Possible frozen hot end fix

Basic Information:

Printer Model:
MCU / Printerboard:
klippy.log

Fill out above information and in all cases attach your klippy.log file. Pasting your printer.cfg is not needed
When pasting code or config snippets use the code tags </> in the menu above

Describe your issue:

two trees sapphire 5
robin_nano-V3.1, stm407 mcu
steel nozzle, tight .4mm
Extruder way out of calibration at default 33.xxxx
I just started another relative small 2 hour print after setting the rotation_distance to 10.5 and then dialing octoprint down to 85% which is looking decent. This is not the main problem however, frozen hotend is turning rapidly into a pita.
And a klipper safety feature, the max_extrusion of 50mm is getting in the way of an idea.

I’d like to pull back 150mm at job end, and restore 151mm at job start, but klipper does not allow that.

The idea is to end a job with enough retraction to get the hot end of the filament pulled 30mm up out of the heat break into cooler bowden tube quickly, then an additional 18mm slowly to keep it from sticking to the innards of the tube. Serialization enforced with M400’s.

Then an indeterminate time later, a new job is started, it waits for temps, homes all, set a 20mm bed clearance is callled for followed by an M400, then a 49 mm extrusion is done, 1 more than retracted to reprime the nozzle to lay down the test line. This is also followed by an M400, then the move to the starting test line is done.

Does anyone think this might be helpful with a constantly frozen hotend?

Thanks.

Oh, come on Gene. Please not back to messing around. How to properly calibrate your extruder, we have discussed here: Extrusion diff between part printed withe Marlin, and one printed by klipper - #4 by Sineos

Not at all needed. Just do a small retraction at print’s end (2-5mm) and be happy.

What the hell is a frozen hotend?

one which is filled with solidified filament, high enough in the heat break to be above any point where a prolonged heating will not thaw it without unplugging the fan so the heat telegraphs better, and even that my not be enough, In which case then take it completly apart to remove the heat break, take it to the gas cook stove, grab it with some suture clamps and heat it till the PETG burns, then remove the ashes. Or, if you possess a small drill bit set in numbered sizes, mic them till you find the 1.84mm one, chuck it up, lower the bed, and gently drill out the cold plastic, with titanium heat breaks, relatively safe if you are a good judge of vertical.

If you’ve never had a frozen hot end in a bowden tube fed hot end, which to control stringing has to have a retract distance setting of 7 to 10mm, something you can avoid in the direct drive extruder because it only needs 1.5 to 2mm of retract. This is the main reason I wind up with direct drive eventually as it does away with the frozen hot ends because it doesn’t pull melted filament back far enough to freeze it up.

So my idea means put up with the stringing from inadequate pullback during travel moves by not pulling above 2mm, and pull it clear back into the tube and a bit more so it cools while in motion but does not freeze the hotend. Theoretically anyway. I’m set for 2.5mm atm, not enough to stop the stringing, but far enough to freeze it up.

If you don’t have a problem with freeze ups, what is your secret that prevents them?

Thanks Sineos

In 3D printing 1337-speak we call this a clogged hotend.

  1. First and foremost get your extruder rotation_disctance calibrated or any other measures will be completely futile
  2. Properly assemble your hotend.
    • The nozzle is never completely screwed into the heat-block (1/4 to 1/2 turn out) and is then tightened with screwing the heat-break against it
    • I recommend to apply some boron nitride thermal compound around the threads of the nozzle (only nozzle!)
    • Re-tighten to ~2N in hot condition
  3. Too hot printing / bad hotend cooling
    • This is called heat-creep and probably the biggest reason for such issues
    • The heat is creeping up and melting the filament in the heat-break or even up in the heat-sink area
    • No retraction or anything what you have proposed will save your back here
    • Better cooling, lower temperature
  4. Staring to print too close to the bed
    • Especially when printing “on the hot side” this can lead to back-pressing the filament upwards instead of out of the nozzle
  5. Too high retraction
    • NEVER retract out out of the hot-zone or even into the heat-break / heat-sink. Your filament will cool down and get stuck there and in subsequent heat-ups not get the needed heat to cleanly come back off again
    • Properly calibrate Pressure Advance as this well greatly help

And, at the danger of repeating myself, PROPERLY calibrate your $%&§ extruder rotation_distance!

I did that by a slightly different means, measuring the amount fed for a 50mm command, then edited the rotation difference by the ratio. at 12.500 is fed 32.xxx twice. so the rotation then should have been 7.5625.

I set it for that. restarted klipper and now measure within a mm of 50mm fed for a 50mm command.
That’s the whole point of this I assume.

