Use caps in series to get lower value?

[JJones]

Is there any reason not to use caps in series to get a lower value (other than cost)?

If you put 2 10uF caps in series, you end up w/5uF...

What I don't understand is, if you connect 2 10uF 100VAC in series, then, why wouldn't the VAC rating be 400? ( x2 caps and 1/2 the value so x2 for that also )

Anyway, I was curious because I've never seen mentioned putting caps in series when discussing speaker crossovers? Seems like this is an easy way to get very precise values adjusting down too (not just up) - I often see putting multiple caps in parallel mentioned, but never in series...

[ThomasW]

http://en.wikipedia.org/wiki/Series_and_parallel_circuits#Capacitors

[stangbat]

Is there any reason not to use caps in series to get a lower value (other than cost)?

If you put 2 10uF caps in series, you end up w/5uF...

What I don't understand is, if you connect 2 10uF 100VAC in series, then, why wouldn't the VAC rating be 400? ( x2 caps and 1/2 the value so x2 for that also )

Anyway, I was curious because I've never seen mentioned putting caps in series when discussing speaker crossovers? Seems like this is an easy way to get very precise values adjusting down too (not just up) - I often see putting multiple caps in parallel mentioned, but never in series...

As for the voltages, also see here:

http://en.wikipedia.org/wiki/Capacitor#Series_or_parallel_arrangements

As for why people don't put caps in series to get lower values, it doesn't make economic sense in this application. The total when putting caps in series is always going to be lower than the largest value capacitor. So you are paying more money for two higher value caps to get a lower overall capacitance. Why not just spend less money on a lower value cap? Series capacitors do give you a larger voltage capacity, but usually that isn't the limiting factor in loudspeaker crossover networks.

[JJones]

The whole point of my idea was to save money, actually...

As for the voltages, also see here:
http://en.wikipedia.org/wiki/Capacit...l_arrangements

I read some of the web pages discussing capacitors from a basic electronics viewpoint, but they didn't address whether there were any issues affecting the sonic quality of series use in speaker crossovers...

As for why people don't put caps in series to get lower values, it doesn't make economic sense in this application. The total when putting caps in series is always going to be lower than the largest value capacitor. So you are paying more money for two higher value caps to get a lower overall capacitance. Why not just spend less money on a lower value cap? Series capacitors do give you a larger voltage capacity, but usually that isn't the limiting factor in loudspeaker crossover networks.

More expensive caps usually are rated higher voltage. Do you think if we compare a 100V cap to a 450V cap, they're be no discernible difference in an audio crossover application? Also, I've got a few dozen each of 1.5, 2.2, 3.3, and 3.9uF metalized poly caps. With the 10uF "surplus" poly caps @ 81 cents each from Madisound, I could add 10s and add/subtract smaller values to get an exact value for probably cheaper (if I could even find the exact value) than buying most other poly caps? For example, a basic Dayton 5.1uf poly cap is $2.19, they don't even have 5.0uF - so, if I want 5uf and I find a pair of my 81 cent 10uFs that equal 5uF total in series it actually costs me a little LESS (though not much less, admittedly), and, I might get exactly the value I want, rather than being ~.1uF off?

I suppose if I want to pursue this any further, I should figure out what kind of program I can use to directly test the quality of capacitors.

I recall one website where a guy tests various diff. types of caps with a scope and how straight or curved the line is shows us how good the cap is...

[Saurav]

One somewhat 'standard' practice is to parallel a smaller and higher quality cap with a larger and lower quality cap to get to the value you're trying to reach. I don't really know the justification behind it... maybe the lower ESR of the better cap reduces the overall ESR of the combination since it's in parallel with the main cap. The rule of thumb I've seen mentioned is to use a smaller cap that's about 10% of the larger cap's value. But don't quote me on this

[ThomasW]

If you look at many of my projects, you'll find I use the GE closeout caps from Madisound. These are as good as the poly PE caps, cheaper than putting 2 surplus 10mfd caps in series. The GE's also have high voltage ratings.

http://www.madisound.com/manufacturers/ge/ge_caps.php

Oh, and here's a not so subtle hint....
Stop putting large blocks of text in bold
None of us are vision impaired.

[---k---]

Dayton caps are rated at 250V. Does anyone think that this is inadequate for a crossover? I don't think there is really a need for caps rated at 400V - is there?

