WEIRD temperature swings

[eriksw 0]

I have an Ultimaker 1 which has been intensively used for almost a year. Worked fine until two weeks ago.

Problem: the temperature reading on the controller is now giving weird readings during printing.For instance, with temperature set to 220, the readings follow in rapid succession (a second or less) saying: 220-217-223-217-222- 216-223-220-223-216 and so on. So: weird hops that can not relate to actual temperature of the nozzle as this cannot fluctuate that quickly. The actual temperature seems to drop, as printing usually starts to fail then.

The weird thing is that it works fine when I don't use the fan. No problems then, but as soon as I switch on the fan (or when it switches on automatically at 2nd layer) the readings become weird as described above. Switching off the fan cures the problem again...

So no broken leads or faulty sensor I would think, otherwise I would have problems all the time, also when printing without fan.

ANY SUGGESTIONS ANYONE ?

And yes, I know that the temperature starts to swing a bit when the fan switches on, that is normal. But I have never seen these weird "jumps" before...

 

[gr5 2,069]

It's common for the PWM signal from the fan to interfere with the temperature probe.

One solution is to disable PWM by using the fan ONLY at 0% or 100%. This is probably the change you have seen recently - the more recent versions of Cura (last few months) turn on the fan more gradually. This is more in response to the UM2 design and not as critical for UM Original.

Another (better) solution is to separate the fan wiring from the temperature probe. You only have to worry about the 50mm or so between the heater block and the board on top of the print head. Usually a distance between the cabling of 20mm is enough. Twisting the fan wires will help.

Another solution (maybe) is to use a copy of Marlin that uses "slow fan PWM". You can do that by unchecking the 4th checkbox from the bottom here. This won't eliminate the problem but it will cause less frequent spikes in the voltage so that hopefully you get less frequent "bad temperature" readings.

http://marlinbuilder.robotfuzz.com/

 

[eriksw 0]

Thanks gr5,

you are right, no problems when starting fan at 100% at 2nd layer. Didn't know PWM was used for controlling the fan speed, it makes a lot of sense that this is the cause. And now I finally understand what that little circuit board is for...

I'll change the wiring - seems like a good idea to use another hole in the printhead for the fan wiring, and I'll make sure to twist those fan wires.

Still strange though, as it didn't have any problems with these intermediate fan speeds before, and I made no changes to it - the problem started "spontaneously".

Is it possible that the output of the probe has become weaker due to aging ?

Or is it just a heat-sensitive resistor ?

 

[gr5 2,069]

The PWM signal on the fan causes the wiring to act as radio antennas and the electromagnetic radiation can interfere with the probe wires which also act as a receiving antenna.

Again, Cura changed how the fan turns on in a recent version. Did you upgrade Cura just before problems started?

There are other common temperature issues. The wiring at the *top* of the print head is supposed to go through a black piece of plastic (delrin) for strain relief. That 20mm length of wire at the top of the head has a very high failure rate after months of printing. You can test by pushing the head to the 4 corners and see if temperature suddenly changes. Also try pushing gently in many different directions on the wire on the head.

The solution for that problem is to replace the wiring or use the "nozzle #2" wiring that is just sitting there doing nothing (replace at both ends of course).

 

[eriksw 0]

Yes I checked the wires already, no problems there.

I have used Cura 14.01 since it came out, without any of this trouble. So that's not it either.

Perhaps the positioning of the fan wiring has slightly changed while replacing the nozzle, but only SLIGHTLY if at all.

Still, I get the impression this probe has been getting less sensitive for a while now, as it seems I have to print hotter than a year ago to get the same results. So any info on probe lifetime expectancy ?

Thanks for your help, much appreciated.

 

[gr5 2,069]

The UM Original has a thermocouple. It's made by connecting 2 different pieces of metal together. I think typically they are twisted together wires.

I'm sure something can affect it. Maybe corrosion? Maybe metal plating? I don't' know.

But you would be surprised how the wiring seems fine and you can poke it and prod it but it still has a problem suddenly for no apparent reason yet replacing with the second set of wires fixes the problem.

 

[eriksw 0]

But if it was these wires, I would have the same problem when printing without using the fan, I thought... Disabling the fan solves the problem completely.

I have been printing stuff without fan for 2 weeks now, no problem with those. And now I know I can also use the fan at 100% without any problems, that seems to work (thanks for that one).

