Trying to add heated bed, Ulticontroller does not show temp correctly

[puromaani 0]

I am having difficulties with modifying marlin to accept heated bed thermistor. The process seemed simple enough, but problems appeared when I realised that with the sensor I use, I need to change analogReference to internal 1.1 Volt.

First the electrical side:

Heating resistors will be powered by and external power source, and the current will be switched on the low voltage side with an solid state relay. The system is grounded and shielded properly. Necessary fuses are added. So I would say that the problem is not here.

The thermistor I use is KTY81-210 from phillips. It differs from normal thermistors in that it has a positive temperature coefficient (resistance goes up with temperature).

I have added 15kohm resistor to R4 (as per http://reprap.org/wiki/Thermistor, KTY82-210 (Philips) (2kOhm SMD)).

I did not do anything to AREF (VREF) as it seemed not to be connected to anything.

I have calculated the resistor value (just to be sure) with basic voltage divider calculations, and by values obtained here: http://www.nxp.com/documents/data_sheet/KTY81_SER.pdf, 15kohm is spot on.

So the electrical side seems to be correct?

I think that the problems are on software side.

I have not connected the solid state relay to ultimakers mainboard yeat, because ulticontroller shows temperature wrongly.

I downloaded Marlin from daids github. From configuration.h I did the necessary modifications to enable heated bed. This seems to be done correctly, as ulticontroller now shows also the heated bed temperature. I had to add new temperature sensor setting (I used #11) as there were no config already for my thermistor.

From configuration_adv.h I only enabled quick_home, everything else is default.

I added new table to thermistortables.h, using the same notation as the other tables there. Table values I took from reprap wiki mentioned before.

Temperature.cpp is where the problem lives. Probably this line:

ADMUX = ((1 << REFS0) | (TEMP_0_PIN & 0x07));

If I read correctly that line makes the AREF to be VCC. For my sensor I need the reference voltage to be 1.1 V.

Doing this modification at the "case 2: // Prepare TEMP_BED" portion:

ADMUX = ((1 << REFS1) | (TEMP_BED_PIN & 0x07));

Only seemed to change the received ADC value about 200 units lower. This is strange. Writing 1 to REFS1 should make AREF internal 1v1?

I put SERIAL_ECHOLN function to temperature.h to line 1066, after temp_meas_ready = true;, to see what data is the ADC getting from the KTY sensor. With 1 written to REFS0 ADC will get about 1600 and with 1 written to REFS1 ADC shows ADC value of about 1800.

That is strange enough, since change from 5V reference to 1.1V reference should rise the value sharply?

The other, even stranger, problem is that ulticontroller shows bed temp to be 0/0° and when I grab the sensor (to elevate temperature) it changes first to '/0°, then (/0° and )/0°. After releasing it slowly goes back to 0/0°.

I would say that the problem is definitely in marlin settings, and I have done something wrong there. Should I upload the whole firmware to some website?

 

[gr5 2,071]

Well.... You could just spend the $3 and get this:

https://ultimachine.com/content/thermistor-100k

Everything you say makes sense. I'm wondering about the 1.1V reference I guess. Maybe it isn't properly going into the ADC mode you think it is. Maybe some code somewhere else prevents you or overrides you in some loop or ISR somewhere. I have done embedded programming before but not with Arduino. I would read about all the ADC modes and see if somewhere something is putting it in a mode you didn't expect.

Did you measure the voltage at the input to the ADC? With a voltmeter? To see if the voltage is as expected?

 

[puromaani 0]

Thank you for your suggestion.

I will buy that sensor from ultimaker to fix the problem.

In the meantime it would be nice to figure out why the firmware doesn't work properly with the other sensor. I will try to measure the voltage at AREF pin. Probably I wont get much out of it, since the software is continuously changing between 5V and 1V1 AREFs. And I don't own an oscilloscope.

 

[gr5 2,071]

I don't think ultimachine is related to ultimaker. Don't get them confused.

 

[puromaani 0]

Yes I realised that after sending last message, my bad.

(btw, this "..post need approval from a moderator.." is kind of annoying)

After doing some tests, I figured out that weird ulticontroller readings came from a simple problem. Measured ADC value was out of limits of the used temperature table.

Changing AREF on the fly does not seem to work, this might be because AREF voltage needs some time to drop from 5V to 1V1. KTY81 sensor works fine, but with 5V AREF the measured range is only about 150 or-so steps. This limits the possible accuracy of the measurement. But for my purposes accuracy of 1 to 2 degrees celcius is close enough. I will use this setup until I can change the sensor to a 100k one.

So the problem is, atleast partially, solved. Thank you gr5.

If someone has figured out a way to dynamically change AREF very quickly, please let me know.

 

[gr5 2,071]

> "..post need approval from a moderator.." is kind of annoying)

I think that will go away after a bit. I'll "like" your post to help speed you on to a non-newbie poster.