On the left is the diagram and formula for wiring in series, and on the right is the same for wiring in parallel. If you do the math, you'll find that the overall capacitance is reduced if you wire multiple capacitors in series, and increased if you wire multiple capacitors in parallel. With our target value of about 660 uF as a starting point, capacitors generally get more expensive as the capacitance increases, and cheaper (and more common) as the capacitance decreases (at least until you get to very low values, then price and rarity might go up again). So for our purposes, we want to wire in parallel to be able to use more commonly found resistors. 660 uF is not a very common capacitance value, but 330 uF is, so I used two 330 uF capacitors wired in parallel for a total capacitance of 660 uF. The particular capacitors I bought are from mouser at this link:

Nichicon Aluminum Electrolytic Capacitors - Leaded 50volts 330uF 85c 16x31.5 7.5LS

I ordered capacitors with 50 volt rating because you want to have a lot of headroom over the actual power that will be going through the circuit so the capacitors don't blow out. 50 is a good minimum, it's fine to go higher, but the capacitors will increase in physical size (they are already pretty big at 50).

### Manufacturing Tolerances

Here's where things get interesting.

If you look at the specifications for the capacitors I linked to above, you'll see that they have a tolerance rating of 20%. This means that even though it's listed as a 330 uF capacitor, what you actually get might have a capacitance of +/- 20% of that value, so from 264 - 396 uF. As you can imagine, this has a pretty large impact on our calculations.

It's not just capacitors that have wide tolerances, the speakers, amplifier, resistors, all of them are manufactured to specs to some degree of accuracy. If you pay a lot of money, you can buy components with very narrow tolerances. But the cheaper option for manufacturers is to have wide tolerances, and rate their products accordingly. For good manufacturers, this usually means they will err on the side of giving you a better product than the rating.

So I ordered 10 of these capacitors (I only need 4, but bulk pricing kicks in at 10 for Mouser, and it's helpful to have spares), and then measured them when they arrived with a multimeter that can measure capacitance. Then I did some calculations to figure out what the actual cutoff frequencies would be (I used this other crossover calculator this time, because it lets you plug in capacitor values and get the responding corner/cutoff frequency, and not just the other way around). Here are the results: