need help on configuring new thermistor

Hello,

I bought a new hotend and it comes with a thermistor that does not have a glas bubble as usual, but is packaged with zylindrical metal housing like the heater cartridges.

Description says: Thermistor NTC 100k (choose 13 in marlin firmware)

Looking at config.g I did not find any similar to those "13" of marlin, but I found the M305 entry.

I changed the existing line for P1 to "M305 P1 R100000 H0 L0" and got a value of 2000°C at powering the printer on.
I removed one "0" and after power on I got 10°C

When I turn the heater on, firmware stops heating with failure message: heating to slow

So I'm trapped. I checked the wiki for M305 but I don't understand all that parameters and I guess, I have no value for any of them.

Any hint is appreciated!

The thermistor resistance in the M305 command is specified by the T parameter in the M305 command, not the R parameter which is the series resistance. Thermistor type 13 in Marlin is a 100k thermistor with a B value of 3950. So use:

M305 P1 T100000 B3950

HTH David

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Yes it worked

Thank you very much!

Did I mention: I love duet3D

Hello,

last days I had strange experience with hotend temperature - last print I did a test with 320° hotend temperature.
I was freightened about the values so I measured the temperature with a multimeter.
That showed, that nominal 320° was really 280°

On heating up the hotend I noted that on lower temperatures the real temperature was above the nominal value - about 2-5°
On higher temperatures the real temperature felt below the nominal value and the gap increases with rising temperatures.

How can I calibrate a thermistor? - Its the 100k that should match marlins nr.13

I just got the hint, that I possibly got a thermistor ATC Semitec 104GT-2 instead of that marlin type 13

The 104GT2 thermistor is one of the presets in the config file generator at reprapfirmware.org. If you want to enter the values manually, you can find the T, B and C values for this thermistor at [duet3d.dozuki.com].

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Puh - that's very high sophisticated stuff. Guess I'm not cute enuf for it ...

I found the values for the 104GT which leaded me to these values:
[firmware-value <> real temperature]
180 186
230 239
300 310
The new calculated B value should be 4563
By trial and error I tuned B value to 4630 (which matched 300 firmware-value by 301 real)

Then I started autotune.

Now I get these values:
[firmware-value <> real temperature]
180 183
230 232
300 296

I guess, I should change C parameter, but I have no idea on how to do that.
I searched wikipedia for Steinhard-Hart math, but my math knowledge is to poor to understand it.

Is there a way to calculate the B and C parameters by the gap between firmware-temperature and real measured values?

The configuration at reprapfirmware.org includes a calculator that will work out the T, B and C parameters from three temperature/resistance pairs. But if you know that it is a 104GT2 thermistor, you can just use the values given on the wiki page that I linked to.

Edited 1 time(s). Last edit at 01/16/2019 03:03AM by dc42.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Quote

But if you know that it is a 104GT2 thermistor, you can just use the values given on the wiki page

I wrote that I found the values. Of cause from the page you guided me to.
The first three value pairs where obtained using the parameters from your wiki.

Before I changed config to your values, I had a gap of 40° - firmware value 320° and real temperature of 280°

So after changing to your wiki values, I got a gap of about 10° - much better than before, but I'd like to reduce the gap even further.
May be the chinese thermistor is not like the one you know as 104GT2 ...
So - to be honest - I don't know the type of thermistor.

I read from the calculator in reprap-firmware in your wiki, but I didn't find any link in the reprap-firmware-configurator. I stepped through all pages.

and after all - I don't have temperature/resistance pairs - and I don't know how to obtain.
When I use the thermistor to regulate the heater, I can't measure the resistance.
So I only have the firmware-temperature and the real temperature.

Isn't there a way to calculate the rest from that two values?

@Brummie: How and where do you measure the hotend T°

---

"A comical prototype doesn't mean a dumb idea is possible" (Thunderf00t)

I borrowed a multimeter that is capable to measure temperatures.
Afaik those multimeters have a deviation less than 1°

The heatsensor of the multimeter is like the usual thermistors - very small head
... and I put that sensor in the gap between nozzle and heatblock.

by the way: I read in the output of the M303 command, that autotuning uses "default" frequency.
I suppose that frequency refers to adc measuring cycle
Is it possible to change frequency?


