Are there any arguments FOR better caps?

[chuckd666]

Where's Mr Smallbear when you need him...

[LightSoundGeometry]

Where's Mr Smallbear when you need him...

guess who didnt read the description and has two of them...two silver micas for 1.20 thought it was a good deal ...oh boy 

[12Bass]

Perhaps this gets into semantics, however, I'm not sure what is meant by "better" capacitors.  That said, I think that sometimes a good argument can be made for using different types of capacitors depending upon their function in the circuit.  For example, I've found that the type of capacitor used to roll off harsh highs in a clipping circuit can make a difference in the sound; silver mica or polypropylene caps seem to do a better job of removing the harsh high frequencies than cheap ceramic discs (which could even be microphonic).

[Bill Mountain]

Sorry to sound like a broken record (flat, round, black thing that spins on a turntable to make sound; obsolete-ish), but until you can say what "better" means, you have no chance of getting "better".

Does it mean:
- higher priced?
- from a company name which is bandied about as "better"?
- smaller?
- bigger?
- sturdier? more manly-looking?
- better colors?
- mentioned in many places on the internet?
- tighter tolerance? (and yep, the next question is "if so, what tolerance is needed?")
- higher voltage?
- more syllables in the parts materials names? more esoteric and harder to find materials?
- not made any more and therefore rarer, and therefore more desirable because they don't make 'em like that any more?
- looks more aesthetically pleasing in the pedal gut shots?
- gathers extra mojo points from people reading the descriptions?
- gathers envy from friends/associates?

Put it another way: you have two caps in front of you on the workbench. One is from a highly acclaimed old-line parts maker, the other is a miniaturized unit from a no-name but huge volume manufacturer. Which is "better"? What would you measure to know?

You may feel like a broken record but to me you're always the voice of reason.  When I say "better"  I mean objectively better whether it's tolerance, reliability, noise, etc.  Whatever a group of learned folks would define as better (for a particular application).  I don't just mean it's more expensive.  For example:  I like to use caps that are close to the stated value, are not leaky or noisy, and are of a high enough voltage that I don't have to worry.  Are these things I can get with cheap components? Sure.  Should I stick with name brand stuff to make part selection easier?  Maybe.  This is the type of thing I'm hoping to get my head around.

I wonder what YOU look for when selecting parts for your commercial pedals (specifically caps for coupling and/or filters).  I always trust the old guard when they say there's no such thing as mojo but I still study the choices they make when they build pedals.  I just like to know why I'm making the choices I'm making and not because I saw someone else do it.

[R.G.]

I like to use caps that are close to the stated value, are not leaky or noisy, and are of a high enough voltage that I don't have to worry.  Are these things I can get with cheap components? Sure.  Should I stick with name brand stuff to make part selection easier?  Maybe.  This is the type of thing I'm hoping to get my head around.

Good - you have the right set of questions.

I was extruded a "quality in design and manufacturing" set of coursework back in the 1980s when the Japanese were taking over the electronics world with high-quality products. The single big item that comes out of that was the answer to the question: what does "quality" mean? The answer is:

"Quality" = "meets requirements".

That seems fatuous and limited until you realize that in the electronics design world, "quality" is a technical term. We were - willingly or not! - wrapped around the idea that customer perception of high quality products tended to be that the products just worked and did what they were supposed to. Of course, nothing is perfect, but customers tended to be forgiving of products where the distribution of "good" versus failing products was about six-sigma wide, sigma being the statistical term for standard deviation. You may want to look at "six sigma" in wikipedia. It's enlightening. Quoting that:

A six sigma process is one in which 99.99966% of all opportunities to produce some feature of a part are statistically expected to be free of defects

As you can guess from the number of 9s and decimal places in that number, the air is pretty thin up here.   

The only way to do that is to start with a very carefully crafted statement of requirements. If you don't have very clear set of requirements, you're unlikely to ever meet them. So there was a lot of pressure on us, the original design engineers to have a really, really clear set of product requirements. And that word "requirements" is the technical term I'm referring to.

So this moves the term from "quality" to "requirements". But that's something you can do something about.

One requirement might be "works without repairs for 100 years". In this case, you're in trouble. Humans don't know how to do this yet. NASA can only design for about 40 years without failing on space probes. So that's a low quality requirement - it's not possible.

A requirement for a pedal might be: "normally runs on +9Vdc, but is not damaged by -9Vdc nor the application of +18Vdc". In that case, you can change the schematic to include reverse polarity protection and evaluate the circuit operation when 18V is applied to be sure that even if it doesn't work right at 18V, nothing burns out. That may mean that you have to use all 25V rated caps, different ICs, internal protection circuits, and higher power-rated resistors for certain resistors that produce a lot of power when the circuit is fed 18V.

