Welcome! Log In Create A New Profile


Heater Calibration questions

Posted by spcmicro 
Heater Calibration questions
September 29, 2008 12:50PM
I am currently finishing up my first extrude head. I am using the Bits from Bytes extruder mechanical assembly with a version 1.2.1 Universal PIC controller board built as an extruder board using a 10K thermistor. The firmware is version 1.1. Version 1.0 gave temperature readings that were bouncing all over the place and 1.2 won't communicate at all. Not a problem though since the 1.1 version seems to function properly.

I ran the heater profiler and plotted my results. Fairly linear over the entire range with the exception of the higher temperatures over 147 - 149 degrees C.

With reference to the profiler instructions, my question is this, what exactly is the process for calculating hm and hb? The text talks about a linear regression and fitting a line but I'm not sure how to achieve that and I want to get it right. I've attached the plot of my data from the test.

The other question I have relates to the ABS and HDPE material that I obtained from the RRF shop here in the states. BTW, I think it would be nice to have a chart of the possible filaments and the suggested operating settings. What are the desired settings for these materials?
open | download - HeaterProfileData.JPG (69.8 KB)
Re: Heater Calibration questions
September 30, 2008 06:57PM
Okay, I'm no expert but...

Now that we have the failsafe in place above...I used this site to calculate the regression coefficients here


The y-axis intercept is one value and ur x-coefficient is the other. Simply plug these into the properties page. I'm not sure which is which as I'm at school right now but the bigger default value in the preferences page will be the y-axis intercept.

AS to your electronics. I've got that same set up working on my machine and have a question. What is your C3 cap value you used on the Extruder board? I had a bunch of trouble with this but finally had someone here give me the proper values of components to use. That could be another source of problems for you.

Hope that helped

Re: Heater Calibration questions
October 01, 2008 08:19AM
For my thermistor I used a 1ufd electrolytic for C3. The assembly instructions had 2 references to using a 1ufd value, once in the bill of materials and once in a note that stated that the value of C2 was nominally 1ufd. This morning I found another reference on the next page of the extruder board instructions that states that the best value for C2 is .1ufd when using the standard thermistor. I stuck with the 1ufd. I am using the standard 10k thermistor that I purchased last year from the RRRF shop.

Using the 1ufd I set the value in the preferences to 0.000001. I can't remember whether this was the default.

The problem I'm having might be on the heater side of things. I'm going to follow the link you sent and calculate the correct values for hm and hb. Thanks for the link...
Re: Heater Calibration questions
October 01, 2008 09:42AM
The PIC electronics can not measure temperatures any higher with a 10K thermistor, you need a 100K thermistor for HDPE and ABS. See the thread here [forums.reprap.org]

Re: Heater Calibration questions
October 01, 2008 11:09AM
Thanks nophead and demented. It looks like I'll have to build another heater barrel because the thermistor is sealed in the fire cement. I'm glad I bought extra parts and I've been through it once already so I know where thee tricky parts are. It sounds like I'll be good to go with the ABS and HDPE using the PIC electronics once I get the new heater built with the 100K thermistor. Is this assumption correct? Is there anyone else out there using Ian's extruder design using laser cut parts and a PIC controller?
Re: Heater Calibration questions
October 04, 2008 03:02PM
Think the easiest way to do first-order-regression is to plot all coordinates on a paper and then draw an averageing line through the points, everyone does this in basic physics.

They use the square metric because of some mathematical simplicity, so you end up with a curve which minimises the error as a sum of all the squares of the distances from the points to the line.

I just looked at the wiki for linear regression, jeez they really know how to complicate things, lots of greek symbols and transposed matrices :-)

In essence, there exists some kind of proof that if you do according to the blah blah algorithm you will end up with an error which is minimized for the square of the distances.

One shouldn't put any magic in the metric choosen, it is simply choosen because it is easy to count with.
Sorry, only registered users may post in this forum.

Click here to login