The slack on the right side is the start of the fan
and temperature needs very long time to get back to nominal temperature.

I would like a quite faster response.

To determine the thermistor specs, you have to remove it from the hot end, attach it closely to your T° probe and dip both in liquid, oil for ex. that you heat at different T° and compare.

Edited 1 time(s). Last edit at 01/16/2019 01:23PM by MKSA.

---

"A comical prototype doesn't mean a dumb idea is possible" (Thunderf00t)

I finally found the temperature-calculator in repetier firmware configurator.
Its quite hidden and I did not find any hint on how to open it.
So first you have to enable pid control and then you have to put the cursor (enter the field) in the resistor field.
Then a window pops up, where a thermistor can be choosen.
After selecting "custom thermistor", the window content changes to be able to enter 3 value pairs.
The window closes as soon as the resitor field looses focus, so its not possible to manually copy the values. But copy&paste works.

Quote

To determine the thermistor specs, you have to remove it from the hot end, attach it closely to your T° probe and dip both in liquid, oil for ex. that you heat at different T° and compare.

Hm, I don't know, what to use to hold liquid at 300° or more - no, not my way!

I used the values from calculator and changed them by trial and error.
I now found a setting, that is quite close to reality.

But pid-control is extremly slow on little temperature changes. When I want faster prints or use fan at different speeds, I'd need a controller much faster.

So I tried bang-bang - but that runs 10° above nominal value and 5-10° below nominal value.
Is there a way to configure threshold values?

PID control can't work faster than the dead time of the heating system. If you have run heater tuning, then the tuning process will have reported the dead time that it measured, and you should have copied this into the D parameter of the M307 H1 command (and copied the A and C parameters too).

You can attempt to speed up the PID response by reducing the D parameter in case the value measured by heater tuning was higher than actual, but if you set it too low then the temperature will oscillate.

Going back to the original question, if you want to adjust the thermistor reading at high temperature to match your thermometer reading, you can do live adjustments of the B parameter by sending M305 P1 Bxxxx where xxxx is the new B value you want to try. Make small changes, e.g. 1% or 2% at a time. When you have found a B value that works, put it in the M305 P1 command in config.g.

Edited 1 time(s). Last edit at 01/17/2019 08:26AM by dc42.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Quote
dc42
PID control can't work faster than the dead time of the heating system. If you have run heater tuning, then the tuning process will have reported the dead time that it measured, and you should have copied this into the D parameter of the M307 H1 command (and copied the A and C parameters too).

You can attempt to speed up the PID response by reducing the D parameter in case the value measured by heater tuning was higher than actual, but if you set it too low then the temperature will oscillate.

Going back to the original question, if you want to adjust the thermistor reading at high temperature to match your thermometer reading, you can do live adjustments of the B parameter by sending M305 P1 Bxxxx where xxxx is the new B value you want to try. Make small changes, e.g. 1% or 2% at a time. When you have found a B value that works, put it in the M305 P1 command in config.g.

The point is that what he measures is not the T° the thermistor is exposed to.

Edited 1 time(s). Last edit at 01/17/2019 09:03AM by MKSA.

---

"A comical prototype doesn't mean a dumb idea is possible" (Thunderf00t)

Quote

The point is that what he measures is not the T° the thermistor is exposed to.

I measure the temperature touching the outside of the nozzle.
May be the inside of the nozzle has different temperature, but I'm quite sure, that the distance from heater to thermistor is as far as from heater to the nozzle point, where I put the sensor to
... so I don't care for the possible difference

When I startet with the wrong thermistor type, my temperartures where off by about 40°
... now I'm as close as 1° - so pretty good (well, for my opinion )

My thermistor setup is now:

M305 P1 T99962 B4810 C6.7e-8

Quote

PID control can't work faster than the dead time of the heating system.

Ok, that's a point, I understand.
What I don't understand - what is the base for dead time, or how is it determined?

