Hi, New to forum. Have CTC and some technical questions on thermocouples

[Raymo]

Glad to see this is a an active CTC forum

I have my CTC about a week now and can print basic shapes etc.. but like many I've had some issues at the start.

My main problem is the extreamly inacurate temp sensors, on the Hotbed and extruders.

As a result I have to run the extruders at 280 (the max in software) to bearly achieve 215 at the head.
The Bed is not so bad I run this at 100 in software to get about 70 degrees measured with an infrared temp probe.
I measured the Bed thermistor at 67K Ohms, (totally wrong value) so I've ordered a 100K instead.

Thermocouples are generally accurate, it's the reading circuit that induces error.

My guess is that CTC may have either populated the PCB with 'wrong value' components or just used very poor quality parts (most likely).

So question:
Has anyone attempted to swap out resistors on the PCB to attain better thermocouple readings?

I'm going to give it a go, but if anyone can chime in.. please do.

And if possible I'll do the same on the thermistor, there is a 4.7k series resistor whick I might play with.

Cheers
Ray
PS, still running Makerware... but will switch to Sailfish as soon as I've resolved these issues.

other issues I've hit...
Poor pressure from the extruder feeders, replacements ordered.
No feeder guide tubes were shipped with my CTC, so I'm having to babysit it.
The supplied CTC filament has a high melt point so I can't use it, using Excelvan and it seems to melt at 210.
The aluminum hot bed is NOT flat.. so glass bed ordered.

[slippyr4]

How are you sure that your extruder temperature reads are accurate? an IR thermometer won't read such a small target accurately - especially given that the heater block is wrapped in ceramic tape. There are no resistors between the thermocouple and the MAX6675 thermocouple driver. The resistors next to that chip are for the SPI bus and will not change the temperature readings at all. The setup is very accurate - a type K thermocouple just is a type K thermocouple.

Inaccurate readings are more likely to do with the mounting of the thermocouple junction itself. They are not electrical.

The HBP on the other hand:-

On my CTC the thermistor is (was) a 100k one (note that 100K is a nominal value at 25 degrees C - you won't get 100k at any other temperature). Yours is almost certainly 100k too. The problem, however, is that CTC chose a part with the wrong beta value. Beta is a coefficient for an equation which reasonable well describes the curve of how the resistance changes over a temperature range.

In the firmware, there is a lookup table of ADC readings which have been calculated with some maths against a 100k NTC with a beta of 4066, and an R1 resistor of 4.7K. If the beta is wildly out, the readings are wrong, and that is what happens with the stock HBP.

So if your new thermistor isn't close to 4066 beta, it's not going to help very much. Also, if you follow through the maths you'll find the the value of R1 doesn't actually matter very much - a 5% tolerance on 4.7K makes little difference. It's not worth replacing R1.

The other thing to consider is that the thermistor measures the temperature of the heater pcb on the underside and that isn't the same as the surface temperature on top of the aluminium build plate. You'll always see lower temperatures there and that is worse with a layer of glass on the top too.

It's not sensible or appropriate to move the thermistor off the pcb because you'd mess with the feedback code which adjust the heater duty cycle.

I've changed the thermistor on mine for an appropriate one. The temperatures measured on the underside of the pcb are within 2 or 3 % of set temperature from 25 to 130 degrees. The top of my build plate (with glass on) is about 110 degrees when the heater is set to 130 (and it's 130 degrees on the underside). You won't achieve much better than this, and I find it's perfectly adequate for good adhesion of ABS (i use abs juice on glass). I run the HBP at 60 degrees set temp for PLA and the build surface is then around 56 degrees. Adhesion is also very good (direct to painter's tape).

I wrote up a post on replacement of the thermistor in a thing at thingiverse: http://www.thingiverse.com/thing:1403267 - it mentions the thermistor I used which is very suited to the firmware in our printers.

[Raymo]

Thanks for your detailed reply.

You raise good questions which put me back to the bench for more testing.

