Spikes on non-connected channels

[vansoest]

Hello,

We are seeing spikes on our non-connected channels. Has anybody an idea what could be the cause of this. The scope PCI-5105 is triggerd on a channel, runs at 60MS/s, al channels are measuring over 1Mohm. When the spike is visable on the non connected channel, it sometimes also aprears on a measurerd channel. The spike is on sample long.

Thanks,

Robert

[Nathan-P]

That is interesting.  Could you post some pictures of the spikes on the connected channel and non-connected channel?  Also, is the spike a time domain spike, or a frequency domain spike?  If its time domain, does it make the signal discontinuous?    Also include the Serial number for your board.

Some things to try in the meantime:

1. Run Self-Calibration

2. Place terminators on the unused input lines.

-Nathan

[vansoest]

Hello Nathan,

Could you help me out and tell me where I can read something about the time/frequency domain?

By termination, you mean something like this : http://nl.rs-online.com/web/p/rf-terminators/7165034/ ?

Here is an image from a spike, both on the connected and disconnected channel. The scope is running on 60MS/s, so the spike is 16,7ns or smaller, not? And I think its strange that one channel goes up and the other down.

The scope board serial is: 00EFECCB

And I have done a self calibration already.

I hope you have a suggestion.

Robert

[Nathan-P]

How often does that spike appear?  Every acquisition?  Every once in a while?  Is it always in the same spot?  

When I mentioned a terminator, I meant a Load terminator (short), like one of these:

http://www.pasternack.com/smb-50-ohm-rf-load-terminations-category.aspx

Also, the graph you posed was a time domain graph.  It plots the voltage input on the Y axis, and time on the X axis.  A frequency domain graph is the same except the x-axis is frequency instead of time.  Its obtained on Scopes by running the acquired data through a FFT (fast fourier transform).  

http://en.wikipedia.org/wiki/Frequency_domain

The spike appears to be just one point, which could mean that its a digitial failure.  But the strange part is that it appears on the other channels as well.  Does it appear on all of the 8 channels, or just some of them?  Which channels?  It could be appearing on the other channels due to cross-talk.  

Does this only happen on this one NI 5105?  Do you have any other 5105 units that you could connect your signals to, this would help verify if the problem is in the NI 5105, or ouside of it.  Could you explain what you have connected to the NI 5105, and how they are connected?  

Also, another test would be to disconnect all your channels, and see if the spike still occurs.  This would eliminate your connected devices, as a source of the spike.

I hope this helps,

Nathan 

[vansoest]

The spike(s) appear once in a while, but on the same spot. (the signal is triggerd by a channel) But I have also seen multiple spikes per channel. Channels 3,4,5,6 seem to be the most sensitive for the spikes. Currently we don't use 0,4 and 7.

We use the PCI-5105 with an input impedance of 1 Mohm on each channel. Because of the measuring range of the PCI-5105, each sensor has a voltage divider (input 50 ohm). The divider's grounds are shared/connected to each other. All but one connected with the default cable (http://sine.ni.com/nips/cds/view/modelpopup/p/pcat/1005/lang/nl ), the other one had a longer cable, 3 m.

This 5105 is the only board we own.

Is there a test to see if it is a digital failure? Like, to increase or reduce the range of a channel? The spike should stay the same, not? I will try to do a test with all sensors disconnected. But it may be nessecery to keep at least one connected, due to the nature of our setup. (we must keep eyes on the process.)

And by the strange part you mean that all channels are also digitized simultaneously?

Ok, the specs of the load/terminator appear to be equal.

The domain type is now clear to me.

Thanks in advance,

Robert

[TimmTheEnchanter]

Hi Robert,

Is this "spike" always of the same magnitude on each channel?

[vansoest]

Yes, or at least if one channel has multiple spikes, they're te same. A measured channel has a smaller spike. And one spike goes upwards, the other could go up or down.

[TimmTheEnchanter]

In other words, multiple spikes could happen in the same acquisition?  Could you show an image of that happening?

I think I understand that the channels you actually are measuring (are connected to something) have smaller spikes at the same time as a channel you are not measuring (is unconnected, but still part of the acquisition).  Furthermore, I'm curious if the spikes are of the same amplitude from on acquisition to another acquisition, or if the amplitude varies among acquisitions.

-Andrew

[vansoest]

Yes, the measured channels have smaller spikes.

The amplitude varies with the "load" of the measured system.

[vansoest]

And further more: the magnitude varies between input impedances.

1 Mohm not connected has higher spiks then 50ohm not connected input impedances

Regards,

Robert