Pitfalls When Wiring Capacitors in Parallel?

[m00dawg]

So this sort of carries on with another post of mine about building a PSU suitable for the MB-SID. I thought I would post a new topic, however, since the question doesn't quite relate.

The design of my new powerboard (which using a center-tapped transformer) is coming along nicely but I'm trying to conserve as much space as possible on my board. As a result, I have had to place my capacitors in interesting configurations. What I was wondering was how much does the order of the capacitors matter and do they need to be in parallel (one after another) or simply on the same wire to have the same smoothing/filtering affect?

I know that connecting capacitors in parallel increases the total capacitance, but I'm more concerned about making sure HF noise doesn't get through and I'm not sure if there's tricks or pitfalls to doing it a certain way.

Based upon my searching and knowledge, I always assumed you go from big to small, so, say, 2200uF -> 47uF -> 330nF -> 100nF -> 10nF. Each one has less of a capacitance, but because they are made of different materials, they have different characteristics. I know the eletrolytics don't filter out HF noise, but are obviously good for smoothing out the AC ripples.  The smaller ones can then filter out the HF noise. But can the sequence be re-arranged to produce the same affect (such as 2200uF -> 330nF -> 47uF .... ).

And can I do something like have my +V coming from my rectifier and have it fan out, making sort of a star, to each capacitor instead of routing the wire to each of them individually? I assumed no :)

Doing either of these means I can save space and probably avoid a two layer board (meaning the bottom can just be a ground plane). But before I make a big mistake, I wanted to make sure I knew more of what I was getting into :)

Also I apologize if this is already on the forums. After some lengthy searching, I didn't come up with an answer to either of the above questions.

[m00dawg]

I realized that it's not immediately apparently what I'm talking about :) So here's a few images. So my question is, are the two boards electrically equivalent (will they filter out the same noise, etc.)?

[stryd_one]

I kinda touched on this in your other thread you linked... I've read that the caps have the effect of handling the current individually even though they're in parallel. I guess it makes sense if you consider that the charge is caused by electrons that can't be in two places at once, but of course they are still all influenced by each one's neighbour's potential to flow. Kinda like, one body of water flowing through a pipe, but still individual molecules of water being effected at each point in the pipe.

I guess that's also why 'the done thing' is to put the big caps at the big end of town, and then pass the lighter-duty work onto the lighter-duty components later on. Personally, I'm inclined to look at this and go "that's how it's been done for decades, it's probably right"

I hope someone can back this voodoo up, or trounce it, with something vaguely scientific ;)

[m00dawg]

Yeah that's sort of what I figured. I haven't seen circuits designed that way when I was looking around so I'm guessing there's a reason that is the case. Other than having to use both sides of the board, so far the only issue seems to be a longer run but I think that's marginal really.

Thanks for your input and help, as always!

[moogah]

Stryd pretty much explained what you are dealing with.  The *right* way to do it is the intuitive way, even though the caps are in parallel you should think of each one being a separate filtering stage and place them one after another.  However, with this design I don't think you'd be able to measure a difference between those two layouts.  I would spend my time perfecting the layout so that all your traces are on one side and the ground plane can occupy the other (this will help with hum as well).  It can certainly be done with the amount of space you have.

Also, as you have it now those two regulators will pretty much be touching each other.  I would re-locate them to the "sides" of the PCB (this will give you *much* more room for various heatsinks).

[m00dawg]

Well, I ended up having a bit more room after I rounded up the measurements (since BatchPCB rounds up to the nearest inch - I figure why not use the space). I added and re-arranged some caps around and was able to fit things in while using the "one after the other" approach. I'm still using bottom traces (sparingly) but have both a ground and isolation plane. I may look and try to see if I can put all the traces on one side. I'm going to build a prototype before I get the real board printed, so I have some additional time to test things out.