Have a look at this thread, it might give you some ideas.

https://www.eng-tips.com/viewthread.cfm?qid=485342

Nest Regards A

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

I'd guess it's either an electrical heater cycling at that rate or the water is being stirred at those times because maybe a pump is cycling then or valves change settings.


What size are the aquariums?
Residential or commercial?
Town location? This can allow us to speculate on possible local effects.

Keith Cress
kcress - http://www.flaminsystems.com

Is the data logger temperature compensated?
Is the ambient temperature varying about 5 degrees as the HVAC cycles?
The simple thermo-couple circuit measures the difference between the hot junction and the cold junction.
It must be temperature compensated for the ambient temperature or the reading will drift with changes in the ambient temperature.
We had an installation where the ambient temperature dropped below the limit of compensation of the controller.
As the temperature dropped below the lower rated limit, the indicated temperature dropped degree for degree.
This was a high temperature alarm we were willing to accept a higher trip point when the ambient temperature was below freezing.

Bill
--------------------
Ohm's law
Not just a good idea;
It's the LAW!

All 4 channels acting the same (with one in the air and three in the water) says it can't be ambient air temp... a large body of water won't change 4+ degrees in 30 minutes just due to HVAC. The very slow, but repeatable pattern also says this is unlikely to be electrical noise.

How about some pics of the area these are in, with pics of any surrounding equipment, window locations (if any), HVAC vents, etc.?

Dan - Owner
http://www.Hi-TecDesigns.com

Quote:

a large body of water won't change 4+ degrees in 30 minutes

So, is the data-logger submerged in a large body of water?

Bill
--------------------
Ohm's law
Not just a good idea;
It's the LAW!

You're thinking perhaps some affect on the main unit? Possible, I suppose... though that seems like a pretty poor unit design if the controller is affected that heavily by temp and the thermocouples attached to it aren't.

Dan - Owner
http://www.Hi-TecDesigns.com

@itsmoked
What size are the aquariums? Two are 10gal (38L) and one is ~50 gallons (200L). The fourth probe is in ambient air.
Residential or commercial? Commercial.
Town location? Melbourne, FL (east coast, south of Cape Canaveral)

@waross
Is the data logger temperature compensated? I don't know. It is not mentioned in the manual. If not, then this logger would only be accurate if the ambient temperature is very consistent? That would make it useless for many applications it would seem.
Is the ambient temperature varying about 5 degrees as the HVAC cycles? This is very possible. It is hot here this time of year, and the AC is running most of the day, and probably much of the night. What is odd about the data, though, is that is doesn't seem to fluctuate in the same way during the middle of the day.
So, is the data-logger submerged in a large body of water? I know you're joking, but no. Only the ends of three of the four thermocouples are submerged in water.

@MacGyverS2000
How about some pics of the area these are in, with pics of any surrounding equipment, window locations (if any), HVAC vents, etc.? The logger is in a room with about 50 small (10 gallon) tanks of salt water and a larger 200 liter tank. There is an HVAC vent on the ceiling near the logger. There are no open windows nearby, but there is a glass door to the outside about 10 feet away from the logger. I have tested it by moving the logger to an office room and just measuring ambient temperature, and got he same results: "normal" temperature drift during the day, then the regular 40-minute, 5-degree sine wave at night. This was a completely different room/setting. I can post pictures tomorrow.

Melbourne, FL.... It's likely something outside your building. Possibly one of various radar systems at the Melbourne Airport or one of the 16 radio transmitters scattered about Melbourne. I believe Palm Bay also has a large electronic warfare facility.

Take it home and see if you get the same results at a geographically different location.

Keith Cress
kcress - http://www.flaminsystems.com

When you moved it into the other room what sensors were attached?
I am with Keith on this, it is likely something external.
Poor input filtering on the unit, or the noise could be being picked up by the power supply section.
But all of it (logger and sensors) inside a steel box overnight and see if you still get it. I bet that you won't.
Shielding and filtering are in order.

= = = = = = = = = = = = = = = = = = = =
P.E. Metallurgy, consulting work welcomed

Stay with the unit past 9 pm.

