Incorrect readings in differential channels

[IñigoP]

Hello,

I have a problem with the ni 9205 using it on a cRIO 9025. I'm using it to measure some sine-wave signals with the differential inputs. The problem is that I have some of the inputs unwired, but I still see the same sinusoid in the unwired channels.

In the image "Captura2" you can see the graph Vrst, where the wired differential inputs (0, 1 and 2) are conected (is the same input). Also, you can see the graph Irst, which corresponds to the unwired terminals.

If I measure the unwired differential channels with the oscilloscope I see the signal on the image.

[Artem.SPb]

Why do you think this is a error?
Contacts in the air, they can have any electrical interference.
If you want to see there zero, short-circuit the contacts.

[IñigoP]

Because I see the same signal in both wired and unwired terminals. I think it can't be noise because of the amplitude of the signal (same as the measured signal), if it was noise the amplitude should be smaller. 

[Artem.SPb · Accepted solution]

see The NI 9205 channels share a common ground (COM) that is isolated from other modules in the system. All channels share a programmable gain instrumentation amplifier and are multiplexed to an ADC.
There are only one ADC for all channels, not one ADC for each channel. This is the reason, why noise is equal to signal on another channel

[crossrulz · Accepted solution]

---

@IñigoP wrote:

Because I see the same signal in both wired and unwired terminals.

---

It is a form of "ghosting", which is what happens when you share an ADC through a MUX and you have high impedance (or open) sources.  The reason for this is because there is a capacitor at the ADC to reduce noise.  But if you have an open source, there is no path to drain that capacitor and it just holds the voltage.

Note that this is not an NI issue.  It is a fundamental issue for all DAQs, DMMs, etc that share ADCs through a MUX.

[Henrik_Volkers]

---

Note that this is not an NI issue.  It is a fundamental issue for all DAQs, DMMs, etc that share ADCs through a MUX.

---

Sorry, NO 🙂  That is common for cheap mux multi channel DAQs ...

Better ones spend an input buffer for each channel 😉 and some even give you the choice for 10/1Meg input impedance   that limit the input drift on open chanels to Ri*Ibias ...

(And the marketing >10GOhm input is also only valid if you don't have to recharge the S&H cap)

[IñigoP]

Thank you for your fast response fellas!

[JÞB]

Another thing to consider. 

You're getting close to the 20V limit.  If the input and I see some evidence of compression in your picture.   A 2:1 divider and a scale might be helpful.