Thermister Temperature Data Logger

Are you looking for something off the shelf or custom. You have some pretty demanding specs for an off the shelf device. Does it have to be thermistor? Is the temp range you specify for just the sensor and if so what is the ambient temperature range of the logger?

Thermistor (in its nature) practically cannot keep for expended time the accuracy / resolution of 0.1 deg C. For above temperature span better is RTD Pt type sensor.

Some of US semiconductor manufacturers offer adequate op-amplifiers + Anal-to-digital converters. As their output can be used a counter or a proper card in a computer. NI + LabVIEW are the best combination of h/w + s/w combination I have used for years.

Holyday Greetings!

You can have all holidays. I have none for me. I will prefer to take last leave sometime later as it is too early for me now.

Idea is to develop a new product which is very compact, highly accurate, self powered for say minimum 5-years of use. The device is to work under high pressure of up to 2500 psi and also need to be corrosion resist, SS316L casing. Yes the entire device should take -40C to 125C temperature range and perhaps need to be more accurate in 0-70C range to about 0.1C and then it can be as bad as 0.2C for the entire range. Thermisters are best choice for high sensitivity and small size. RTD can also be used and I have a plan to use PT-1000 or PT-2000 or even PT-5000 in some devices. Thermisters have come of age now and getting 0.1C accurate in glass encapsulated is just possible to be fit for almost any environment.

There are few problems here.

1. Accuracy 0.1C 0-70C and 0.2 for -40C to 125C range

2. Precision - 12-bit

3. Long battery life

4. Large data storage with 12-bit precision

5. SS316 capsule that can take 2500 psi pressure

6. Mechanism to start the data logging when required

7. Dumping data using RS232 or USB or some kind of serial port

8. Non-contact data transmission, unique ID, real time tagging for data

10. Affordable for mass production - 100,000 parts per year

11. To be produced within an year time frame

Temperature is inert thing. Therefore making one metering per second is enough.

So if you will use for example - TI microcontroller MSP430F447 (there is others):

• Ultralow-Power Consumption:
• Active Mode: 280 µA at 1 MHz, 2.2 V
• Standby Mode: 1.1 µA
• Off Mode (RAM Retention): 0.1 µA

• Five Power Saving Modes
• Wake-Up From Standby Mode in less than 6 µs
• 16-Bit RISC Architecture, 125-ns Instruction Cycle Time
• 12-Bit A/D Converter With Internal Reference, Sample-and-Hold and Autoscan Feature

And if we assume that processing time will be about 100ms for one metering (including spline interpolation of temp. sensor characteristic - it's necessary for high accuracy measurement) and other part of second is Standby Mode (not take into consideration). The off-duty factor will be 0.1. And it's very approximately we assume that Active Mode will get 1mA for processing time (100ms) taking into consideration of the off-duty factor it will be about 0.1mA during 1s. And so for bettery having 4A/h (take it for no particular reason) it will be about 40000 working hours. It's about 5 years.

I'm sorry if there are some inaccuracy in my thoughts I did it anyhow only as example attainability of goal.

Reduced error at 0.06 and temperature derating are certain trouble for the development.

But anyway this idea is realizable in hard.

Luck!

MPS430 uC is a very good choice for low power measurement as it has internal 24-bit ADCs also in some models. I will sure into this and is valuable information. In 2007 atmel will also come out with low power fast uC with internal ADC but their plans for SRAM is not so good. Best idea should have been with large SRAM. Atmel engineers say that SRAM take very large space compared to Flash so they prefer more flash and less SRAM. I have some battery backed SRAM from Texas and Maxim but these are all very big size. Now FARM may be a better idea but their power rating does not sound so good. Data reliability is another problem in SRAM. I reveived some valuable information on that in another discussion header from NASA engineer. I think using some technique, low power SRAM data can be made reliable.

Hello Shyam,

We manufacture loggers for the electrical industry, in embedded systems with built-in temperature sensors. Microchip have all of the electronics that can cover your temperature range already; we are up-the-curve on Microchip processors etc. so that would be OK. We have experience in techniques for filtering, linearising and interpolating to achieve your required accuracies, low power consumption etc

It would seem to me that there could be more of a problem with a battery to operate over this temperature range reliably for 5-10 years! Maybe battery technology is where our expertise runs out, but I've had a look at: http://www.batteryuniversity.com and in particular http://www.batteryuniversity.com/partone-15.htm

where it would seem that lithium-polymer with dry solid polymer electrolyte is suitable to operate from 60-100degC. So assuming it could be stretched to 125degC, then there is still a problem to operate down to -40degC. So perhaps a second battery is required to do this???

NeilJ

Lookfor Tadiran LTH-5903 battery that can go from -55C to +125C range. These also have long life to 30000 hours. There are other models also in the same range with different current ratings. Problem comes at low temperature when battery can not deliver current. High temperature is often good for current but shrtens life span. Data logger normally is not operated in extreme conditions but used for recording such conditions, once a while.

Hi,

may be the battery problem (use it at -40C) can be solved by using thermo-insulating material inside box (there are modern materials with good thermo-insulating behaviour) and also auto heating system (gets some power but not so much as it can seem).

It's necessary to do temperature model.

Luck!

At lower temperature, internal resistance is high, and hence current capacity decreases and voltage drops down to 2.7V from 3.6V for 10mA operation at -30C.