Negative spikes, not sure what is going on

Hi all

I got a strange one for you all. I'm using this board: [www.tindie.com]
mega board is from: [store.arduino.cc]

I have tried this to fix the issue:

• PID tune
• New wires for thermistor all the way from Ramps to Thermistor
• New thermistor cartridge from [e3d-online.com]
• New heater cartridge also from [e3d-online.com]
• did i mention that i did a PID tune?
• rechecked that all wires in screw therminal are secured in place and no wires are sticking out.
• rechecked that all screws in screw terminals are tight
• tried to switch arround the thermistor to plug with no noise on

a temp graph from octoprint is attached

since the noise seems to stay with the same thermistor could it be that? i mean i just got a brand new one i would not expect it.

could it be the capacitors? a bad SMD solder job? could it even be the FET that makes the noise ?

what i will try tomorrow is temporarily using a no name 100K thermistor and connect it to same wires so the only difference is the thermistor. Only need to crimp on a plug

Edited 3 time(s). Last edit at 06/26/2018 07:07PM by boelle.

How to explain this phenomenon. Do not know it it is even possible.

You will note you have Red, Blue and Orange signals.
The Red (T0) has three downward spikes
The Blue (Bed) has two smaller downward spikes
The Orange (T1) has 2 medium sized Upward spikes
Spikes occur at -26.75, -24, -20.75, -18.50, -14, -10.25, -7.75 (approximate time intervals) this has a time of approximately every 2 1/2 minutes.
This is the one that makes me think it is external, T1 it is turned off and still spiking from 34.6C to 54C. This almost has to be noise.

T1 is just a chamber temp sensor so i dont care about that one

but yes noise is the only one i can think of as blue (bed is always the most noise free not matter what thermistor input i connect to

what i cant figure is where the noise should come from and how i could try and get rid of it

IDK if Marlin is polling the AD converters or uses interrupts? If it's polling, then maybe the AD-conversion is still ongoing while Marlin polls the value?
Interrupt handling routines are also tricky, especially with a constant data stream from Arduino to OctoPi.

Quote
o_lampe
IDK if Marlin is polling the AD converters or uses interrupts? If it's polling, then maybe the AD-conversion is still ongoing while Marlin polls the value?
Interrupt handling routines are also tricky, especially with a constant data stream from Arduino to OctoPi.

I dont know, that is why i thought of using a no name 100k thermistor to test again, even thou i dont expect any difference since


rememeber that BLUE (bed) is almost perfect except for a few small spikes, while the other 2 are close to useless


prints come out fine thou, so its more where noise could come from and try to rule that out first

i do have an LCD connected and the temp on that does not jump as much. But again i want to start with ruling out noise


EDIT: when not printing there are no spikes

Edited 1 time(s). Last edit at 06/27/2018 03:28AM by boelle.

in octoprint it was set to get temp every 5 sec when not printing and every 2 sec when printing

since octoprint does not control temp that did not make sense to me, so i have now changed it to every 5 sec while printing

and autoreport (if marlin even supports that) i have now set to 0 to force it to use the the pooling intervals

hmm

i'm just testing things out

heating up the bed did not get my any noise, BUT when it came to temp and started to flash the LED on the heat bed i got small amounts of noise on the Orange and Red lines

since the bed does not have big temp swings i did not see the noise since octoprint has the a light blue target line on top of it masking the noise


so there are noise on all 3 lines and i know that changing the inputs arround does not make a change

could it be the capacitors on the thermistor inputs?

hmm

maybe switching noise from the FET's ?

while i'm adjusting 1 layer height i can as well change to bang-bang and not pid just to see if that makes a difference

Edited 1 time(s). Last edit at 06/27/2018 04:54AM by boelle.

progress..... i think

i changed bed to bang-bang and much of the noise is gone


still need to get 1st layer right, with a 0.25 nozzle its a pain as you need to adjust very little to go from ok to not ok

I'm sure the LCD values are filtered or averaged. Otherwise we would only see dancing pixels. Might be, that the data sent to octoprint are raw sensor values without bandwidth filtering or averaging.
The capacitors on Ramps might be too small or mislabelled crap. We've seen that in the past too often..

Quote
o_lampe
I'm sure the LCD values are filtered or averaged. Otherwise we would only see dancing pixels. Might be, that the data sent to octoprint are raw sensor values without bandwidth filtering or averaging.
The capacitors on Ramps might be too small or mislabelled crap. We've seen that in the past too often..

about [www.tindie.com] being crap i dont think so, it's not made in china and cost quite a bit more

just got first layer perfect in the center of the bed and the short test print did reveal that using bang-bang for the bed made it much better


i tried bang-bang for the nozzle but i have to experiment with max current on that one as it overshoots just enough that it can stabilize enough for the print to start

but will report back what i figure, but all good suggestions are welcome, but i think we should concentrate on switching noise from the fets for now

Let's look at this from a different point of view.

