Measuring small dV signal over 600VDC bias

[mechaman1]

Hi!
I am an engineer at traction power substation. In our system 13.8kV is stepped down and rectified to 600VDC, which is fed into DC network feeders.
The amperage is registered through Shunt resistor in series with feeder cable at the feeder 4000Amp and rectifier 8000Amp breakers. Amps are calculated based on 4000Amp - 50mV drop at 600VDC. Analog 0-50mV voltage is wired into Crompton Analog gauge, which receives differential voltage from shunt.
The substation analog gauges don't have data storage capabilities. I wanted to measure Amps, peak, average demand through data recording oscilloscope. We need to measure fast changing in milliseconds differential 0-150 mV signal biased at 600VDC relative to ground. The probes must be safe for CAT IV 1000V job. I looked into Fluke 190 series oscilloscope with insulated inputs 1000V and 8 bit resolution. Not sure if vertical resolution accuracy is enough. Also, I think PC based oscilloscope has larger data storage memory.
Could you recommend what Pico Oscilloscope, bits resolution and probes to use? What is most appropriate for this job.

[bennog]

A 8 bit scope will do at +/- 1000VDC = 2000V top-top 7.8V per LSB.
And usually the LSB has a lot of noise on it.

You will need at least a 14 bit scope then you wil have a LSB of 122 mV
With the enhanced resolution you can go to 18 bits with a LSB of 8 mV

If you want to do a beter yob then you can get a stable DC supply of 600V and measure the difference with the 600V DC you want to know the variation of (be sure to use a properly rated differential probe)

I would use a picoscope 4444 with a picoConnect 442 probe and a TA301 AC/DC current probe.
So you dont have to worry about the battery in the probes.

Edit: checked with my PS444 and 442 you can do a 600V DC offset for the +/- 250V range
so you get a LSB of 1 mV



Benno

[Gerry]

Hi mechaman1,

First of all the PicoConnect 442 Probe is NOT suitable for your application. The Probes are CAT III not CAT IV. I would suggest searching for a reputable supplier of CAT IV probes on Google.

The PicoScope 4444 has a relatively high noise floor so although it will give you an improvement over 8-bits it may not be a significant enough improvement. Also, a fast changing (small signaI, relative to the larger signal) could get filtered out using our Resolution Enhancement. So, ideally, an inherently lower noise 12-bit Scope may prove to be the better option. You may be better off with a PicoScope 4424, if your measurements will be single Phase, or if you are going to be using the 2 wattmeter method of measuring average demand for 3-phase. If you will be using the 3 wattmeter method for 3-phase then the 8-channel PicoScope 4824 would be a better option.

Regards,

Gerry

[mechaman1]

Thanks for replies.

I will be measuring current flow due to train acceleration through the series resistor shunt at the substation breaker in order to set properly protection relays. Typically, single train (6 cars) pulls 1000-2000Amps (12.5mV-25mV) at the beginning for few (2-3) seconds, then load ramps down to steady state 500amp. Typical DC Breaker set at instantaneous 4000Amps (50mV differential voltage at relay) trip.The load is shared between two breakers at each end of the power section. During short circuit fault the current flow at the breaker may be lower than train load current at long distance. dI/dt relays are used for that purpose to differentiate between high inductive impedance of the train motor and low inductance cable fault dI/dt cable > dI/dt motor.
Would PicoScope 5443D with 16 bit resolution and TA-042 1400V 1:100 differential work? What is minimum signal resolution could be measured in differential mode (1mV desired )?
I don't expect a lot of noise at 600VDC because from observation analog gauge needle moves fast in response to load without jitters flutter.
Sincerely,
Akmal

[mechaman1]

Good morning,

Would it be possible to connect output of high voltage differential probe into bnc-bnc preamp to amplify small differential signal to make it visible? Or the Noise will blanket the signal?

V/r,
Akmal

[Gerry]

Hi mechaman1,

O.K. now that you've explained your application in more detail, I can see there is no AC usage involved (or phases to consider), just transients. As you're measuring a differential voltage at the relay, to get the precision that you want, you need to be able to reject the 600V DC signal level during the measurement, which differential probes, with a common mode voltage in excess of 600V would do for you (to answer your question, the TA042 is a 1400V CAT III probe, so it has the common mode rejection, but not the CAT IV protection that you need, so you will need to find an alternative, as we don't have one). You could then use a 12-bit PicoScope as you would only need to measure between 1 and 50mV (or even 1 and 150mV as you mentioned in terms of the transient). Even if you lose 4 bits of resolution to noise (as you would be using one of the smallest input ranges) you would then still have 8 bits of Effective Resolution allowing you to measure between 1 and 200mV.

You need to be aware that at this level of voltage reduction differential probes typically don't provide very accurate measurements, i.e. 1% or 2% accuracy are typically quoted specs. So, you would need to look for accuracy in the specification, because it could mean that although you would have the precision you need (e.g. you would be able to detect the 1mv difference between getting a measurement of 50mV followed by a measurement of 51mV) you may not have accuracy to go with it (e.g. at ±2% accuracy, which typically means "% accuracy of reading", you would only be able to say with any certainty that the first measurement was between 49-51mV, and the second between 50-52mV).

If it turns out that you can't find the Probe that you need, then your only other practical option that I can think of would be to use a precision high voltage isolator (you mentioned the need to measure a fast changing signal from 0-150mV in milliseconds. If you need to capture this then, if we say that it represents a worse case scenarion of 15-135mV in 2ms, then that would be a Rise-Time of 2ms, representing a bandwidth of 175Hz which should be slow enough for any kind of isolation method, i.e. transformer, opto, etc).

A Bandwidth of 175Hz would require a sample rate of 875 samples per second which would mean that you could use the Streaming mode of our PicoScope 6 software to stream the data directly to the PC with no need for a large buffer on the Scope itself. This would give you almost 14 hours of continuous monitoring with the ability to capture the transients as well. If this is not enough time for logging the activity, then you could use an 8-bit scope such as the PicoScope 2204A to log the data continuously at, say 1kHz, for much longer than 14 hours, using our PicoLog 6 software, and use the 12-bit Scope to capture the transients with a trigger (and automatically save the data to a file with a time stamp, and then be ready for the next transient).

To answer your other questions:
You could use an instrumentation amplifier to amplify the signal in between the differential probe and PicoScope input channel (which would help with the noise of the input channel itself) but then finding a cost-effective suitable quality instrumentation amplifier module is not that easy to do.
The 5000 series PicoScopes have noisy inputs in comparison to the 4000 series PicScopes, so you have to use a lot of post processing to try and remove the noise for the smallest measurements (which could wipe out your transients, as previously mentioned). I'm assuming that you were interested in them because of their deep memory which would be good for long term logging, but you wouldn't necessarily need a deep memory PicoScope as you could use the memory of your laptop (as mentioned). So the 5000 series PicoScopes (even if you could effectively use the noise reduction) would be overkill for your Specific application.

Regards,

Gerry