• Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Electronic Thermometer

Status
Not open for further replies.

Dr.Mandrake

New Member
I am going to work on an electronic thermometer which detects ambient temperature and output to a seven segment display. Range is 0C-99C and it should be able to display upto one decimal point accuracy. eg. 12.5. I searched through net and found temperature sensor ic's readily available such as LM35,LM334 etc. I like to avoid bar graph display and LCD panel displays. Display should be very active during dark and I prefer seven segments. Any inputs are welcome. Also I prefer to avoid uC's but in my limited knowledge in electronics, I think the temperature sensing IC's can directly output voltage changes and the complicated thing is to add voltage to display functionality.
 

jbeng

Member
Use a digital voltmeter chip such as the ICL7107 with the LM35 as the input. Here's a link to a schematic you might be able to modify to suit your needs...
 

tcmtech

Banned
Most Helpful Member
If you go the ICL7107 IC route you will be money and time ahead to go on eBay and buy the LED 3.5 digit digital voltmeter from Asia engineer for around $10 with free shipping. They use an improved version of the same ICL7107 circuit so all you need is to tweak the input voltage divider to work with your temp sensor.
Plus they have a panel mounting case on them too!

I have been doing that to make them into amp-meters and to be able to read different voltages than what they come set up for.
They are fairly easy to modify for any input of Milli Volts or higher. And if you smoke the IC they are socket mounted so all you do is buy another ICL7107 IC for around $3 on eBay and change it out!
 

Brevor

Member
I made several digital thermometers years ago. The first ones I built used the ICL 7107, These work good but the calibration process is time consuming. Later I switched to using a microcontroller and the DS1820 temp. sensor, no calibration required.
 

ericgibbs

Well-Known Member
Most Helpful Member
Actually, the LM35 gives a linear output, 100 mV / °C.

Just amplify that and throw to an ADC converter + Decoder + BCD driver.
hi Hayato,

The LM35 is 10mV/Cdeg output.:)
The accuracy is only +/-0.5Cdeg and the OP wants an accuracy of 0.1Cdeg.

3v0 and I have already suggested the LM35 solution but the OP didnt want to go that route.
 
Last edited:

Mikebits

Well-Known Member
hi Hayato,

The LM35 is 10mV/Cdeg output.:)
The accuracy is only +/-0.5Cdeg and the OP wants an accuracy of 0.1Cdeg.

3v0 and I have already suggested the LM35 solution but the OP didnt want to go that route.
At this point, what can ya do?
 

ericgibbs

Well-Known Member
Most Helpful Member
At this point, what can ya do?
hi,
Basically the OP has got to make up his mind which solution he wants::)

1. DSxxxx to say a PIC with 3*7seg LED's [ he wants 7 segs as the display]

2. LM35 to a ICL 7106/7 adc ic with LED drivers. [this would be my choice]

3. LM35 with bar graph display.

Regards.
 

Dr.Mandrake

New Member
Wow thanks very much for your inputs!
jbeng, thanks for the information regarding ICL7107. Brevor, why you told that the calibration and finetuning is difficult for ICL7107.
ericgibbs said:
2. LM35 to a ICL 7106/7 adc ic with LED drivers. [this would be my choice]
Regards.
I couldn't much understand the circuit posted by jbeng, but I think that it uses thermistors as sensors and you are suggesting an LM35 in place of that. Correct. Also I afraid, it's CPU temperature sensor mainly, and requires range settings for different environments. I want to avoid any post-implementation calibration and sensing range is 0C-99C.
Eric, do you think LM35 is capable of providing .1C accuracy. Then I'll stick on that. Anyway I want to avoid any uC here. How can I modify it for LM35?
 

KMoffett

Well-Known Member
...do you think LM35 is capable of providing .1C accuracy...
Leno,

Did you look at the datasheet for the LM35?
http://www.national.com/ds/LM/LM35.pdf
Check the "Electrical Characteristics" on page 3, for "Accuracy" and "Nonlinearity". Even at a single temp, in the middle of its range it's only to 0.2C...1.0C at the extremes.

What are you doing that you need that kind of accuracy over that range? Even physically coupling a sensor to the CPU will result in variations of degrees C.

Ken
 

ericgibbs

Well-Known Member
Most Helpful Member
I want to avoid any post-implementation calibration and sensing range is 0C-99C.

Eric, do you think LM35 is capable of providing .1C accuracy. Then I'll stick on that. LM35?
The accuracy of 0.1C is not possible, but a resolution of 0.1C is.

As you are measuring the room temperature, I would have thought that 0.5C 'accuracy' would be acceptable, but display the result to 0.1C.

The LM35 outputs 10mV/C so for 99.9C thats ~1Volt.

The ICL 7107 [LED driver version] has a base sensitivity of 200mV, so so by choosing the correct input resistors 100.0C should be possible.

Get the ICL7107 datasheet.
 

Attachments

Last edited:

transistance

New Member
LM73 can measure within your desired range, but it has +/-1.5C accuracy @ -25C to +115C range.
 

tcmtech

Banned
Most Helpful Member
For countless normal day to day applications I see so many people wanting sensitivity and resolution capabilities that are 10 to even 100 times greater than whats practical or even necessary.
In real life terms every time you want 10 times greater sensitivity or resolution your device becomes 10 to 20 times more complex and 10 to 100 times more costly!

Unless your doing scientific research those levels of accuracy and resolution are totally unreasonable.
If your trying to measure your CPU temp on a computer just get one of the free hardware monitoring programs. They will tell you your core temperature and everything in real time and with a far greater accuracy than a passive sensor on the external case is ever going to be able to tell you!

When passively measuring temperature, environmental drift is going to limit you to a realistic +/- 5% margin of error. Being able to compensate for that environmental influence is what will drive up the complexity and cost many many times over!
 

Hayato

Member
hi Hayato,

The LM35 is 10mV/Cdeg output.:)
The accuracy is only +/-0.5Cdeg and the OP wants an accuracy of 0.1Cdeg.

3v0 and I have already suggested the LM35 solution but the OP didnt want to go that route.
Thanks for the advice, Eric!

Under the given circunstances I considered that his meaning of "accuracy of 0.1 °C" was, in fact, "resolution of 0.1 °C"
 

Brevor

Member
Brevor, why you told that the calibration and finetuning is difficult for ICL7107.
Not difficult just time consuming. (1-2 hours) It only matters if you are making many.

The ICL7107 is a good circuit, the units I built 20 years ago are still working today operating 24 hours a day 365 days a year.
 
Last edited:

tcmtech

Banned
Most Helpful Member
How is calibrating that difficult? If you use the standard that water freezes at 0 C and boils at 100 C It should not take more than few minutes to fine tune one!
Or at least thats How I have always calibrated temp sensor devices.

Although I am now temped to make one by modifying a ICL7107 based volt meter to work as a temp meter.
 

Sceadwian

Banned
The temperature water boils at depends on your local barometric pressure. And the temperature of ice is dependent on the impurities in the water, but yeah you could use them as general guidlines, though I would definitely check your local barometric pressure to determine the boiling point at least, that's the most flexible. Some cooking recipes have different directions for high altitude cooking because water doesn't boil at the same temp.
 
Status
Not open for further replies.

Latest threads

EE World Online Articles

Loading
Top