CR4 - The Engineer's Place for News and Discussion®

Previous in Forum: Flow Control Logic Case   Next in Forum: Durian Ripeness Instrument
Close

Comments Format:






Close

Subscribe to Discussion:

CR4 allows you to "subscribe" to a discussion
so that you can be notified of new comments to
the discussion via email.

Close

Rating Vote:







18 comments
Active Contributor

Join Date: Aug 2011
Posts: 13

Accurate Temperature Sensing

03/29/2012 6:35 PM

Is there some way short of a $1000 platinum resistance probe to electronically measure liquid temperature with repeatable accuracy to better than +/-0.1C (say +/- 0.02C or so) in the 20 degree C range? I am trying to control a coolant loop to a maximum drift of less than +/-0.1C and I don't think that I can control it to better than I can measure it! Maybe a glass bulb thermometer with an electronic eye to read the column height?

Register to Reply
Interested in this topic? By joining CR4 you can "subscribe" to
this discussion and receive notification when new comments are added.

Comments rated to be "almost" Good Answers:

Check out these comments that don't yet have enough votes to be "official" good answers and, if you agree with them, rate them!
Guru
Hobbies - Automotive Performance - New Member Popular Science - Weaponology - New Member

Join Date: Oct 2008
Location: Want to be: 34° 34' 21.60" N, 92° 55' 42.28" W Really am in Arizona
Posts: 29363
Good Answers: 1097
#1

Re: Temperature Sensing

03/29/2012 7:33 PM

How big?

What kind of load?

Etc. etc.

__________________
Luck comes and goes. Skill is forever.
Register to Reply
Active Contributor

Join Date: Aug 2011
Posts: 13
#13
In reply to #1

Re: Temperature Sensing

04/02/2012 11:25 AM

Hi Lyn, The heat load is 350W and should be very stable. The temperature controlled region is convoluted but extends for about a meter.

Register to Reply
Guru

Join Date: Mar 2007
Location: at the beach in Florida
Posts: 9693
Good Answers: 524
#2

Re: Temperature Sensing

03/29/2012 7:36 PM

If you could calculate the heat load accurately, you could program an automatically compensating solid state cooling circuit...In other words all temperature measurement and compensation is automatic, according to heat load anticipated by energy used.. a certain amount of energy used creates a certain amount of BTU's triggering a certain amount of energy to counteract...

__________________
Life is like riding a bicycle. To keep your balance you must keep moving. A.E.
Register to Reply
Guru
Engineering Fields - Power Engineering - New Member

Join Date: May 2007
Location: NYC metropolitan area.
Posts: 1757
Good Answers: 234
#3

Re: Temperature Sensing

03/29/2012 10:10 PM

Others have done it:

http://www.ilxlightwave.com/propgs/laser_diode_temperature_controller_5900.html

but I sense (pun intended) that you don't want to spend really big bucks to solve your problem. We really don't know much about the process/product being cooled, so more detail would help, until then the biggest problems I see are "sensor placement" and thermal mass/inertia.

Why not try multiple sensors and some sort of averaging circuitry to reduce the sensor placement problem, and adding some thermal mass to the enclosure/subsystem to provide some flywheel/damping effect to reduce the swings. I like the idea suggested above of using a some sort of really smart controller to monitor what's happening with the environment of your black box to anticipate the those effects.

Do space constraints permit surrounding your box with a coolant jacket? If so then you could minimize the external environmental effects and use a chilled air loop to incrementally tune the enclosure temperature. All this depends upon adequate modeling and an intelligent control strategy.

__________________
Curious minds want to know, engineering minds get answers....
Register to Reply
Guru

Join Date: Mar 2007
Location: City of Light
Posts: 3624
Good Answers: 161
#4

Re: Accurate Temperature Sensing

03/30/2012 2:25 AM

Is that the same problem you mentioned a couple of days ago ?

Register to Reply
Active Contributor

Join Date: Aug 2011
Posts: 13
#10
In reply to #4

Re: Accurate Temperature Sensing

04/02/2012 10:54 AM

Hi Nick. Yes, it is the same system. However, we have opted for a chiller system instead of the direct phase change system. The heat load is still very stable. I just wanted to add a small resistance heater to increase the temperature stability of the coolant coming out of the chiller. But the chiller already seems to be more stable then we can easily measure, so I don't think that there is much of a point in doing so.

