shift in RTD Reading

Hey all, Using an RTD to measure temperature in a fluid flow line. When a VFD gets turned on to start a fan (600V) the temperature reading drops by
~4ฐC. Using National Instruments Fieldpoint RTD module to read temperature. It is my understanding that noise that causes an offset like this has a DC component. Is this true? If so, Why (or how)would a variable frequency drive cause DC level noise? Any ideas on how to remedy this?
Dan W
Add pictures here
โœ–
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Thu, 27 Nov 2003 22:20:03 -0500, the renowned Dan Wright

Somewhere there is a DC voltage being produced, that much is true.

Poor electronic design of the signal conditioner is causing nonlinearities in the circuit to rectify the high frequency noise, yielding a DC offset.

Hard to do over the internet. Maybe ask National Instruments as it is their fault. You could also try shielding the VFD wires to the motor, for example by running them through a conduit. Certainly don't run the RTD wires too close to (or through the same conduit as) the motor wires.
Best regards, Spehro Pefhany
--
"it's the network..." "The Journey is the reward"
snipped-for-privacy@interlog.com Info for manufacturers: http://www.trexon.com
  Click to see the full signature.
Add pictures here
โœ–
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
If possible and not doing, use differential input.
Are you using twisted shielded pair wiring, all shields grounded to one point. Is signal wiring and power wiring ran in parallel and unshielded?
Look for ground currents or some sort of RC filter or rectifier being formed by the hook up and charging up some capacitance.

Add pictures here
โœ–
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Dan Wright wrote:

I have seen this symptom, not only with temperature transmitters, but pressure and other transmitters, also. It was very scary to see almost every measured signal in the power house shift somewhat the first time we turned on the first PWM drive in the old power house. Apparently the sensitive input circuits act as AM radio receivers and detect the high frequency noise and add some of that DC to the measured signal.
We ended up adding quite a bit to the price of the drives by including both line and motor lead filters (transformers and 5% line chokes on the input and custom made three phase common mode chokes on the T leads), in order to eliminate or reduce this distortion to tolerable levels. But we did experiment with some transmitter filters with fair success. We tried things like putting a ferrite bead on both the RTD cable and the 20 ma line, both at the transmitter. These are available in two piece cores with a plastic snap closure, so they are easy to try. But wound common mode chokes worked better. However, in out situation, it was impractical from a time and labor standpoint to add a pair of common mode chokes to each of hundreds of transmitters. So we spent money on big hardware to stop the noise at the source.
--

John Popelish

Add pictures here
โœ–
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
But where does the DC level signal Come from? Out of curiosity of course... DanW
John Popelish wrote:

Add pictures here
โœ–
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Dan Wright wrote:

Ant high gain signal entry path that is also nonlinear is likely to rectify the high frequency noise into a DC addition to whatever else it is amplifying. This is analogous to the detector function in an AM receiver.
Almost any solid state amplifier will do a little of this rectification when its inputs are overloaded with high frequency noise.
--
John Popelish

Add pictures here
โœ–
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
So my question then becomes, What can I do to eliminate this DC offset, provided I 1) am shielding my instrumentation cable properly, 2) keeping the instrumentation wiring as far away as possible from the VFD power wiring? I can deal with pure AC noise, as this averages out to 0. My RTD signal is running from the UUT to my NI FieldPoint RTD DAQ module. Here, I believe the hardware is pulsing a 250 uA signal, and the 'return' voltage is measured across the RTD.
In the relativly short itme I have been in the instrumentation / control work force (5 years) this is the first time I am totally stumped. How to deal with a DC offset, when you are measuring a DC signal to begin with? I honestly value all of this groups input on this matter.
Dan W
John Popelish wrote:

Add pictures here
โœ–
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Dan Wright wrote:

The spiky input gets into your instrument by capacitive (or direct) conduction. This can happen because the high frequency signal will not equalize out to a single voltage at physically separate points, even if they are all connected together (as building steel is) due ot the propagation speed of light. So if the RTD is mounted on a piece of (probably grounded) equipment some distance from the transmitter, which is some distance from the signal collection hardware (which is also probably grounded) and all the signal paths are enclosed or near grounded metal, there is still plenty of opportunities for high frequency ground noise to get inside the signal paths.
Since it is not usually practical to isolate the entire signal path from ground, except at a single point (and impossible to isolate it from capacitive and radiated coupling), the best you can do is try to reduce the escaping noise at its sources and/or attenuate it at the points where it is being rectified into an offset.
I am not familiar with the abbreviation UUT. But I am willing to bet that the DAC is a node that is subject to this rectification effect. You might try an experiment of putting a common mode choke on the signal lines as close to the DAC input as possible. This common mode choke is best placed such that the excitation current for the RTD dopes not pass through it, because the wiring resistance compensation will not take its IR drop into account. I am not familiar with the details of tour digitizing hardware to know if this is possible or not.
Do you have a way to detail your hardware setup on a web page (including some photographs and schematics)? The possibilities are so great it is hard to make useful generalizations.
--
John Popelish

Add pictures here
โœ–
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
One more quick question: If this is DC being added to the signal, why, then, is the rtd reading dropping? With the increase in signal level caused by this RF Rectification, I would expect the temperature reading to rise, no??
DanW
John Popelish wrote:

Add pictures here
โœ–
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
DC could be positive or negative. Depends on which direction the diode faces that is causing rectification to occur.
Walter.