Example: what requirements are there for a bypass capacitor? Answer: the cap must withstand the required power supply voltages, including any reversals, or over-voltages, for the stated life of the product, and must have a six sigma probability of surviving for the stated life without sudden failure or drifting outside its capacitive tolerance range, ESR or ESL specifications in that time.

OK, that translates to "define the working voltage, overvoltages, and reversals; then define the min and max capacitance tolerances, and ESR/ESL, as well as the lifetime requirement for the product." Only then can you evaluate the capacitor's quality in the application. You evaluate this by taking your sheet of requirements and the datasheet of the capacitor being evaluated and comparing them, line by line, along with reading the cap's datasheet on lifetime and product quality guarantees, as well as the manufacturer's statements on ISO certification and so on. If the maker of the part will not certify it to last six sigma for the product lifetime, you cannot decide that the product meets requirements if it uses that cap.

Example: pulldown resistor. A pulldown resistor needs to withstand the environment and anticipated signal size for the stated lifetime, while not going less or more than the necessary resistance values. Since guitar level and pedal level signals are quite small, and the necessary resistance is just "significantly less resistance than the coupling cap it's pulling down", the requirements to meet on this resistor are pretty trivial, and a quality job of designing a pulldown resistor is probably to pick the cheapest, easiest-to-source part that fits in this huge window. I can hypothesize that in some circuits, a quality pulldown resistor might be a heavy, black pencil mark.

This line of thinking is where you need to be going. Quality is "meets requirements".  Buy things that last far longer, are more sturdy than needed, or are made of fancier materials than needed, and you waste money that could be spent on the parts that are really under stress.

Final thought: is solid gold high quality?

Depends. It's GREAT for making jewelry or resisting corrosion. It's so malleable that it's worthless as a structural material. A solid gold I-beam would be worthless for the purpose. An engineer specifying a solid gold I beam would - and should! - be fired for not understanding requirements and specifying to them. Cheaper components that do meet requirements are the best choice. Knowing what matters and what does not matter as requirements is the hard part. Picking parts once you know what matters is easy.

[12Bass]

Depends. It's GREAT for making jewelry or resisting corrosion. It's so malleable that it's worthless as a structural material. A solid gold I-beam would be worthless for the purpose. An engineer specifying a solid gold I beam would - and should! - be fired for not understanding requirements and specifying to them. Cheaper components that do meet requirements are the best choice. Knowing what matters and what does not matter as requirements is the hard part. Picking parts once you know what matters is easy.

Perhaps it is worth bringing up that engineers and accountants often take a different approach toward component selection.  For example, many of the audio interfaces that I have disassembled use polarized electrolytics as audio coupling capacitors instead of film, likely because film capacitors of that size would be more expensive.  In every case where I have replaced or bypassed an electrolytic with a high quality film capacitor there has been an easily noticeable improvement in fidelity, one which is measurable as lower distortion in loopback testing.

[Bill Mountain]

God Damn! R.G.  Nail on the f'n head.  Every time.

I'm always hung up on the wrong thing.

I like the idea of setting requirements for each component in a circuit.  It may sound nerdy but that actually seems like a fun exercise.

I've looked at Six Sigma before in school (I work in Procurement) but never thought about how it could apply to the stuff I do in my free time.

Thanks again for the detailed post!

[amptramp]

In one contract job I had, we were required to do a tolerance analysis that fed into a stress analysis.  There were some eye-opening things:  If you have two 1% metal film resistors forming a voltage divider and you take into account tolerance, temperature coefficient of resistance and aging, you could end up with a 5% variation in voltage at the divider.  You would pick the tolerance value that gave the worst-case stress and use it in the stress analysis which took into account derated values of voltage, current and power.  Derating, for those who are not familiar with it, is a policy of using devices only up to a certain percentage of their published rating.  For example, you might derate a resistor to 60% so if you had a dissipation of 300mW, you could use a 0.5 watt resistor but if it was 301 mW, you had to move up to the next size which is usually 1 watt.  The same thing was done with resistor voltage - you might have a derating of 80% so a tiny resistor rated at 200 volts could only be used up to 160 volts - above that, you either needed to change the resistor rating or put resistors in series.

Sometimes, you would run into problems where you could not determine what the rating was - what is the inductive kick of a relay turning off?  Since the inductance of a relay when energized is higher than that of a relay that is not energized (since the magnetic pieces when pulled together have less of a magnetic gap than when it is off), this was something that could not always be measured and some snubbing was needed.  I had one case of this that started me doing complete circuit analyses for every board I was doing these calculations for.  One person objected and the program manager defended me, saying that it was a lot cheaper to find problems on paper than after the circuit boards had been laid out.  The guy who objected hired me after this gig because he was having problems with a project at his own company (we were both contract consultants) and recognized that maybe a complete analysis of his stuff was needed on another program.