So I calibrated a Fluke thermocouple probe with Ice and boiling water, it's accurate within 2 Degrees.
Then I recorded the following:

		Measured
	Set to	IR sensor	Fluke TC
Hot end	280	231	240
Bed	100	74	69

As you correctly pointed out, the IR is not as accurate.
But I'm confident in the Fluke.
The thermal paste is the white paste used for PC CPU's.

I fairly sure the CTC software reads +40 degree error in the Extruder reading and a +30 Degree error in the Bed reading.


To fix the bed, I shall order a new thermistor as you have defined.

But for the hot end.... I'm sure why the error is there.

So as you suggested... it may be a poor contact with the thermo couple (but a poor contact would only lower the reading, not increase it ?)
Anyway, I applied thermal paste between the CTC thermocouple and the hot end head.
But the results were the same, still a +40 degree offset between setting and actual temp.

More work to be done in this area...


might there be a lookup table for the thermocouples as part of the Firmware?
I'm on Makerware right now...

[slippyr4]

I've never studied the thermocouple sections of the source code. However, I doubt it: that chip handles all the thermocouple reading and cold junction compensation and provides a reading of temperature in degrees C readable over SPI.

Where are you putting your fluke probe? Can you get it under the same washer as the CTC thermocouple at the same time?

What temperature do you get on the fluke when you set a target temperature of 190 or 200? It might be that the heater can't provide enough heat to maintain 280 degrees. That's pretty hot.

I've never tested my extruders to be honest; I only experienced problems with the HBP. I use excelvan filament, like you, and I find it works well at a set temp of 230 for PLA and 230-240 for ABS (depends on the colour). It doesn't really matter what the set temp is, if the plastic performs ok.

Have you tested both extruders? Do they perform similarly?

[Raymo]

Spend the evening fitting the pollow bearings x 4... wow what a tricky job to do without taking it fully to bits... but managed to slide the bars out the sides

Anyway, on with the temp issue...

To measure the temps I placed the probe at 3 different points which all read much the same:
1, under CTC thermocouple
2, at junction between brass head and block
3, directly on the heater element
All read the same within 2 degrees


Good idea abourt the load to maintain 280, so I tested a setting of 230, but again the Fluke measured 199 to 200,
Intrestingly this is only a 30 degree error.

Both extruders were the same reading within 1 degrees, so unlikely to be a faulty component as it's on both extruders.

Next I tested 200 degrees setting and measured 172 degrees on both extruders, an error of 28 degrees, similar but the delta is dropping

Finally I tested 100 Degrees setting (I know te Fluke to be accurate here), the result was a reading of 97 and 98 degrees on left and right extruders respectively.

So now all three thermocouples ( two CTC's and the Fluke) are accurate at 100 degrees (close enough), the errors start to occure at elevated temps.
Since both CTC probes are similar, I'm suspecting my Fluke!!!

[Raymo]

Interesting article here
http://circuitcellar.com/cc-blog/the...thermocouples/

The thermocouple measures difference not absolute temp.

There is a hot junction (measurement end) and a cold junction (at terminal block).

The difference is read as voltage. The terminal block is assumed to be at zero degrees.
In real application a chip is needed to measure terminal block temp and subtract it from the reading, thus making the reading more accurate.

However for this to work, it assumes the chip and terminal block are at the same temp.

Looking at the mega board cooling design, it's clear that hot air from the stepper drivers is driven over the chip and junction block.

This may be creating a difference in temp between chip and junction block.

I plan to test this theory out later today.

.

[Mjolinor]

You can get your cold end temperature by shorting out the hot end thermocouple.

[Raymo]

Not sure I follow, isn't the hot end already shorted out?

[Mjolinor]

No, the hot end is producing a voltage with the thermocouple. If you short it out then it doesn't make any voltage and the cold end thermocouple (or the equivalent cold compensation circuit) temperature is shown on the display, it is probably 25 degrees C.

[Raymo]

Ok, thanks.
I've examined the PCB layout and have come to the conclusion that there is no substancial temp offset between the chip and terminal blocks... though it's not a good design cooling wise... the CTC has the hot air blowing over the thermocouple terminal block.

I've moved my attention over to sailfish... see my other post on this as I have small issue with the head printing off the hot bed.

Cheers,
Ray