Author Topic: thermistor time constant  (Read 1585 times)

0 Members and 1 Guest are viewing this topic.

Online SimonTopic starter

  • Global Moderator
  • *****
  • Posts: 18118
  • Country: gb
  • Did that just blow up? No? might work after all !!
    • Simon's Electronics
thermistor time constant
« on: January 07, 2020, 07:27:22 pm »
I am looking at this datasheet: https://www.te.com/commerce/DocumentDelivery/DDEController?Action=showdoc&DocId=Data+Sheet%7FGA10K3MCD1%7FA%7Fpdf%7FEnglish%7FENG_DS_GA10K3MCD1_A.pdf%7FGA10K3MCD1

the time constant is 200ms but what exactly ho they mean by time constant? is this the time to fully match the new temperature or is it our old friend tau from RC circuits of 66% ?
 

Online floobydust

  • Super Contributor
  • ***
  • Posts: 7679
  • Country: ca
Re: thermistor time constant
« Reply #1 on: January 07, 2020, 08:08:46 pm »
I've always seen it defined as the thermal time constant, to reach 63.2% of the temperature difference.
You can cheat this spec, depending on the fluid used (water vs oil) and if there is a stirrer.
Tiny bead thermistors are around 10 seconds at best, so this 0.2 sec response time is very fast, almost too fast...

"Thermal Time Constant simply put, under zero conditions, is the time it takes a thermistor temperature sensor to change 63.2% of the total difference between the initial and the final body temperature; when subjected to a step function change in temperature. In simple terms, it represents in time, how long it takes a thermistor to recover up to 50% of its initial resistance."

It might be a guillotine tester they use?
https://www.ametherm.com/blog/thermistors/thermal-time-constant-ntc-thermistors
 

Online SimonTopic starter

  • Global Moderator
  • *****
  • Posts: 18118
  • Country: gb
  • Did that just blow up? No? might work after all !!
    • Simon's Electronics
Re: thermistor time constant
« Reply #2 on: January 08, 2020, 07:39:25 am »
So theoretically I don't have to wait 5t, knowing what the last reading was allows me to make sense of the next reading. If I take a reading 0.2s later I know that this is 63% of what the temperature actually is but that assumes that the last reading had stabilized after 5t so I would still need to wait that time or this gets really complicated as I would need a history of readings and use an algorithm to model the response of the thermistor say over the last 10t period.
 

Online floobydust

  • Super Contributor
  • ***
  • Posts: 7679
  • Country: ca
Re: thermistor time constant
« Reply #3 on: January 08, 2020, 08:39:13 am »
Ideally, the sensor response time is faster than the system (time constant) for stability in PID control etc. if that is what you are doing.
If the system changes temperature faster than that (sensor), feedback control gets sloppy as you are always lagging, always behind, and need 1 second at least to get accuracy.
You can model the sensor's thermal characteristics (time constant) and attempt to predict (add lag compensation) but it never works well for me. The problem is the system gain (that a model is built upon) might change and make closed-loop control go unstable, if you are modeling it. A classic example is a change in airflow past the sensor, that controls a heater. Higher airflow decreases thermistor response time (due to better heat transfer to it) and usually system gain goes down due to the heater being less effective with higher mass flow.
I might be wrong here but I have made a few oscillators by accident.
 

Offline Ice-Tea

  • Super Contributor
  • ***
  • Posts: 3229
  • Country: be
    • Freelance Hardware Engineer
Re: thermistor time constant
« Reply #4 on: January 08, 2020, 08:58:54 am »
I image that if you would measure every 0.2s and use the 63% value that would be absolute worst case/largest possible change (for when you just applied a fresh step). Not a bad thing to work with, not sure if keeping a ledger and releasing further algorithms would bring much to the table?
 

Online SimonTopic starter

  • Global Moderator
  • *****
  • Posts: 18118
  • Country: gb
  • Did that just blow up? No? might work after all !!
    • Simon's Electronics
Re: thermistor time constant
« Reply #5 on: January 08, 2020, 09:53:05 am »
I image that if you would measure every 0.2s and use the 63% value that would be absolute worst case/largest possible change (for when you just applied a fresh step). Not a bad thing to work with, not sure if keeping a ledger and releasing further algorithms would bring much to the table?

Well As i see it unless i take a measurement every 5t I cannot say for certain what the temperature is an what I measure can still be out of date as a new temperature will already be in the system and what I read is at best 0.2s old but then this will happen on every sample. So if I sample every 0.2s I need to know the measurement history back by a sensible amount of samples in order to interpret the latest measurement which can either be really good or go horribly wrong and be quite some math for a micro controller to compute on what could be multiple inputs.
 

Offline 741

  • Frequent Contributor
  • **
  • Posts: 406
  • Country: gb
    • Circuit & PCB Design (small PCB quantities OK)
Re: thermistor time constant
« Reply #6 on: January 08, 2020, 12:06:55 pm »
The 63.2% will be 1-1/e.

Could it be you want to measure the slope? Say you measure at 1/10 the time constant.
  • If the slope is flat-ish, you are close to an accurate measurement
  • If the slope is +large, you have a large under-estimate
  • If the slope is -large, you have a large over-estimate
If the above is logical, "it should be possible", hopefully, to quantify and express temperature in terms of  slope and resistance.

Online SimonTopic starter

  • Global Moderator
  • *****
  • Posts: 18118
  • Country: gb
  • Did that just blow up? No? might work after all !!
    • Simon's Electronics
Re: thermistor time constant
« Reply #7 on: January 08, 2020, 12:26:53 pm »
You mean measure the rate of change, yes that makes sense. But when arriving close to the setpoint things get tricky. The system would have to slow down and take accurate measurements 5t apart.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf