This covers the basics : http://www.avx.com/docs/techinfo/dielectr.pdf . Polystyrene doesn't lend itself to metalization and has a low dielectric constant so you need a lot to make a cap, which makes them both expensive and large. Also polystyrene melts at a low temperature and is very hydroscopic, both not good for assembly or life expectancy. However its very stable over time and temperature.
Teflon handles heat and moisture well, but has pinholes forcing the use of several layers when making a cap, and again metallization of Teflon is problematic (doesn't stick) so they are big and difficult to make. It has the lowest leakage and DA making them good for sample and hold applications (which seem to be going away).
NPO (COG) ceramics are not piezoelectric which makes them good for stability and low loss, however low cap value per unit volume.
At microwave frequencies the low loss of Teflon and NPO make them the preferred choice but GHz response in audio is of questionable value (except to marketing).
Polypropylene lends itself to metallization, handles heat better, is not quite as temperature or time stable as the above and much cheaper to manufacture.
None of the above will have distortion much above -160 if implemented well. You are more likely to get distortion from the termination of the foil or the connection to the metallization than from the dielectrics with those materials.
Teflon handles heat and moisture well, but has pinholes forcing the use of several layers when making a cap, and again metallization of Teflon is problematic (doesn't stick) so they are big and difficult to make. It has the lowest leakage and DA making them good for sample and hold applications (which seem to be going away).
NPO (COG) ceramics are not piezoelectric which makes them good for stability and low loss, however low cap value per unit volume.
At microwave frequencies the low loss of Teflon and NPO make them the preferred choice but GHz response in audio is of questionable value (except to marketing).
Polypropylene lends itself to metallization, handles heat better, is not quite as temperature or time stable as the above and much cheaper to manufacture.
None of the above will have distortion much above -160 if implemented well. You are more likely to get distortion from the termination of the foil or the connection to the metallization than from the dielectrics with those materials.
One minor quibble- polystyrene is not very hygroscopic, generally about 5 times less moisture uptake than PET, and similar to that of polypropylene. This is because of the absence of polar groups like the ester moiety in PET.
It is a combination of cost and size and mechanicals of the machinery. A machine is bought to make a certain range of values.... this establishes the physical properties/dimentions of the machine's design and that is forever fixed. So, it can make only a certain maximum diameter size (value). It takes another machine to make a larger size range of values. With polystyrene - large values mean a large physical size making machine is needed. There is little demand for such large size film caps. The demand is zero for them. And at 30-100K USD per machine, it aint going to happen.
THx-RNMarsh
I recall having seen a ~1960 book on dielectrics with a graph of loss factor vs equilibrium moisture content - for polystyrene even the ~0.1% of absorbed water molecules increased the measured loss factor by 5-10x at audio frequency
googling around today mostly gives RF/digital circuit speed measurements
one interesting comment was that polar dielectrics show less effect per water molecule - presumably because they are partially bound by the polar interactions
of course by absorbing a whole lot more the do show greater moisture effect overall
switching and motor run polypropylene get pretty big - somebody is making them
googling around today mostly gives RF/digital circuit speed measurements
one interesting comment was that polar dielectrics show less effect per water molecule - presumably because they are partially bound by the polar interactions
of course by absorbing a whole lot more the do show greater moisture effect overall
switching and motor run polypropylene get pretty big - somebody is making them
Last edited:
Film grade PS will have a moisture uptake of 0.05% or so, compared to 0.03% for polypropylene (using ASTM D570). That's more than an order of magnitude better than PET.
Thanks everyone for the answers to my questions. Demian, that was a very nice set of factors you presented, that answered most of the question by itself. I guess my choice of polypropylene as xo caps was not misplaced. I always used 100V or greater caps and many times these were in parallel so way on the overkill side of things.
Those stranded copper lead caps that John spoke of were made by someone I know, now retired who make them by the customers demand, he thought them rather silly but wasn't going to turn down the money. His name was John Strickland but I can't think of the name or his company right now. It was located in the San Fernando Valley. He sold the company and most of the winding machines and that company is still in business but at a different location. He used all German films for his capacitor manufacturing. I should have taken one of the winding machines when he offered them up but didn't see any need to make my own caps at the time. Silly me!
Those stranded copper lead caps that John spoke of were made by someone I know, now retired who make them by the customers demand, he thought them rather silly but wasn't going to turn down the money. His name was John Strickland but I can't think of the name or his company right now. It was located in the San Fernando Valley. He sold the company and most of the winding machines and that company is still in business but at a different location. He used all German films for his capacitor manufacturing. I should have taken one of the winding machines when he offered them up but didn't see any need to make my own caps at the time. Silly me!
PET, polycarb are hydroscopic. PP, PE and PS are not hydroscopic.
100v caps would be Ok if the power amp power supply was +/- 50v. Otherwise use 200v rated caps in speaker XO.
!00 v metalized is used as a cheap idea...... once your p-p signal (maybe bass) exceeds it, all that happens is some arc over inside occurs and an area of metalization - plate material - is burned off. But no short occurs. Over time, the total plate area is reduced and the XO phase/freq is compromised. Some times just changing out the cap with same type and voltage rating will give an audible improvement.... but it wasnt magic.