[JJones]

If you look at many of my projects, you'll find I use the GE closeout caps from Madisound. These are as good as the poly PE caps, cheaper than putting 2 surplus 10mfd caps in series. The GE's also have high voltage ratings.
http://www.madisound.com/manufacturers/ge/ge_caps.php

Oh! Well, I didn't expect to find any better value than ones in the "Surplus" section, and I had looked at that page before too. Thanks for the reminder. I wish I had got some of those with my last order - it's no fun to pay $8 - 9 shipping if you're just getting a couple $ worth of caps!

I'd figure 100V would be plenty, but then again, I haven't done any testing. How high are the voltages in dynamic peaks in "typical" home music listening?

[Saurav]

Been a while since I dealt with this stuff, so take this with a grain of salt...

With series LC circuits, the voltage across each component (the capacitor or the inductor) can actually be higher than the voltage across the combination. Sometimes, a lot higher.

According to SPICE, voltage across the capacitor and inductor reach a peak somewhere around 70 volts! This is quite impressive for a power supply that only generates 1 volt.

...

The expected values for capacitor and inductor voltage are 100 V. This voltage will stress these components to that level and they must be rated accordingly. However, these voltages are out of phase and cancel yielding a total voltage across all three components of only 1 V, the applied voltage.

http://www.allaboutcircuits.com/vol_2/chpt_6/3.html

How high are the voltages in dynamic peaks in "typical" home music listening?

Not counting resonant circuits... you can get a ballpark idea. If your speaker is a nominal 6 ohms and you're sending 100W into it, that works out to a nominal 24V. That's the voltage drop across the binding posts, so the drop across an individual component will be lower.

On the other hand, if you wanted to be paranoid, you'd consider that this 24V is the 'applied voltage' to the crossover, and the calculations show that applying 1V can cause 100V across the cap, so if you apply 24V...

Again, like I said - grain of salt.

[Amphiprion]

The reason they don't have 5.0uF caps is because caps are commonly only available in the E24 series (5%) series of values. A single 5.1uF cap is only 2% off your desired value of 5.0uF. Putting two 10uF caps in series to get 'exactly' 5.0uF with 5% tolerance anyway isn't really worth the money.

ETA: just saw Saurav's post. I've used extremely high Q resonant L/C filters to get into the many hundreds of volts range at a previous job. Some here may even remember me posting about it. It had a 12V input IIRC. I sincerely doubt this would ever happen in a loudspeaker crossover, getting Q's that high takes a LOT of effort.

[Saurav]

Agreed, that was an extreme example. I was just pointing out that voltages across individual components may be higher than the voltage across the whole circuit. This is (or was, to me) somewhat counter-intuitive, and so I figured it may not be something that everyone automatically thinks about. Would it be that hard to exceed 100V across a cap if you're sending in 100+ watts into the crossover?

[---k---]

Would it be that hard to exceed 100V across a cap if you're sending in 100+ watts into the crossover?

I thought your previous post said yes, but color me confused.

[Saurav]

Maybe I'm the one who's confused Let's try again... here's what I think.

If you don't have any LC resonant sections (like notch filters) in the crossover, then it would take a lot of power to reach those voltages.

However, if your crossover has resonant sections, then individual components can be exposed to voltages that are higher than the input voltage seen by the whole crossover, and in that situation, I don't think it's out of the question that it might get close to 100V.

Hopefully Mark will correct any errors in that statement.

[JonP]

The whole point of my idea was to save money, actually...
I read some of the web pages discussing capacitors from a basic electronics viewpoint, but they didn't address whether there were any issues affecting the sonic quality of series use in speaker crossovers...

I suppose if I want to pursue this any further, I should figure out what kind of program I can use to directly test the quality of capacitors.

I recall one website where a guy tests various diff. types of caps with a scope and how straight or curved the line is shows us how good the cap is...

To your question about issues affecting sound quality... there are a few things that might.

Every cap has a certain amount of internal losses and or resistance. This is stated as "dissipation factor" and "ESR". D is mostly a function of dialectric type, electrolytic, ceramic, polyproplyne, etc. ESR is equivalent series resistance... the virtual internal resistance, and results more from how the cap is constructed/what it's made from, etc. Also, there is a small (to vanishingly small) amount of distortion created by non-linearity, which varies with cap type.

