ben76

Yes you can obviously use two 10,000uF capacitors.

Make sure you order 4 identical capacitors. Somewhat like batteries, it is best not to mix makes and models because this breaks the symmetry of the circuit.

Good Luck,
Ben

xPat,

Your diagram is nicely drawn and very clear. However, as you suspected, it is not correct. You should not have any wire connecting the negative terminals of the capacitors to each other.  This wire makes the capacitors sit on the cable like a bird on a power-line :-)

Below is a corrected diagram:
 capacitor configuration.JPG   40.76K   42 downloads

Indeed, it looks like this stuff is new to you. Please ask a friend to help you, or do not do this at all unless you are certain that you can do this without making any mistakes.

I have just looked in DigiKey at capacitors of 4700uF / 100Volt (5000uF is a non-standard value).
Remember that you will need two of these per speaker.

If you do not mind soldering these should work fine:
http://www.digikey.c...6110-ND/132316.

If you do not want to solder you can use these (they cost double though):
http://www.digikey.c...3302-ND/2095916

One more thing, I have just noticed that in post #8 I wrote by mistake that when connecting two identical electrolytic capacitors in series to create a non-polar capacitor one gets a capacitor of double the voltage rating of each individual capacitor. This is obviously not correct since the whole point of this configuration is that when faced with AC voltage only one capacitor carries the full voltage burden (the capacitor whose polarity matches the instantaneous AC signal polarity).

In short, when connecting two identical electrolytic capacitors in series as described in post #8 one gets a capacitor of half the capacitance and the same (not double) voltage rating.

Please note that I have fixed post #8 so it is now correct.

xPat, as Carlspeak said the capacitor you linked to in DigiKey is a polar one and you will need two to make a non-polar one!

Let me repeat:
********
One polar capacitor per speaker would explode violently in this application!
********

Unfortunately, non-polar capacitors of this size (10,000 uF) are hard to come by. For example, the largest non-polar capacitor that PartsExpress currently has to offer is only 500uF. Thus, you will probably have to connect two capacitors as I explained.

One more thing. I see that your original question regarding the size of the capacitor was not directly addressed (except indirectly by providing the LST diagram which uses 5000uF). In my replies I have used 1000uF as an example in the calculations, but this value is way too small .

The capacitor and the speaker act as a voltage divider where the AC voltage from the amplifier that reaches the speaker is proportional to the impedance ratio between the speaker and the capacitor. The bigger the capacitor the lower its impedance and the more 'invisible' it is, i.e., the more energy goes to the speaker as intended.

The impedance of both the capacitor and the speaker is different at different frequencies. For the speaker the connection between frequency and impedance is complicated. Saying that a speaker has an impedance of 4ohm or 8ohm is a crude simplification. For some speakers AR actually provided a graph showing impedance vs. frequency. For practical purposes at audio frequencies, if we ignore the parasitic ESR of a capacitor (don't worry about ESR - it should be pretty low for a new modern capacitor - I mention it so that the experts will not complain), the impedance of a capacitor goes down linearly with increasing frequency. I.e., double the frequency, half the impedance.

So, what does this all mean? Very simple: we have to design for the worst case (the frequency for which the capacitor impedance is highest compared to the speaker's impedance) which would for all practical purposes be the lowest frequency the speaker can produce.

Let's consider your toughest client: the AR9 (because it has low impedance and can sound very low frequencies), which has appreciable ability to go as low as 20Hz. According to the AR9 manual its impedance at that frequency is 8ohm (it is worse - i.e. lower - at higher frequencies, but at these frequencies the capacitor will have a very low impedance, so we are fine). If you use a capacitor of 1000uF it will have an impedance of almost 8ohm at 20Hz. Thus, the ratio of impedances is 1:1 and the speaker will get only half the voltage the amplifier is outputting at the 20Hz portions of music (I remind you that most music has negligible components at 20Hz, but it makes no sense to cripple an AR9 with such a capacitor). At 40Hz the capacitor's impedance will be 4ohm and 2/3 of the amplifier's output voltage will go into the speaker. Not too terrible, but not good. Note also that we have created a non-linear response curve for our combined speaker-capacitor - the lower the frequency the lower the volume we hear. Not very good.

If we go with a 5,000uF capacitor (like the LST) its impedance at 20HZ is only 1.6ohm, and the speaker will get 5/6 of the amplifier's output voltage. At 40Hz it will get 10/11 of the voltage. This is very good. If you go with 10,000uF the speaker will get 10/11 of the voltage already at 20Hz. There is hardly any point trying to improve on this.

In short, 5,000uF is very good. 10,000uF if you want to go all the way. Also, I suggest that you go with capacitors rated for at least 100 Volts, which by the way is what the AR9 uses in its crossover.