Add pictures here
โœ–
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Could be a bit of corrosion on the connectors also.
Michael

offset,
keeping
control
Add pictures here
โœ–
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Sat, 29 Nov 2003 17:55:46 -0500, the renowned Dan Wright

My opinion- put some industrial grade 4-20mA transmitters in there, and toss those RTD cards into the nearest dumpster.
Best regards, Spehro Pefhany
--
"it's the network..." "The Journey is the reward"
snipped-for-privacy@interlog.com Info for manufacturers: http://www.trexon.com
  Click to see the full signature.
Add pictures here
โœ–
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
I guess the noise source may be too noisy. We have used other VFD's and I never have seen anything like this. WRT using 4-20 mA xmitters, we have tried this too, and the same problem is seen, ie the temperature drops a few degrees. we also use 4-20 mA pressure transducers too. these show the same symptoms, ie the pressure reading drops by a few psi. The problem is there whether it is a low level RTD reading, or a 'high' level 4-20 mA signal.
Again, thanks for the insight.
DanW

Add pictures here
โœ–
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Sat, 29 Nov 2003 20:25:34 -0500, the renowned Dan Wright

The nuclear weapon is to put isolators in there. Have you tried that?

Best regards, Spehro Pefhany
--
"it's the network..." "The Journey is the reward"
snipped-for-privacy@interlog.com Info for manufacturers: http://www.trexon.com
  Click to see the full signature.
Add pictures here
โœ–
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Are your all grounds at the same potential or do you just think they are? Could very well be a ground loop problem. I really think you have a grounding problem, maybe in the VFD. VFD's are hi power inverters using square waves and thus very noisy by nature, over a broad spectrum.
You could use isolated and maybe amplified signal conditioning at the signal source with differential inputs and a current loop output to your differential instrumentation inputs.

Add pictures here
โœ–
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
I'd just like to thank every one for their input the past couple of days. There are some things I want to try next week when I'm back to the grind.
DanW mindspringnews wrote:

Add pictures here
โœ–
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
mindspringnews wrote:

With a 600 V VFD, this is probably a pretty big motor - 25 - 100+ Hp. With a VFD of that size operating, "ground loops" are unavoidable. The magnetically induced currents in every conductor (whether wire, water pipe, structural steel, rebar, air conditioning ducts, etc.) will be substantial, and will couple to EVERY conductor within reach of the radiated field. Tight cabling in conduit of the input and output cables will reduce the field somewhat, but beyond a certain point, it becomes very difficult to suppress the EMI. At that point, you have deal with it in the instruments that are affected by it. Supressing the (high frequency) AC components with filters is almost always successful when you get a filter configuration that shunts all the induced currents to a safe grounded node. As long as this RTD conditioner uses DC excitation to sense a DC signal from the RTD, filtering should be pretty straightforward.
Jon
Add pictures here
โœ–
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Dan Wright wrote:

The most productive step would be to put a low-pass filter at the input to the RTD conditioner module. (It is also possible that interference is getting into the current source that powers the RTD sensor and altering the excitation current, in which case those lines need filters, too.) Generally, a Pi filter with a modest series inductor, say 100 uH, and a capacitor to ground from each end of the inductor, say .1 uF, should take care of the kind of frequencies that would cause the rectification errors. In worst-case situations, it is sometimes necessary to match the component values of the L and C to get the greatest common-mode supression without inducing a possibly greater differential mode signal.
Jon
Add pictures here
โœ–
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
I had same problem using 4-20 ma speed control signal to the VFD and using 4-20 ma pressure transmitter on the same PLC. I switched the VFD control signal to 0-10 volts and it fixed the problem. I think that VFD input was grounded in a way that there was voltage difference between VFD and pressure transmitter.
Good luck
Martin Gagnon

--
Outgoing mail is certified Virus Free.
Checked by AVG anti-virus system (http://www.grisoft.com ).
  Click to see the full signature.
Add pictures here
โœ–
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Polytechforum.com is a website by engineers for engineers. It is not affiliated with any of manufacturers or vendors discussed here. All logos and trade names are the property of their respective owners.