Register to Reply
Guru
United Kingdom - Member - Indeterminate Engineering Fields - Control Engineering - New Member

Join Date: Jan 2007
Location: In the bothy, 7 chains down the line from Dodman's Lane level crossing, in the nation formerly known as Great Britain. Kettle's on.
Posts: 22848
Good Answers: 591
#5

Re: Accurate Temperature Sensing

03/30/2012 3:51 AM

At what point in the liquid? The top of many liquids is cooler than the bulk by more than 0.02degC as a result of evaporation from the surface!

If this is truly a coolant loop, then the critical thing is the temperature of the process material, and not the temperature of the heat transfer fluid.

__________________
"Did you get my e-mail?" - "The biggest problem in communication is the illusion that it has taken place" - George Bernard Shaw, 1856
Register to Reply Score 1 for Good Answer
Guru

Join Date: Nov 2010
Location: Virginia, Georgia, Idaho
Posts: 849
Good Answers: 26
#8
In reply to #5

Re: Accurate Temperature Sensing

03/31/2012 7:33 AM

very good answer, because in fact in any cooling loop, the coolant temperature gradiant will be variable, and (the best) delta control systems are process focused, not coolant specific temperature focused.

Register to Reply
Guru

Join Date: Mar 2007
Location: City of Light
Posts: 3624
Good Answers: 161
#9
In reply to #8

Re: Accurate Temperature Sensing

03/31/2012 12:46 PM

I fully agree that is exactly what I wanted to insist on: one does not measure fluid temperature but reference temperatures at the exchange wall.

Register to Reply
Guru

Join Date: Mar 2007
Location: City of Light
Posts: 3624
Good Answers: 161
#6

Re: Accurate Temperature Sensing

03/30/2012 8:01 AM

If it is the same system then I do not see where is the problem. What do you intend to control : the electronics, the cooling air flow or the fluid refrigerator temperature ?

Your goal is to have your electronic module at 23 ± 0.1 °C. If you use the scheme I mentioned which only needs a rough temperature control of the heat sink ( let say 10±1°C measured at the wall of the heat exchanger to air flow) and you modify your air flow so that the electronic module has its temperature in wished range then you measure the module temperature since the air has to be cooler in order to allow a heat transfer. This is the fastest control loop.

In both situations you do NOT need to measure in the fluid but on the surfaces.

In both cases a pt1000 sensor which costs a few bucks will satisfy the demand.

If you want a stronger signal then build a bridge with 2 pt1000 and 2 konstantan or nyckelin 1 kΩ resistances. The supply has to be low voltage in order to avoid own heating. If the sensors are correctly bond on a conductive surface the contact error is extremely small.

Register to Reply
Active Contributor

Join Date: Aug 2011
Posts: 13
#11
In reply to #6

Re: Accurate Temperature Sensing

04/02/2012 11:15 AM

Hi Nick. We do have provision to use feedback to modify the airflow. We physically can't directly measure the temperatures of the components that we care most about, so, since the thermal loads in this system are inherently stable, I wanted to keep the cooling system as stable as possible as well. That would allow us to use the temperature sensors nearby to detect any long term drifts in the system and change the set point of the coolant to correct them.

Register to Reply
Guru

Join Date: Mar 2007
Location: City of Light
Posts: 3624
Good Answers: 161
#14
In reply to #11

Re: Accurate Temperature Sensing

04/02/2012 11:26 AM

The component is on a support and has a convection surface to have an exchange with air flow. You can bond a PT1000 (3x2.5 mm) on this convector. The error you make could be the temperature drop within the convector which I presume is metallic so that it has a high conductivity and thus a high diffusibility. I doubt that the temperature difference is more than a fraction of a degree. In general the thermal resistance is about 99% at the convection surface in the boundary air layer.

Register to Reply
Guru
Engineering Fields - Instrumentation Engineering - EE from the the Wilds of Pa.

Join Date: Feb 2006
Location: middle of nowhere, Pennsylvania
Posts: 949
Good Answers: 25
#7

Re: Accurate Temperature Sensing

03/30/2012 9:45 AM

Where are you buying your RTD's???????

I just quoted a small tank with a temperature control circuit for the heating jacket. The total cost for the RTD/cable/temperature control panel meter with display was less than half what you are saying the RTD will cost.

I know you can buy an RTD and transmitter assembly (without display) for a sanitary use with a somewhat costly triclamp fitting and a sanitary finish for less than $300. A good display can be had for $150 or so. We buy that set-up nearly every week here. Make that an industrial type, which I am not too familiar with, and the price will drop.