Something to keep in mind is that the HVAC may just be keeping up with heat removal during the day but at night it can catch up, but then heatsoak pops the temp up again. That soak will gradually decrease overnight.

It is certainly possible there is internal compensation, but it may not be matching the cold junctions temperatures.

OP's link to logger specifications shows 0-50C operating range

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert! https://www.youtube.com/watch?v=BKorP55Aqvg
FAQ731-376: Eng-Tips.com Forum Policies forum1529: Translation Assistance for Engineers Entire Forum list http://www.eng-tips.com/forumlist.cfm

I built a quick and dirty amplifier for a system (too quick). It worked great in the lab but when I took it out into the field far from most electrical interference to do the job I had the weirdest periodic results in a pressure monitoring application. Every 5 seconds the output went crazy. After trying everything myself and two other engineers could think of we realized the anomaly was in sync with the 85 ton 5MW early warning radar on Mount Umunhum across the valley from us.

It took half a box of tinfoil and an hour of fiddling to finally reduce it enough to mostly work. Not a recommended solution. It wouldn't work in the case where there are thermocouple antennas leaving the shielded space only to bring in the interference. It would need to be addressed in the entire system's design. You'd need to change a few of several aspects if you're having the same sort of external issue.

Keith Cress
kcress - http://www.flaminsystems.com

My thinking:
During the day, the AC is running full out.
In the evening, the heat load reduces and the AC cycles on and off.
There is something wrong with your compensation circuit or your connections that is affecting the compensation circuit.

MacGyverS2000: While not as challenging as quantum physics, measuring the voltage output of a junction without affecting the measurement is nontrivial.
Imagine that each ire material has a unique temperature/voltage curve.
When dissimilar wires are connected together, a voltage equal to the difference between their respective temperature/voltage curves will be developed.
The voltages generally diverge with temperature changes so the hot junction will develop a different voltage than the cold junction.
The voltage difference between the hot junction and the cold junction may be used to determine the temperature difference between the junctions.
So that voltage will rise and fall with changes in the ambient temperature of the measuring junction.
Hence temperature compensation is required to determine the actual meaured temperature.
One means of compensating is to use a temperature sensitive circuit to bias the voltage difference developed by the junctions.
Another possible compensation method, which may be advantageous when several junctions must be compensated is the use a temperature controlled heater to maintain the reference junctions at a controlled temperature above the highest expected ambient temperature.
The failure of such a compensation circuit could be expected to make the outputs vary as the ambient temperature varied as the A/C cycled.
Check your water temperatures with a known good thermometer. All of your readings may be in error.

Bill
--------------------
Ohm's law
Not just a good idea;
It's the LAW!

Bill.... I'm sorry, but after all my external considerations I'm going to have to agree with your logic.

The cycles are just about exactly what I'd expect out of A/C cycling (~30 minutes) in a well-coupled water filled room. The tanks subsequently heating the room back up with an almost fixed cycle rate due to temp controlled tanks trying to heat the room up to higher temp (82°F) than the A/C would be set to. During the day the A/C stays on all hot day with the lighting on and people fanning the door.

Ocellaris; Before trying it somewhere else try this. Take a cardboard box a couple of times larger than the logger. Maybe stand it on it's end. Stuff some insulation in the back of the box, some packing foam or wadded paper or hacked Styrofoam. If you can, put a jar of water in next for thermal mass. If you can't manage a jar of water use something heavy like a brick or a pound of metal. This will thermally decouple the room from the logger. Slice the box just enough to get the wires in to the logger without drafts. Put some more insulation in front of the logger so the logger is sequestered in the box with the thermal mass and away from what's happening thermally to the air in the room.

Run it for two days. I'll bet the cycling is completely gone. If that works it would still leave the logger accuracy as suspect as being poorly compensated.

Keith Cress
kcress - http://www.flaminsystems.com

Or just stick it in a mini cooler, like the kind used as a lunchbox... that should more than adequately compensate for room temp variations over such a short scale. If the problem goes away, the problem is temp... if it doesn't, it's likely electrical in nature.

