Electrical noise from VFD affecting Temperature Probe 2

[veryconfused]

Firstly, I have no idea if I am posting this question in the correct forum, hopefully it's correct. Secondly, I am not an engineer in anyway. I have an issue with EMI from a 3 phase 480v VFD affecting a shielded temperature probe and was hoping somebody on the forum might be able to offer some advice on how to solve my issue.

If I leave the temp probe hanging in the air then the EMI has no affect on the temperature reading. However, as soon as I put the temp probe into the water I start getting temperature spikes. If I only needed to measure the air temp things would be fine, sadly this is not the case.

Does anybody have any ideas on how I might fix this issue? I have ordered some ferrite rings to install on the output side of the VFD. Also, I tried placing a small metal rod in the sump and grounding it with a wire but that hasn't helped.

Any help/advice would be greatly appreciated.

You can see the red temp probe in this image

Position of VFD relative to temp probe

Graph showing temp probe in and out of the water.

 

[edison123]

I would say defective probe since you have the same ones working underwater in other places. Swap the probes and see if the problem goes with the probe or stays with the VFD.

PT100 is a time tested reliable probe with linear resistance characteristics for temperatures up to 200 deg C in case you are looking for an alternative.

Muthu

http://www.edison.co.in

 

[electricpete]

> @Electricpete - the probe is made by avtech, they don't have the exact ones listed on the website as these are slightly customized. I have the same probe elsewhere, in other systems, fully submerged in water and don't have this issue. I can certainly try having the probe higher up in the water column.

Yes I would recommend trying that. You can continue to troubleshoot vfd as a source, that may well be it. But I would not rule out water intrusion just because you have another one working fine without vfd. Water's ability to penetrate into circuits is not necessarily predictable / repeatable. We have many outdoor instrumentation circuits (not submerged but damp) and every once in awhile we get a problem on one of them traced to moisture. They don't all fritz out at the same time.

=====================================
(2B)+(2B)' ?

 

[RedSnake]

Freshwater is dimagnetic repels magnetism, while saltwater do not.
It would be a easy test putting the probe in a glas bottle with freshwater just holding it in the tank to see if it makes any difference.
And I am a bit curious of the result.

It's also important that the motor and pump is properly potentially grounded.
This is not the same as protective earth PE.
Since most of the equipment is plastic this might not be so and that the motor cables are shield and connected to the bonding circuits.
The red square is the inverter.

Best Regards A



“Logic will get you from A to Z; imagination will get you everywhere.“
Albert Einstein

 

[Comcokid]

Redsnake has a good idea. Use fresh water in a bottle with the probe inside as a test. This will insulate from any stray DC or low frequency current to the probe.
Don't use salt water in the bottle or the bottle will be more like a capacitor to the tank water. A rough estimate of a 3" dia bottle 5" deep filled with salt water in the salt water tank is around 700 pF. Stray VFD switching currents of 100 kHz or higher will not be attenuated much by this capacitance.

 

[veryconfused]

Edison123 - I actually bought a new probe as I thought this might be the issue, however, the new probe acts in the same way.

Redsnake - I had to drain the tank to do some work on the system. As soon as I get the tank refilled and the system running. I will test out the probe in a glass jar of freshwater and update you with the results.

I will update in a few days. Hopefully, this can get solved.

 

[electricpete]

I think I found your probe here (photo below). Three observations:

1 - It talks about "ambient temperature" measurement, which to me means air temperature, not liquid temperature. They mention silicon coating to protect it from the elements, ok that's good but to me it's still not the same as saying this thing is designed for immersion.

2 - It looks like it terminates in a telephone jack. Maybe it's telephone cable? We certainly don't have any instrument circuits like that at our plant. From my understanding most instruments should typically be twisted shielded pair for noise immunity, I think phone cable is just a twisted pair. That could be another problem. Maybe try making sure that cable is not routed close to any power circuits/equipment. (is anyone else familiar with this type of cable and whether it is suitable for instruments?)

3 - it's called a digital temperature sensor. I don't really know what that means, it's as if they're suggesting there might be some digital circuits inside that cylinder with digital signal in the cable. All I'm familiar with is RTD's and thermocouples with analog signal in the cable.

=====================================
(2B)+(2B)' ?

 

[itsmoked]

Nice sleuthing Pete!

Keith Cress
kcress -

http://www.flaminsystems.com

 

[3DDave]

Flat phone cable is not twisted. This allows it to be used with those small crimp connectors. RJ 5?? Memory fails.

Certainly "outdoors" is not "submerged in salt water" though it might outperform it's target application.

 

[electricpete]

> For some extra info, this is how things look when the VFD is off but the probe is still in the water

The plot shows spikes and then constant temperature. Are you saying:
[ul]
[li]1 - the vfd was off the whole time[/li]

[li]or[/li]​

[li]2 - the the vfd was turned off during the middle of the plot (spikes before turnoff, no spikes after turnoff)[/li]
[/ul].

=====================================
(2B)+(2B)' ?

