Common Ground for DSC

[EricWilliams]

Can multiple digital inputs on the DSC be connected to a common ground pin, say DIO Pin 1, or must all of the signal pins be connected to their corresponding ground pin?

[Tom Line]

Conceptually, if you take a digital multimeter and check between the ground pins and discover no resistance, then they are shorted together on the PCB the same way the Driver Station is. However, each individual pin is limited to a certain current, and will fail if subjected to too much.

This is a question that should be asked on the formal Q&A.

[Al Skierkiewicz]

Eric,
Although electrically that would seem to work, you would need to follow other electrical system rules for combining wires, insulating same and connecting to the DSC. The wiring convention provided is designed for you to be able to use commonly available PWM cables for interconnect. In general digital inputs will come from different locations on the robot, it would seem that a common point for digital inputs is contrary to good practice. Additionally, using the intended interconnect method will give slightly better noise immunity by providing a separate path for noise on each cable directly to the DSC rather than comprimising all digital inputs with a common ground outside of the DSC. Finally, and more importantly, a common point outside of the DSC gives a single point failure for all digital inputs if the wire breaks or becomes disconnected.

[EricWilliams]

Originally Posted by Al Skierkiewicz

Eric,
Although electrically that would seem to work, you would need to follow other electrical system rules for combining wires, insulating same and connecting to the DSC. The wiring convention provided is designed for you to be able to use commonly available PWM cables for interconnect. In general digital inputs will come from different locations on the robot, it would seem that a common point for digital inputs is contrary to good practice. Additionally, using the intended interconnect method will give slightly better noise immunity by providing a separate path for noise on each cable directly to the DSC rather than comprimising all digital inputs with a common ground outside of the DSC. Finally, and more importantly, a common point outside of the DSC gives a single point failure for all digital inputs if the wire breaks or becomes disconnected.

We will be creating a a few sensor bus lines that combine sensor wires from key areas of our robot, and therefore will only be using the PWM MOLEX connector for connection from the bus breakout to the DSC. Using a common ground (at least one common ground per bus) would greatly reduce the complexity of our bus system. At most, we would be connecting 4 components (encoders) to one ground and 3 components (encoder + limit switches) to another ground. Would the added vulnerability to noise in this system be non-negligible for digital signals?

[Al Skierkiewicz]

Eric,
Any one signal can affect the other signals that are combined. It is unlikely that the noise component would cause problems in your configuration but anything is possible. I still point to the single failure. A defective ground takes out as many as four of your sensors, while individual wires may only take out one. I would like to understand the term "buss" in your configuration. Are you referring to common power supply connections or summing several sensor inputs to one connection?

[EricWilliams]

Originally Posted by Al Skierkiewicz

Eric,
Any one signal can affect the other signals that are combined. It is unlikely that the noise component would cause problems in your configuration but anything is possible. I still point to the single failure. A defective ground takes out as many as four of your sensors, while individual wires may only take out one. I would like to understand the term "buss" in your configuration. Are you referring to common power supply connections or summing several sensor inputs to one connection?

Thank you for all of your help on this issue so far, your expertise is greatly appreciated. We won't summing the signal over one conductor, but using a multi-pole connector (Molex, D-Sub - still deciding) to facilitate wire routing and a implementation of a quick-disconnect system.

[Al Skierkiewicz]

A Dsub works great in this application and it is one of our connectors of choice.

[EricVanWyk]

The DSC's GPIO port has an extra power and ground pin to make it easier for you to do exactly what you are asking about. You can use a long single row connector to pick up as many signals as you want, plus power and ground.

Each pin is rated for 750mA. This may require you to use multiple pins if you are powering something beefy.

Note that the rules do not require the ground path isolation/redundancy strategy that Al recommends. In some cases, it is absolutely necessary for reliable operation. In other cases, it is superfluous. You must use your own judgment to determine what is best for you and then live with the results.

For reference, I do a bit of work with signals up in the low GHz range. Here, grounding strategy is critically important. On the other hand, I've gotten away with a lot of slop in the tens of MHz range. It all depends on the application. Also, when in doubt add bulk capacitance.

In my personal opinion (which you may ignore as you please), I think starred ground returns for the GPIO will have a negligible impact on noise, but I am warming to the idea of the redundancy.

[EricWilliams]

Thanks for all your input guys. We've made our original problem irrelevant by going with a 57 position circular connector from Tyco Electronics. At $20 for the receptacle and plug together, the coolness factor was just too high to pass up. With 57 pins, we'll have plenty of positions to run discrete grounds for each of our components.

[Japper]

Originally Posted by Tom Line

Conceptually, if you take a digital multimeter and check between the ground pins and discover no resistance, then they are shorted together on the PCB the same way the Driver Station is. However, each individual pin is limited to a certain current, and will fail if subjected to too much.

This is a question that should be asked on the formal Q&A.

I was having problems with my DSC yesterday as I was connecting the ground (-) of Digital Inputs 3 and 4 together and routing the signal pins of these two inputs to limit switches as the DSC has internal 10K pull ups (to clarify, my configuration is 2 limit switches with a common ground taken from Dig in 3 & 4).

For some strange reason when the limit switch connected to Digital input 4 line was closed, my signal (RSL) light went off and the robot was disabled.

I disconnected both limit switches and with an ohmeter checked to make sure that I was not shorting the (+) pin and all check out fine for both switches...

I plugged the limit switch that was originally connected to Dig Input 4 into other Digital lines with the same results and then disonnected Dig Input 3 and it not longer exhibited this problem.

I plugged Dig Input 3 back in and the problem re-appeared so I swapped Dig Inputs 3 & 4 and the problem went away but I am not sure if my limit switches are working yet and need to play with the software a bit yet to find out...

Very peculiar and this puzzles me... I wonder if there is anthing else that I should check?

[EricVanWyk]

You were correct in looking to see if you were shorting the 5V supply. Next time, take a look at the 5V LED.

You said the robot became disabled. Did the DS report "disabled" or did things stop moving?

A DS report of "disabled" would point me to a problem further "upstream". Perhaps your code around that particular switch does something funny - does it kill/starve the watchdog?

If it doesn't report "disabled", I'd examine and re-examine your wiring.