leaves PP as the most likely candidate, which, it is quite good, so that's fine.

Small values are occasionally available, but you may find C0G ceramic the better option.  C0G is essentially ideal; typical parts can have Q over 3000.  It's so high it's hard to measure.  Available down to fractional pF, which, you'd have a hard time even making a film cap as such at those values (it'd be mostly body/lead capacitance, not the dielectric as such).

Values 1nF and up, I use a mix of types, as C0G starts going up in price compared to X7R that remains flat until 100s nF (in low voltages), and, it's nice to have film around when breadboarding (cheap like X7R, none of the nonlinearity; good for audio purposes).  PE or PP is fine for signal purposes, again unless you need high Q.

Large values simply get expensive, no matter the voltage; C0Gs can't be made in low voltages (50V ratings basically mean 10s or 100s nm dielectric thicknesses I think?), and the same is true of film caps.  Yield and production rate suffer; extreme surface area means high chance of dielectric breakdown, and lots of layers means more time spent in production.  Films up to a couple uF and 50 or 100V are still reasonable, but good luck with like 10s or 100s of uF.

Larger values, consider using one of the "poorer" types, as a compromise.  For example, electrolytic can be used at AC by using a bipolar type, or pairs in anti-series (with clamp diodes or a bias supply); tantalum are excellent timing / sampling capacitors for LF/DC purposes (well behaved within rated voltage range, lowish leakage); etc.  Adjust circuit impedances to suit the ratings; for example, electrolytics aren't bad as coupling capacitors, even at fairly high frequencies (MHz) as long as you don't mind the modest and variable impedance (some ohms ESR + ESL; negligible in something like a wideband RF preamp).  Or for LF highpass applications like signal coupling, the nonlinearity of electrolytic isn't much of a problem as significant (AC) voltage is developed across the capacitor only around the cutoff frequency.

Ironically, ceramic are quite good for large values at very low voltages, simply because e.g. 47uF 6.3V (or even 4V) X7R and X5R, say in 1210 chip size, are a commodity item -- used en masse in cellphones, PCs, etc.  Production probably sucks (very thin layers of dielectric, and very many of them), but because it's competitive and at scale, that's just part of the process.

Also mind that, you won't find something cheaper than whatever the minimum material plus production plus amortized capital cost is; but you can always find something more expensive, whether because it's a boutique part, specialized purpose (especially certain approvals?), low production volume, going obsolete (or gone, even -- old stock), or etc.  Compare modest-size ceramic chips (0805 to 1210 package, say) to large-chip or lead-frame types; I don't know the economics of the lead-frame kind, but I assume they're low production and therefore just stupendously expensive in general.  Maybe they come down in quantity, or from China, I don't know.  See also, most anything for military purposes: wet-slug tantalum, hermetically sealed electrolytic; or oddities like poled ceramics.

I suppose there's also the trifecta of large, low-quantity and boutique, for applications like induction heating, power factor correction and pulse generation; the catch with these is, because they're so large, the economics work out better in general -- you're buying a few thousand bucks worth of capacitor anyway, so the fact that it's sealed with love into a hermetically welded metal can isn't the most expensive part of it anymore.  And you occasionally see them come up for cheap on auction, if you need to snag one say for that quarter shrinker project you've always drooled over.

Tim

[TimFox]

Polyester/Mylar vs. polypropylene:
Polyester has a higher dielectric constant than polypropylene, therefore less material and winding are needed than with polypropylene, and the result is physically smaller, all other things being equal.
Unfortunately, polyester is mediocre in dielectric loss and dielectric absorption compared with polypropylene (which is one of the best dielectrics), so polypropylene is preferred for critical applications.
Polystyrene is also good, but melts easily.  Teflon is great, but expensive and hard to find.
Ceramics: NP0/C0G dielectric is also excellent, but the other common materials have higher loss and are non-linear (capacitance decreases with applied voltage).  Those other dielectrics have higher dielectric constant than C0G, so larger C values can be had in small packages, if the higher loss, etc. can be tolerated in the application: read the data sheet for the specific part number (it’s quite interesting).

[MathWizard]

Ok I'll have to spend a bit more and get real COG and NPO ceramics. I'm still sticking to through-hole, and/or copper-clad/deadbug stuff tho. But yeah, I forget how much cheaper and available some SMD stuff is, and that size relates to some frequency effects.

I thought Mylar was better at RF than common ceramics, I knew they are for audio. I've noticed higher output voltages from my low RF circuits with them. I'll compare their ESR and Q next time too.

I'll have to look this stuff up again, but I was thinking the Mylar hold their capacitance better across varying voltages, and so in all my little RF projects lately, I've been using them whenever possible. The ones I have from a real retailer, sure are a lot closer to their rated value, on my LCR meter, than most my cheap ceramics, most of which are old cheap ebay stock.

