MUFF WIGGLER Forum Index
 FAQ & Terms of UseFAQ & Terms Of Use   Wiggler RadioMW Radio   Muff Wiggler TwitterTwitter   Support the site @ PatreonPatreon 
 SearchSearch   RegisterSign up   Log inLog in 
WIGGLING 'LITE' IN GUEST MODE

Decoupling capacitors, poly-film vs. ceramic?
MUFF WIGGLER Forum Index -> Music Tech DIY Goto page 1, 2  Next [all]
Author Decoupling capacitors, poly-film vs. ceramic?
jonbstevens
Hi Muff Gang,

I have a question about the .1uf decoupling caps that are often used in conjunction with an electrolytic capacitor to decouple a module from its power supply.

From reading around on the internet i understand that ceramic's may be preferable in high frequency (Mhz) applications, but in our humble audio circuits is there any practical reason to choose ceramic over polyester film? or to expect ceramics to do a better job?

Should i be breaking out the solder sucker and replacing the film caps i've used with ceramics?
e-grad
jonbstevens wrote:
Should i be breaking out the solder sucker and replacing the film caps i've used with ceramics?


I don't see any reason for doing so since decoupling caps are no part of the audio path.

Anyhow, ordinary ceramics are not preferable over films caps. NP0 type ceramics might do marvels yet more often ceramics are chosen if small value caps are needed.
Dr. Sketch-n-Etch
Concur. For decoupling, it doesn't matter that much. I use 10uF electrolytics and 100nF multilayer ceramics on each rail, plus the occasional additional 100nF pair if I have a particularly noisy (read: digital) chip on the board, or if the board is excessively complex. I've never had any problem with rail noise.
daverj
You might want to get some .1uf ceramics to use in the future, since they're usually a fair amount cheaper and smaller. But there's no need to change the ones you've already used. The films will work fine.
jonbstevens
Thanks for the replies. My oscillators and self oscillating filters are putting a little ripple onto the power rails, at normal levels it's inaudible but when I crank up the gain to unrealistic levels on my mixer I can hear the unpatched oscillator/filter bleeding into the output of other modules.

It's not really a problem, but i guess it's eating at me a little bit. I was thinking about simple fixes -- like the fact that i used film caps instead of ceramic for decoupling. I was also thinking of replacing the 10uf electrolytics with larger tantalum capacitors, 47uf tantalums at Jameco are not too expensive, and adding ferrite beads inline to the +/- 12v inputs to all of my modules. Beyond these things I don't know if there's much else i can do.

Is a little bit of ripple like this normal in all modular synths?
daverj
The ferrites aren't going to help at audio frequencies. They work in the Mhz range. You are probably better off with a 10 ohm resistor and 100uf cap. Even that only gives you about -3db at 160Hz. (-20db at about 1600Hz)

Increasing the large caps on both the source of the problem and the module picking up the problem will help. Large caps on the power distribution board can also help. So will moving the source and destination modules apart on the distro, or to separate distros.
emdot_ambient
I use a lot of polys for decoupling chips, simply because I found them cheap. Just go with whatever's cheaper from your supplier.
ultrashock
e-grad wrote:
jonbstevens wrote:
Should i be breaking out the solder sucker and replacing the film caps i've used with ceramics?


I don't see any reason for doing so since decoupling caps are no part of the audio path.

Anyhow, ordinary ceramics are not preferable over films caps. NP0 type ceramics might do marvels yet more often ceramics are chosen if small value caps are needed.
afaik, 100nF are not easy available at NP0/G0G types. but X7R are pretty widespread
jonbstevens
Thanks again, i've got some larger tantalum caps on order to replace the 10uf aluminum electrolytic's. While i'm working I may just put in the ferrite beads for sake of "completeness".

Anyone care to comment on the value in replacing general purpose aluminum electrolytic capacitors with tantalums in situations like this? Is it worth the trouble? Aside from their tendancy to short closed, and explode, they're better parts right?

