Computer Display?

[HenrikOlsson]

Hi Ed,
In your case I'd keep the GND connection as short as possible, the connection to the positive rail then becomes whatever length is required. Then place the decoupling capacitor right across the supply pins of the device or as close to the supply pins as you possible can.

/Henrik.

[Ramius]

Hi Henrik!
Thanks! One idea was to use 18ga buss wire for ground while 22ga for power. If I understand correctly then the negatives leads of the capacitors would be short and the positives would be long? Wouldn't this be allowing any spikes or signals to be on the positive supply rail? Thanks, Ed

[HenrikOlsson]

Hi Ed,
Not that I think it matters in this application but if you're going to use different gauge wire use the heavier gauge wire for the ground bus and the thinner for the power bus. The connection from the IC to the GND will be kept short, the connection from the IC to the power rail perhaps a little longer. The capacitor should be placed across the power supply pins of the IC and its purpose is to act as a local "power reservoir" for the IC as well filtering noise on the supply rail, bypassing it "down" to GND.

It's very important but really not that complicated for the general stuff we're tinkering with.

Here's a link to an article describing and showing how and why and here's a more technical app note from Analog Devices.

/Henrik.

[Ramius]

Thanks Henrik!
I basically understand what the intent is and from all the examples show the caps are mounted with equal lead lenghts which is where I became confused. (http://www.vagrearg.org/content/decoupling breadboard setup figure 4) Of course the ground will be connected directly to the ground pin of the I.C. It just seemed that with pulse widths indicating a frequency of 2.4 mhz that having parallel capacitor wires (caps on the 3.0 volt side and caps on the 5.0 volt side spaced about an inch apart and at a higher than ground potentials (3v and 5v) you would be inductively coupling the signals thus the need for a better understanding. Best, Ed

[HenrikOlsson]

Hi Ed,
I don't really understand but I can say that there's no signals in that circuit with a frequency of 2.4MHz. The sensor is supposed to work with a 32768Hz clock signal (which I'm guessing you're generating with the CCP module in the PIC) and SPI communication runs at a couple of 100kHz at most (depending on the clock frequency of the PIC).

Understanding is great but at this point you are just overthinking it. Supply power to you chips using as short wires as you can, place a capacitor across the supply pins of each chip - done. Voltage regulators usually wants/needs capacitors on both the input and the output, read the datasheet for the device in question.

/Henrik.

[Ramius]

Thanks Henrik and you are right I probably am overthinking this and just trying to do the best possible job posible. About the 2.4Mhz, the Saleae Logic Analyser shows the pulses to and from the sensor. When you look at the pulses from the sensor the pulse width +5v to zero is narrow and the Logic Analyser says the frequency is 2.4 Mhz. Yes, I agree regulators need both input and output capacitors or they can oscillate, pluse you need a "reserve" for powering devices. Thanks again. Best, Ed

[HenrikOlsson]

Hi Ed,
If there really ARE pulses with a frequency of 2.4MHz then I'd say that's a problem, note that the width of the pulses alone doesn't say anything about the frequency. Can you post a screenshot and tell us exactly where you're probing?