Is there a method to programati​cally detect when nothing is connected to the physical channel?

[busarider29]

I have searched the knowledge base a learned that the floating voltages that I am seeing on my graph when nothing is connected to the physical channel is normal for NI devices.  So is there a way to programatically detect if there is nothing connected to the physical channel??  I am using a cDAQ9184 with a 9205 AI module.  I'd like to disable my wafeformgraph or the wafeformplot of the graph via coding if there is nothing connected to the physical channel in question.  I posted this question to the Measurement Studio forum and I was directed by NI to post the question here as it relates to the DAQmx drivers.

Thank you

[Dennis_Knutson]

It relates to the hardware and not the driver. Not odds it unique to NI hardware. You can read an article on ghosting at http://digital.ni.com/public.nsf/allkb/73CB0FB296814E2286256FFD00028DDF

If you have users that connect/disconnect channels, I think the best thing is to ask the user what they have done before starting the acquisition. I have no suggestions for a robust automatic method.

[busarider29]

I will have technicians on the floor that will be plugging and unplugging transducers as well as engineers using the application for testing.  In most cases, every channel will not be in use or wired.  When that is the case, I will have floating voltages showing up on each channel's respective WFG.  This is not a problem but I can see technicians and engineers scratching their heads and wondering what's "wrong".  Asking them every time they wire up something or unwire a transducer before starting the acquistion just isn't practical as it will be very often.  I can't be standing out there or having them come to me every time they connect/disconnect a transducer.  If I can disable the WFG or the plot of the WFG when nothing is plugged into that channel, it would prevent any confusion.

[John_P1]

---

@dm@Means wrote:

If I can disable the WFG or the plot of the WFG when nothing is plugged into that channel, it would prevent any confusion.

---

If I walked up to your test station and saw the measurement drifting up to one of the rails or reporting the same measurements as the previous channel I would probably think to myself "hmmm... let's check our connection, it looks like the channel is floating".

If I came to the same test station and I saw the graph was disabled or one of the channels was missing, I would probably assume there was some sort of software error or I hadn't configured the software correctly.

That being said, it isn't really possible to distinguish anyway since an arbitrary input signal might very well be behaving in the same manner as the floating channel.  If you know what your input signal *should* look like, you might be able to come up with some sort of detection algorithm if you really want to--but then again if the channel doesn't meet the criteria of the algorithm and is missing from the graph you wouldn't have any way to diagnose the issue.

If it were me I would instead provide a means for the operator to just set (through the software) the channels they will be acquiring from.  It shouldn't be unreasonable to ask a technician or engineer to specify/connect the signals that they wish to measure.

Best Regards,

[busarider29]

For your first scenario, there is probably maybe 3 engineers here (myself along with two other EE's) out of 50+ that would even know what a floating voltage is and what that means.  90% of our engineers are MEs.  Right away they would think they had something wired wrong or there was something wrong with the program.  And then what do you think they're gonna do next?  BINGO....come and hunt me down and ask me "What's wrong??  Why is it showing this voltage on the other WFG when I'm not even using that channel???"  Now, repeat this scenario 30+ times until everyone realizes its normal when nothing is plugged into the other channels.  Then at that point, everyone will want me to "fix" it so it doesn't do that anymore.

Now, if I disable the WFG if nothing is plugged into the channel (if its even doable, which at this point it appears not), of course I would program an indicator of some sort on the WFG to indicate to the user that the hardware is not detecting anything plugged in to the channel, therefore the WFG is disabled.  Essentially, do what I always do when programming and disabling a control - tell the user WHY.  So then what do you think the Engineer/technician is going to do in this scenario?  BINGO.....check the channel on the DAQ to be sure they have something plugged into it.  And if they are not even using that channel and using a different one, they will simply go about their business because they'll already know why its disabled.  And then everyone is happy....YAY!

[ChrisR.]

There may be something you can do, depending on what your transducers and expected signal levels are.  For example, suppose your expected signals are all in the ±100mV range, and that you scan a dedicated 10V channel between every transducer measurement.  Then if a transducer is disconnected, the floating channel will tend to follow the 10V, and will very likely exceed the expected ±100mV, so you could catch the problem.

Doing that will also dump a little charge from the 10V into each transducer, which may or may not cause a problem, depending on the nature of the transducer.

[Dennis_Knutson]

It's not unreasonable to require a 5 or 10 minute lesson to everyone that would be using the software. It's not unreasonable to provide a written manual at the test station that everyone can refer to.

What sort of system do you have with the connections? Is it an easy visual indication? Are they easily physically accessible? Can you easily connect a ground connection when a sensor is removed?

[crossrulz]

---

@ChrisR. wrote:

There may be something you can do, depending on what your transducers and expected signal levels are.  For example, suppose your expected signals are all in the ±100mV range, and that you scan a dedicated 10V channel between every transducer measurement.  Then if a transducer is disconnected, the floating channel will tend to follow the 10V, and will very likely exceed the expected ±100mV, so you could catch the problem.

Doing that will also dump a little charge from the 10V into each transducer, which may or may not cause a problem, depending on the nature of the transducer.

---

Why go with 10V?  Use AGND!  You want to remove charge, not add to it.

[BuddyHaun]

If it is critical that you be able to detect disconnected channels I would recommend using 4-20 ma transducers and matching DAQ hardware. When you have 0 ma the transducer is disconnected. That is one reason that 4-20 ma is popular in industrial automation. disconnected sensors are easily detected.

[Henrik_Volkers]

- Ask the Operator before

- Use plugs with a internal switch/contact for auto detect

- Use current loop,IEPE, TC sensors that have open/shortcut detection

- use DAQ hardware with 'intelligent' connectors ... the automobile industry here use it a lot.... higher channel cost pays against robust, fast,.., ..... setup  

You can't make it fool proof because fools are so genious 😄

For multiplexed DAQ cards, one migth test with internal mux channels (AGND,AIx,Vref,AIx)  and measure something like settle time and make an 'educated' guess of the source impedance

Yeah... ghosthing is not a bug... it's a feature 😉

wonder if there is patent claim:  using the residual charge of the sample and hold capacitor in conjunction at least two known input channels to detect a open third input channel ..... 

(if not, this is prior art know :D)