Have I got a problem?

[philipy]

I'm not sure if I have a problem or not, so any thoughts are welcome.
This is an Octoprint screenshot of the temps of hotend and bed on my Creality CR20pro when printing a temp tower.
Both lines show constant very small flickers which I don't remember seeing before but then again I haven't looked for it!

The actual temps are flicking between about +/- 0.5 deg of the set temp, which is probably what those little flicks are recording. So, have I got a problem developing and if so any idea what it might be?
I should add that actual prints seem to be OK.

Screenshot 2024-07-12 09.45.23.png (573.32 KiB) Viewed 10654 times

[ge_rik]

I don't know, but I suspect the blips are actually to do with a conversion from analogue readings from the thermistor to the digital values handled by the software. I imagine that in reality there is a steady and uniform drop in temperature from start to finish.

Rik

[philipy]

Rik, Sorry, I may have inadvertantly misled you. The steps in the graph are simply the 5 deg temperature drops at each stage of the temp tower print. What I was referring to is the slightly jagged 'straight line' if you look closely, as opposed to a flat straight line.

[Paul_in_Ricky]

I wouldn't worry about such small variations. There's bound to an amount measurement error in the hardware, plus as Rik says, there'll be software conversion and display errors too. At 200+ half a degree is only a 0.25% error.

[philipy]

Thanks Paul.
I'm not really worried, I just happened to read something recently and only when I looked closely did I see the minor flicker, which had me wondering.
It prints fine, which is all that really matters.. until it DOES go up the spout!

[GTB]

The actual temps are flicking between about +/- 0.5 deg of the set temp, which is probably what those little flicks are recording. So, have I got a problem developing and if so any idea what it might be?

The last figure in any measurement is uncertain. My opinion is the machine designer was kidding himself if he thought the hardware could read a temperature to the nearest 0.1degC. Most digital readouts of one sort or another flicker in the last digit when a reading is somewhere near the middle of that digits range. Which is why I prefer analog meters.

If your maths is still up to it, this teaching note on uncertainty of measurement covers the subject.

https://www.deanza.edu/faculty/lunaedua ... ties2A.pdf

The target analogy at the end is commonly used to explain the difference between accuracy and precision, but a statistician colleague where I worked had a simpler way of putting it. "Precision is something we measure. Accuracy is what God knows....."

Have fun,
Graeme

ps. I looked up the specs for a typical thermistor. Accuracy was quoted as 1%. That's just the sensor, there will be tolerances on the electronics as well. Correct practice is to add all measurement uncertainties. If I had to quote that bed temperature in a report, what you'd get would be a result of 60 +/- 2 degC (ie. it's somewhere between 58 and 62).

[GAP]

I have 2 printers running right now using Octoprint and I can see similar blips on the temperature lines, the bed temp more so than the nozzle temp.
I didn't even notice them till I looked real close after reading this thread.
Short answer is no I do not think you have a problem, if you do I have the same one.

[philipy]

Cheers for the replies guys.

As I said, I didn't really think I had a problem because my prints are fine, but something I read did make me wonder.

[Paul_in_Ricky]

ps. I looked up the specs for a typical thermistor. Accuracy was quoted as 1%. That's just the sensor, there will be tolerances on the electronics as well. Correct practice is to add all measurement uncertainties. If I had to quote that bed temperature in a report, what you'd get would be a result of 60 +/- 2 degC (ie. it's somewhere between 58 and 62).

I think this is why so many people have problems with FDM printing. One might hope that our machines are reporting true figures, but manufacturing tolerances mean that can't happen. So we have to test and adjust as we see end results.

It would help if manufacturers gave correct spec for their products. I've just been trying out a new filament from Sunlu; The web page for it lists two different temperature ranges on the same page and if you download their technical data sheet for it, it's different again. Although that spec is only really applicable for one nozzle size anyway.

Exasperating stuff.