I attach a simplified diagram of an arduino circuit in which I have attached 10µF 16V Tantalum Bead Capacitors to the 5v and 3.3v pin outs for decoupling purposes. Having done so I am finding that I measure 4v on my sub-circuit supplied by the 3.3v pin. The 5v pin still reads 5v.
If I remove either the 5v sub-circuit or the capacitors, the voltage from the 3.3v pin drops to the expected 3.3v.
I am using an Arduino Duemilanove.
I'm not sure if this is expected behaviour so I would be grateful if anyone could point me towards an understanding of what I am seeing. I recently fried an Arduino Due and this may have been the cause because the 3.3pin was supplying voltage to the digital I/Os.
Is there any load on the 3V3 supply? If not put a 1K R from 3V3 to GND and see if it goes down.
10uF seems large for decoupling logic levels, the 3V3 originates at the FTDI chip and already has a 100nf decoupling cap and it's only rated 50mA, remove just the 10uF on 3V3 and see if V goes back to normal.
It is not expected behaviour. In fact if it is wired up as you show it is impossible behaviour. So you must not have wired it up like you think you have. Maybe some wires are touching or something but there is no way a capacitor can increase the voltage output.
The 3.3volt of that Arduino is extracted from the FT232 (serial) chip.
Don't know if it likes a low ESR (tantalum) capacitive load.
It could oscillate, and give the wrong reading on your DMM.
Try to look at the voltage wih a scope if you have one.
Don't expect a lot from that 3.3volt pin (~50mA absolute max).
Leo..