When I start autotuning with default pwm settings, I get this message:

Heater 1 model: gain 361.7, time constant 213.2, dead time 5.5, max PWM 0.60, calibration voltage 24.1, mode PID, inverted no, frequency default

when I start autotuning with pwm-setting of 1, the message is this:

Heater 1 model: gain 399.6, time constant 139.5, dead time 11.0, max PWM 1.00, calibration voltage 24.1, mode PID, inverted no, frequency default

so shorter time constant means longer dead time?

Quote

You can attempt to speed up the PID response by reducing the D parameter in case the value measured by heater tuning was higher than actual

Hm, don't understand this.
I take the result from autotuning to adapt M307 and M301 commands.

Which means - from this result of autotuning:

Heater 1 model: gain 399.6, time constant 139.5, dead time 11.0, max PWM 1.00, calibration voltage 24.1, mode PID, inverted no, frequency default
Computed PID parameters for setpoint change: P5.6, I0.144, D43.6
Computed PID parameters for load change: P5.6

I changed the commands in config.g to:

M307 H1 A399.6 C139.5 D11.0 S1.00 V24.1 B0
M301 H1 P5.6 I0.144 D43.6

When you look at the temperature-graph some posts above, you'll see, the initial elevation is much more rampant than the heating after switching on the fan.
bang-bang always has the same rampant heating, but temperature threshold values are way to far from nominal value. So if it would be possible to configure the threshold-values I would like to change to bang bang control.

Remove the M301 command from config.g. The PID parameters are set automatically from M307, and different ones are used depending on whether the hot end is heating up or maintaining temperature.

The time constant and dead time should not vary so much depending on the maximum PWM. My guess is that the time constant obtained when PWM=1 is more accurate, but the dead time of 5.5 is closer to the true value than the figure of 11, which is much higher than for a typical hot end.

Edited 1 time(s). Last edit at 01/18/2019 02:53AM by dc42.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Quote

Remove the M301 command form config.g

Thanks for the hint!

Works better now:


by the way: what's the time unit of dead time?

If you find a way to reduce dead time of full pwm (1), I'll be glad to hear from you!

Dead time is measured in seconds. You can edit the M307 command if you want to try reducing it.

I have it on my work list to improve the measurement of dead time in the tuning process.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Quote

Dead time is measured in seconds

You're kidding - aren't you?

I know the words "dead time" from full bridge drivers, but those drivers have timeunit of microseconds. Slow drivers might have timeunit of milliseconds, but timebase of seconds?
I can't beleive it.

When I look at the specs of the atmel processor, it is capable of doing 1Msps on ADC. Well, that might be quite optimistic, but tear it down by ten, than the adc will still do 100kSps.
What is so slow, that you need a dead time of 11 seconds? ... or even 5 seconds?

In 11 seconds the whole printer might burn down and nowbody gets rid of it?

Before you wrote the time unit, I thought Duet3D is rock solid and reliable.
Now I'm in doubt.

May be, I don't understand the dead time meaning here.

In my opinion PID control of hotend is the most vital control of a 3D Printer. So it can't be fast enuf. Ever.
I'm off socks

The dead time is determined by the heater cartridge, heater block and thermistor, not by the processor. Think of it as the time taken for heat from the cartridge to reach the thermistor.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

I understand the confusion. This seems to be one more case of a term used elsewhere that's different for this application. (If you've been around a while, you'll see lots of ... discussion about the term "jerk.")

When I saw the term "dead time" my first thought was the same as yours, but since we aren't dealing with AC waveforms, or switch-mode power, I had assumed that once again, we're dealing with something else. It does make sense that there is a delay between when you switch on the heater and the thermistor (thermocouple, PT100, PT1000, or whatever temperature sensor you're using) starts to warm up, and if stable temperature is desired, you need to know what this is.

---

MBot3D Printer
MakerBot clone Kit from Amazon
Added heated bed.

Leadscrew self-built printer (in progress)
Duet Wifi, Precision Piezo parts

Quote
SupraGuy
I understand the confusion. This seems to be one more case of a term used elsewhere that's different for this application. (If you've been around a while, you'll see lots of ... discussion about the term "jerk.")