__________________
They asked me if I was all right - I said, "well, no, I am half left."
Register to Reply
Active Contributor

Join Date: Aug 2011
Posts: 13
#12
In reply to #7

Re: Accurate Temperature Sensing

04/02/2012 11:20 AM

Hi Phys. Maybe you found what I was looking for. What RTD's are you using? The inexpensive and rugged sensors I found were in the 0.1 to 0.15 accuracy range at best.

Register to Reply
Guru
Engineering Fields - Instrumentation Engineering - EE from the the Wilds of Pa.

Join Date: Feb 2006
Location: middle of nowhere, Pennsylvania
Posts: 949
Good Answers: 25
#15
In reply to #12

Re: Accurate Temperature Sensing

04/02/2012 1:52 PM

I am guilty of what I accuse others of doing to me when I post a question - not reading the post thoroughly. I read 0.2, not 0.02 so, yes, I am looking at the same RTD's you are referencing. Only the SPRT class devices will get this accurate and I have never priced them - apparently you have. For pharmaceutical work, Class A is accepted.

A Class A, if factory calibrated at your desired 20 degrees C, may get you there. The 0.1 accuracy range is across at least 150 degrees, normally.

__________________
They asked me if I was all right - I said, "well, no, I am half left."
Register to Reply
Active Contributor

Join Date: Aug 2011
Posts: 13
#16
In reply to #15

Re: Accurate Temperature Sensing

04/02/2012 5:43 PM

Hi Phys. Before I started looking for these, I had no idea that temperature was such a hard thing to measure. Digital thermometers showing tenths place precision are ubiquitous -- but now that I think about it, I don't recall any that show hundredths place precision, so maybe that should have been a clue.

Nicks suggestion of PT1000 RTD's look like they come with something called 1/10 DIN standard, which is supposed to work out to +/- 0.03C at 0C. That would probably work for me. I just find out this morning that the physicists who came up with this spec in the first place, might not be able to tell the temperature of the critical components that accurately anyway.

Register to Reply
Guru
Engineering Fields - Instrumentation Engineering - EE from the the Wilds of Pa.

Join Date: Feb 2006
Location: middle of nowhere, Pennsylvania
Posts: 949
Good Answers: 25
#17
In reply to #16

Re: Accurate Temperature Sensing

04/03/2012 8:57 AM

The real catch here is the conversion to a readable number - I try very hard to avoid multiple readouts from the same RTD, because they will NEVER match. Unfortunately, on our water heating skids we often are required to have a display on the RTD/transmitter, a temperature control panel meter with display, and a PLC driven HMI that displays temperature. When this happens, I can guarantee that none of them will ever match unless you display to only the whole number, no tenths and certainly not hundredths place. So if you get an RTD that is accurate to 0.02°C, the display you are using will introduce far more tolerance variance than that.

__________________
They asked me if I was all right - I said, "well, no, I am half left."
Register to Reply
Power-User

Join Date: Feb 2007
Location: Gdansk, Poland
Posts: 100
Good Answers: 2
#18

Re: Accurate Temperature Sensing

10/16/2012 1:47 PM

If You will keep Your sensors always in room temperature, without temperature excursions +-30degrees , quite many hermetically sealed Pt sensors will be stable enough. Most of the PT sensors I am using (Rosemount, HP ) in biological range (0...50'C) exhibit annual drift below 3mK at ice melting point. The rest of the circuit will introduce bigger stability issues than sensor itself. But with stable load, as You wrote, stability <0.1deg should be possible.

By me, the key to success is in proper costruction of heat path. I would use liquid thermostat (with high heat capacity, isolated from ambient temperature) in that path, between thermostated object and ambient.

Consider also heatpipe with boiling temperature (pressure inside) set to Your requirement. It could be relatively cheap solution, but not sure if accurate enough.

Register to Reply
Register to Reply 18 comments
Interested in this topic? By joining CR4 you can "subscribe" to
this discussion and receive notification when new comments are added.

Comments rated to be "almost" Good Answers:

Check out these comments that don't yet have enough votes to be "official" good answers and, if you agree with them, rate them!
Copy to Clipboard

Users who posted comments:

czgut (1); lyn (1); nick name (4); PFR (1); Phys (3); PWSlack (1); RAMConsult (1); SolarEagle (1); tsaravich (5)

Previous in Forum: Flow Control Logic Case   Next in Forum: Durian Ripeness Instrument

Advertisement