Thanks for helping a newb.
J3RK
I use 10uF or 100uF electrolytics for the power filtering (near power input,) and ferrite beads. (or occasionally 1/2W 10R-22R resistors) Then I use (usually) .01uF mylar (Wima red-box style) caps to decouple at chips, or other noise-sensitive locations. I didn't choose those particular ones for any other reason except I can get a small lead spacing, cram them into tight spaces easily enough, and they look cool. Mr. Green w00t I've used ceramic discs before, and the typical green mylars as well, and they all seem to do just fine. I've never tried tantalums because of the shorting thing.
andrewF
jonbstevens wrote:


Anyone care to comment on the value in replacing general purpose aluminum electrolytic capacitors with tantalums in situations like this? Is it worth the trouble? Aside from their tendancy to short closed, and explode, they're better parts right?


On paper tanties look good, in reality I wouldn't let one anywhere near my modular. They are just too fragile.
At least when you have a short on a circuit you can be sure it is the tant, makes for easy trouble-shooting.
Dr. Sketch-n-Etch
I made an executive decision when I started putting my modular together to go with all 10uF electrolytics instead of 100uF because I didn't like the idea of putting so many large caps in parallel on the rails. However, it's easy enough to try the larger caps and see if this eliminates your ripple. The 10-ohm resistor will probably do more to help. I also concur on the ferrite beads: they're pretty useless at audio frequencies.
megaohm
andrewF wrote:
jonbstevens wrote:


Anyone care to comment on the value in replacing general purpose aluminum electrolytic capacitors with tantalums in situations like this? Is it worth the trouble? Aside from their tendancy to short closed, and explode, they're better parts right?


On paper tanties look good, in reality I wouldn't let one anywhere near my modular. They are just too fragile.
At least when you have a short on a circuit you can be sure it is the tant, makes for easy trouble-shooting.


I agree.
You can get electros that are nearly as good, too.
daverj
Tantalums have low impedance at very high frequencies. Great for digital. For audio rate modules an aluminum in parallel with a ceramic will work fine as bypass. Tantalums will not survive even a very short reversal of voltage. When a tantalum blows, they tend to short. When an aluminum blows they tend to go go open.

If you use a tantalum in a modular, you really should use a series diode to protect it from plugging the module in wrong. But if the circuit isn't designed for that, it might cause other issues because of the voltage drop.
Dr. Sketch-n-Etch
I've never used a tantalum in my modular (except perhaps in my very first module, before I became aware of all the hoopla). Other than offering large capacitances in a small package, I don't really see why one would ever want to use them. They may have slightly better specs than electrolytics, but I don't think they're worth the trouble. I certainly haven't missed them.
neil.johnson
As Dave says, ferrite beads are for RF noise rejection, not audio.

At audio frequencies you need bulk capacitance and some reasonable resistance to get any useful rejection. Local per-chip decoupling (100n typical) is to ensure op-amp stability by reducing the effective inductance of the supply rails.

First, consider DC. Series resistance is limited by voltage drop caused by the DC load current. If we try to limit that to less than 500mV (in Euro land we live on starved 12V rails!) and allow 100mA load then that limits the series resistors to 5 Ohms. Remember, this is the DC drop, not yet considering AC.

Now consider AC. We need an impedance less than 5 Ohms at the lowest frequency we're trying to reject. Lets aim for 100Hz, and an impedance of 5 Ohms as the very upper limit giving -6dB rejection. That gives us 330uF as the nearest preferred value. More is better. Next value up would be 470uF. Also, choose low ESR aluminium electrolytics, such as Panasonic FM.

Now, if you know your module is taking less, say 50mA, you can double the series resistance and either lower the capacitance or benefit from better noise filtering.

Some experimentation may be necessary. As one concrete example, my AHB mixing desk has 22R/100uF per rail on power entry to each channel PCB, which has three TL072 op-amps.

Cheers,
Neil
L-1
In general, film type capacitors are not useful in power supply decoupling applications because they are generally wound, which increases their inductance. They are more often used in audio applications where a very low capacitance vs. voltage coefficient is required.
http://www.analog.com/static/imported-files/tutorials/MT-101.pdf
neil.johnson
In general ... I'd disagree. For example, the popular Arcotronics R82 series of white-box polyester film caps are non-inductively wound, and specified for decoupling applications. Datasheet here:

http://www.rapidonline.com/pdf/10-3272e.pdf

Or maybe something a bit cheaper:

http://www.rapidonline.com/Electronic-Components/Metallised-polyester- capacitor-71096

Again, non-inductive wound.

