# questions on separate grounding

Apologies for still not entirely understanding electrical current, but i hav 2 questions about ground.

1. If I can have a resistor on EITHER side of an LED, could i also tie all of their cathodes together, then resistor, then ground? I am guessing not, since otherwise there would not be much need for bussed resistor networks. But why/

2. Hard drive and other 12/5v combo power supplies typically have 4 pins: 2 for +12/ground, and 2 for +5/ground. How come they do not simply use the same ground between the two?

On your LEDs are you tying all cathodes to each other and the anodes to each other so all light at once. Or are the cathodes tied to common and the anodes independently switched. If each is independently switched then current flow will vary with number of LEDs on. One resistor could not be matched to current limit all or partial lighted LEDs.

spicetraders: On your LEDs are you tying all cathodes to each other and the anodes to each other so all light at once. Or are the cathodes tied to common and the anodes independently switched. If each is independently switched then current flow will vary with number of LEDs on. One resistor could not be matched to current limit all or partial lighted LEDs.

Oh duh, I totally did not think of that but now that you mention it seems obvious. thanks!

anyone know about 12v/5v hard drive power supplies and why ground is separate?

Generally speaking the reason you run separate grounds to devices is that it divides up the ground currents. For example, a 3.5" hard drive would use 12V to spin the platters and 5V for the logic. The platter is a motor, so we run separate grounds back to the power supply in case the motor makes electrical noise you don't want that coupled into your logic circuits. Make sense?

rmetzner49: Generally speaking the reason you run separate grounds to devices is that it divides up the ground currents. For example, a 3.5" hard drive would use 12V to spin the platters and 5V for the logic. The platter is a motor, so we run separate grounds back to the power supply in case the motor makes electrical noise you don't want that coupled into your logic circuits. Make sense?

Yes! Thanks. So if I had limited wires and no concern for noise, I could tie the 5 and 12v ground together, right?

If the two lines are truly power supply "grounded" return lines or common grounded then running one single line for your "ground or return line" is workable provided its size matches the total current flow. But not all powers supplies have the return lines and grounds being one and the same. Earth, chassis, power and data grounds may or may not be connected depending on the manufactures design and needs. So if you have a power supply in mind that you want to run just a single wire for returns. Test it , look at the schematics, look at the circuit it was built for, and the manufactures specs. Most of the time there will be no issues.

As far as the 4 pin connector on a computer goes, the grounds are tied together at both ends.

The pins are rated at some maximum current, and the wires each have a certain resistance and maximum current. You have current coming in over two pins, and going back out over two pins. Hard drives also used to draw a bit more current than they do now. But in the end, I'm not sure it is any more than symmetry.

As far as the 4 pin connector on a computer goes, the grounds are tied together at both ends.

Just don’t use that as gospel. In cases where you have “Analog” and “Digital” supplies, they are NOT tied together at both ends. For the best noise floor, you will find that an Industrial A/D card will have the two grounds tied together in a “star” connection at the A/D converter. Not sure if those rules are followed in the Arduino world.

I am -only- answering for the hard drive power connector inside a desktop computer.

rmetzner49 is correct, you cannot apply that across the board for every power supply. Analog and digital circuits, when done correctly, use separate ground lines.

In the Arduino world, it is a huge tossup. There are too many people with too many levels of experience designing things. Even the official Arduino boards make no effort to isolate AVcc from Vcc, as suggested in the documentation for the Atmel AVR chips used on them. But they are meant as general purpose breakout boards for instruction, not high end microcontroller boards.

Even the official Arduino boards make no effort to isolate AVcc from Vcc

It's probably adequate for 10-bit A/D converters when one LSB is 5mV. As you go up in resolution, so does the need for robust power and ground routing and decoupling.

My former workplace hired an engineer that was laying out a 24-bit A/D on a board with a PIC Micro. When he got all done and grounded the input, there was about 12 bits of noise (numbers unsteady) on the output. One wonders if the guy even read the data sheet for a recommended layout. At those levels you also don't run clock and data lines UNDER the A/D.

Only after reworking some 40% of the board, I was able to get 20 bits.

So if I had limited wires and no concern for noise,

You never have no concern for noise. Noise stops circuits working, it is always important.

Sometimes a little noise is good, you can use it to increase the resolution of an ADC.

Let's go back on topic.

blah44: Apologies for still not entirely understanding electrical current, but i hav 2 questions about ground.

1. If I can have a resistor on EITHER side of an LED, could i also tie all of their cathodes together, then resistor, then ground? I am guessing not, since otherwise there would not be much need for bussed resistor networks. But why/

Do not apologise for asking questions if you have something you do not understand.

This question is more complicated than you might think. You are only addressing the cathodes, not what happens with the anodes. If you share all cathodes and a resistor, you should also share the anode with the same signal.

The resistor's presence is meant to control the current through the LED (and the resistor, as they are in series). So you should look up the current and the forward voltage for that LED, and then use the resistor to meet those specifications.

If you have components in parallel (instead of in series), the current will go up. Two (almost) equal LEDs in parallel will result in a double current figure. So that means you need to use a different resistor.

The complication here is this:

What if you have 5 equal LEDs, cathodes tied together, a single resistor to GND, and anodes controlled (whatever way) separately. Now your resistor has to be changed whenever you change the combination of LEDs to be lit.

This also means, that if you are using some circuit that has all LEDs lit at the same time, if one ceases, the others will suffer and cease too soon after that.

So there's two good reasons to use separate resistors.

If you have to use multiple LEDs to be lit at the same time, look into putting them in series. You need the number of LEDs times the forward voltage to supply the set, but only once the forward current. If one LED ceases, the other ones will also no longer be lit, but they will not be destroyed.

Good explanation, but what I did not see is the reason you don't just parallel the LEDs and expect them to both work using one resistor. This is not a good way because LEDs are NOT matched for forward voltage (Vf). When you have a significant mismatch, one will light brighter than the other and this isn't desirable.

When you connect them in series. the current will be constant through both LEDs which is presumably the desired effect.

Not only that but if you are controlling the LEDs independently the brightness will depend on how many are lit.

Both are mentioned, but might need further explanation.

I did tell about looking up the forward voltage and current. I didn't tell that this forward voltage can also be measured, and that it is likely to see differences. Using LEDs from the same production batch will likely get you as close as possible, but there will be differences. Then again, you'll have a hard time finding matching resistors. If you want to take it this strict, you'd need to take that in consideration too.

The changing brightness when controlling independently isn't mentioned directly, but still hinted to in the part where i tell that the resistor has to be changed to whichever combination will be active. This is true if you desire an optimal brightness of the LEDs. Of course changing the resistor isn't a practical solution, compared to using a separate resistor to each LED.

Of course one could always mention that if you are using a fixed voltage for the LEDs you could run LEDs with internal resistors so no external resistor is needed.
Again provided each LED has the same operating voltage.

spicetraders: Of course one could always mention that if you are using a fixed voltage for the LEDs you could run LEDs with internal resistors so no external resistor is needed. Again provided each LED has the same operating voltage.

No this is total crap, you always need something to control the current, a resistor is the simplest. Every LED is different and the forward voltage drop changes with current, individule LED and over the age of the LED.

We do get the odd "you don't need to use a resistor with an LED" type of person here and they are most not welcome as they propergate incorrect information for beginners. We try and keep this forum a rubbish free zone.

And some do not read what was said before posting.

I guess you never heard of internal resistor LEDs.

Or maybe you just spit for the fun of it.

spicetraders: I guess you never heard of internal resistor

yes I have but that is not the same thing as you said is it. Do not wind up if you do not want the recoil.