PCI-5102 defect

[darkxceed]

Hello,

I am wondering if I did something wrong with my 5102 digitizer.

It worked for some weeks in the production but now it doesn't work, just a noisy half sine wave.

After inspecting I found 3 wires that where burned out on the PCB.

When in the production I put 5,15,-15,24 and 78VDC and I measure the AC component of it to see if there is not too much noise.

Can someone tell me what I did wrong?

KindRegards,

Bart

[ninlmartijn · Accepted solution]

Hello Bart,

thank you for your forum post! The NI 5102 has a measurement range (with a x1 attenuation probe) from +/- 5V. The overvoltage protevtion is (DC + AC, and AC with maximum of 10 kHz!) +/- 42V.

Best regards,

[darkxceed]

Hello,

So what does that mean?

Is it protected till 42V but at 80V the PCB can burn out?

Is the input isolated from ground?

For now I have made a 3:1 devider, now only 80V/3 thus 26.6V will be put on the scope input.

Will this be a solution?

Bart

[ninlmartijn]

Hello Bart,

thank you for your respond.

Mostly a attenuation probe is used to divide the voltage. The following table can be found in the help file from the digitizer;

Input Range
Gain	X1 Probe	X10 Probe	X100 Probe
1	±5 V (default setting)	±50 V	±500 V
5	±1 V	±10 V	±100 V
20	±0.25 V	±2.5 V	±25 V
100	±50 mV	±50 mV	±5 V

Besides the attenuation probe you are able to set the right gain. Without attenuation the digitizer was made for a +/- 5V voltage range, the 42V (only with AC 10 kHz) is a specification that your hardware is protected agains these type of voltages. Because of the burned traces I would highly recommend you to verify (calibrate) if the card is working within the specifications.

Best regards,

[Matt_E.]

Hi Bart,

The 42V peak spec means that the input to the channel (center pin of the BNC) can withstand up to 42V relative to the board's ground.  Since the maximum input range of the 5102 is +/-5V this means that if you put more than 5V on the center pin of the BNC (or whatever the maximum voltage of the range you have selected) the amplifiers in the signal chain of the board will go into saturation.  In the board will report it's maximum voltage reading of the range  you are in but will not be damaged as long as the signal is less than 42V.  In general it is not good practice to do this because it takes time for amplifiers to recover for saturation, meaning that even once your signal comes back within range there will be a period of time where your measurements are not accurate.

The input channels of the 5102 are not isolated for each other of from the board's ground.  You should treat the BNC's outer shell as a ground connection and make sure that the other end of your cable which connects to the board's BNC shell is tied to the same potential as the board's ground.  The 5102 input channel is configured in a pseudo-differential manner.  What this means is that the input stage has a differential amplifier that senses the potential at the BNC shell to allow the board to remove small DC offsets caused by ground current loops from the measurement.  To achieve this the BNC shell is hard tied to ground and can float a small amount.  The input amplifier can not withstand large DC voltages on the BNC shell so clamp diodes are used to make sure the shell stays within a diode drop of the board's ground potential.  The intent of this type of input is only to correct for ground current offsets so it is important to make sure the voltage on the BNC shell stays within +/-0.6V to prevent damage to the clamp diodes.

Using a voltage divider can work to extend your input range but it can result in a loss of bandwidth due to the filter created by the tope resistor and the combined capacitance of the digitizer input and your cable.  If all you are trying to measure is DC then you may not care about this bandwidth loss.  If you do require the bandwidth then your best bet would be to use a 10x probe.  Which ever approach you use you can enter the attenuation factor of the probe or divider into the "probe attenuation" control on the niScope configure vertical block.  This then allows you to enter in your vertical range you want (100Vpp for example if you were using a 10x probe on the 5102 and wanted the largest range) and the software will automatically select the appropriate range of the digitizer.

Since you said you saw burnt traces on your board it is most likely that the damage was caused either by exceeding the 42V peak input voltage on the center pin of the BNC or by applying a voltage to the BNC shell.

Hope this helps answer your questions!

Regards,

-Matt

[darkxceed]

Hello,

I only want to see the AC component of the signal.

Now I made a voltage devider with a 2M resistor, so the voltage measured is devided by 3.

This way the DC voltage will be below 42V.

My setup is simple, 5 power supply's 5, 15, -15,24 and 78VDC, they are switched directly to a programmable load and the scope card then measures the noise on the output of the powersupply.

The burnout must be because of the 78VDC, or maybe because badwireing of that supply, giving 78V on the outershell of the BNC and GND on the inner pin, could this be true?

Maybe I have to use a isolated scope or something I could put between the powersupply and scope so that it is isolated.

Does NI have something like this?

Bart

[Matt_E.]

Hi Bart,

Given that you are trying to measure noise putting a resistor in series with the input is probably not a very good idea.  The 5102 has an input capacitance of up to 10pF (not including any sort of cable you connect between the resistor and the input).  With a 2Meg resistor in series with this capacitance you would have less than 10kHz of bandwidth.  If your supply has any noise outside of this bandwidth it will be attenuated by the filter you have created and you will measure less noise than is actually on the line.  

