i will be receiving a Turbosource upper intake manifold in the next few days and it will give me another opportunity to compare the dynamic response of a thermistor and thermocouple. i have attempted to do this previously but wasn't able to get the thermistor working properly... maybe the wrong calibration.... or it was just bog slow..

the reason for this post is i would like to be advised as to which thermistor is considered the fastest? i will pick one up and install it next to my thermocouple and we will have a race. film at eleven.

thanks for your help.

 

Measuring the dynamic response time of any type of thermal sensor (thermistor, thermo-couple or whatever) is inherently difficult. Laws of thermodynamics and such. For that reason, I'd suggest doing this as a DIY laboratory project instead of attempting this on an intake manifold/engine to ensure you'll get accurate and repeatable test data out of the effort.

Ideally, to do this measurement, you'll want to simultaneously expose all sensors being tested against a calibrated delta-T step function, and measure the time response of each sensor. Here's a strawman "how to" procedure for this sort of test, feel free to critique if I'm missing something:

1. Obtain two "calibrated" temperature measurement mediums to test your sensors against - one can be a bucket of ice water (~0*C), the other can be a pot of boiling water (~100*C). In both cases, have a 3rd independent temperature measurement device handy to verify the temperatures of each medium while you are testing your sensors.

2. Build a rig to mount the two (or more) sensors under test so that they can all be simultaneously dipped/exposed to your calibrated temperature mediums. Electrically, you'll need to rig up a means to simultaneously monitor the electrical responses of each sensor versus time, suggest an oscilloscope so you can see each sensor's electrical response to the step temperature change vs. time; e.g., voltage or resistance measurement vs. time.

3. Start your test by exposing both sensors under test to the colder (ice/water) medium, and wait until both sensors are reading 0*C or the actual measurement reading from your 3rd temp measurement device. That is when both sensors have reached equilibrium, and the "race" can start...

4. To start the race/test, at time = 0, hit the stop watch and quickly take the sensors out of the ice bucket medium and drop them into the boiling water medium and leave them there, while still monitoring the electrical response of each sensor on the O-scope. The stopwatch is used to null out the inherent measurement error of the delay in moving the sensors from one medium into the other (a second or two?). Since you just simultaneously exposed both sensors under test to a nearly instantaneous ~100*C change in temperature, watching & recording the time it takes for each sensor to respond to and stabilize at the higher temperature is the measurement that answers which sensor has the fastest response.

 

if one had an old ECU laying around, one could hook the sensors up, and run a data log. it would also be pretty easy to put the two competing sensors on a bracket, so you just dip them at the same time?

 

^That's the idea - dip them both in at the same time so they both see the same abrupt delta-T (change in temperature). You can do a cold-to-hot delta-T as I described, or hot-to-cold delta-T; in theory, it shouldn't matter which way you do it, but perhaps some sensors might respond better in one direction than the other? So if you have the time/resources, it would be best to test in both delta-T directions. Using an ECU to log the electrical sensor output vs. time data would also simplify things greatly and automate the data logging, especially if said ECU has calibration curves to support all of the sensors under test.

 

As someone who has actually used ice points for drift checking laboratory platinum temperature sensors and done temp measurement, enclosure testing training etc, I'd strongly recommend using as long a plugged cylinder or test tube in a large Thermos or cold box/boiling pot/thermos with 3rd test sensor to confirm your "zero" and "hunny" is stable (even if not right at zero/100, being a comparative test), preferably with something like a small milk throther head (magnetic driven puck may be another option if it doesn't induce noticeable signal offset) stirring air within the base of the tube, you want to test the combined thermal/output response lag of the sensing elements exposed to moderate airflow not liquid medium or solid contact with the element given 2 orders of magnitude higher density of water on top of the higher heat coefficient of water, unless you are testing for an immersed coolant, oil or fuel temperature application. Into water they will change much, much faster, positive current devices (thermistors/resistive coil types) will read slightly different in a medium which removes more energy at a given temperature too.

After saying all of that, just for seeing step response, not getting accurate scale point data, a 2 or 4 channel thermocouple to use 2 as test stream stability references, a stabilised hair dryer/heat gun output (not just turned on), and a fridge or freezer or ice box on fan cycle or with a fan running internally will probably be much easier for most people to set up at home and more likely to actually reflect sensor exposure conditions in a car intake for transient response purposes.

bonus points, mount sensors in/on piece same material as intake/elbow, preferably slightly thicker than mount as the whole thing will be exposed to airstream as opposed to one surface in an intake.

I's suggest tired fridge that doesn't actually hold 4C or the outlet stream of an air conditioner to low or medium heat settings on a hair dryer is more reflective of real world transients than a 100 Celsius degree step too.


It's a worthwhile test but given there is thermal mass in intercooler and intake and much of the transient fuelling behaviour is from the pressure and flow change in the manifold and wall wetting conditions (funny how a lot of these problems go away with predictive MAP and wall wetting fuel models) I don't think it quite as important as some people make it out to be for tuning.