But as soon that PWM starts going, it starts acting up. So it must be that interference problem, I'm sure.

I'll rewire the fan wires tomorrow and I'll keep you posted on the result.

Thanks

Erik

 

[anon4321 16]

The thermocouple is two different metals pressed together held in be a ceramic insulator.

How do I know this? I subscribe to the philosophy "Don't turn it on, take it apart!" (please read that quote aloud in an Australian accent).

Here's a pic:

I doubt that anything has happened to the sensor.

The issue with TCs is the voltage differential they produce is very small and needs to be referenced to something called the cold junction temperature. This is done on the amp board and both aspects mean the process is very sensitive to EMI from the fan.

Actually, I doubt that it is the PWM as when the fan is on @ 100%, there should be no PWM.

Most likely, the noise comes from the fan's electronics. It's possible that even at 100%, there is still a minor (1/255) off time that results in PWM noise.

 

[gr5 2,069]

But as soon that PWM starts going, it starts acting up. So it must be that interference problem, I'm sure.

 

You are very likely correct.

 

[eriksw 0]

UPDATE

I have changed the path of the fan wires (through another hole in the printhead), twisted them, and made them shorter ( as short as possible).

I have also shortened the wires of the thermocouple.

So now the fan and probe leads are much further apart, minimal 20 mm as suggested.

Unfortunately: only a slight improvement. Still these fluctuations when using fan between 0 and 100 %, although a bit less violent. But still no fix...

I think there are 3 parts that might need replacing, either:

1. the cooling fan. Is it possible this has started to produce more interference due to aging ? It sounds ok to me...

2. the thermocouple. Can't find any info on the life span of these. Do they wear out ? It has been used for approx. 1500 print hours now...

4. the small amplifier circuit board. Perhaps it is not amplifiyng the signal (enough) anymore ? Or would I not be able to print fine with fan at 0 or 100% if this was the case ?

Your help is greatly appreciated !

 

[gr5 2,069]

You said it started when changing the nozzle? Maybe the wires were damaged slightly? If the resistance increased between the thermocouple and the small board then it would be more sensitive to noise.

My guess is the wire is almost broken. Probably right where it goes into the heater block.

You could measure the resistance through the thermocouple with a volt meter. I've never tried this but I think it should be close to 0 ohms. If it is over 1000 ohms then that might explain the problem - you could jiggle the wiring and push and prod the wiring until the resistance comes back down. That could both fix the problem or find the bad spot.

 

[eriksw 0]

Ok thanks, I'll check that. Might take until monday before I get a chance...

I'll keep you posted.

 

[eriksw 0]

UPDATE

Measured the resistance of the probe: 1,5 Ohm, so no problem there. 99% sure no wires got damaged while changing nozzle...

I also twisted the wires of the probe before reconnecting them. Alas, no fix.

So that leaves the fan and the amplifier circuit board, I think. I'm tempted to change the whole lot in one go, but then I would not find out what is causing this problem, so I'll be patient and work it out one by one.

Another clue that leaves me clueless: I just tried printing a vase with fan speed at 40 % to see what would happen. Temp reading started jumping about again, but I left it to see if it would settle. It didn't. When I noticed that the print had been failing for quite a few layers (got distracted, I have a shop to run...) I saw that the temperature reading was on average about 230, while print temp was set to 240....

I would assume that the jumping of the probe readings around the set temperature would cause the actual nozzle temp to be roughly the same, close to the set temperature. If the probe reading is too low (as was the case just now) I would expect the heater to be on all the time, causing actual nozzle temp to be way too high. But I'm quite sure the failure was caused by the nozzle temperature being TOO LOW, causing under-extrusion. WTF ??

 

[gr5 2,069]

So that leaves the fan and the amplifier circuit board, I think.

 

It could also be the darlington transistor that turns the fan on and off. It might be turning it off too fast/abruptly which causes higher frequency radio signals. I don't know - it seems more likely that an aging darlington would switch more slowly thus getting hotter but creating less high frequency radio.

 

[gr5 2,069]

Oh. Or the 5v signal that goes to the small circuit board might not be a steady 5v. You would need an oscilloscope to check that. It think you should just go for 100% fan. I always do 100% fan for PLA and the only reason I turn it on gradually is so the nozzle PID controller doesn't let the nozzle temp drift too much as the fan comes on.

 

[eriksw 0]

Helpful info, thanks again.