Dan - Owner
http://www.Hi-TecDesigns.com

I have insulated the logger in a cardboard box with packing paper. The box isn't large, but if insufficient ambient temperature compensation is the culprit, I should at least get a reduction in the noise, if not outright removal, correct?

See pictures of the system below.
All water is saltwater (marine aquariums). The thermocouple ends in water have been insulated with hot glue, applied by myself.

Thermologger in insulated box. The end of thermocouple #4 is on top of the box, measuring ambient air:


The current location of the box (red arrow) vs where the datalogger used to hang on the wall (yellow circle). Note the AC vent about the area:


Another angle:


The how the thermologger used to hang on the wall. Note this picture only shows 1 attached thermocouple, but the current logger in the box has all 4 thermocouples attached.


End of thermocouple #1 in a 10 gallon tank:


End of thermocouple #2, also in a 10 gallon tank:


End of thermocouple #3, in a 200L tank:


Wide view of the 200L tank and surroundings:

Thanks Keith. I value your opinion.

Bill
--------------------
Ohm's law
Not just a good idea;
It's the LAW!

Anytime Bill.

Ocellaris; Outstanding. This should be educational, thanks for following thru. Looks good. That A/C vent and the position of the logger are dead-ringers for Bill's theory.

You have a fun place there. Reminds me of my 40 years running largeish saltwater aquariums.

Keith Cress
kcress - http://www.flaminsystems.com

Thank you all for the replies.
I recorded some new data with the setup shown in my previous post. Unfortunately, the box did not remove the noise.
I also recorded some data using an identical datalogger in an un-air-conditioned warehouse, and compared it to the original data from the air-conditioned office (original post). There was no noise in the warehouse data.

It seems that the AC cycling is the problem, but the insulation that I used was insufficient. Does the entire length of every thermocouple need to be insulated?

Another thing that is odd: Two of the channels are not getting the same degree of noise. The thermocouples are not identical:
Thermocouple #1: A K-type extension wire with the end stripped and twisted together, then covered with hot glue. In a 10-gallon tank.
Thermocouple #2: A K-type thermocouple, end covered in hot glue. In a 10-gallon tank.
Thermocouple #3: A K-type thermocouple, spliced with extension wire (for length), end covered in hot glue. In a 200-liter tank.
Thermocouple #4: A K-type thermocouple, end covered in hot glue. Measuring room temperature.
Channels 1 and 2 are not getting the same noise (except for channel 1 before the insulated setup - but still to a lesser degree than channels 3 and 4).
Any ideas on what is going on here?

Also, the noise seemed to get worse for channels 3 and 4 after insulating the logger (the chart scales below are identical for easy comparison). Thoughts?
Edit: Now that I think of it, it might have just been a hotter series of days outside.

Shown below are the results for:
1) Original water tanks/room temp setup with uninsulated datalogger.
2) The same setup with the insulated box, pictured in my previous post.
3) Room temperature sitting on an office desk in an air-conditioned room.
4) Room temperature in an un-air-conditioned warehouse.

1) Original uninsulated setup


2) Insulated setup


3) Air-conditioned office


4) Un-air-conditioned warehouse

Let's try another experiment... coil up all of the probe wires and place the probes within the same insulation box as the main unit. Wrap the entire thing (unit and probes) in a wrap or two of aluminum foil.

Let's see if all of the probes measure the same temp and the same swing amplitude when it's all in a close-knit bundle.

Dan - Owner
http://www.Hi-TecDesigns.com

Quote:

Does the entire length of every thermocouple need to be insulated?

No.

Quote:

Any ideas on what is going on here?

I'm not seeing anything as you've described in the T/Cs themselves that would be any particular issue.

Quote:

Thoughts?

hmmm We have a brain twister here.. Dan's experiment is worth a try. What's going to happen is after enough data lands here something will jump out at us.

All these various tanks... Since you're measuring different ones I'm going to assume they are not running in a series flow loop and that you are individually heating them? How? With standard thermostatic aquarium heaters?

