NI 9205 (Dsub) Card keeps getting knocked out!

[CMSys]

Hello all, 

I am having an issue on one of my testing benches with a NI 9205 card that's plugged into a cDaq 9188 chassis that keeps getting knocked out by what I believe is a transient voltage. The voltage is created by a spark gap in excess of 16kV where its is voltage divided down to a readable level between 4-10V.

When looking at the signal through an O-scope, there are some spikes in both the negative and positive peaks (wave behavior is that of a dampened harmonic oscillation/ typical spark trace https://www.crypton.co.za/Tto%20know/Ignition/burn%20time.html ) that last less than a microsecond that are pretty high. But none the less, the card should be able to handle it accord to the isolated voltage specs (btw it varies from d sub, to screw terminal, to push in terminal). 

I tried using a TVS diode (p/n 15KPA33A) on each of the 6 voltage dividers, but my program was not able to read the voltage. What other component could I use to clamp the voltage (such as an appropriately sized MOV) to protect the card, if this even sounds like the issue. 

Thanks in advance!

[santo_13]

Wait, how is the 9205 related to your spark? don't you have sufficient isolation between the card and your high voltage?

If no, you need to redesign your connections to properly isolate your HV source or be ready to spend for more damaged cards.

[CMSys]

I am reading the voltage produced by a magneto (similar to a cars distributor) that is sent over high tension leads to a spark gap... hence why I added the link to a page delving into the voltage trace of a spark plug. The high power is reduced by a voltage divider (700Mohm and 200kohm resistors) to low voltage signal. 

[AnalogKid2DigitalMan]

Are the resistor rated to the peak voltage levels you are generating?

They could be internally breaking down or arcing over and hence not correctly limiting the voltage you are inputting into your DAQ.

I would at least pass the output of your divider network into a high bandwidth galvanic or opto-isolated module to prevent any high voltage spurious transients from getting into you DAQ card in addition to providing ground isolation.

-AK2DM

[CMSys]

Thanks for the reply, I am going to attach the voltage divider being used. It is difficult to imagine us going beyond the breakdown voltage on the resistor, but there could be a manufacturing defect. I was under the impression putting a TVS diode in parallel with the R2 (the one we are reading across) would clamp the voltage but instead I did see a reading, but I believe that is because the program is trying to read a certain voltage (I really need to see what's happening on the acquisition loop in the program, haven't had time!). Tried flipping the diode both ways, thought maybe it was reaching the turn on voltage for the diode to start conducting. Now thinking maybe a MOV resistor. I've never heard of those components but will do my research!

https://microohm.com/highvolt/pt.html

[crossrulz]

Another thing to consider is that DAQ inputs tend to not like high impedance sources.  I prefer adding a simple op-amp (voltage follower) after a voltage divider to give the DAQ a nice low impedance source.

[AnalogKid2DigitalMan]

Regarding the divider, that was much more than I was expecting. Good choice!

-AK2DM

[santo_13]

You had chosed a TVS with Vbr 33V, any specific reason for that despite you're trying to measure <10V?

I would have gone with something very close to 10V.

Alternatively, you could use a voltage follower using OPAMP in between the voltage divider and the DAQ to safeguard the DAQ.

[CMSys]

I choose this because I had it in our inventory... and the continuous voltage the card can sustain is 60VDC, so it should be fine. I believe that one of the issues, and I cannot believe I missed this, is that the waveform is AC and the diode was.. well.. clipping off the waveform at the zero crossing point. So, what do we think about going with the same diode but using a bidirectional diode? 

Thanks