As you say I can use the printer with fan at 100, which I also do normally, but there's 3 problems with that:

1. The gradual start of the fan over the first few layers is very nice, for the reason you mentioned. But now this is only causing more trouble...

2. The printer is right next to a single pane window and a shop door that is letting in cold gusts of wind regularly, so I found that during the cold winter period it was better to use the fan at 40-50%, else it was cooling too much (even though I closed off the printer with pvc windows and a beautiful self-made harmonica roof).

3. It bugs me. I want to know the cause of the problem, for mine and hopefully other people's sake.

Where is that darlington transistor hiding ? Is it on that little circuit board ?

 

[Daid 304]

Just double checking, isn't the fan blowing on the head? This can also cause huge temperature swings. Even if the airflow is just going over the tip.

 

[gr5 2,069]

This can also cause huge temperature swings.

 

He said it was fine at 100% fan.

 

[eriksw 0]

Daid, thanks for joining in. For the record: I'm not a "newbie", but I forgot my forum login details so I made a new account. I have been printing practically 24/7 since last august. I run Printed In Space, the Amsterdam 3D print shop.

My printer was doing fine (at all fan speeds), now I have trouble with PWM interference, but haven't been able (yet) to find the source of the problem...

 

[eriksw 0]

GR5,

could you tell me where to find that darlington and how to check it ?

 

[gr5 2,069]

For um original it's Q4. Kind of in the middle of the schematic on the left. It looks like a transistor on the schematic.

It's labeled clearly on the circuit board and it's positioned near the fan connector. It is a 3 lead device.

Where to get circuit diagram:

ULTIMAKER1

The circuit diagram, and board layout are here:

http://reprap.org/wiki/Ultimaker%27s_v1.5.7_PCB

There is a zip file at the top. It contains the "brd" file which is the layout. Also the "sch" file which is the schematic.

Both files can be opened by eagle software which is free:

http://www.cadsoftusa.com/download-eagle/

ULTIMAKER 2 SCHEMATIC - click "raw":

https://github.com/Ultimaker/Ultimaker2/blob/master/1091_Main_board_v2.1.1_%28x1%29/Main%20Board%20V2.1.1.pdf

 

[gr5 2,069]

how to check it? Oh - you can't really - you just replace it - it's only a few cents. But I doubt it's the darlington. I'd check the 5v supply going to the print head first (with fan running at 50%) to see if the voltage looks really spikey and ugly on the oscilloscope (lots of noise).

Or replace all the relevant wiring with shielded wires. The shielding will block a large amount of the radio waves. This would be the 3 wires to the test head, the 2 wires to the probe and the 2 wires to the fan.

 

[eriksw 0]

Thanks for all that info.

I'm not in a rush, as I can use it with fan at 100. So I'll take the time to replace some parts until the problem is solved.

I don't have an oscilloscope so I'll need to find one I can borrow.

Replacing the wires with shielded ones seems like a solution that is not really dealing with the cause of the problem, as it used to print fine with the normal wires. As a test, I have wrapped aluminium foil around all the wires, but I did not notice any improvement.

If, as you said, chances are that an aged darlington will only emit LESS noise, I'll leave that one until I have no other options to try. I could switch the main board from my other printer to see if that solves it, I'll try that some time this week.

But I think my first bet would be to change the fan. It seems to me that this is the one element that is most likely to be worn out after such extensive use. I noticed today that it did not start to move when I set the speed to 51 %, it only started after I gave it a little push...

So perhaps it is the fan itself that is emitting more radio noise during PWM, due to aging ? Or is that not possible ?

 

[anon4321 16]

It's possible if as the fan gets older, it draws more current. More current will result in "stronger" noise.

 

[gr5 2,069]

As a test, I have wrapped aluminium foil around all

 

radio waves (electromagnetic waves) have both an electrical field component and a magentic field component. If you block either component you will reduce the waves somewhat.

A mu metal such as iron or non-stainless steel (one that magents stick to) will block most of the magnetic component but aluminum doesn't affect magnetic waves. Aluminum won't block electric fields very well either unless you ground it. So I wouldn't expect that to help without grounding the aluminum.

 

 

I noticed today that it did not start to move when I set the speed to 51 %

 

Interesting. Mine starts spinning closer to 30%. If you hold the fan still with your finger does the temperature behave itself any better? Or worse?