[Benta]

A small note:
Polystyrene caps are history since 20+ years.. PS film hasn't been produced since then. NOS parts can be found.
PP is an excellent alternative, but prolly not for RF.

[TimFox]

Ok I'll have to spend a bit more and get real COG and NPO ceramics. I'm still sticking to through-hole, and/or copper-clad/deadbug stuff tho. But yeah, I forget how much cheaper and available some SMD stuff is, and that size relates to some frequency effects.

I thought Mylar was better at RF than common ceramics, I knew they are for audio. I've noticed higher output voltages from my low RF circuits with them. I'll compare their ESR and Q next time too.

I'll have to look this stuff up again, but I was thinking the Mylar hold their capacitance better across varying voltages, and so in all my little RF projects lately, I've been using them whenever possible. The ones I have from a real retailer, sure are a lot closer to their rated value, on my LCR meter, than most my cheap ceramics, most of which are old cheap ebay stock.

“Common ceramics”:
Ceramic capacitors get an undeserved bad reputation for audio:  this is due to old disc ceramic capacitors that used bad dielectrics (X7R, Z5U,  etc.j for values above 500 pF or so.  With MLCC construction (SMD or with wires added for TH) C0G can be had above 10 nF at reasonable voltages.
When comparing against ceramic, you must specify the dielectric: they are very different.

Mylar is mediocre for audio; polypropylene is better.  At RF, you worry about the self-resonant frequency;  the parasitic inductance is determined approximately by the total length (including leads) of the capacitor and that makes the resonance with the capacitance.  Below SRF, polypropylene will have lower ESR than Mylar.

[coppercone2]

I think a C0G SLCC would be best for low value decoupling

[MathWizard]

Any recommendations on where to buy parts from in Canada ? I'm just looking on LCSC, I've bought stuff there before, and just in the category "Ceramic Disc Capacitors ", there's hardly any small pF caps in stock. The 3pF caps are rated 1kV, and are ~$0.25 each. Thats crazy, but thats the only 3pF caps they have in there own category.

A lot of parts they say to request a quote, so I'm guessing it's not stuff they really want to deal with.

Same with JFET's, especially through-hole. And the ones I was looking at, some are $3/each, and not really that special.

I'll check Digikey and Mouser again, usually I find stuff more expensive there.

[vk6zgo]

A small note:
Polystyrene caps are history since 20+ years.. PS film hasn't been produced since then. NOS parts can be found.
PP is an excellent alternative, but prolly not for RF.

In earlier days Polystyrene was regarded as "just the duck's guts", & for low values that was pretty true, but as capacitance & voltage rating increased, they became somewhat inductive.

Back in the early 1970s we replaced two cruddy old Hunt's metallised paper 0.47uF capacitors which were part of the "Sub modulator" of a 1959 Marconi high level modulated TV transmitter with two nice, "high tech looking" Polystyrene caps of the same value.

When we put a video sweep signal through it, the result looked like a fish---a nice notch filter.

We had to get some other type, which ended up being two then brand new, but old tech, waxed paper caps, which worked without a hitch till the transmitter was retired in 1991.

[vk6zgo]

; tantalum are excellent timing / sampling capacitors for LF/DC purposes (well behaved within rated voltage range, lowish leakage); etc.
Tim

Back in the day, tants were regarded as the answer to relatively high capacitance value, along with small size problems.
My work made an "in house" "vertical synch detector" which was used with remote analog TV sites.
The Studio could turn the Tx on or off by inserting or removing the vertical synchs from the video signal.

As vertical rate was 50Hz, a RC network was used of the correct time constant, using a nice highish value bead ceramic.
With the first such installation, we started getting "no start" or "sudden switch off" alarms, so headed off on a road trip to the site.
Pulling the device out of the rack & testing it failed to show a fault, so back it went into service, where after around half an hour, it promptly turned the transmitter off.
Out on the bench, it worked perfectly, but with an oscilloscope monitoring various parts of the circuit, we noticed that heating it caused what looked like a radical change in capacitance, hence failure of the D.U.T, & cooling it with "canned cold" brought it back.
We replaced the tant with a hulking great "Polyester greencap" & the device was solid as a rock!

[T3sl4co1l]

JFETs are mostly obsolete now, but you can play with them if you like. I haven't shopped for disc caps that small before, but I can say SMT chips are abundant, *shrug*.

