Sharing signals between boards, reference voltage?

Thread Starter

Flow

Joined May 30, 2010
37
Hey there!

I have two boards with independent power supplies - a microcontroller dev board and a sensor board I'm designing just now. The microcontroller is to read the sensors output.

Now, I'm not sure about the two power supplies. Should the sensor board provide the reference voltages, or can I be sure that GND is on the "same" potential?

It'd be alright if the difference is a few millivolts, anything larger than that could mean though that the microcontrollers ADC might have to handle voltages larger than its VCC. That might just damage the microcontroller...no?

Does anyone have any experiences with this?

Thanks alot in advance!
 

beenthere

Joined Apr 20, 2004
15,819
Two rules apply - the grounds must be common and the ADC must not be exposed to an overvoltage. See the data sheet on the uC to see any suggestions about range and reference voltage.
 

Thread Starter

Flow

Joined May 30, 2010
37
Two rules apply - the grounds must be common and the ADC must not be exposed to an overvoltage. See the data sheet on the uC to see any suggestions about range and reference voltage.
"The grounds must be common"... hm, the microcontroller board has its GND on the connector. Couldn't I just use that GND for my board while still getting the voltage from the power supply? I'm using a LM217 to get a fixed voltage anyway.

Would that work?

What voltage are you using for the uc, and for what is the voltage for the sensor?
The microcontroller uses 3.3V. The sensor board is fine with anything above 7.5V.
 
One thing -- check that there is no ground loop -- no significant current flowing if you connect the ground of one board to the ground of another through, say, a 100 Ohm resistor. (I'd like feedback from others on this recommendation.)

Another thing -- definitely use a current-limiting resistor on the input the ADC. That way, if there *is* an overvoltage, the current capability of it will be small. Also check that the ADC has limiting diodes and what their current capability it. If not, then add them to your circuit. These, with the resistor, will limit the damage that overvoltage can cause to your ADC. Something like a 10 K Ohm resistor would be fine for most ADCs. Some Schottky diodes to the ground and to the Vcc rail would do well to limit the voltage within 0.2 or 0.3V of the rails, and you should not see damage.

Third thing -- you *should* connect the two grounds. You may connect them through a resistor, such as 10 or 100 Ohms, for protection sake, but you should connect them.
 

Thread Starter

Flow

Joined May 30, 2010
37
Thanks radachowsky!

Third thing -- you *should* connect the two grounds. You may connect them through a resistor, such as 10 or 100 Ohms, for protection sake, but you should connect them.
Hm, but what's the point then? The potential voltage difference between those grounds would still be there then.

I was thinking about not using the ground from the sensor boards power supply. This would level out any differences and I wouldn't need to think about overvoltage caused by different ground voltages. Wouldn't that work?
 

beenthere

Joined Apr 20, 2004
15,819
I was thinking about not using the ground from the sensor boards power supply. This would level out any differences and I wouldn't need to think about overvoltage caused by different ground voltages. Wouldn't that work?
That would probably not work at all. One thing that circuits must have is a common reference point. The voltage out of your sensor can't be measured by the uC unless the ground is common.

This is very theoretical. Can you give some more information as to the nature of the sensor and its power supply?
 
Hm, but what's the point then? The potential voltage difference between those grounds would still be there then.

I was thinking about not using the ground from the sensor boards power supply. This would level out any differences and I wouldn't need to think about overvoltage caused by different ground voltages. Wouldn't that work?
Yr welcome...

The point is to connect them together, through a direct signal path... not making assumptions about how their power supplies are connected or not, or isolated or not. Also, a more direct path than through the house wiring or whatever. There would be more noise if you don't connect them directly.

The point of a resistor is to limit current if there is a ground loop, to prevent damage. It also lets you sense whether there is a ground loop.

If there is no ground loop, then the operation will not be affected at all, because at low current levels, the 10 or 100 Ohm will not make a difference. If there is a ground loop, then you have other problems and the difference of voltage offset will be the least of your worries, and at least you know about the problem.

That's my philosophy on it.

I think a lot of professional DAQs and other devices do the same -- connect the ground through a resistor of moderate value.
 

Thread Starter

Flow

Joined May 30, 2010
37
Umm, but why shouldn't there be a reference point?



What it boils down to I guess is how the power supplies work. Do they work when 1) their ground is not used and 2) when there's more current flowing back than flew in...
Mh my knowledge is really lacking in that regard.

Can you give some more information as to the nature of the sensor and its power supply?
The power supply is one of those unregulated Voltcraft ones. The sensor board consists of several sensors plus circuitry. All it needs is -any- voltage between 7.5V and 30V.

Thanks alot so far everyone! :)
 

BMorse

Joined Sep 26, 2009
2,675
Basically, you just need to connect the power lines to your sensor board (+ and Ground), then connect the power supply you are using with the UC (+ and ground) then connect BOTH those grounds together at a common point....

Common ground.jpg

B. Morse
 

Thread Starter

Flow

Joined May 30, 2010
37
Ah, so that's basically what radachowsky was saying, except that he suggested a 10-100 Ohm resistor between that GND to GND connection.

Is it essential?

This is for a big class project at university and I'm entirely responsible for this part of the system. A PCB board is part of it, so screwing up is really not an option :(.

Thanks!
 

BMorse

Joined Sep 26, 2009
2,675
You could use a 10 ohm to connect the 2 grounds to be on the cautious side, but in my opinion, it shouldn't be necessary....

B. Morse
 
I would connect both ground together. We are talking low voltage power supplies. If you connect them through a resistor and there IS a ground loop (which I doubt) then you would be defeating the purpose of joining them together as they would no longer be common. If you are worried about the 7.5V of the sensor doing harm to the μP then you could connect it through an opto coupler. If you feel there would be too many losses through one you could try using a voltage divider on the I/P of the μP.
 

GetDeviceInfo

Joined Jun 7, 2009
2,095
Connect your commons together. The less resistance the better, and that includes wire length and solid termination.

The problem you may run into with two supplies is compounding error with drift / regulation variance between the supplies. More specifically, the variances between sensor ref and your ADC ref.

The sensor, converter, and ref loop should be as one, to mitigate drift.

An alternative would be to utilize a sensor that has everything onboard, only having to digitally transmit it's data to the micro.
 

BMorse

Joined Sep 26, 2009
2,675
I have a question. What voltage are you using for the uc

----------------------------------------------------------------------

If you read through the post you will see he said 3.3 volts and atleast 7.5 for the sensor module....
 
Last edited by a moderator:

BMorse

Joined Sep 26, 2009
2,675
Morse, I wouldn't send the output sensor to the micro device without a proper voltage devider. (see post#10)

Alberto

Yes, I know... I was just covering the common ground question the OP was asking about, I have not said anything about how to connect the sensor output to the uC.... But yes, the OP needs some kind of voltage divider to feed the output of the sensor to his 3.3 volt uC, or he will toast the input pin or the whole uC....

B. Morse
 
Top