Empty spaces on boards

[keyboardcowboy]

Almost all boards have empty capacitor spaces,how can it be found that which cap should go there and is there any advantage in filling these spaces

[Tom41]

Re: Empty spaces on boards

These empty spaces usually result from the same basic board stencil being used for various different kinds of boards. Depending on what the board will be handling, it may or may not require caps in any given space.

I doubt you would see any performance increase by putting caps into those spaces, on the contrary - it might draw so much power that the board becomes unstable!

[gonzo0815]

Re: Empty spaces on boards

Usually there is a version -in most cases a super deluxe one and very expensive wich is fully populated.
So if you know something about the design, you can assume what should be there.

Additionally, the VRM desing is mostly not designed by the board manufacturer, rather by the supplier of the VRM control chip.
So you can google for the data sheet and the typical application.
Then you can decide, if it is due to a component change e.g. some 3300uF caps instead of more 2200uF ones or if it is to reduce manufacturing cost.
In the later case, may be it could be worth to add the missing caps.

[MD Willington]

Re: Empty spaces on boards

Engineering samples:

We have some boards that are like that, we use them for development, while the customer gets a board that is only stuffed to the BOM for the final product.

[gdement]

Re: Empty spaces on boards

You can use a multimeter to see which caps it's in parallel with. You can also check the voltage on the pads while the board is running.

I've populated some extra capacitors on a couple Asus P2B-F boards, and it did help them handle heavy loads a little better.

[999999999]

Re: Empty spaces on boards

Sometimes a board is designed with more cap locations than caps to allow the manufacturer to use more than one kind of cap. If their supplier runs out or they find a better spot-price, they are covered and can use the alternate parts even if not of exact same spec.

I have seen boards where cap locations unused, caused a problem. In particular I recall one MSI motherboard (unfortunately I forget the model #), which was a late PC133 or early DDR266 memory based chipset. The memory was stable but barely so at 133MHz memclock rate, but after adding caps to empty spots on the PCB, the otherwise unchanged system then overclocked by over 10MHz more. Funny thing was that I wasn't even interested in overclocking so much as I was playing around with it and checking stability margins and I'd found it alarming that it couldn't be o'c by much at all, before becoming instable with memory that had previously exhibited it was stable at higher speeds in another system. I'm not comfortable deploying a system that is barely stable even when brand new, I insist on it being stable at a little higher than the actual clockrates it will run at, too.

Tom41 and gdement have good points. Checking voltage while it's running will give you a voltage range you need to meet, but you dont' want to indiscriminately add several thousands of uF capacitance which can result in a larger start-up current required from the PSU (which to an extreme level, could prevent the PSU from rising to running voltage fast enough), or that surge current through other parts can stress them unnecessarily. IOW, if/when I add caps to unused locations, I like to stick with a moderate to low value, not putting the biggest cap (highest uF) value in that would fit in the holes). If you've been studying other boards with similar designs you can get a better idea of what capacitance range seems reasonable.

There's also the age-old saying, "If it ain't broke don't fix it". If you see no reason (no potential problem like caps overheating from current rise of overclocking) to add the caps, if it works fine as-is, it is not expected to be a worthwhile mod to do.