Tim

[Kleinstein]

Disc capacitors can be a bit hard to get, as they are getting less and less used. One may still get lucky and may even consider sources like Ebay for a sets. There is a resonable chance to get NOS instead of fakes as the prices are moderate and NOS can still be around, espeically outside China. The values around 100 pF (what one may need for audio) are still available at the major distributors.
Smaller, more hobby oriented shops may be an option too - though I don't know in America.

For the tiny values there is just no way around SMD, alone from the parasitic capacitance of the wires. A quick hack for a few pF an below are 2 parallel wires, cut to a suitable lenght (~ 10 mm/pF)  - however the typical PVC isolation is a rather poor dielectric.

[Zero999]

JFETs are mostly obsolete now, but you can play with them if you like. I haven't shopped for disc caps that small before, but I can say SMT chips are abundant, *shrug*.

Tim

What are you talking about? Cut open any electret microphone capsule and you'll find a J-FET.

[MathWizard]

Don't ham radio people still use JFET's all the time ? For hobby use with discrete parts, is there a direct successor to JFET's ? How low do discrete MOSFET input capacitance's get ?

And by disc ceramic capacitor's, I'm just talking about dirt common looking ceramic through hole cap's. I'm not liking Digikey, too many parts not sold in small quantities, or not stocked, or overly expensive, and hard to add a bunch of parts at once, from 1 page.

I don't have a good way to cut traces in or out of copper clad boards, so trying to do SMD on that would be hard for now. And I don't really want to go messing with etching for now either. And I like building stuff, and I often have to change my circuit's, and IDK proper SMD/PCB design, and I'm not making anything worth ordering a PCB for.

I should be living near an electronics market, I'd walk down there and stop at a fresh food market aswell.

[T3sl4co1l]

JFETs are mostly obsolete now, but you can play with them if you like. I haven't shopped for disc caps that small before, but I can say SMT chips are abundant, *shrug*.

Tim

What are you talking about? Cut open any electret microphone capsule and you'll find a J-FET.

Not sure if you missed that "mostly" there, but I wrote it for a reason.

Tim

[T3sl4co1l]

Don't ham radio people still use JFET's all the time ? For hobby use with discrete parts, is there a direct successor to JFET's ? How low do discrete MOSFET input capacitance's get ?

Well, sure, but hams aren't nearly as technical as you might like to think; most of them are, pick up a radio, plug in and start chatting.  A few build kits, or even work from schematics -- passed on down over the decades, with often few if any changes; and of them, rarely changes based on core design principles.  There are some quite excellent engineers out there, license in hand, but they are few and far between -- as with any population, the 80/20 rule applies, and you need to apply it several times in succession to get down to the highly technical core.

I hope I'm not being unfair or anything; I had the same surprise many years ago, as I slowly realized some of my favorite tools (COIL.EXE comes to mind, and http://hamwaves.com/antennas/inductance.html nowadays -- COIL is cited on this page, actually) are the exception rather than the norm, and in fact, not representative of the technical prowess of hams in general.  That is, there are many highly-technical practitioners out there who are hams, but the converse is not true.

Another case in point, you also see 2N3866 quite often, which is also mostly obsolete; only Central is making them right now I think -- basically the zombie graveyard of parts too old to die, and you pay for it in kind (several bucks for what's little more than a 2N4401's worth of silicon in a can).

Discrete MOSFETs sadly don't get much smaller than this one, https://www.digikey.com/en/products/detail/rohm-semiconductor/RUM001L02T2CL/4004581?s=N4IgTCBcDaIE4FcC2AGFBGEBdAvkA and it's still a pretty big device at 3.5 ohms; compared to say a 3N170's 200 ohms that's comparable to CD4000 output transistors, or, compare to 74HC's (and most CMOS devices today) 30-70 ohms.

("Big" in the sense of, you put enough MOSFET cells in parallel, you fill up the die reducing Rds(on) while increasing Q's and C's.  A logic gate might be a single cell's worth, if even that.  Good RF MOSFETs are painfully small; so small they're hard to use, even with responsible ESD practices in place!)


The confusion is fundamentally not about device availability, but what is providing that basic functionality anymore.  The world has moved on, no one needs (or wants, in quantity) things like this -- there are better solutions now, integrated ones.  "Better" might be a matter of opinion, but on the mass market, only one matters: cost.

Unfortunately, it doesn't help much with developing an individual understanding of the circuits within those ICs.  But since you're just buying a few to play with, it's not a big cost in the grand scheme of things, even if it means buying old stock or salvaged parts.  Kinda sad I guess, but it is what it is.

So, to wit -- hams don't use JFETs, they use MMICs, integrated analog front ends (even ye olde NE602 is long since obsolete, though there are similar parts still around), and integrated synth tuner-on-a-chip, or SDR (software defined radio) the whole way.  At least, those hams that are either using these (whether they know it or not) in commercial equipment, or those specifically designing boards using them.  Hence the popularity of SDR and high density digital modes in some bands these days, for example.