 

[veryconfused]

Electricpete - The VFD was only turned off after the temp reading returned to normal

3DDave - It's a RJ11 connection. The wire used is alpha wire, type 78104

 

[LionelHutz]

The shape of it makes me think it's a Dallas DS18B20 sensor like this.

https://www.robotshop.com/ca/en/ds1...MI54aC1-bs8QIVC43ICh2acARSEAQYBiABEgKTkfD_BwE

 

[electricpete]

For some extra info, this is how things look when the VFD is off but the probe is still in the water[/quote[

The plot shows spikes and then constant temperature. Are you saying:
1 - the vfd was off the whole time

or​

2 - the the vfd was turned off during the middle of the plot (spikes before turnoff, no spikes after turnoff)

The VFD was only turned off after the temp reading returned to normal

thanks. It's an important answer and I want to make sure we understand exactly what you're saying. Is it scenario 1 or scenario 2?

=====================================
(2B)+(2B)' ?

 

[veryconfused]

I turned the VFD off around 09:00. So spikes before turnoff, no spikes after.

 

[electricpete]

ok, that clears it up nicely. Sorry for being dense. That definitely steers us towards vfd noise rather than towards moisture.

=====================================
(2B)+(2B)' ?

 

[electricpete]

So from Lionel's link and the telephone wire I guess it's a digital serial connection.

Maybe move the probe and telephone cable further from motor and any power cable.
The 2nd photo showing some panels on the wall, that is all power and vfd control, right? I sure hope the temperature digital signal cable (telephone wire) doesn't go anywhere near there.
Where does that temperature digital signal cable go? I would trace it back to whatever digital unit it connects to and look for where it might be close to power circuits. Is it the same computer unit as the others that are working fine or is it a unique unit? (if unique unit then expand your search for things that may be interacting with that unique unit and its grounding).
It does occur to me that serial communication sometimes includes data integrity checks like a parity bit. There might be something that could be done on the software side to discard corrupted data if you talk to your vendor. (although with so many jumps down to exactly 10 degrees, I wonder if that is the default value it goes to when it senses "bad data")

I'm getting far out of my depth though. I'm not an instrument guy and that goes double for digital instruments... I blame it on my employer and their glacial speed of adopting new technologies.

=====================================
(2B)+(2B)' ?

 

[RedSnake]

Early in the thread he wrote that the spikes disappeared when he took the probe out of the water.
If the interference had come from the probe cable being to close to the power cable to the motor and especially if it wasn't screened the spikes would still be there even if the probe is in the air.
I think we can exclude that one.

And the Avtech "telefon" cable does not look exactly as his, the one on his probe looks more round.
He wrote that it was a Alfa wire 78104 it looks like this.

Moving the probe further from the motor might help, as you suggest.

Best Regard A

“Logic will get you from A to Z; imagination will get you everywhere.“
Albert Einstein

 

[RedSnake]

Unfortunately, I'm stuck using the type of probe, otherwise I would try something else.

Why is this?
Is it a complet system?
There still must be a brand or something either for this probe or the system that evaluates it or some other data.
Most machines follow some kind of standard, so that there would be only one probe that would work for this seems unlikely.
If it is not a one of a kind system.

Best Regards A

“Logic will get you from A to Z; imagination will get you everywhere.“
Albert Einstein

 

[veryconfused]

RedSnake - I'd be willing to try another shielded digital temp probe that I can connect with a RJ11 connector. I wouldn't know where to look to find one. You can certainly hook up other probes not made by avtech to this system. Most of the probes I connect to the avtech unit are not made by avtech.

electricpete - The digital temp cable doesn't run anywhere near the power for the VFD. It's in its own conduit, well away from it. I can send more photos when I get back to work. I may even have some on my phone. I have other avtech units in other rooms, all using similar probes.

 

[3DDave]

Probably not the VFD. Probably from whatever the VFD operates.

 

[electricpete]

> He wrote that it was a Alfa wire 78104 it looks like this

ok, good that looks a lot better than telephone wire.

> If the interference had come from the probe cable being to close to the power cable to the motor and especially if it wasn't screened the spikes would still be there even if the probe is in the air.

Yeah I forgot about that. I can believe capacitive coupling on the high frequency components. Water has a dielectric constant about 80 so it gives much better capacitive coupling in water than air. I'm not sure if the salt makes a big difference unless we are suspecting resistive coupling which doesn't seem as likely to me but who knows. I don't think it affects the next steps which one we suspect.

We talked about moving the probe further, that's good.

We talked about moving the probe up so only portion of probe is submerged. I think that's good although I don't know how much of the probe needs to be submerged for accurate temperature sensing of a liquid (and OEM might not know either because it's not a liquid temperature sensor). At a minimum get all of that cable out of the water as much as possible since there can be capacitive coupling to the cable too, which would be minimized by getting it out of the water.

Some other things were mentioned to try to reduce emissions from the equipment.

Beyond that the solution that comes to my simple (non instrument guy) mind would be put progressives trial and error involving shielding by grounded structures. move probe close to a large existing ground plane. put a grounded barrier between the suspected source (the motor) and the probe/submerged cable. If that didn't work maybe try enclosing the probe / cable in something like a grounded can with enough openings to allow flow for temperature sensing. It might slow down your temperature response a bit but from your plots it looks like the system temperature doesn't change very quickly so that might not be an issue. It's admittedly a messy nonstandard solution idea but worth at least as much as you paid for it ;-)

I guess from all the discussion it's logical the motor is submerged but I wanted to verify that. Is it some kind of submersible motor? Is it that cylinder in the first photo the motor? What does the motor drive... a pump?

=====================================
(2B)+(2B)' ?