I know That when I received my 12 volt power it was adjusted to above 12volts to 13.8 volts, I was getting strange effects. (My 24 volt 3d Printer supply was also set high.)
Re-adjusted the Supplies to the correct voltage 12V for the 12 volt supply and 24 v for the 24 volt supply, and the effects were reduced considerably.
("However I also had Controller replaced under warranty, because out of 4 controllers it was acting up causing the others to act up. Does crazy stuff at any given moment")

Anyway I digress
It could be that if you have your supply voltage set to high, that when the heaters first switches on there is a spike because the supply can not keep up.

All the above information is from my two 3D Printer and may not be what is causing your delima. Though wanted to put the information out for your consideration.

Its an ATX supply that i got spec. for the reason that it can supply 48A on the 12V rail (single rail)

i just tried to do bang-bang on the nozzle and that removed even more noise


but it was valid info non the less, you did not know what PSU i had

i have narrowed this one down further

tried new thermistor with complete new wiring to RAmps
tried another PSU
Tried another MEga board


logical conclusion and the only bit left not changed out is the Ramps board

it will have to make do as i just got 2 genuine Arduino Due and will build a Ramps FD v2.3


remember that Ramps FD is only unsafe if you got the first version that Geeetech spammed the market with even thou they where told it was not safe. V2.3 is safe,

There's probably a bad contact between Ramps and Atmega? Or between thermistor and Ramps. Cheapskating has a price...
If you can locate the pins, you can slighty bend them to increase contact pressure.
I used to use graphite-powder/grease mix to prime my high current connectors ( upto 1000A ). Not sure, if that's suitable for those pins?

its not cheapskating on my part as i picked an european and more pricy build board

[www.tindie.com]

but yes, bad contacts can make for noise too soo will look at it after breakfast

need to do some prints today anyway so i get time to test if its between thermistor and ramps or if its between ramps and mega

btw.... i plan to build a Ramps FD v.2.3, i was lucky enough that i got 2 arduino due from arduino.cc so they are not cheap knockoff's

i have allready compiled a BOM for it and wonder if i have picked good enough parts

so a small side track: if anyone have the knowledge please have a look at my BOM here [docs.google.com]

and best is to report back in this post: [reprap.org]

just to make a fresh list here is what i have tried this to fix the issue:

PID tune
New wires for thermistor all the way from Ramps to Thermistor
New thermistor cartridge from [e3d-online.com]
New heater cartridge also from [e3d-online.com]
did i mention that i did a PID tune?
rechecked that all wires in screw therminal are secured in place and no wires are sticking out.
rechecked that all screws in screw terminals are tight
tried to switch arround the thermistor to plug with no noise on
tried new mega board
tried another PSU
tried with a noname 100K thermistor directly to ramps board, wires are just long enough
checked that thermistor pin headers make good contact by bending them just slightly apart

its still the same so my next thing to look at are pins between the mega and ramps

if that fails the conclusion but be that the FET for the heater is bad somehow and makes noise since the noise is still on the nozzle heater even thou i switched arround the thermistors (and changed it in firmware too)


i use latest bugfix from marlin (2.0.x)


new graph attached

hmmm.... could it be the poly fuse that is going bad?


EDIT: since my prints are in PLA i did a print without the bed heating, its not as bad but still "too much"

its an 3 hour print to will snap another graph at the end of print where heat has throughly evened out


and yes.... i have a 80mm fan blowing on the Ramps board


EDIT2: 40 mins left and not looking much better, new graph attached

Edited 2 time(s). Last edit at 07/22/2018 10:06AM by boelle.

3 hour print done, most strange thing is that when it did the last few top solid layers the graph became much more acceptable but not perfect

after dinner i will check if the connectors for the X and Y steppers are good

graph attached

This may be a Marlin bug: https://github.com/MarlinFirmware/Marlin/issues/11323

yeah, it was about the last thing left

will put wires and pins file back to what it should be and then have a look at that link

so far thanks very much for that link, i took 2 files from that PR mentioned and now everything is back to what it should be

will let test print run for 30 mins just to be sure

much improvement for the nozzle but now i get positive spikes on unheated bed

will still call this a success

FYI, the newly released 1.1.9 version of Marlin has that fix included.

yep and so does bugfix 2.0.x as i have been using for a while

:-D

actually i did some testing of the code for this part here: [github.com]