Once the stress analysis is complete, the reliability analysis is a breeze because you can load your reliability program with accurate stresses.  These programs, such as Relex, automate calculation of failure rates per MIL-HDBK-217F (at least it was still F when we were doing it) and you can get a failure rate in failures per million hours and the reciprocal of failures per hour, the MTBF or Mean Time Between Failures.  Quite often, the company asking for reliability analysis from an outside contractor could not tell the Relex operators what the circuit was doing, so they would have to re-do the analysis, costing time and money and often getting wrong results from wrong analysis.  The way it was done on this job made it fast, and more importantly, correct and you had full documentation of the design.

Some of this may apply to the designs we do here.  Even if you don't need it, it is nice to know how things are done in a well-run aerospace program and you may decide to apply some of the lessons we learned.

[stallik]

Fascinating stuff. Just wondering how build quality factored into this? Or is professional, automated circuit building reliable and constant enough not to need factoring in? Just thinking of my own amateur builds where an otherwise 'reliable enough' circuit can be rendered useless by a poor solder joint

[R.G.]

Perhaps it is worth bringing up that engineers and accountants often take a different approach toward component selection.

That is very much true. Accountants get fired if the dollars don't add up - literally can't be accounted for.    Company executives get fired if the price of the stock goes down. Engineers get fired if the products don't work when delivered, or are delivered late, or can't be manufactured because they specified parts that can't be obtained, or don't make the accountants'/execs' estimates of cost.

An engineer has a professional obligation to tell the accountants and execs when the A&Es have set requirements that can't be met. If you let the accountants tell you to buy components that are too cheap to meet the dollar goals, you're failing as an engineer. If you let the accountants or execs demand that you design something that does not meet their end-product requirements, you're failing as an engineer. If you don't tell the A&Es that you can't get there within the constraints of time, money, and delivered quality, you ought to be fired. If you can't tell whether you have a chance of making all three or not, find another job on your own while you have time.

The designer is the only one on the team that can say that X distortion figure, Y power, and Z cost can or can't be made with existing materials and techniques. The A&Es don't have a clue about that. And whether or not they know it, the A&Es can't do it without the techies to keep them grounded in what's possible - and not.

For example, many of the audio interfaces that I have disassembled use polarized electrolytics as audio coupling capacitors instead of film, likely because film capacitors of that size would be more expensive.  In every case where I have replaced or bypassed an electrolytic with a high quality film capacitor there has been an easily noticeable improvement in fidelity, one which is measurable as lower distortion in loopback testing.

You're confusing performance with quality. If there was no requirement for lower distortion or better frequency response, the added expense of the film caps was wasted. If you demand (er, require...  ) that distortion be below thus-and-such, then you may have to go with film. If you the audio interfaces you have disassembled didn't meet your requirements for low distortion, you have a requirement for a different interface. Obviously, the designers and manufacturers of the interfaces didn't have your needs for lowest loopback distortion as a requirement.

As a designer, if you know, through training and experience, that using more expensive film caps makes distortion lower, you're free to only use audio interfaces that meet your requirements. It's only higher quality if you had a need for it.

Let's talk horses. Is a quarterhorse or a Clydesdale a higher quality horse? Both from the very best bloodlines and breeding, both immaculately fed, maintained and trained. Which is higher quality?

Depends. If you require the horse to pull heavy loads, or haul around a 240lb man with another 150 pounds of plate armor and chainmail, I'm guessing the Clydesdale is going to get the nod. If you are working cattle, you're going to think very little of the Clydesdale's ability to CARRY the silly cow, and opt for the quarterhorse. And neither is going to get around a racetrack like a thoroughbred.  Probably neither will cost as much as a highly-considered thoroughbred, either, but that's neither here nor there.    

Performance is what a person, animal, plant, or thing can just barely do. Quality is whether you picked the one that can meet your needs. As we used to say when we were given a new, even more ridiculous schedule, performance, and quality set to design, if the job was easy, they'd have hired cheaper people to do it.

[PRR]

> what the heck is that thing used for?... at 1kv ?

Traditional Mica was not split thinner than 500V. You don't get more slices per block, you just get more damaged slices.

1KV was another very popular rating. (Mica could be split thick for 10+KV rating, handy in large transmitters.)

Yes, TODAY we can buy Mica caps at 50V. Everything else got small, Mica had to also. But remember most our pedals are Very Vintage designs and constructions. Like re-building 1965 Mustangs to 1965 specs. And a modern 50V Mica would look teensy in an otherwise vintage recreation.

That's very likely just-like a cap somebody used in the 1960s-1970s, because it was standard radio/TV parts, always in stock.

