Ikea probes

Hello,

Let me start by introducing myself...
I'm Tom, live in Belgium and found this nice forum after researching "pit controllers"

I'm planning on building a couple of HM myself in the near future. Just waiting on the boards to arrive by slow boat from China... All the electronics are waiting in a box from Mouser...

Now I've read in the Wiki that Ikea probes are not the best probes to be used because they are "wired backwards". This I find strange as inside the probe sits a thermistor and those don't have polarity... So could someone please explain to me why they are wired wrong?

Also found a thread on a Dutch forum about someone who changed the thermistor inside the Ikea probes to a "NTC MF58 4320 B 204 200K ohm ±3% thermistor". These have a high tolerance for temperature and are small enough to fit inside the probe. So I ordered a couple together with a few 2.5mm mono male jack's so I could rewire the entire probe and save a couple of euro's
What are the thoughts about this approach?

I have so many questions, but am reading up on this part of the forum and already found some answers so maybe most of them get answered before my boards arrive. When I start/finish soldering I'm probably gonna need some help, but we'll see that when we get there

Well...ok...Maybe one more... I've read an older post about PID tuning. In Marlin (yes have a 3D printer ) there is an autotune feature for the PID algorithm and I thought why not implement this in HM? Has this got something to do with the Bias that is introduced? Or am I missing something?

Regards,

Tom

The IKEA probes are fine and like other probes, the metallic "sheath" is actually one of the two conductors that forms the circuit. There is one wire inside the sheath and the braid forms the other. The way that they fit into the jacks, we'd put 3.3V into the sheath and ground the internal wire. That works just fine... until the sheath touches anything that can carry current and it will effectively take the thermistor out of the circuit, or put another resistance in parallel with the thermistor. The FANTAST device itself does not use a simple resistor divider like HeaterMeter does, I can't remember how it works but it uses some sort of RC circuit and times pulses to find the cutoff frequency which will tell you what the resistance is.

PID autotune simply doesn't work very well for this application. What autotuning relies on is that the response of the system is completely dependent on the input. Our input is air and our output is temperature. The control variable is actually the fire, which we can not control predictably as the conditions change over the course of a PID tuning period. If you autotune the Marlin where sometimes you're blowing on the hotend and sometimes you aren't, you're going to get very bad results because you've added a variable it doesn't know about.

Let's say on our first PID tune loop we add X quantity of air, which ignites Y quantity of coal, and heats the smoker mass itself by Z degrees. Then we turn off the air and wait for the temperature to drop, but that doesn't extinguish our coal completely so when we start adding air again, it will take less of our X variable but our Z degrees will also help the temperature to recover more quickly. This isn't just academic because I've done probably hundreds of hours of attempting to get any usable working autotuning into HeaterMeter but the values are always garbage.

You can see this for yourself by turning off I and D and just running P and trying to find the P where you get stable consistent oscillations


You're going to find that even if you find the right number that works for one oscillation, the very next oscillation will be different in gain or period. You can check these numbers by mousing over the last high or low peak in the graph and it will tell you the gain and period. If you find something you think is close to right, you can look in the system log and it will have the calculated PID autotune values there and you're good to go. It gets you in the ballpark, but you already were in the ballpark when you started because the defaults aren't terrible.

It works pretty well if you hook the heatermeter output to a SSR and an immersion heater stuck in a bucket of water, the PID loop is easily tuned because we're controlling the power going into the water and measuring the output. I've even used a heatermeter to control a 3D printhead with a 30W heater cartridge and a 100k glass bead thermistor. Works great! But adding air to a fire does not behave this predictable proportional manner.

An answer of the creator himself

Thank you for the informative answer about PID tuning. I completely understand now why this hasn't been implemented over the years...

About the probes... do other probes use two wires inside the metal casing? If so I could order myself some glas insulated wire, rewire the inside of the fantast with the high grade thermistor and be done with it.
If the above statement doesn't apply and all probes use the metal casing as a conductor... which polarity does the casing have?

Just looking back on the link of the thermistor I provided I'm thinking if it is posible to use a 200K thermistor with HM?

P.S. thank you for creating such a nice device! I bow my hat down to you sir :b


Regards,

Tom

In the complete rework of the webui that's coming next design iteration, I'm going to expose some of the PID tuning stuff just so everyone can see how mediocre it is (and then I suppose get requests for implementing 34 different PID autotune algorithms!)

Most probes have a single wire insude then use the braid as ground. This works better because usually what you're going to be touching the braid to is some other ground so the potential between them and the HM ground is minimal. It can create some ground loop noise but we try to filter that out both in hardware and software. The fantast probes work fine if the mono jacks are just rewired tip positive rather than shield positive as they are now. As I think you've seen elsewhere though, they're not that great of probes to begin with.

Any thermistor can be used with HeaterMeter, but you'll need to get steinhart coefficients for it because we don't use Betas. You might also need to swap out the 10k 1% resistors in the HeaterMeter with another value to shift the probes best response range to where you want it. You can see how the resistance affects the range with this quick graphing tool I made.

Why would you go through all the trouble to modify an Ikea Fantast probe when Maverick ET72 are readily available for an affordable price?

Because the cheapest I can source them is about 24€ + shipping...
I live within 5min of an Ikea and there the fantast costs 7€80. You do the math
If the only mod I need to do to them is change de mono jack to get them to work then this is acceptible. I only plan to use the HM for low-n-slow and for this purpose the fantast has never let me down.
If for one reason or another the probes don't work as they should (ie, give irradic readings) I'll first change the thermistor and only as a last resort I'll considor buying 100€'s worth of probes...

Bryan Mayland said:

Any thermistor can be used with HeaterMeter, but you'll need to get steinhart coefficients for it because we don't use Betas.

Click to expand...

I've found this Steinhart calculator from from Stanford Research and after making some measurements and inputting those one gets the coefficients.
I've found the two files where to input the new thermistors ( conf.html and hmcore.cpp ), but how to edit the cpp file as this is already a hex file within the rPi image...?

You don't need to edit the source code, you just go to the web page in the UI, select "Custom" thermistor type and it gives you the A B C fields for the Steinhart values. Enter them and hit save and you're good to go.

That's so much easier!!

Tom De Bie said:

Because the cheapest I can source them is about 24€ + shipping...
I live within 5min of an Ikea and there the fantast costs 7€80. You do the math
If the only mod I need to do to them is change de mono jack to get them to work then this is acceptible. I only plan to use the HM for low-n-slow and for this purpose the fantast has never let me down.
If for one reason or another the probes don't work as they should (ie, give irradic readings) I'll first change the thermistor and only as a last resort I'll considor buying 100€'s worth of probes...

Click to expand...

Hmmm, €24 sounds like a little bit too much. I've send you a PM.