Your input needed: genuine high quality thermistor vs genuine high quality PT100 in 3d-printing

Your input needed: genuine high quality thermistor vs genuine high quality PT100 in 3d-printing application.
Besides the ability to read temperatures above 300°C, are there more advantages compared to the thermistor? Maybe more precise readings? Or better durability? Or something else? And why does the amplifier board cost 15$? -.-
Right now, for me it is “printing above 300°C for additional 20$” - Am I missing something?

I love the cartridge format, but now that E3D have cartridge thermistors, that is a not really relevant anymore.

I believe the same, once E3D went with the cartridge thermistor the only reason to get a pt100 is if you need the higher temps.

Why not thermocouple? I still don’t understand why E3D went to PT100 instead of type-K (which lots of printers already used).

There are resolution differences due to non-linearity of output; thermistor has high resolution in the range of interest but also larger error.

E3D offers PT100 and thermocouples, on top of thermistors. Thermocouples are very prone to receiving noise. Pt100 seems a bit more durable than the thermistors, but not a lot. The only reason I use PT100 is I was planning to go high temp, which finally happened a few months ago.

@Jeff_DeMaagd I’ve never had a noise issue with my three thermocouple machines. Main issue is getting the TC tip in electrical contact with the hot block, which is a problem with thermistor wires too. (Cartridges solve this problem at the cost of slower sensing.)

You can also get cartridge TC but larger mass around TC will slow response. Typical TC are more involved in signal as their signal is much smaller then PT100 or thermistors. We use TC in equipment we build but that is for large range of type T as we need to read -100 c to +200 C.

@Peter_Hanse In this case, TC and PT100 are nearly identical to the user… both need an amp daughterboard or a controller that comes with an amp chip onboard.

RepRaps usually use thermistors simply because the MCU’s ADC can read them via a voltage divider without an amp chip. That saves $3-4 on the controller board versus using an amp. But that has caused lots of issues over the years with things like low MCU ADC resolution, noisy sensing in Smoothieboards, thermistor reading lockups in Beaglebone Blacks, etc.

TCs are bad due to the wiring being brittle and subject to fatigue.

@Andrew_Diehl_Landru Not if you use stranded TC wire. (Admittedly more expensive.)

@Ryan_Carlyle ​ everything we have been able to get ahold of eventually breaks. We had a nearly 100% failure rate at 2 years for all machines in the field. Where do you get your stranded wire from and what strand count/size?

Just a couple (pun intended) of points.

With a thermocouple you should be mindful of how you terminate the compensating cable if accuracy is your bag.

Second, PT100’s which I work with daily, are still sensitive to noise. Just use a screened cable and ground the screen at only one end.

With a proper screened cable you’ll have to think about routing because they have a fairly stiff character. Keep your bend radius large if you want a long cable life.

In conclusion: many disadvantages for the advantage of higher temp-readings :-/

On the topic of noise: The @E3D_Support PT100 board has pretty brutal hardware low-pass filters in the input circuit, so I’d think noise of any reasonable level would not be an issue.
(Image from https://wiki.e3d-online.com/wiki/E3D_PT100_Amplifier_Documentation )

@Andrew_Diehl_Landru A few stranded TC options:

  1. Performance 3-d sells premium TCs with M3 thermowell tips as aftermarket replacement for Makerbot Rep2/R2x printers. It’s 24ga but I probably would have done 28ga if it were me. Also works with pre-cartridge E3Dv6 blocks that had the M3 thermistor fixing point on the side. https://www.p3-d.com/collections/replacements-parts
  2. Carl Raffle sells a matched hot block + thermocouple cartridge that is compatible with Makerbot 4th gen and E3Dv6 hardware: https://shop.raffle.ch/shop/tecto-heater-block/
  3. Here’s 100ft of TC wire on Amazon if you want to cut your own to length: https://www.amazon.com/Thermocouple-Lead-Wire-24AWG-100Ft/dp/B007IBG6CK

@Rene_Jurack It’s not just higher temp readings. Benefits of EITHER option over thermistors:

  1. Dedicated amp chips offer higher resolution and more reliable operation without needing extensive oversampling or filtering (eg 14bit native versus 10bit on an Atmega or noisy 12bit on a LPC1768). This works best if you use SPI to talk to the amp, like the Mightyboard and Duet do. Having the amp output an analog voltage to the MCU ADC like the E3D PT100 board is throwing away a lot of performance.

Sidenote: I would actually advise against TCs and PT100s if you’re going to hook the amp up to an Atmega 10bit ADC. The low ADC resolution ends up being an issue. But that’s the controller board’s fault, not the temp sensor itself.

  1. Every PT100 or Type-K thermocouple in the world has the exact same calibration curve, so there’s no worrying about which thermistor part number you have, or whether your firmware is using good values for the lookup table or equation coefficients. They’re all accurate to about +/-2C and precise to about +/-0.25C. Every time, no calibration required. In comparison, a thermistor with a standard pre-coded firmware lookup table or beta value is often off by 2-10C in the printing temp range and possibly >30C outside the normal range.

  2. Output is nearly linear across the entire working temperature range, meaning you don’t run into temp sensing issues outside the intended temp range for the thermistor voltage divider circuit. (This is mainly an issue for cold-temp operations like printing PCL at 50C where the hot end thermistor resolution is terrible, or turning on the printer in a cold garage where low heatbed temp can trigger firmware temp errors.)

Then, I personally prefer TCs over PT100s because they’re insanely robust and hackable in the RepRap spirit. A thermocouple is literally just two wires of different metals, with some end-dressing for protection. If you cut a 6ft thermocouple in half, you get two 3 ft thermocouples! You can also buy any type-K thermocouple off Amazon and cut off the tip to dress it the way you want. If your TC is damaged, you can cut it back, strip and twist the ends together, put a little kapton tape over the tip for insulation, and you’re back up and running. Whereas if your thermistor or PT100 lead wire breaks close to the tip, it’s toast.

I can plug my multimeter’s thermocouple probe into my printer and use a heat-and-wait command to delay printing until the chamber is warm enough for ABS. I can plug my hot end thermocouple into my four-channel temp data acquisition unit. (http://paxinstruments.com/products/sku927/)

The main issue with TCs in my view is that you have to run the TC wire all the way to the TC amp for the sensing to be 100% accurate. The TC only measures the difference in temperature from one end of the TC wires to the other. So if the amp chip isn’t adjacent to the cold end of the TC wire, it may not read properly. Some printer designers have screwed this up (looking at you, FlashForge!) and had bad results with TCs because they spliced a shorter TC wire over to copper wire within a hot enclosure. That only works properly if the TC amp is also inside the enclosure.

Beware of PT100 “sensors” from AliExpress - the example is below. It makes sense to put in PT100 is to reliably print at 250-280C.
But when these are sealed by regular epoxy resin that burns at 270C, it is a problem. Some sensor resistive elements are rated for 450C, but when placed in the tube and sealed by epoxy, they fail at 250C.

The “high temperature sensor” pictured below is even worse - the platinum axial low accuracy resistor there is rated for 150C max and isolated with teflon tubing - see what happened with the right one exposed to 300C for 15 minutes.

When selecting PT100 sensors, make sure that the vendor explains how they are manufactured, that they are sealed with high-temperature silicon and have thermal-conductive filling. And what warranty for high temperature operation they give.

Interestingly, the 50K B=3950 glass sealed thermistors can nicely work up to 300C, and with the proper 2.5K resistor in parallel they are nearly linear in 200-300C range.
Unfortunately, only few 3D printer vendors know about thermistor linearization techniques…

1 Like