Error in ELVIS Oscilloscope Readings

[Tom_J.]

Hello,

We are running ELVIS Traditional software and are using the origional ELVIS units.

We are getting some erroneous readings on the oscilloscope.

When we input a square wave into the scope directly into Channel A, it displays fine.

But Channel B shows an output (shown in attachment) whether a probe is attached to the BNC connector or not.

If we turn Channel A off, Channel B goes to 0V as it should.

We never had this problem before and it is happening on all of the ELVIS units in our lab.

Could it be a software problem?

Thanks,

Tom J

[Kyle_A.]

Hi Tom,

It sounds like you have an undercompensated probe you are using with your ELVIS boards.  Check out <link no longer exists> on the subject.  You should have a trimming potentiometer on your probe that you can use to increase compensation and sharpen your square wave corners.  Hopefully this will resolve your issue.

[Kyle_A.]

I just noticed you said you are seeing this signal whether it is actually connected or not.  That is very strange behavior, did you recently do a software upgrade on your system?  It is exactly replicating your FGEN FUNC_OUT signal in an undercompensated way, do you have any other code that is connecting your Function Generator to your CH B?  If you connect your function generator to CH B and look at the BNC on CH A, do you see the same result?

[Tom_J.]

HI,

Tried what you suggested, connecting the function generator to CH B and looking at the BNC on CH A and got the same waveform result on CH A.

If I turn off CH B, then CH A goes to zero.  Same problem in reverse.

I don't believe we have any other code running that would cause this and it is happening on all 20 of our lab computers.

Some have the data cards inide the computers and some are the USB type.

The only thing that has changed is that we upgraded our computers to 64-bit machines and have installed Windows 7.

I also tried downgrading from ELVIS Traditional 3.0.5 to ELVIS 3.0.1.  Same problem.

Thanks for your help,

Tom

[charlesy]

Hello Tom,

It looks like you are seeing an artifact of multichannel settling. In the original ELVIS, the scope channels are routed to analog inputs on your DAQ card (the PCI or USB device). That DAQ card has a single Analog to Digital Converter, so if you want to measure more than one channel, it uses a multiplexer to route each input (one at a time) to the ADC. When the multiplexer switches to a new channel, the signal on the input drives the ADC to its voltage.

The problem happens when the source impedance of a channel is very high (or infinitely high in the case when nothing is connected). A high-impedance source doesn't have enough time to dissipate the charge on the ADC from the last channel, so you see crosstalk from one channel to the next.

For example, when Channel A is driving +10V and Channel B is open, the ADC charges up to +10V while the switch is connected to Channel A. When it changes over to Channel B, there's nothing driving it and nothing to dissipate the charge left over from the +10V signal, so Channel B appears to measure +10V also.

The situation is exacerbated by scanning quickly. If you slow down the acquisition rate (timebase adjustment in the ELVIS Scope), the effect will get smaller -- although an open channel will take a long time to discharge the ADC.

There's a more thorough explanation of this phenomenon in the MIO manuals. Check out page 4-7 under Chapter 4 of the M-Series manual:

http://www.ni.com/pdf/manuals/371022k.pdf

In ELVIS II, we added scope buffers. These are unity-gain amplifiers that drive the inputs of the multiplexer. In that case, it doesn't matter what the source impedance is that you hook up to the inputs, we'll always drive the multiplexer with a low impedance. Additionally, there is a 1 MOhm input impedance to ground. That gives you a proper divider with 10x attenuating probes and also reliably discharges floating inputs back to 0 V. You could add these circuits on your ELVIS 1 protoboard if you need attenuated probes or want better settling behavior.

I hope this is helpful.

Charles Y.

National Instruments