RS, Farnell, Mouser, Digikey all stock similar types of polyester film cap. Arcotronics, AVX, Panasonic, Kemet, and so on all have very informative datasheets.

Two things ceramics do have going for them are much better capacitance density, which can mean a smaller/lighter PCB, and better temperature endurance. Especially so with surface-mount components where the ceramic can withstand the soldering temperatures better than surface-mount film capacitors. On the other hand, beware the ceramic dielectric type: Y5V is very cheap, but make sure you specify a high voltage type otherwise what you think is 100n is actually much much less by the time you apply 12V to it.

So, in general, read the datasheet of the component you're thinking of using.

Cheers,
Neil
rarara
anyone care to link to a photo of decoupling caps fitted? think i have it correct but not sure.

I have power arriving in the form of Pin1:12v, Pin2:Ground, Pin3:6v and am wanting (has been recommended for the purpose) to put an electrolytic and foil cap (in parallel presumably) on each rail. Am I just connecting like this?:

Pin1 ---<foilcap>---Pin2
Pin1 ---<elec.cap>---Pin2

Pin3 ---<foilcap>---Pin2
Pin3 ---<elec.cap>---Pin2
ashleym
Quote:
My oscillators and self oscillating filters are putting a little ripple onto the power rails, at normal levels it's inaudible but when I crank up the gain to unrealistic levels on my mixer I can hear the unpatched oscillator/filter bleeding into the output of other modules.


Are you using screened cables to carry audio from the PCB to the panel? Are your earths nice and low impedance? Are any modules worse than others? Are they the modules next to VCAs?

Just having a think about where the bleed comes from
ashleym
double post. sorry
Eric G
A good rule is to place bulk caps (33 - 220 uF) near the board's power connector. Then place 100nF decoupling caps very near circuits that handle switching, i.e. high frequesies, like comparators (oscillators have comparators inside) and logic. If the board is large, then add more bulk caps.
The type of caps are not really critical on audio boards.
More important is to have solid power and ground connections to each consumer, and also to their decoupling caps!
Resistors on the power lines are not recommended, as you get variable voltage drops which can jeopardise your function.
Jarno
rarara wrote:
anyone care to link to a photo of decoupling caps fitted? think i have it correct but not sure.

I have power arriving in the form of Pin1:12v, Pin2:Ground, Pin3:6v and am wanting (has been recommended for the purpose) to put an electrolytic and foil cap (in parallel presumably) on each rail. Am I just connecting like this?:

Pin1 ---<foilcap>---Pin2
Pin1 ---<elec.cap>---Pin2

Pin3 ---<foilcap>---Pin2
Pin3 ---<elec.cap>---Pin2


Yes, filmcap parallel to electrolytics.

I might add that I often use 1206 SMT caps (not film), because usually you can find a spot where you can tack them over the rails. Also, really close to IC's, you can fit two or three of them underneath a 8 pins DIL, to tame even the wildest of opamps.
rarara
OK, here is my artistic impression showing my molex connector sitting on the PCB. If i want to add a capacitor as described does it just link one of the say 12v pins with the middle ground pin as shown:



presumably also, if the cap has polarity the -ve side will be ground?
frijitz
jonbstevens wrote:
Thanks for the replies. My oscillators and self oscillating filters are putting a little ripple onto the power rails, at normal levels it's inaudible but when I crank up the gain to unrealistic levels on my mixer I can hear the unpatched oscillator/filter bleeding into the output of other modules

What is the magnitude of the ripple?

Quote:
It's not really a problem, but i guess it's eating at me a little bit. I was thinking about simple fixes -- like the fact that i used film caps instead of ceramic for decoupling. I was also thinking of replacing the 10uf electrolytics with larger tantalum capacitors, 47uf tantalums at Jameco are not too expensive, and adding ferrite beads inline to the +/- 12v inputs to all of my modules. Beyond these things I don't know if there's much else i can do.

In my experience, ripple is usually caused by poor power distribution, ie ground loops. How long are your power supply leads?

Ian
MUFF WIGGLER Forum Index -> Music Tech DIY Goto page 1, 2  Next [all]
Page 1 of 2
Powered by phpBB © phpBB Group