Ignition spark noise interference

[PKP]

Hi, My project involves using sensors to collect data in a shock tube that employs an automotive (>35kV, 175mJ of energy sparks) and capacitive discharge (2kV, 2Joules of energy arcs) ignition system to initiate combustion of fuel+O2. However, the signals collected by the NI DAQ is overwhelmed by noise caused by the ignition sparks/arcs, as high as 5V or perhaps more because the noise signal is saturated and cut off at 5V.
Background: The DAQ consists of a PXI-1042 chassis with two PXI 6133 cards. The DAQ is connected to a PC located in another room by a 30m fiber optic cable (MXI-4 cards). The sensors are connected to the PXI-6133 cards by means of BNC 2110 panels using the BNC coaxial Analog Input sockets. Pressure transducers (piezo type) output is connected through a signal conditioner to the BNC2110. Other sensors include a load cell and accelerometer (also piezo type, through the signal conditioner) and type K thermocouples connected directly to the BNC2110 panel.
The igntion sparks are discharged at varying rates from 10 to 50 Hz as desired. The required signals from the sensors also vary at these frequencies. Therefore a Low pass filter is out of the question. The sensors when connected to a digital oscilloscope did not show any noise.
All cables (sensor wires, coaxial data cables, and the power cables of the ignition) have been shielded and grounded. The BNC2110 sockets are set to "Grounded Source". However, the noise is still very strong at 5V. Even the channels that are unplugged on the BNC2110 show the same noise pattern up to 5V, which correspond exactly with the firing of the sparks/arcs. We tried putting the DAQ within a grounded metallic box and moving it outside the room, but still the noise is not alleviated.
The only thing that worked is to reduce the current in the ignition by connecting a series resistor thereby reducing the power of the sparks. This way we are able to get good sensor readings. However, the capacitive discharge delivers currents as high as 2000A, again causing a noise deluge. Also higher igntion energy yields better combustion, so we would really like to use higher ignition currents.
Does anyone have any insights as to what we can do to cut down the noise, if I am doing anything wrong or can you just point me in the right direction?
Thank you very much
Philip Panicker
GTA
Aerospace Engineering
University of Texas at Arlington

[e-Cash-Money]

Hello Philip,

From a hardware standpoint it seems like you have done as much as you can to eliminate the noise. I'm wondering if you have tried implementing any sort of software filter to remove the noise pattern. The only thing that concerns me is when the amplitude of the noise rails your signal. In these instances, we won't be able to do anything in software to retrieve the original signal since it is railing at the maximum input voltage. I was also wondering if you have tried measuring your signals in differential mode instead of as grounded signals. I don't have a BNC 2110 in front of me right now but I thought I remembered seeing a switch that toggled inputs between grounded and differential mode. I hope this helps!

[PKP]

Thanks for the reply.
I have not tried differential mode, for two reasons. One is that most of the transducers' casing is the return/negative/gnd terminal, which when fixed into their sockets inherently grounds the terminal as all metallic fixtures of the hardware as well as the building are grounded. Secondly, the number of channels would then be reduced in half if used differentially.
But the odd thing is that in the presence of noise, the DAQ output seems to contain no signals at all. For example pressure would be slowly varying, but in the presence of noise, no pressure signals are visible in between the railed noise signals.
Today when we changed our dc motor for one with higher hp, we noticed even higher noise. All motors and induction machines with commutators are prone to spark noise.
Thanks and any suggestions would be highly appreciated.
Philip

[PKP]

I forgot to mention that we are using DAQ Assistant VI to input the channels into Labview. The channels are selected as "Voltage" sources and the only options available in Terminal Configurations are "Differential" and "Let NI-DAQ Decide". The Reference to Ground (RTE) option and the others, such as AISENSE, are faded and not available for selection.
Could it be this diffential mode of terminal configuration that is causing the noise to be so magnified because the DAQ is referencing it with respect to some other reference and not ground? Spark noise would be common mode noise affecting all cables and terminals. Would using a different VI to input the channels work better?
Thanks
Philip

[PKP]

We may have solved the ignition spark noise problem. On the BNC2110, there are switches located right next to the analog in sockets that can be set to either "Floating Source" and "Grounded Source". Initially we had them all set to "GS" and then we got very high noise levels. Today when we set them to "FS" the noise levels dropped significantly down to a few millivolts. One would think that since the transducers have one terminal grounded, the switch should be at "GS". But the "FS" setting seems to work much better in alleviating common mode noise.
Philip