I'd like to get opinions in regard to this. I've been given an assignment in which I need to find a way to hook up mains voltage (up to 600Vrms) to a Labview CompaQ Daq with module NI 9201. It accepts a -10V to 10V signal, and provides isolation up to 2,300Vrms from other modules in the system.

Generally, most people say do not use voltage dividers from mains unless you know exactly what you are doing. I'd like to use a voltage divider to reduce the high mains voltage to a -10V to 10V signal. Would this setup be considered 'safe' given the high isolation voltage of the module?

 

Are you trying to measure the mains voltage, or just sense the presence or absence?

Ken

 

Why can't you use a transformer? It may be more expensive but it will be much more safe.

 

We are looking to measure the voltage for a certain period of time to verify that there are no sags or overvoltage. Test was performed before by a meter, but we are looking to record the data using this module, instead of monitoring it.

 

Why can't you use a transformer? It may be more expensive but it will be much more safe.

I think this is a possibility. I only went down this road because a voltage divider was initially suggested. From a raw data point of view, what sort of gotchas would using a transformer do to the signal besides reduce it? Add any distortion?

Any suggestions of a transformer that would reduce 100:1? (1000V max to 10V)

 

I think this is a possibility. I only went down this road because a voltage divider was initially suggested. From a raw data point of view, what sort of gotchas would using a transformer do to the signal besides reduce it? Add any distortion?

Any suggestions of a transformer that would reduce 100:1? (1000V max to 10V)

In theory - no distortion. In practice, it will introduce distortion, due to parasitic/imperfect properties of the transformer. If you want to accurately measure mains then a precision voltage divider might be necessary, but they are expensive for higher voltages. How precise are you needing it to be?

 

How precise are you needing it to be?

As far as I understand it, it looks like the standards are between 1% and 2.5% depending on the voltage under test. That doesn't seem like a ton of room for error.

 

I'm looking at using these, particularly part #USVD2-B10M-010-02

http://www.caddock.com/Online_catalog/Mrktg_Lit/TypeUSVD_HVD.pdf

to feed into the module. Does this seem reasonable? Is there any way to use this in conjuction with other parts to make this safer?

 

Since you'll be inputting the voltage into a high impedance circuit, there will be little current draw. Thus, some small transformers will fit the bill. If you have a problem finding a commercial transformer to do the job, remember you can use two small units cascaded to divide the voltage.

Building on that idea, it might be worth considering making your own transformers from some ferrite toroids. Find a wire with e.g. Teflon insulation that will withstand the rated voltage, then put 10 windings on one side and 1 winding on the other. Cascade two of these and you're done. Probably 15 minutes of work or so -- at least, it's worth a try...

 

I'm looking at using these, particularly part #USVD2-B10M-010-02

http://www.caddock.com/Online_catalog/Mrktg_Lit/TypeUSVD_HVD.pdf

to feed into the module. Does this seem reasonable? Is there any way to use this in conjuction with other parts to make this safer?

These dividers all have very high resistance. The resistance/impedance of the DAQ module will be in parallel with one leg of the divider which will adversely affect the accuracy of the divider.

These dividers are all listed as DC devices. There must be a reason they're not specified for AC.

I haven't used every DAQ out there, but most of the ones I've seen are DC input devices.

 

These dividers all have very high resistance. The resistance/impedance of the DAQ module will be in parallel with one leg of the divider which will adversely affect the accuracy of the divider.

Good point. As far as I can tell the datasheet doesn't specify the input resistance of each channel.

I haven't used every DAQ out there, but most of the ones I've seen are DC input devices.

Each channel can swing between +10 and -10 in reference to a common. So I take that to mean it can accept a 10Vpp AC signal without issue.

Since you'll be inputting the voltage into a high impedance circuit, there will be little current draw. Thus, some small transformers will fit the bill.

If I was able to get the voltage divider working, would it be advisable to add a 1:1 transformer to the line for protection?

Like this? http://ww2.pulseeng.com/products/datasheets/SPM2007_61.pdf

It might negate the 'precision' aspect of the divider.

 

Each channel can swing between +10 and -10 in reference to a common. So I take that to mean it can accept a 10Vpp AC signal without issue.

You can not make that assumption. You can take a potentiometer and swing between +/- 10 volts DC with common in the middle. If this unit does accept AC voltages, there will be a frequency response specification for it.

 

http://sine.ni.com/nips/cds/view/p/lang/en/nid/208798

From my limited experience with NI modules, the common on the inputs are not isolated, and are at building ground. Tying the inputs directly to a mains line, even through a voltage divider, is just inviting Murphy's Law at best and a Darwin Award at worst. If just looking for surges and sags, not fast spikes, a stepdown transformer would be the way to go.

Ken

 

Precision for the circuit should not be a requirement. You can build it then calibrate it.

There is nothing wrong with a voltage divider. I guess a lot of these guy's do not remember the old tube days. The divider can be built using good high voltage techniques (space out components, mount them solidly, all bare wires should be insulated, etc...) and tested, then placed in it's own metal casing. You could make it esily with a two row, screw connection barrier strip, like this:

).
And a small aluminum box to house it in.

I'd use a 2MΩ and 20KΩ 1/2 watt resistors. Like I said, do not worry about precision or loading. Just do a good job of calibration. I assume the data will be collected by a computer so calibration will be no problem.

 

http://sine.ni.com/nips/cds/view/p/lang/en/nid/208798

From my limited experience with NI modules, the common on the inputs are not isolated, and are at building ground.

They are isolated but I think there is only a 250v common to to ground maximum.

Yeah, given that and that the input impeadance is 1meg, I'd go with a transformer. If you are monitoring for votage sag over time, distortion should not be a problem.

 

http://sine.ni.com/nips/cds/view/p/lang/en/nid/208798

From my limited experience with NI modules, the common on the inputs are not isolated, and are at building ground. Tying the inputs directly to a mains line, even through a voltage divider, is just inviting Murphy's Law at best and a Darwin Award at worst. If just looking for surges and sags, not fast spikes, a stepdown transformer would be the way to go.

Ken

OOPs!
Went back and read the isolation spec:
"Isolation Type Ch-Earth Ground Isolation "
"Up to 2,300 Vrms isolation (withstand), up to 250 Vrms isolation (continuous)"

Ken

 

You can not make that assumption. You can take a potentiometer and swing between +/- 10 volts DC with common in the middle. If this unit does accept AC voltages, there will be a frequency response specification for it.

As far as I can tell there is no frequency response listed. I will look into it. If this module is ill-equiped to handle what I am doing then I'll rewind. Thanks for the warning.

Yeah, given that and that the input impeadance is 1meg, I'd go with a transformer. If you are monitoring for votage sag over time, distortion should not be a problem.

I've been looking for an appropriate transformer, 100:1, and I'm having a hard time finding one. Would a voltage divider into a 1:1 transformer be an option?

EDIT: The -3dB point is 420kHz. It should handle 60Hz fine I would think.

 

The spec sheet lists: "500 kS/s aggregate sampling rate"
So, would that be 500000 samples/second/8 channels = 62500samples/second/channel
or 1041.6 samples per one 60Hz cycle?

Ken

Input impedance:

Resistance 1 MΩ

Capacitance 5 pF

 

After some confusion, I will actually be working with the NI 9125, not the 9201.

http://sine.ni.com/nips/cds/view/p/lang/en/nid/208793

 

So, 1666 samples/60Hz cycle and a differential input.
600VAC = 2048Vp-p. Allow headroom for surges and spikes.

Ken