Measuring Millivolts

janner_10

Lifetime Supporting Member
Join Date
Dec 2014
Location
Tewkesbury
Posts
1,301
Background:

We have a customer that wants to start measuring the Oxygen content of syngas. All quite straight forward using a NGK Lambda Sensor. Bearing in mind, we aren't doing any wiring as the plant is in the States and we are in the UK, we are just adding the software remotely to their / our existing system.

We will be measuring the output in mV from the sensor (into a 1769-IF8U) and then scaling to Oxygen content. The output from the sensor will be roughly -6 mV to +30mV for their application.

Problem:

The distance between the sensor and the Analogue Input module, is roughly 25m to 30m.

My proposal to the customer is to invest in a signal conditioner local to the sensor to convert mV to mA and then read this at the PLC.

Naturally he knows best and just thinks that we can read mV in the range above on a cable run of 30m.

I think it is practically impossible due to the volt drop and such a small value to read, he doesn't.

Any thoughts?
 
Without looking at the specs of the card, assuming it's 0-10V and 16 bit resolution, you'd be using 240 out of a possible 65535 counts, assuming a 36mV range. That's 0.36% of the input range.

I will bet tomorrows lunch on it not working well. The signal will be all over the place.

I'm not familiar with that sensor but does the same company not make a transmitter to sit between the sensor and the control system? I only work with the water industry so our dissolved oxygen gear is a bit different, but we would never try connect such a sensor direct to a PLC.
 
The card is set to read -100mV to 100mV and the resolution in units at this range is +/- 0.000003 at 60hz, so I'm not concerned about that.

My question is theory based about reading small values of mV over a 30m cable run (csa @ 0.75mm2), which I think is nigh-on impossible.
 
I cannot state any standards that specify it, but I instinctly feel that measuring such small voltages over such a distance will be extremely noise sensitive.
Grounding and shielding should be done extremely carefully. And cable must be twisted pair.
Or, ... as you suggest yourself just place the signal conditioner as close as possible to the sensor.
 
There is no appreciable voltage drop in the signal wires because the input impedance is going to be quite high and the current therefore quite low.

The problem is potentially noise from EMI and RFI. Fluorescent lights, motors, power supplies, even relay coils will produce interference. When you start out with a mV signal it doesn't take much to influence the reading.

The cost of a signal conditioner is insignificant compared to the cost of screwing around diagnosing and fixing a problem.
 
Agree with Tom, no voltage drop. The issue of noise will be the biggest issue. Is the input module single ended? I would use an isolated diffenential signal conditionier to then output to the IO card. Being differential, both + and - signals should hopefully float up/down together due to noise but the differential voltage between them should still be same. Noise on a singled ended input does not work very well since the - side will be not allowed to float around.
 
Use a signal conditional (convert Mv to Ma) put the signal conditional as close to the sensor as possible. This way the bulk of the cable run will have already been convert to 4-20ma .

Just another way to get the job done.
 
As mentioned, signal noise is an issue. It's simpler to use a signal converter, and easier for those at the site to troubleshoot.

But if the customer would rather pay you to program than buy some hardware and get it wired in ...

If the O2 sensor signal does not change quickly (can use measurements over several seconds to generate one good value), some heavy filtering on the input may work out. It could be as simple as an average or as complex as a digital filter.

If the average does not work ... it's a bit of a pain, but using samples that are taken at an EXACT sample rate will work for a digital low pass filter.
 
If this oxygen sensor is a zirconium oxide sensor that has to be hot or heated to produce an oxygen reading, then you should know that that type of sensor is a high impedance output, which can be 'loaded' by connecting it to a 'relatively' low impedance analog input.

The loading effect is an error proportional to the parallel resistances of the sensor output and the AI input, so a correction factor can be applied.

It was not uncommon some years ago for vendors of those zirconium sensors to offer their signal conditioner which provided 'standard' 0-5V or 4-20mA output.
 
We will be measuring the output in mV from the sensor (into a 1769-IF8U) and then scaling to Oxygen content. The output from the sensor will be roughly -6 mV to +30mV for their application.

Problem:

The distance between the sensor and the Analogue Input module, is roughly 25m to 30m.

It can be done, I have seen it work with load cell cables to an electrical panel 30m away. It requires a differential signal, through a well shielded twisted pair cable. For the voltages you mention (-6mV up to 30mV) you may want to use a load cell amplifier/digitizer or a modular PLC with special input module designed for Wheatstone bridge setups (strain gauges). E.g. Wago have the 750-491 module for "analog input; Resistor bridges (strain gauge)".

A loadcell setup can reliably measure 2500kg loads with smaller than 0.5 kg accuracy. Typical loadcell sensititivy is in the range of 2mV/V, so a 2500kg loadcell on a 5V excitation voltage would yield 2 * 5 = 10mV at max load. A change of 1 kg yields a delta voltage of (1 / 2500) * 10 = 10 / 2500 = 1/250mV = 0.004 mV = 0.000004V. It may sound crazy, but this can be reliably measured with the correct hardware. It makes no sense whatsoever to try and use something like a regular 0..10V analog input. Unless you apply an analog load cell amplifier near the sensor that scales the orignal signal up to that range.

As others have said, keep the cables away from other cables (especially motor cables on VFD drives).
 
Gentlemen, as usual great advice, which will be taken on board.

I am flying out in 10 days to make other changes they have decided they want, which is fair enough as it's all T&M, they are prompt payers and I get a week/10 days away from the kids, drinking lovely APA!

Incidentally, if anyone wants an adhoc job of maintaining changes to the plant on a T&M basis, they are looking for a States based control company / firm / one man band, to take over from us, as understandably, to travel from the UK to the USA is quite cost intensive. They now have the .acd, .apa & .sk files at their disposal, but need local(ish) assistance. They are based in Santa Fe, NM. PM me and I will give you the details of the decision maker.
 
Another option may be a hang an inexpensive PLC near the sensor and have the other PLC read that via modbus or the like.

I've done exactly that with a Automation Direct Click with a thermocouple card set to the +/- 39mV range. Just a shade over $200. I'm not sure what a signal conditioner is cost wise. I went with the Click solution as it's 4 channels and I was able to monitor 4 load cells with it. That particular thermocouple card seems to be a bit slow, 1 update per second or so if memory serves me, with all 4 channels being used. No program in the Click, just reading the DF registers directly.
 

Similar Topics

Hi everyone, I don't know much of PLCs but it happened that i need to connect Leuze AMS358i Ethernet laser measurement to Productivity 1000...
Replies
0
Views
1,049
Hello everyone! Firstly I'd like to say that I'm new to this forum and also new to PLC programming world altogether, that said I really would...
Replies
27
Views
4,768
I am looking for a way to determine level in a silo. The material is called Perilite which has a very light density. Bulk density could be...
Replies
10
Views
2,722
In our water cooling tank, I've installed and successfully wired up 4 float switches & updated the PLC (big thanks to @parky for that). As a...
Replies
46
Views
11,271
Hello, I have an application where I need to measure the salt level/volume in a salt saturator. This is a fairly large tank where we load bulk...
Replies
21
Views
6,459
Back
Top Bottom