Signal Conditioning Question: Of Data Loggers, Transmitters and Hydaulic Systems

1. Always use shielded signal cable. Connect the shield solidly to ground, but ONLY on the transducer end. The other end should be insulated from ground. There are times when you might be able to get away with unshielded twisted pair, but unshielded, untwisted signal wire is just asking for problems.

2. If you have a calibration source available, try putting a straight DC milliamp signal into the data logger and see if the spikes still occur. That will confirm or eliminate the data logger as the noise source.

3. Similarly, connect a good quality digital multimeter directly to the transducer output. Set the DMM on DC milliamps scale, and use the max/min recording function to check for spikes out of the transducer. If no spikes show up, connect the DMM in series with your signal loop and compare the meter results with the data logger readings.

4. Check the quality of the power supply output. If it spikes, the transducer output will follow.

5. Check the power supply and grounding of the data logger. Interference can be intruduced from that end as well.

6. Is it possible the pressure is actually spiking? If so, hook up an old chart recorder in parallel with the transducer (on the pressure side), and compare the results.

7. The sample frequency of the data logger may be fast enough that you are seeing "ripple" which was not caught by the chart recorder.

Many thanks pwr2thepeople, I will try your suggestions out and revert. Our works are closed for the day now; 0700-1500 GMT so will not be before tomorrow

Cheers!

HydroScot

I disagree with the advice to ground the shield on the transducer end. The shield and all other ground conductors should be connected ideally at a single point, which is typically at the power supply or electronics to which signals are being sent. Ground loops are buggers and can inadvertently be created, and different ground paths can cause spikes to appear in small signals -- even in 4-20 mA current loops. Improper grounding of pump motors, for example, can actually jerk the ground potential around, and be the source of the problem.

is it a good practice to ground (bond) the 4-20mA shield together with the MAIN power supply (230VAC)?

Hi pwr2thepeople, et al, I have been checking out the wiring based on yours and others input and have 2 areas of concern.

1. The cable to the transmitter was connected to the earth terminal at that end via the shielded strands with the other end (shielding) also connected to the logger (C terminal) which in turn is wired (unshielded) back to the earth terminal on the input of the power supply. I have removed the connection from the transmitter earth and find no difference in the spiking. From what I can gather from your point 1. this should be grounded/earthed at this end and removed from the logger end. I will try this!

2. The power supply is a switching mode (FTR Novus) din rail module with an output voltage of 24Vd+/-1%. The transmitter requires a supply of 9-30Vdc. Do you see a problem with this?

One other point of note is that by way of comparison we have fitted a different logger in parallel with the original system. This is a battery driven logger with its own special cable and plugs. The read out is quite robust, and when downloaded through Excel, the graph shows no signs of spiking. However, to keep the battery charged we have plugged it into the mains, and although no problems in itself, causes even more spikes with the original logger. This is proven by disconnecting the second logger from the mains and running from the battery only.

In case you wonder why we are bothering with the original unit when the second logger gives such good results there are several reasons.

1. The setting up and downloading of the second unit is quite ardous.

2. The original logger sofware set-up and downloading not only is very simple but the graph is in real time and in a clear format.

3. The price of the second system is 3 times that of the original, and I know that you get what you pay for but we are trying to develope this for use by personnel who have even less computing skills than myself. Downloading to Excel is not the easiest task and open to subjective input or mistakes. The original systems own software cannot be inerfered with.

Any further comments would be appreciated and I will revert with details of findings after grounding at the transmitter end.

thanks again to all replies

"The transmitter(s) (Druck type) have a 9-30vdc input and an 4-20mA output. Originally we had the power supply next to the data logger but have now separated this via a 5' length cable and the output cable from the transmitter to the logger is shielded."

If I am reading this correctly, then the 4-20ma output is shielded; but you do not say anything about the 9-30vdc inputs. The voltage inputs are more likely to be "trashed" by EMI than the 4-20ma (one of the reasons it is so popular).

You might also consider placing a noise filter at the point where the power enters the logger.

Reply to Kilowatt0 and Bill,

The only shielded cable we have used is to connect the Transmitter to the PS+ and to 2 terminals on the logger (1 and C). Between terminals 1 and C we have connected a shunt resistor.

We do not have shielded cables between the logger and PS or the mains supply to the PS.

The earth/ground from the transmitter is also connected to C on the logger which is also wired to the PS- (9-30vdc) output and also to the earth/ground on the 240v input, which itself is earthed via the incoming mains cable (not shielded).

Hope this makes sense!

Terminating the shield only at the transducer end is a common thing to do. The idea is to help eliminate possible ground loops but still provide electrostatic shielding of the cable. Now, here is the hitch. It won't work for high frequency interference. You will have to terminate the shield at both ends of the cable for this case. If your problem doesn't go away with only one end of the shield terminated, try both ends. Clamp-on ferrites are worth trying because they are inexpensive and easy to install. They help knock down high frequency common mode noise.

I agree with #3. I have rarely seen a 4-20mA trashed by EMI. You have to do some egregious practices to mess it up. I would concentrate on the voltage connection between the pressure transducer and the transmitter (if they are not an integrated unit) and try various configurations of shielded cable, separation from noisy cables, equipment etc.

You may want to try clamp-on ferrite (Type 43 Fair-Rite for instance) on the cable between the PS and the datalogger.

Single point grounds are not always the Holy Grail of EMI/EMC. They can help with some problems but not all. Watch out for isolated differential inputs to the data logger. Sometimes the unit "feeding" the data logger can "float" outside the common mode voltage range of the input device if it isn't referenced to ground. With isolated DC/DC converters providing power to sensors, this can happen quite easily. Common mode capacitors to ground can help shunt AC noise and one meg or higher resistors can help keep the transducer from "floating" away.

The thing to keep in mind is to try all the "usual rule-of-thumb" practices. And when that doesn't work, don't be afraid to try things not normally recommended. Good luck with your quest.

If they are using DC/DC converters, they could be bouncing the ground with the running freq, filtering would help. An RF shield along with the shielded cables would help, tied to a chemical ground, which most labs don't have. Something a good EMI lab would have. Single point ground back at the pwr source would help as well. Then there is the PC ground, which could be adding noise. One could remove the return grd from an o-scope then measure the shield grd for noise, as well as the pwr return side for grd bounce. The site might also have a very noisy AC feeder return. And depending where the system is set up, even solar noise could cause this. One could separate the grds of different equip by using optical isolators as the data transfer method

Thanks for all the input, although some of it is conflicting, and I was intending to attach a sketch of the wiring diagram but have not been succesful as the ms word or pdf file was not recognised. Any help as to how to do this would be appreciated.

Meanwhile, It has since been suggested that the lead (approx 10 ft long) from the sreened cable currently earthed via the PC frame may be acting as an antenna and therefor contributing to the problem. We have been advised to make this cable as short as possible to ground (which means in our case, drilling a hole in a concrete floor and inserting a copper rod). Any further thoughts before we do this?