So if you series them, the ESR and other effects would add, and the overall cap would have higher than a similar single one of the same value. The same probably goes for the distortion effects.

Now, these things are pretty small... and may or may not be even measurable down at our pretty slow audio frequencies. (at least with the better cap types) So, it might not be a big issue, and the cost issues would be the main thing. But I wanted to throw that out there for completeness.

Check the app notes of the big IC manufacturers, I think at National or Analog, and Maxim as well... there was some stuff on caps and distortion, and other testing. Try Bob Pease's articles for fun reading and I think a few pertinent ones if you want to try building something.

[Amphiprion]

Hopefully Mark will correct any errors in that statement.

Nope, you're good. I will elaborate, though

The perfect example of where this phenomena would occur is, like Saurav said, in a notch filter. In a perfect LC filter with absolutely no series or parallel resistance, the voltage theoretically approaches infinity at the node in between the 'L' and the 'C' when driven at the resonant frequency of the notch (where the frequency null is in normal use).

This is why it is very, very important not to hook up a crossover with a without a load (a dummy load will do fine).

Take a look at the woofer crossover for Zaph's old L15 design:



Attached are SPICE simulations of the voltage at the output (green) and the voltage at the node in between the L and the C (blue) for the cases where there is no load vs. an 8 ohm load. I was even surprised at how large the resonance was when I included some sizeable DCR in the inductor (plots not shown). Plots are in dB referenced to a 1V peak amplitude sine wave.

The lesson here is LC notches can create some huge voltages if you don't damp it out with a load. With load and everything wired correctly, it'll probably never be a problem.

Oh, and I can't make a post with pictures without a pic of the dog. This one comes from the first day I got her from the pound when she was just one year old. She just turned 7 this month.

[Amphiprion]

One somewhat 'standard' practice is to parallel a smaller and higher quality cap with a larger and lower quality cap to get to the value you're trying to reach. I don't really know the justification behind it... maybe the lower ESR of the better cap reduces the overall ESR of the combination since it's in parallel with the main cap. The rule of thumb I've seen mentioned is to use a smaller cap that's about 10% of the larger cap's value. But don't quote me on this

Just did

IMO this is an invalid technique with regard to loudspeaker crossovers. It does have use in audio (and all other) electronics, such as active crossovers. It is standard practice when designing a printed circuit board to use a bulk electrolytic capacitor at the power entry to the board, or just after the voltage regulators. These electrolytics can only provide adequate power supply decoupling (meaning low impedance for AC signals from power to ground, to get ripple and noise off the supply lines by shunting AC current to ground) up to about 1MHz. Above 1MHz, ion mobilities in electrolytic capacitors are too low and prevent them from functioning properly, and they become inductive. Audio opamps, however, have bandwidths of approximately 10MHz or so (depending on the type) when used at unity gain. Because of this, film or more commonly ceramic capacitors are placed immediately at each opamp (or any other integrated circuit) to provide good bypassing into the tens or hundreds of MHz, depending on the capacitance value. This helps ensure no oscillations or other undesired behavior.

So the entire point of bypassing (in general) is to make up for the deficiencies of one capacitor with regard to frequency by using another capacitor to affect a different range of frequencies. Using a much smaller capacitor in parallel with an 'inferior' capacitor will not affect to a significant degree the behavior of the first capacitor in its intended frequency range (in a passive crossover, this would be near the crossover point).

Now bypassing one type of film or ceramic cap with another type of film or ceramic cap is actually a tremendously BAD idea. In so doing, you've just made a C-L-C circuit due to the series inductance of the capacitor leads.

Didn't we just learn what can happen with undamped LC circuits?

This is more of an issue in bypassing high frequency devices on PCB's, where a designer might parallel a 100nF cap with another 1nF cap in an attempt to give better supply bypassing to a device. This can and usually will cause a resonant spike in between the two primary resonances of each cap and provide worse bypassing (higher impedance to ground) in between the primary resonances of each cap.

But now I'm getting WAY off topic...

[---k---]

Didn't we just learn what can happen with undamped LC circuits?

Nope. You're making my head hurt. :P

[evilskillit]

Questions about BoM

Responded to the wrong thread, please delete.