BTW - I have known speaker maker to measure 30% loss in C for caps which had very thin metalization on 100mfd film cap due to plate oxidation...... yes, it had stranded leads. caps were OK when bought newly made but after >1 year setting on the shelf (NOS) they measured -30%. Speaker company was also located in Washington state (rains all the time there... think high humidity).
Bi amp/tri-amp or use film and foil for high current apps -- like bass/mid range in XO. But be sure they are rated at 200v or more. because they short (permanently) when over voltaged.
THx-RNMarsh
-RM
100v caps would be Ok if the power amp power supply was +/- 50v. Otherwise use 200v rated caps in speaker XO.
!00 v metalized is used as a cheap idea...... once your p-p signal (maybe bass) exceeds it, all that happens is some arc over inside occurs and an area of metalization - plate material - is burned off. But no short occurs. Over time, the total plate area is reduced and the XO phase/freq is compromised. Some times just changing out the cap with same type and voltage rating will give an audible improvement.... but it wasnt magic.
BTW - I have known speaker maker to measure 30% loss in C for caps which had very thin metalization on 100mfd film cap due to plate oxidation...... yes, it had stranded leads. caps were OK when bought newly made but after >1 year setting on the shelf (NOS) they measured -30%. Speaker company was also located in Washington state (rains all the time there... think high humidity).
Bi amp/tri-amp or use film and foil for high current apps -- like bass/mid range in XO. But be sure they are rated at 200v or more. because they short (permanently) when over voltaged.
THx-RNMarsh
-RM
Last edited:
I think I know who did that 🙂 Funny thing just last night I discovered that my MMT arm was wiggling in its bearing. Thank God for the forums in 5 min I found a blow by blow on a repair where I thought it was toast.
Some very low compliance MCs will shake the bearings loose on even the better arms. HAppened to me with an Audionote Io in a Wheaton IV some years ago. [Also had SME 12" 'chatter' with Ortofon SPUGT/E due to vertical bearing bounce]. The bearings were incapable of staying at the pre-set tension.
Then I built a high mass unipivot….that did the trick; until a friend's brat of a son actually dropped the arm onto a record from quite an elevated angle. Bent the cantilever, but fortunately that very nice late Kondo san not only replaced the cantilever and stylus, but also totally rebuilt the cartridge for only £350.00 - including return post costs! Still got it and still a great cartridge.🙂
(Embarrassed) I should have got that right. I think the Polystyrene and moisture issue was more on how they were made for cheap applications. Those variations seemed to be wound with no casing at all and could not be washed in the usual PCB process. Any unprotected cap would have similar issues.
Why be embarrassed? Polymer science is my professional specialty. I have learned one or two things about electrical engineering from you, after all.
Speaking about the measurements and subjective expertize, I’ve found some strange results of measurements of five various amps, ordinary THD and noise IMD distortion coefficient were measured.
The noise measurements were as the following:
“The test signal is generated by white noise generator. The white noise signal is limited by a bandpass filter to 3 .. 12 kHz bandwidth, which is used as an input test signal.
IMD products of the band-limited noise signal after amplifier are selected by low-pass filter (30 Hz – 1,2kHz) and measured with a voltmeter (rms Uc voltage).
Numerical factor of the noise-signal nonlinearity Knoise is expressed in decibels ratio of the Uc to the voltage of reference signal ( Un ) from built-in generator : Knoise=20lg (Uc / Un )”
Amp No 1 2 3 4 5
GNFB dB 78 72 81 19 14
THD % 0,01 0,02 0,01 0,1 0,13
Knoise % 9,8 9,3 10 0,9 0,8
Subjective quality 2 2 1 5 9
The amps No1…4 are SS amps, the No5 is tube amp. What is strange, is it real that big growth of distortions for SS amps, to 9,8%, for white noise measurements?
The noise measurements were as the following:
“The test signal is generated by white noise generator. The white noise signal is limited by a bandpass filter to 3 .. 12 kHz bandwidth, which is used as an input test signal.
IMD products of the band-limited noise signal after amplifier are selected by low-pass filter (30 Hz – 1,2kHz) and measured with a voltmeter (rms Uc voltage).
Numerical factor of the noise-signal nonlinearity Knoise is expressed in decibels ratio of the Uc to the voltage of reference signal ( Un ) from built-in generator : Knoise=20lg (Uc / Un )”
Amp No 1 2 3 4 5
GNFB dB 78 72 81 19 14
THD % 0,01 0,02 0,01 0,1 0,13
Knoise % 9,8 9,3 10 0,9 0,8
Subjective quality 2 2 1 5 9
The amps No1…4 are SS amps, the No5 is tube amp. What is strange, is it real that big growth of distortions for SS amps, to 9,8%, for white noise measurements?
IPDiA Silicon Capacitors | Mouser
Interesting product. Some lines claim to have X7R capacitance / volume but C0G dielectric performance.
Interesting product. Some lines claim to have X7R capacitance / volume but C0G dielectric performance.
😎🙂 That is a nice test !! -- more like real music being used and similar to a multi-tone test. Easier for anyone to try. Interesting results, too. Can you give more details about the amp topologies?
THx-RNMarsh
Last edited:
Very interesting, the only problem is the cost at 10X the cost of a MLCC (0.1uF 0201 case, probably the most used cap in the world these days) and with a few hundred on the average board!!!
- Status
- Not open for further replies.