And to be clear, like, JFETs show up in all kinds of things still -- but by far the mass quantity of them is in ICs, for front ends (e.g., JFET-input amps) and bias circuits.  There are some token applications still, like electrets as above, or some radio front-ends (CPH3910 for example is widely available and would do a fine job at AM BCB), or bespoke things like low-noise or high-impedance amplifiers, electrometers (probably mostly CMOS though?), or etc. custom stuff; but these are all low quantity examples, and only like automotive stuff is going to amount to any kind of production quantity (millions).

And JFETs are completely obsolete for analog switching, variable resistance, mixing, etc.  Most of those applications went down even back in the 70s, when ICs came out to facilitate those functions.  JFETs have been paradigm-leaped a few times over now, as analog discrete solutions (e.g. synth panels) gave way to analog integrated (op-amps, OTAs, BBDs, etc.), to digital integrated (FM synth), to wholly software-defined (e.g. a MIDI library + sound font does exponentially more than an electric piano of the 70s, maybe 80s too, ever could).  Today, mixing is best done by IC (at least matched arrays, but proper precision multipliers are still available, if rather boutique), or again, by wholly digital signal chain; and switching is done by CMOS analog switches (or sometimes, opto-MOS relays).

I should be living near an electronics market, I'd walk down there and stop at a fresh food market aswell.

That would be quite lovely; and they'd likely have an assortment of, not only vintage capacitors to choose from, but a fine deli, fresh bread or pastry, perhaps a flower shop?  Or a book store to sit down and peruse the technical section of?

Alas, those days are mostly over in the US (and, by extension, CA to a similar extent).

Tim

[factory]

Any recommendations on where to buy parts from in Canada ? I'm just looking on LCSC, I've bought stuff there before, and just in the category "Ceramic Disc Capacitors ", there's hardly any small pF caps in stock. The 3pF caps are rated 1kV, and are ~$0.25 each. Thats crazy, but thats the only 3pF caps they have in there own category.

A lot of parts they say to request a quote, so I'm guessing it's not stuff they really want to deal with.

Same with JFET's, especially through-hole. And the ones I was looking at, some are $3/each, and not really that special.

I'll check Digikey and Mouser again, usually I find stuff more expensive there.

As your in Canada, Justradios is an option, they have Mica in 1pF steps from 1pF up to 12pF; https://www.justradios.com/

David

[jonpaul]

0.3 to 1000 pF best quality are old dipped silver micas, avail in 1%..5% tol, VERY low drift, inductance, dissipation factor.

Useable at RF...VHF. Sampling, audio

At Marantz in 1960s and Sequerra 1970s we used Mial polystyrene caps  for fine audio,

High power SMPS used special low ESR plastic caps with high current ratings and 0.250" AMP tabs.

Jon

[mawyatt]

A good source for various capacitor types (surplus), they have Mica, Polystyrene, Polypropylene and NPO types, as well as many other components.

https://www.surplussales.com

https://www.surplussales.com/homenew.html#Capacitors

Best,

[CatalinaWOW]

Perhaps you can take a baby step into SMD by buying/designing a carrier that doesn't have a ton of capacitance or inductance.  It is easy to imagine something that would end up resembling a disc capacitor. Not elegant, but lets you do SMD on a very small scale to start.

[Zero999]

C0G capacitors are available in through hole. There's no reason not to use them.
https://content.kemet.com/datasheets/KEM_C1045_AXIAL_MOLDED_C0G.pdf

I would just use surface mount, even on a strip board. Part sizes from 603 to 1210 (805 is optimum) fit nicely between the tracks. There's no reason not to.

[TimFox]

MLCC C0G ceramic capacitors are, indeed, available for TH mounting.
Usually, they are a 0805 or similar size part with wire leads and coating.
For vacuum tube designs or replacement, they may have too low a voltage rating:  the discs used with tubes often were rated at 500 ot 1000 VDC.
Ceramite (originally Sprague, now Vishay) still makes 1000 V NP0 discs, but only up to 270 pF (excellent parts).

[MathWizard]

Perhaps you can take a baby step into SMD by buying/designing a carrier that doesn't have a ton of capacitance or inductance.  It is easy to imagine something that would end up resembling a disc capacitor. Not elegant, but lets you do SMD on a very small scale to start.

Of my SMD parts, 1206 size and sot-23 BJT's seem it fit ok between the pads on my perfboard. And all the parts I checked, matched very close to their rating. So now I'm going to make a common-base Clapp oscillator, and compare it to the through hole part version I made last week.