[whoisalhedges]

Unsurprisingly, it seems R.G. has the bases covered here.

Of course there are arguments for better caps - depending on what you mean by better. The "better" input cap for a Rangemaster is different than the "better" input cap for a bass envelope filter, for instance. If you're building an amplifier, 16V electrolytics for power filtering are definitely "worse."

Beyond that... there's plenty of controversy as to whether there's any benefit beyond the obvious (voltage rating, correct capacitance value, microphony) in a circuit like a stompbox. Near as I can tell, the preponderance of evidence suggests that there's no significant audio advantage - not saying that's absolutely true, but it seems to be probably the case - to "mojo" caps.

BUT - even if that's the case... even if that tropical fish, paper-in-oil, shiny Sozo/Roederstein cap doesn't offer any sonic improvement, depending upon your desires, it might still be "better!" Chefs say you eat first with your eyes... stands to reason you hear with your eyes, too. When it comes to buying things, I have a certain look I want; when it comes to building things, too. I often use funky capacitors in my builds - not because I'm going after some elusive sound, I worked that out with greenies on the breadboard already. But I'm putting my time, effort, and money into this; and it's a hobby, not a business. If I'm ever in a position to turn a profit doing this, well, sure: all things being equal, I'll use the cheaper part. But for my overdrive, for instance, it's a Si BJT sandwiched between a pair of FETs... I have those boring old plastic JFETs, and sure I could have used a TO-92 2N4123 (as prototyped) for the BJT; but this guy just looks better: http://smallbear-electronics.mybigcommerce.com/motorola-house-number-6431/ And, since it does the same thing....

So, yes. There are as many reasons to use better caps as there are definitions of "better." There are arguments against it, too. BUT - really, if you're not pressed to turn a profit, I think you should just wire up the circuit you want. The happier you are with the build, simple human psychology would suggest, the happier you'll be with the sound. Make yourself proud! It may not (hell, probably won't) sound any different - but you'll have fun unscrewing your pedal and showing people your work.

[R.G.]

You're correct - if your requirements are "it makes me happy to do this" or "I like these because they're prettier", that's perfectly valid, too.

There are some guidelines from the techie side. Reading a lot of datasheets gets you to where you can make some generalizations. Here are a few of mine:
- don't use electrolytic caps anywhere you want the time constant/filter frequency to be set to anything other than "WAY bigger" or "WAY smaller" than some cutoff frequency. They drift too much, even if you can hand pick one that has the right capacitance today. It won't be that value next week.
- don't count on carbon comp for magic tone; the magic is there, alright, in the form of voltage coefficient of resistance, but you need signals in the 100s of volts to make it prominent.
- film caps do have differences in their secondary and tertiary effects. But don't specify something other than polyester/mylar unless you can write down in one clear sentence what the difference is in your circuit without using the words "better", "best", or "tone".
- same as above for wire; unless you can clearly state exactly why commercial copper hookup wire won't do, use commercial copper.
- capacitors have dielectric absorption; know your circuit to know whether this matters or not. If you don't know what DA is, you have no way of knowing if it matters or not.
- small signal silicon transistors are remarkably, astoundingly alike in how they sound. This is another way of saying that the circuit matters more than the transistor type number.
- use inductors only when you can't do anything else. Likewise transformers.
- it's OK to try to do something that you're sure no one else would ever have thought of - but don't be disappointed if it blows up, melts down, or just sits there mocking you.   

There are others, but this is s good start.

[smarkalet1]

Just because we're on the subject, here's Joe Gore displaying the lack of difference in sound between various caps.
https://youtu.be/dEr-66DR8PM

Sent from my LGMS330 using Tapatalk

[12Bass]

A couple of relevant links:

http://diyaudioprojects.com/mirror/members.aol.com/sbench102/caps.html

http://conradhoffman.com/cap_measurements_100606.html

[Fast Pistoleros]

i've experimented a little with circuits using the cheapest possible parts available; results varied from not working at all to a lot of unwanted noise.

[amptramp]

A couple of relevant links:

http://diyaudioprojects.com/mirror/members.aol.com/sbench102/caps.html

http://conradhoffman.com/cap_measurements_100606.html

I am very familiar with the first link - in fact, I have quoted it in posts here before.  The second one was more interesting, particularly the poor performance of Mylar/polyester.  A lot of my film cap stock is unidentified yellow or tropical fish capacitors, so it would be nice to understand just exactly what I have been using.

[R.G.]

And it would be very interesting to come up with a repeatable electrical test so that
(1) the oddities of a given cap could be measured
and
(2) the oddities of a given dielectric cap could be measured on a statistically significant sample of that type
and
(3) a random cap could be identified by its oddities and the body of measurement data.