AD594 Thermocouple Interference and Ground Capacitance Problem

Thread Starter

delmasli

Joined Sep 17, 2009
17
Hello All,
I designed and assembled /almost completed a machine for industrial gas sampling. Main units of this machine is GLCD, 2 Electronic Cards (Controller card and Analog Measurement Card), Switch Mode Power Supply, Frequency Controller and 1KW asynchronous air pump. Analog electronic board/card contains 7 sensors (3 pcs pressure, 2 pcs temperature/LM35 and 2 pcs thermocouple). Controller Card contains 3 MCU (8722 x 2 and 4685). Frequency controller controls the pump, and MCU's controls the frequency controller. When the pump is off all the sensors works perfectly. But when the pump starts to operate thermocouple measurement results become unstable / when other sensors are still ok. The thermocouple ICs are AD594/595.
It is obvious that pump creates interference by its electromagnetic field and also by electrical connection lines where these interferences are aslo amplified by AD594/595. I've assembled choke coil between connection of analog card (where AD594/595. located) and frequency controller. Thermocouple results (AD594/595) become much more stable. Then to make signal much more stable I've add LC filter to both AD594/595 inputs and outputs. I've reduced the noise from input cables. Much more clear output but still have a problem.
Main problem is that. As I explained I've two twin Thermocouple IC AD595 and AD595. Both supply voltage is 12V DC and ground. I've shortcut the thermocouple inputs of these ICs (pin1 and pin14). When I look at the outputs one IC1 is something like 240mV (room temperature) but IC2 is something like 90mV. I thougt that there could be problem with IC2. Then I removed both and put each to the other's place. Now IC2 is 240mv and IC1 is 90mV. It seems that there is design problem over PCB.
Both ICs Pin1 is connected to the ground. But there is priority problem. when IC1 240mv and IC2 90mV, IC1's line from thermocouple first visits pin1 of IC1 and then connected to the ground. IC2's thermocouple first visits ground and then connected to pin1 of IC2.
When I connect the IC2's thermocouple first to the pin1 output voltages became equal / both are 240mV.
Before solving this the biggest problem was :
When pump is off, IC1 output was 240mv and IC2 output was 90mV. When pumps starts to operate IC1 output was something 500mv and IC2 output was something 600mV. Magnituded became higher. When pump was turned off IC1 comes to it previous value of 240mv but IC2 comes to unpredictible value. Sometimes 400mV or 450mv, sometimes 390mv, sometimes 100mv. Is there anyone can guess or explain something why this is happening?
For example then the pump stops and IC2 output is 450mv I shot down the system and also pulled off the electric cable from line. Wait for 3-4 mins. then started the system. IC2 output is still 450mV. In real it must be 240mV.
Anyone can guess what is happening or any idea? Could it be Ground Capacitance differentiation??
I know it is not easy to realize what is happening from my writtens. Circuit and all the system has to be inspected and experimented but may be someone inside you could experienced similar problem?
 
Last edited:

JDT

Joined Feb 12, 2009
657
Without going into detail, the best way to avoid trouble like this is to completely isolate your controller and measurement cards from the power electrics.

Use an isolated power supply for these cards.
Couple the output signals to the power circuits using opto-isolators, transformers or mechanical relays.
Enclose your sensitive electronics in a earthed metal box.
Don't connect any part of your sensitive electronics directly to anything - not even ground.
Filter noisy power devices (motors, triac controllers, etc.) at source.
 

rjenkins

Joined Nov 6, 2005
1,013
Hi,
the 'frequency controller' (Inverter?) should have a power filter directly at it's power input, plus the cable to the motor should be screened, with the screen connected to a good earth directly at the inverter.
The better ones have metal cable clamps specifically for terminating the cable screens.

Use 'CY' type power cable, or specific servo drive rated, screened, power cable & make sure the motor is also earth bonded.

Make sure all the different parts of the equipment metalwork are properly earth-bonded, with wide copper-braid earth straps bridging any joints.

Ensure all 'dirty' earths (from power electronics, motors etc) go to one common earth bar so no ground currents circulate anywhere else.

Ensure all power cable 'sets' (eg. the three wires in or out of a contactor) are twisted together to avoid cross-coupling in any trunking etc.