Keith Cress
kcress - http://www.flaminsystems.com

It's a plastic enclosure? Judging from where it comes from I'd guess there isn't much shielding. If you build a shield you might try attaching the shield to the (-) of the battery with a wire. Aluminum foil would help with the electrical part of any airborne noise. To stop the magnetic portion I'd try some sheets of mu-metal.

Thermocouple extension wire does come in shielded and/or twisted wire. You might want to try that.

I also doubt these units use low pass anti-aliasing filters.

We had an instance where an oil heater would go into high temperature shutdown several evenings a week.
It coincided with the railway picking up rail cars on a track adjacent to the heater.
When the train was backing in to couple onto parked cars, the brakeman would be checking the distance remaining to contact.
He would be holding the transmit button down and counting off the distance to the engineer.
Our operators did some testing with our radios and found that a a radio signal would drive the indicated temperature on the controller slowly upscale until it tripped off on high temperature.
You may want to try looking for any intermittent source of interference.
My first response will still be to look for some component that is common to the compensation circuits of all four chanels, but I would be aware of and watching for any possible source of interference.

Bill
--------------------
Ohm's law
Not just a good idea;
It's the LAW!

Quote (MacGyverS2000)

Let's try another experiment... coil up all of the probe wires and place the probes within the same insulation box as the main unit. Wrap the entire thing (unit and probes) in a wrap or two of aluminum foil.

Thank you. I will try this next and post the results.

Quote (itsmoked)

All these various tanks... Since you're measuring different ones I'm going to assume they are not running in a series flow loop and that you are individually heating them? How? With standard thermostatic aquarium heaters?

Thermocouples #1 and #2 are in single 10-gallon tanks that are part of 2 separate series flow loop systems (you can sort of see how the systems are set up in the 2nd and 3rd pictures of my 4 Aug 21 13:43 post). Both of these systems are cooled by chillers at their respective sump tanks. The chillers will kick on when the water temperature reaches 1 degree F above the set point, and they turn off when the temperature drops to the set point. Each system is about 400 gallons in total volume. Heating is not necessary because the pumps and UV filters contribute more than enough heat, necessitating cooling even when the room temperature is below the target system temperatures.
Thermocouple #3 is in an isolated 200L tank that is heated by a standard thermostatic aquarium heater.
Thermocouple #4 is measuring room air temperature.

Quote (BrianE22)

It's a plastic enclosure? Judging from where it comes from I'd guess there isn't much shielding. If you build a shield you might try attaching the shield to the (-) of the battery with a wire. Aluminum foil would help with the electrical part of any airborne noise. To stop the magnetic portion I'd try some sheets of mu-metal.

Thermocouple extension wire does come in shielded and/or twisted wire. You might want to try that.

I also doubt these units use low pass anti-aliasing filters.

The units are relatively inexpensive - something like $96 each. I'm going to try placing everything in a foil-encased box as MacGyverS2000 suggested.

If this removes the noise, then one question I will have will be why is the EM interference only affecting 2 of the 4 channels? For example, channels 2 and 4 are using identical thermocouples, yet channel 2 has no noise, while channel 4 has a lot.

Swap the thermocouples. See it the noise stays with the thermocouples or the channel number.

I would think that the reason for sensor 1 and 2 not having as musch "noise" is that you have a cooling systems there that keeps the temperature in check.
What are the base temperatur you wont keep?

And I would guess that the air temp sensor is in the same room as sensor 1 and 2 and the 10/400 gallons tanks?

And my only explanation for why the 200 gallons tank has a musch higher temp values then the other 3 sensors is that it is in a room with afternoon and and evening sun.

And that it is the ventilation that is doing this.

A sinus wave that peaks ones every hour, is hard to achieve with electrical noise I would think.
And they do not have exactly the same amplitud or hertz either.

Best Regards A

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

How about a simple test. Put two probe in a crushed ice bath for a couple of hours during the time that the fluctuations occur.
One ice bath adjacent to the tank and one close to the data logger on short leads.

Bill
--------------------
Ohm's law
Not just a good idea;
It's the LAW!