If you design and/or build your own RF or audio equipment beware that the choice of the type of capacitors you use can have unexpected negative consequences. Here are some examples learned from experience.
For RF circuits type CGO or silver mica ($$$) are good choices. Other caps can drift with temperature and can produce Q killing losses in these circuits. In one design changing from Z5U ceramic discs in a RF filter to CGO ceramics reduced the loss by several percent.
A recently constructed PLL for frequency synthesis worked well but the RF spectrum revealed spurs and sub-harmonics which should not have been there. Measurements with a scope showed 36 nanoseconds of phase jitter when operating at 1670 kHz. The cause of this jitter was the use of a tantalum capacitor in the PLL loop filter. Replacing this with a polypropylene cap resulted in no measurable jitter and a clean spectrum. The tantalum cap was causing a high degree of noise to be injected into the circuit.
An audio phonograph preamplifier design was spoiled by the wrong choice of capacitors. The design used high quality low noise op amps and seemed to work well but a subsonic rumble was discovered when the signal was viewed on a scope. The problem was caused by the multilayer ceramic capacitors used in the circuit and it was gone when these caps were replaced with mylar film and polypropylene types. The lesson learned is that capacitors can produce noise.
One lesser known property of some capacitors such as electrolytics and ceramics is that the capacitance can change with the applied voltage. One such ceramic capacitor was observed to decrease in capacitance by over 50% with a DC bias of 5 volts. Many times this is of no consequence but if, for example, the device is frequency sensitive such as a filter it can be detuned by this effect.
Most of the time capacitors work as intended and one need not obsess about this but if unexpected artifacts, especially noise, are seen in circuit operation then the cause could be the type of capacitor used.
Neil
I remember in my stereo repair days noise being generated by faulty caps. A little spritz of cold from a freaon can resulted in silence.
John, That's interesting. Was the spritz repair permanent?
I think the problems I mentioned with audio applications result from trying to pack too much capacitance into too small a volume such as the tantalum and multilayer ceramic designs. The dielectrics used in these have some strange electrical properties, probably due to the extreme thinness of the materials. Some even show the piezoelectric effect.
Nonetheless, I don't miss the old wax dipped paper caps of yesteryear (even though I made some money replacing them). The wax drippings on the bottom cover of a chassis was usually a giveaway.
Neil
Thanks Neil for this often overlooked information.
Here is a nice tech note on the subject: http://www.millertechinc.com/pdf_files/TN095%20Capacitor%20Noise.htm
The clear winner is the C0G (a.k.a. NPO) ceramic capacitor. COGs are the best for lowest ESR and excellent temperature and voltage stability, but they are only affordable for relatively low voltages up to a couple hundred volts and relatively low capacitances of up to about .1 uF. X7Rs are the next choice for ceramics, but only if you want to cut cost.
Next bst is silver mica. These are the old standard from the tube days because they are available in high voltage ratings. They have a slightly higher ESR than C0Gs.
Next are film capacitors. They have even higher ESR, but are affordable for relatively high capacitance values and high voltage rating. They are good for audio, but not so good for higher RF frequencies. Designers should check the data sheets for frequency limitations.
Long live C0G ceramic capacitors!
This discussion brought to mind an article about capacitors in audio saved several years ago, and I spent this entire evening hunting for it.
Turned out the article is in a filefolder labeled "Electronics," and comes from Radio World Feb. 19, 1997, titled "Hear the Sound of Capacitors by Jim Somich.
Opening paragraph: Capacitors and quality audio have never been the best of friends. Studies claim that each type of capacitor imparts a subtle distortion to audio passing through. Even power supply filter caps contribute distortion because the power supply path is really part of the audio path.
Subsequent discussion is highly interesting, but I'll skip to...
Conclusions: Things you can do to improve the performance of audio devices using electrolytic capacitors -
1) Bridge all power supply filters with large film capacitors, the larger the better. Be sure not to exceed the voltage rating of the film caps.
2) Remove audio coupling electrolytics wherever possible. Many are not needed. Determine this by experimentation. Every electrolytic out of the signal path will improve performance. The determining factor will be the DC offsets present in the circuit without coupling caps.
3) Remove electrolytics used as input and output capacitors wherever possible. Many of these also are unnecessary. Determine by experimentation. Measure the DC output voltage after removing interstage and input/output caps. If the DC offset is under 250mV or so, you will not experience any problems without the caps. But remember that DC offset is cumulative until the next coupling capacitor.
4.) Where capacitors absolutely are necessary, consider replacing them with high quality film caps. This becomes a size and cost problem in some circuits. Metalized film caps are somewhat smaller and cheaper than pure film types. Experiment for the best sound quality. Some brands sound better than others.
If you can totally eliminate one half of the electrolytics in any device, replace a few with film capacitors and bridge the power supply filters, you will reduce hysteresis substantially and improve the definition in the audio device.
End of quotations from article.
Was the spritz repair permanent?
No Neil. Generally the racket came back in a few seconds. Sometimes however, it took several hours.
An automatic spritzer triggered by increasing noise level.
Thanks, all, for adding to the "body of knowledge" on this topic.
John, I thought so.
Phil, The linked article is a nice overview of this.
Carl, The advice in your quoted article "Remove audio coupling electrolytics wherever possible" is something to consider but I have not had problems using these for audio coupling of line level audio. The problem I had was in a magnetic phono preamp where the signal levels were microvolts as opposed to volts and the problem was due to multilayer ceramics and not with the electrolytics. However, electrolytics can change value with applied voltage and this would introduce distortion so it is something to be aware of. This would not be much of a problem if the DC bias was constant though as is often the case "it depends".
Usually it is not necessary to shunt the power rail electrolytics with film caps but there is a good technical reason to do so in RF or high frequency circuits. Large electrolytics often do not effectively pass RF currents and a common design practice is to shunt them a .1 uF and a .01uF film or ceramic capacitors which will bypass the high frequencies. Not doing so can result in unwanted oscillations. This is also good practice for low level audio or instrumentation circuits where high frequency power supply hash can be a problem.
As a matter of practice, I bypass the + and - supplies at each op amp IC with a .01 uF cap (as is commonly done with digital ICs) and have not had noise or instability problems.
Neil
Using the freeze mist is a common trouble shooting aid for finding thermal problems.
Neil shed light on an interesting fact about using electrolytic capacitors in audio paths: "...electrolytics can change value with applied voltage and this would introduce distortion so it is something to be aware of."
The Jim Smich article "Hear the Sound of Capacitors" (2-19-97 Radio World) gives further discussion on this point.
Quoting: An electrolytic capacitor performs at maximum efficiency when it is fully formed, or holding the maximum possible charge of which it is capable. A 100 V electrolytic cap in a power supply running at 15 V never becomes fully formed and operates at sub-par efficiency. In effect, there is less filter capacitance than if the capacitor was fully formed. The hysteresis distortion is related to how deeply the capacitor is formed. The more deeply formed, the worse the hysteresis.
Quote continues: This explains why lower voltage value electrolytics always sounded worse in our listening tests. :End Quote.
A lot to think about with capacitors.