It sounds like even though your DC voltages are somewhat large the AC voltages you actually want to measure are probably small.  If this is the case then you don't really want to be attenuating at all because you will be attenuating your small AC voltages and loosing resolution.  Your best bet might be to use AC coupling to remove the DC offset from your AC noise signal.  This will allow you to use smaller ranges of the digitizer so you get a good full bandwidth measurement of the noise.  The only thing you loose by AC coupling is the ability to measure the DC output of your supply.  If this is important to you KHz could use the second channel of your 5102 along with a 10X attenuation probe or the resistive divider you have made to measure the DC only.

The 5102 has built in AC coupling and even though the AC coupling capacitor on the board can withstand the voltage you have in your setup it probably would be best to add an external capacitor instead of using the one built on the board.  The reason for this is that if you are applying >42V DC to the input in AC coupled mode and then accidentally switch back to DC coupled mode with software you could damage the input.  If you place your own capacitor on the input you won't have this problem.

I would recommend using a 15nF ceramic X7R capacitor for your AC coupling.  Make sure that it is rated to a voltage of 2 times what you expect to apply to the input.  This capacitor will form an RC high pass filter with the 1Meg input resistance of the digitizer but the cut off frequency will be ~10.6Hz.  If you need to measure noise even lower in frequency than this you can use a bigger capacitor.

Here is one part that could work but there are many others:  VISHAY  VJ1206Y153KXCAT   Mouser Link

 As far as whether your need isolation or not this really depends on the supply you are trying to measure.  If the outputs of your supply can be referenced to the same ground as your digitizer then you shouldn't need isolation.  (i.e.. if the power supply's ground could be tied to the computer's ground with a wire without damaging your supply).  I would take a digital multi-meter if you have one and measure between your supply's ground and the ground of your computer when both are turned on but the digitizer is not connected to your supply.  If you don't measure a voltage >0.6V then odds are your supply ground could be tied to the same ground.  To prevent potential damage to your digitizer you can tie both your supply's ground and the computer's ground to a common earth ground through external cables such as 12 gauge building wire.  Again this is all very specific to your exact setup so I won't be able to tell you exactly what will work for your needs.

Hope this helps,

-Matt

[darkxceed]

Hello Matt,

I have thought about a capacitor.

- The measurement was always in AC coupling ofcourse, but even in AC coupling, will the scope not be harmed when putting a signal >42Vdc?

- In AC coupling(high pass filtering) there will also be a high spike at the beginning, could this also not harm the scope?

If this all will not harm the scope I will put a 15nF capacitor you recomended in serie on the input of the scope.

About the supply's, some have common ground others not.

Bart

[Matt_E.]

Hi Bart,

As long as your DC blocking capacitor is rated to more than the DC voltage you are applying to it then the DC component of your power supply's output will not harm the digitizer.  The input of the digitizer can be modeled as an ideal amplifier input with a 1Meg resistor in parallel with the input and ground.  Therefore in the steady state your DC signal at the digitizer input will be at the sample potential  as the board's ground and nothing will be damaged.  You do still need to make sure that the Component of your signal (noise or anything else) stays below the 42V. 

You are correct that there is still the case where you connect your digitizer's input to the power supply that can generate an edge that would make it through the blocking cap.  Depending on the ramp rate of your supply you could connect the digitizer before turning on the supply to reduce the speed of this edge and minimize the spike.  You could also add a relay between your power supply and your input with a large value resistor in parallel with the relay contact to filter the edge.  You would leave the relay off until the supply has been turned on and enough time has passed for the DC value to settle on the DC blocking capacitor than turn on the relay to allow your full bandwidth signal through.  (See attached diagram AC coupling with filter.gif)Basically in this setup you can reduce the magnitude of the spike by increasing R1.  The larger you make R1 the longer you will need to wait for the value to settle before turning on the relay.  The obvious downside to this approach is that it requires more external components plus you would need to provide a way to control the relay.

In reality it sounds like most of your supplies are less than 42 volts so this wouldn't even be an issue for those supplies.  Even for the ones that are you could be OK because the 42V rating is really intended to apply to DC levels which will cause constant heating of the clamp diodes in the 5102 input.  Short duration spikes would be unlikely to cause problems.  If you are still concerned another approach would be to remove the relay and R1 and put your own clamp diodes between C2 and the BNC input.  You would connect these diodes to rails that are larger than your expected signal but smaller than the input range of the device (+/-5V).  You can see a diagram of this configuration in the attached file AC coupling with clamp.gif.  By setting the voltages to these clamp diodes less than +/-5V you ensure that your external diodes will activate before the ones on the board which will protect your board's input.  This method is easier to deal with than the really in the sense that it requires no control but will require you to provide the power rails for the clamp.

Hope this is helpful

Regards,

 -Matt