Likewise, any small signal or analog wiring should be screened and/or twisted pair.

The whole low-level electronics system should have it's 0V grounded to a single 'clean' point if possible.
The thermocouples should either be fitted in grounded (bonded) metalwork, or if they are floating connect one lead to the electronics 0V with a 1K resistor at the AD594.

This is critical, the AD594 has a limited common-mode range and if the thermocouple floats or has a voltage / noise imposed outside the common mode limit, you will get silly readings.

It's quite practical to put the low level stuff in the same enclosure as the high power gear, as long as you are carefull with the wiring layout & correct grounding.
 

Thread Starter

delmasli

Joined Sep 17, 2009
17
Thank you all for your quick and useful answers,
Due to your recommendations I focused on earthing technics and noise effects. I red the Analog Devices Application note of "AN-369 Thermocouple Signal Conditioning Using the AD594/AD595".
It is written that "...Whether internally generated or induced by radiation from a source, noise becomes one of the limiting factors of dynamic range and resolution. Solving noise problems involves eliminating the source and/or shielding...." and "..Noise may be injected into the AD594/AD595 input amplifier when using a long length of thermocouple..."
I draw my circuit schematic:



In this diagram, pin codes are different from AD594/595 cause I modified some another IC but it is not a remarkable problem.
Anyway by using simple LC filtering, line noise can be suspended. And also at this application note of AN-369 some different input filter circuit with resistor is given which (is same or similar - I think -) Mr. Jenkins told in his previous message. Even though I use thermocouple or shortcut the pin 1 and 14 the noise still exhibits.
AN-369 says that ".. If the output still exhibits noise, it may be entering via the power supply." Now the noise problem seems to be caused earth effect of power supply.
What I've done yesterday to solve.
I removed both earth's of Power Supply (frisby 400W) and Frequency Controller (Delta VFD-L AC Motor Drive) from line. Now only L and N's are connected to AC 220V line. There is an EMI filter between Motor Drive and PowerSupply.





This schematic is what I've done recently. As Mr.Robert Jenkins wrote about dirty Ground effects, I realized that my earth is not clean. And both EMI and 220V Input filter don't have any effect over grounding. They just give permision to pass it though the Power Supplies. After removing connetion of Earth from both Powersupply and Frequency controller (AC Drive) AD594/595 output voltage became much more stable. When the Motor/Pump is off (AD594/595 inputs are shortcut) outputs are similar and something like 150mV, When the pump is on (operating) outputs are 170mV-180mV.
It is strongly possible that the major output voltage failure of AD594/595 was caused of Grounding. But still there is voltage differantiation between operating pump and non-operating pump.
I found that (which was obvious) electromagnetic field around pump effects line and ICs. Field is strong near the pump, Multimeter probs are being effected and they give wrong and unstable voltage values near the pump, and correct constant voltage results away from pump. Even my hand near the pump (5-10cm far, but not touching) distorts the magnetic field and output voltages of ADs are being effected (jumps from 180mV to 210mV). Electromagnetic wave creates current density over conductives. I tried to remove voltage over the pump alluminium cover by earthing it. But values jumps from 150mv-200mv to incridibly high 1000mv-1500mV. And also my device (pump and all the electonic equipments are located inside alluminium constructed box) box absorbs little amount of this e-magnetic field and over these alluminium (something like h=400mm w=400mm l=400mm cubic box composed of profiles and plates 1mm thickness) box little amount of current density occurs. Again earthing the box (as JDT recommended in his/her previous message) makes the output values (ADs) unstably 500mV-600mV. Do I need a superior earthing method? Or do I just need to keep it away from earthing? It is critical that this device is going to work in industrial areas (fabrics, power plants etc)

Kind Regards,

Deniz ELMASLI
 
Last edited:

rjenkins

Joined Nov 6, 2005
1,013
Problems with interference from inverter drives are quite common. Most makers have pulications covering the specific wiring methods to minimise any problems.

