Temperature monitoring in the lab

[king.oslo]

Hello there,

I came to realize that I am lacking decent ambient temperature measurement in the lab. It would be good to get some 0.5% or better accuracy in the lab. Data-logging would be good. What are your solutions?

Thank you for your time.

Kind regards,
Marius

EDIT: I understand that the title may sound misleading. I do not want to check the temperature of my whole lab at once, I just want to make spot checks in small areas. Thank you.

[Kibi]

I have an Arduino wired up to the internet, Thingspeak to be precise. I use DS18B20 sensors.

https://www.thingspeak.com/channels/3355

[ftransform]

I built a RTD circuit from texas instruments app-notes that agreed with a mercury thermometer with no calibration (at least at lower temperatures).
A thermistor or I2C compatible chip would probably be good enough and easier to implement though since you do not require a large temperature range.

This one is particularly easy to implement if you do not want to bother making a more complicated interface circuit with ADC and friends.
http://www.mouser.com/ProductDetail/Microchip-Technology/TC74A0-50VAT/?qs=sGAEpiMZZMuaKyt/jIB/prGm71f6OXg/
Not the most accurate though.

I assume you mean to measure ambient temperature. If you want to measure chips then buying a multimeter that is thermocouple compatible of fluke 51/ [52 (logging)], (if you wanna spend more money and get more accuracy, probobly)

[MikeK]

0.5% or better accuracy?  Won't there be more than that between different spots of the room?

[king.oslo]

An assembled solution or a kit would be best. I have too many projects already  M

0.5% or better accuracy?  Won't there be more than that between different spots of the room?

Perhaps, but I only care about the temperature in the area where the instruments are sitting at the moment.M

[MikeK]

Oh, okay...gotcha.

[Harvs]

0.5% or better accuracy in the lab.

My question would be half a percent of what?  Typically when we talk about a percentage error for a measurement device, we'd be talking about % of Full Scale Range.

In temp measurement its normally more meaningful to talk about +/- xx degrees rather than % of something else.

The most common ambient temp measurement devices I've seen are +/- 1^oC absolute.  Is that enough for your application?

Edit:  I'm interested in whatever you come up with, I've been thinking the same as you for a while.  I've been thinking of using one of the two RPi's I've got sitting around to achieve it.

[ftransform]

BTW for test equipment you may want to consider temperature indicator strips (as shown in daves extreme harddrive teardown video).

They are cheap and require no power but they will display the correct temperature. You can stick them on the case of your equipment.

[robrenz]

Unless you are metrology lab you can practically ignore the accuracy and focus on stability of temperature. High stability temperature control of a room is extremely difficult. Trying to remove the natural vertical temperature gradient in a room is one of the most difficult aspects. 
If you can't control your room temperature to within +-1 deg (which would be very good) why worry about if it is accurate to +-0.5 deg?

I would suggest logging temperature over 24 hour periods in your lab along with doing a temperature map of variations within your main work area including vertical positions. Then work on things to try to make the variations less. Trust me, anything you have now to measure temperature is more than adequate.

I am not trying to be abusive here, just a reality check.

[NiHaoMike]

I once visited a company (for an interview) that built their labs as thermal chambers within thermal chambers, with yet another thermal chamber for the DUT. Both the inner and outer chambers were kept at 25C year round, and the DUT one was either held at 25C or varied as needed. They also designed the air circulation to account for the heat coming off the equipment.

Where I work now, the thermal control is much more lax and there are many spots that are as much as 5 degrees C or so cooler than most parts of the room. It looked really funny with some of the workers dressing up for cold weather in the middle of summer.

At home, I don't have anything accurate enough to require much temperature control so energy savings are the main deciding factor. It often can go as high as 28C (82F) during the summer and as low as 17C (63F) during the winter. I had it go all the way up to 31C (88F) once (unoccupied) in order to perform a thermal stress test on a new PC.

[king.oslo]

Thank you.

My 10V reference output change before my eyes as the temperature varies.

If I am going to plot how it changes with temperature, I thought it would be nice to know the temperature where it is sitting. That way I can put degrees c/kelvin on my axis. I also thought this probably would be useful for many things in the future. Data logging would make it even more flexible.

Harvs: percent of reading. So for 25*C I think it would be good to know the temperature to better than +-.125*C

I care about accuracy so that I can relate whatever I am using it for to someone else, with at least a rough degree of precision.

Thanks.M

[NiHaoMike]

If it's just one device or a few devices, use a Peltier to stabilize the temperature.