It's a very rare thing to see someone post temp results with environmental conditions and probe accuracy, so at best it's a shot in the dark on how accurate their results are.
First off, I think some knowledge is needed to understand how thermister type probes work since they are ubiquitous in watercooling and modding. Using the most basic explaination, thermister temp probes work nonlinearly using changing resistance values of the thermister which is measured by the control unit and then interpolated using the Steinhart-Hart equation. So basically, complete accuracy is all but impossible given the inherent error of the control device, probe and interpolation of results.
This is why Your temps could be the same even knowing they are 1-4C different from Your friend.
I have been thinking of a way to cheaply calibrate ones temp probes without the $150+ costs of conventional calibration labs and this is what I thought would work the best with the least out of pocket expense.
NOTE: I did all the work for this post and calibrated My personal oral thermometer in Fahrenheit and converted all those readings into Celcius since the oven at work only has Fahrenheit readout.
Here's My idea, goto the drugstore and buy a cheap oral thermometer (here's the one I have at home) http://www.drugstore.com/products/prod....pid=71276&catid=1332&trx=29384&tab=0#0
Make sure it clearly states the accuracy range of the probe....this is very important to get the best results otherwise we are back to square one. I actually took this one to work and put it in our calibrated oven (accuracy +/- 0.05F) and it read 100.1F when the oven was set to a stable 100.0F...that's perty damn sweet for a $10 thermometer. Now that we have a standard with which to calibrate our probes we can continue.
Since most of the temps people are reporting are around 80-120F (26.7-48.9C), the idea of using an oral thermometer is best since they are specifically designed to be accurate in this general range and should give the best results with the lowest cost (see thermister description at top). We all know the average person is 98.6F so You make a good "oven" with which to calibrate Your temp probes
Now take Your temperature with the oral thermometer (stick it under Your tongue and breathe through Your nose until it's done) and immediately after it's done (most of them take a few seconds to register the results and will beep) stick Your temp probe to be calibrated under Your tongue, breathe through Your nose again (I hope You washed it well hehehe) and wait the same amount of time as it took for the oral thermometer to register Your temp and record the results. It's best to do this 3 times per probe (with all readings within one decimal place of each other to exclude human or probe placement error) and average the results into one final reading.
With this method everyone should be able to achieve accurate and repeatable results time and again to +/- 0.5F (0.3C)....finally accurate CPU/GPU readings.
Lets all make a stand against erroneous and inconsequential temperature readings and calibrate our probes today!!!
If You have a better test methodology or a cheaper "accurate" standard please post it here.
Lets try make this thread a sticky so that we can finally start believing some of the temps we read!
Lazien
EDIT --- Nonlinearly means the thermister does not increase or decrease proportinally in resistance as temperature rises or falls. There will be spikes and valleys throughout the entire thermisters usable range making temperature predictions very difficult from resistance readings.
First off, I think some knowledge is needed to understand how thermister type probes work since they are ubiquitous in watercooling and modding. Using the most basic explaination, thermister temp probes work nonlinearly using changing resistance values of the thermister which is measured by the control unit and then interpolated using the Steinhart-Hart equation. So basically, complete accuracy is all but impossible given the inherent error of the control device, probe and interpolation of results.
This is why Your temps could be the same even knowing they are 1-4C different from Your friend.
I have been thinking of a way to cheaply calibrate ones temp probes without the $150+ costs of conventional calibration labs and this is what I thought would work the best with the least out of pocket expense.
NOTE: I did all the work for this post and calibrated My personal oral thermometer in Fahrenheit and converted all those readings into Celcius since the oven at work only has Fahrenheit readout.
Here's My idea, goto the drugstore and buy a cheap oral thermometer (here's the one I have at home) http://www.drugstore.com/products/prod....pid=71276&catid=1332&trx=29384&tab=0#0
Make sure it clearly states the accuracy range of the probe....this is very important to get the best results otherwise we are back to square one. I actually took this one to work and put it in our calibrated oven (accuracy +/- 0.05F) and it read 100.1F when the oven was set to a stable 100.0F...that's perty damn sweet for a $10 thermometer. Now that we have a standard with which to calibrate our probes we can continue.
Since most of the temps people are reporting are around 80-120F (26.7-48.9C), the idea of using an oral thermometer is best since they are specifically designed to be accurate in this general range and should give the best results with the lowest cost (see thermister description at top). We all know the average person is 98.6F so You make a good "oven" with which to calibrate Your temp probes
Now take Your temperature with the oral thermometer (stick it under Your tongue and breathe through Your nose until it's done) and immediately after it's done (most of them take a few seconds to register the results and will beep) stick Your temp probe to be calibrated under Your tongue, breathe through Your nose again (I hope You washed it well hehehe) and wait the same amount of time as it took for the oral thermometer to register Your temp and record the results. It's best to do this 3 times per probe (with all readings within one decimal place of each other to exclude human or probe placement error) and average the results into one final reading.
With this method everyone should be able to achieve accurate and repeatable results time and again to +/- 0.5F (0.3C)....finally accurate CPU/GPU readings.
Lets all make a stand against erroneous and inconsequential temperature readings and calibrate our probes today!!!
If You have a better test methodology or a cheaper "accurate" standard please post it here.
Lets try make this thread a sticky so that we can finally start believing some of the temps we read!
Lazien
EDIT --- Nonlinearly means the thermister does not increase or decrease proportinally in resistance as temperature rises or falls. There will be spikes and valleys throughout the entire thermisters usable range making temperature predictions very difficult from resistance readings.
Facebook
Twitter