Some examples here:
http://www.clpa-europe.com/documents/documents.php?do=send_file&id=44

http://www.danfoss.com/NR/rdonlyres...iderationsVFDElectroMagneticCompatibility.pdf

http://library.abb.com/global/scot/scot201.nsf/veritydisplay/4f1197aefcf2a91ac125744f003ba3be/$File/FactFileTDM1_Bearing%20currents_RevB_EN_lowres.pdf

http://www.dem-uk.com/jkcm/cache/fl0003482.pdf

Note: add-on ferrite cores on the motor cable at both ends can help with HF interference.
 

Thread Starter

delmasli

Joined Sep 17, 2009
17
At the moment Ground of the Pump is not connected to the Inverter's Ground (and as well the inverter's ground's to power line's ground).

Due to some readings I drew this schematic.



As I understood that ferrite cores could help me to solve this interference problem. Do they (these ferrite cores) help me if I just use normal (not screened) power cable between pump and inverter and between inverter and line socket (220V multiplier). Some texts say that use switch mode powersupply and inverter not at the same line. But I don't have any choise. This is a compact device and only 1 power cable. On the other hand when I tried to give power to switchmode PS and inverter from the same line but different locations (5meters between sources / Power line jacks). Distortion of AD594's decreased.

Anyway; disconnected ground seems a momentary solution and I don't want to keep the motor away from ground. In the laboratory environment, only (most) EMI is scattered by Motor. But at the fabric and power plant zone where the device is pretended to work, Motor is going to sustain much more EMI than that it exposes. So I need it grounded and I dont have any chance to select if the ground is dirty or clean (due to customer factory working conditions).

I found the cable connection technique of ferrite core from a site.



I'm not sure about blocking all the HF by ferrite cores, but I gonna try. For example does the size of these cores effect the filtering ratio and efficiency? Such as big cores better, smalls are not enough etc. Because the current density over the pump shell (cover) is too much and when I connect its shell to earth, Swich mode power supplies nearby become unstable. Most documents says that current desities over the cables can be solved (minimised) by using ferrite cores.

Is it convenient to use both ferrite cores and screened cables in the same system? Am I thinking in wrong way?
 
Last edited:

Thread Starter

delmasli

Joined Sep 17, 2009
17



I'm not sure how many turns of cable do we need that it is passing inside the core. I did my own and illustrated as it is shown above. At the end by using Heat Shrink Tube, cable becomes much more rigid.
 
Last edited:

Thread Starter

delmasli

Joined Sep 17, 2009
17
According to the Lab.Conditions it seems that ferrite beads have solved the problem. Now Pump and the Switch Mode PS both have ground (earth) connection. In the room temp, when the pump is OFF both AD594/595 say 240mV. When the pump is ON they are 260mV-270mV / 2-3C (reminder: they were silly unstable 650-1000mV when the Pump was ON). And that is enough for me and my application. We will see what is going to happen in high EMI zoned Power plants, and Fabrics.
Thank you both rjenkins and JDT for your response.

Kind Regards,

D.ELMASLI
 
Last edited:

GioD

Joined Mar 20, 2009
30
I had the same problem.
Instead of the AD594 i use OPAMP. The behavior with the inverter noise is the same.
I solve the issue shielding the Thermocouple. This solution work fine, but I use a non standard connector, in order to connect the shield.
Now I want to improve the system, so I'm thinking to build a new circuit.
I read somewhere that the best way to manage a signal is to maintain the same impedance along signal path. So, I think that the problem is that the thermocouple has a low impedance and the OPAMP circuit has an high input impedance.
My actually circuit has an high gain and the thermocouple is like an antenna.

I see that also "delmasli" solved his problem, but not perfectly.

What do you think about ?

Witch is the best OPAMP configuration in order to have a low input impedance ?
 

timrobbins

Joined Aug 29, 2009
318
Have you been able to find screened thermocouple cable. Your application is probably low enough in temperature to maintain the screen up to and around the tip. You may need to even use double screened - the outer screen to your equipment PE, and the inner screen to your AD594 opamp star ground point. Screens are floating at the tip end.

You may also need to check that your thermocouple cable is not run in the same duct, or close to any power cable going to the motor, or any open air current carrying wires within the motor.

Ciao, Tim
 

GioD

Joined Mar 20, 2009
30
Thanks for your reply, Tim.
My idea is that shielding using screened cables is a solution, but not a good solution.
The best thing is to use a normal cable and build a circuit that don't fell the noise.
Any idea ?
 
Top