When I saw the term "dead time" my first thought was the same as yours, but since we aren't dealing with AC waveforms, or switch-mode power, I had assumed that once again, we're dealing with something else. It does make sense that there is a delay between when you switch on the heater and the thermistor (thermocouple, PT100, PT1000, or whatever temperature sensor you're using) starts to warm up, and if stable temperature is desired, you need to know what this is.

The context here is a heating control system. To mention a full bridge is irrelevant. Besides the second is the unit of time, a micro second is NOT a different unit.
So, sorry, not that much confusion when people proceed logically and certainly not when it has been said a few posts before (see DC42 post where he states "PID control can't work faster than the dead time of the heating system.").

Note, I am reacting to the unjustified reaction by the OP and I find DC42 to be very patient .
So if this printer burns down, end of the story, too bad

My last post here.

Edited 1 time(s). Last edit at 01/18/2019 01:46PM by MKSA.

---

"A comical prototype doesn't mean a dumb idea is possible" (Thunderf00t)

Quote

The dead time is determined by the heater cartridge, heater block and thermistor, not by the processor. Think of it as the time taken for heat from the cartridge to reach the thermistor.

Thank you for the explanation.
@SupraGuy
Thank you for your understanding!

I now understand the meaning of "deadtime" in 3d printing context, but I don't understand the result from firmware autotuning.

The time for the heat to reach thermistor depends for sure from the material of the heatblock. If the material would be the only dependency-factor, deadtime must be constant.
So may be, it depends on the power of the heating cartridge too.
Then it must vary on different pwm settings.

But I don't see any reason, why deadtime with 60% power is half the time of 100% power.
Does this make any sense?

I made some more tests yesterday, as I thought - possibly I have a misunderstanding of the pwm-factor too.
But I don't.
My current heating cartridge is quite poor - it takes 2:30 minutes to heat from 21° to 300°
When I start autotuning with pwm of 0.3 the cartridge is not able to reach any temperature above 150°
So using pwm setting of 1 gives the most power to the heating cartridge.

What am I missing here?

Why is the deadtime with 100% power double the time from 60% power?
Is there something wrong with my printer?

Edited 1 time(s). Last edit at 01/19/2019 12:05AM by Brummie.

You are right, the dead time should be constant. It's the measurement of the dead time by the firmware that isn't working well.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Quote

You are right, the dead time should be constant.

Just to ensure, that I got you right:

This means, that the deadtime depends on the material of the heatblock only?
So when I buy a heater cartridge more powerfull - I won't get a faster control anyway.
So max print speed is not limited by the heating cartridge but by the material of the heatblock? Right?

Hot ends with very large nozzles and heater blocks such as the Super Volcano may need additional heating power to handle the high filament melt rate. Ordinary hot ends don't.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Quote

Hot ends with very large nozzles and heater blocks such as the Super Volcano may need additional heating power to handle the high filament melt rate. Ordinary hot ends don't.

That was not my question.

I replaced all parts that influence print speed - so when I end print-tests with very poor speed, I need to ask, whether Duet3D could be the source of lazyness.

What is the actual problem you are having?

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

On my A6 with stock hard- and firmware I was able to tune printparameters, that I got good results with PLA at 120mm/s

Now with Ender-3 I wanted to reach similar results.
I changed to Duet3D as the stock firmware was not able to save changed configuration.
M500 was supported, but only until next reset

Together with the duet3D I bought a Titan Aqua Extruder - as I'm printing ASA and therefore need higher temperatures.
Recently I got a link to a really good test object - 4 cubes in different distance. Mainly intended for retract purposes.
But it showed me a lot more.

With my current setup I can print a 20mm cube in spiral mode at 120mm/s and result is ok.
As soon as I turn off spiral mode, print failures arise on all ends.
I had to lower temperature and speed - but I stopped at 55mm/s. Stil not perfekt, but I don't want to lower speed any further.
Filament is said to be tested at 120mm/s ...

So - I don't have any idea, how to speed things up. Could it be related to firmware?
I don't know.

First of all, I have to change to a firmware, that works reliable. Any other issue is just nice to have.