Orientation of polarised caps when using a bipolar power supply

[j_flanders]

I'm adding a spring reverb to a SS amp that has no reverb and I'm using the schematics from Elliott Sound Products: http://sound.whsites.net/articles/reverb.htm
So far it all works, but I'm confused about the orientation of the electrolytic caps and actually about (polarised) caps in general when I start to think about it.
First my actual problem/question:
On his schematic the C2  electro cap is wired with + towards  ground:


However, when looking at other schematics I see electrolytic caps with - towards ground:




Yet on other schematics, Marshall MG15 CDR for example (sorry, schematic is PDF), I see no cap at all in that spot.

So, who/what is right?
Also, can I just hack my way through and always use two (double valued) polarised (electrolytic) caps in series (+ to + or - to -) and never worry about orientation again? Given that I have room and money to spare.

There is a small note on his page about the caps:

Note that in many of the circuits shown below, I have used standard polarised electrolytic capacitors, including in locations where there is no polarising voltage. This is (perhaps surprisingly) perfectly alright provided the voltage (AC or DC) never exceeds about 1V. In all cases where polarised electros are shown the actual voltage will be less than 100mV. The exceptions are Figures 5 & 6, because the voltage across C2/C4 may exceed the 1V threshold.

Figure 5 is this one:

So, what is a polarising voltage?
Also, I thought caps block DC and let AC through, so orientation should only matter with regard to DC.
Or, in our 9V, single/split power supply we use caps to 'avoid' DC from running out of our circuit to ground, hence the caps, which value also determines the HP cut off.
Do I need caps at all in such places when dealing with a bipolar power supply?
Does one ever need a polarised cap at all? Is the polarisation just a side effect with electrolytics, which happen to be smaller and cheaper than equivalent values in non polarised caps.

[thermionix]

With the bipolar supply, there is no 1/2Vcc bias voltage like we usually use in 9V pedals.  So you basically have 0VDC on either side of that cap, and orientation doesn't matter.  A bipolar (NP) cap would be fine obviously, but a little more expensive and difficult to find in a range of values.

[PRR]

This:
"I have used standard polarised electrolytic capacitors, including in locations where there is no polarising voltage. This is (perhaps surprisingly) perfectly alright provided the voltage (AC or DC) never exceeds about 1V."

Getting specific: the NE5532 has NPN inputs. A 220K bias resistor on the +In pin will cause about -0.04V of input offset, which reflects to the -In pin. The polarity shown is indeed "correct", not that it matters because <1V. The TL072's input current is about zero, so any same bias resistor causes <0.01V offset, and we can forget about it, throw the cap either which way.

If you must be fussy: build all but the caps, put your meter where the cap will go, keep your red and black straight, that's the polarity it wants to be. (Most but not "all" audio circuits will DC-bias without caps.)

[highwater]

If you must be fussy: build all but the caps, put your meter where the cap will go, keep your red and black straight, that's the polarity it wants to be. (Most but not "all" audio circuits will DC-bias without caps.)

I might start doing that even when I *am* sure; polarized caps tend to be the last thing I solder-in anyway (since they're almost always the tallest components), so there's not much of a downside.

[ElectricDruid]

So, what is a polarising voltage?

It's the DC that the cap sees. One end of the cap is usually at a different DC voltage level to the other, but we want to connect the two points together to pass a signal from one to the other - that's what the cap is for.

Also, I thought caps block DC and let AC through, so orientation should only matter with regard to DC.

Yes, exactly.

Or, in our 9V, single/split power supply we use caps to 'avoid' DC from running out of our circuit to ground, hence the caps, which value also determines the HP cut off.
Do I need caps at all in such places when dealing with a bipolar power supply?

You need caps when you're passing signals from one part of a circuit to another, and *where those two bits of circuit have different DC levels*. That's the same for single supplies as it is for dual supplies.

One way it's easy to spot circuits designed by electronics beginners is when they've taken several "building block" type circuits and joined them all up with DC-blocking caps in-between, even when the circuits have the same DC bias level. Sometimes you see circuits where they've got an output cap from one circuit connected to an input cap on the next building block! Newb giveaway! The important thing to think about is what the DC bias is on each end of the cap. If it's actually the same, you don't need the cap at all. For example, joining a 9V op-amp drive stage biased around 4.5V to a op-amp tone control biased around the same level. If the bias is different, you need the cap. For example, you need the cap if you're joining the output from an op-amp input buffer (biased around 4.5V) to a PT2399 delay chip (which runs from 5V and is therefore biased around 2.5V).

Does one ever need a polarised cap at all? Is the polarisation just a side effect with electrolytics, which happen to be smaller and cheaper than equivalent values in non polarised caps.

No, one never *needs* a polarised cap. If we had cheap, small non-polarised caps available in all values, we'd use those. However, for larger values (>1uF or so), it's the case that polarised caps are typically both cheaper and smaller, so we use them instead. The polarisation *is* just a side effect of the construction, as you put it. It's not a feature we'd design in if we could easily avoid it.

[j_flanders]

Ok, great. Thanks a lot for the replies and explantions. This clears things up for me.