This is NOT with the stock heat break, but with a POLISI3D brand #X0020PYKZ titanium heat break which stops the capricorn about 4mm into the top of it where it is cold enough the teflon survives the heat instead of slowly going away, emitting phosgene gas as it does, and of course leaking both down the nozzle threads and up the heat break threads, dripping off to hit the work piece and wrecking the build by jamming the head, causing a home loss…

That’s one of the MUST DO’s if using PETG, and probably ABS or any other such higher temp plastic.

For what I’m doing, PLA is like hercules glass, shatters into sugar when stressed, PETG isn’t any stronger, but much much more resilient. And it ages well, and should be usable 100 years after I’m gone.

Here is an early pix, the new screw in front is now about 100mm longer, of what I’m trying to make, but it took better than 2 weeks to print the plastic wrapped around that 50mm diameter hard maple screw.


The red is regular PETG, the black is PETG with 20% carbon fiber, the rings are re-enforcing the maple against handle torque. The blue bit is TPU, a bumper to keep from knocking off the the red knob on the ends of the handle.

That is not the final design but close. The final nut design is about 3 more threads engaged longer. The housing has grown another butterfly joint between those 2 to hold it together. I wrote the gcode that carved that screw by hand, its a buttress thread, 2 start so faster closing.

Thanks Sineos. Take care and stay well.

Change of subject so to speak, I’ve mentioned that I have a huge, multi-kilowatt autoformer that can make 63 volts AC to feed bed heaters a considerable amperage. All controlled by the on-off of the mcu cards bed voltage output of 24volts dc, control by two huge 60 amp SSR’s wired so one turns on the mains power to the transformer, and another also switched by that same 24 volts from the card, that sends the 63 volts output back to the printer that sent the 24 volts as a heat the bed signal. This all works quite well until I note the frequency of that signal, effectively acting as a PWM control at somewhere around 15 hz.

But at that high a rate of switching, its possible that an unwanted dc current could enter the scene, saturating the core of the transformer, causing it to loose its inductance, and consuming a breaker blowing amount of current. That is quite
undesirable. If the frequency of this PWM could be reduced, to say 1 hz, the dc currant would be considerably
reduced to where it would no longer be a concern. 50% heat would then be 1/2 second on, 1/2 second off, and by the time the beds thermal mass is used to smooth out the temp, regulation to .1C should still be obtained while this potential dc component drops to a negligible value.

As a side effect, and school time for you, on the ferrite core of the transformer, if that occurs long enough to take the core above its curie temperature, the magnetic properties of the core disappear forever, or t least until the core is returned to Arnold and re-annealed to restore its magnetic properties, so it is as far as fixing it is concerned, trashed forever, In my career as a broadcast engineer I’ve see it happen twice both times in HP made gear, their engineers apparently never got the memo and it nearly bankrupted them. And some of the ferrites have a curie point well below the boiling point of water, so its effectively destroyed w/o even discoloring the paint on it. The only clue the tech has is that it blows $50 worth of transistors instantly, and then the fuse that is supposed to protect the transistors. Measuring the inductance of the coils and finding you have 50 feet of wire, and a microhenry of inductance is the only clue. The ohmage of that 50 feet of wire is normal, no shorts, the wire didn’t have time to heat before the transistors blew again, followed by another 50 cent fuse.

So, is there a means to slow the beds PID to 5 or 10% of its present speed? A clock divisor that could be adjusted in the software somehow?

This is likely a question only the board designer could answer, but I obviously don’t know who else to ask.

Thank you, Sineos, take care and stay well.

Cheers, Gene.

How is it “possible that an unwanted dc current could enter the scene”? SSRs, regardless if they use TRIACs or Thyristors, will ONLY PASS AC. I don’t know where you get the idea that DC current can enter into the circuit at levels that will affect the transformer.

The default PWM frequency of Klipper is 10Hz and will work fine with an SSR although the control will be more granular than you would get with a MOSFET controlling DC.

A 300mm square 3D heated 3D printer bed consumes between 250W to 300W when heating. 64VAC at 60A is 2.7kW RMS - that’s roughly ten times as much power. What kind of bed are you using that requires that much power?

You’ll definitely see your electrical bill jump each time you run the printer.

[mykepredko] mykepredko https://klipper.discourse.group/u/mykepredko
January 27

How is it “possible that an unwanted dc current could enter the scene”?
SSRs, regardless if they use TRIACs or Thyristors, will ONLY PASS AC.
I don’t know where you get the idea that DC current can enter into the
circuit at levels that will affect the transformer.

Inadvertant timing sync with the AC, always triggering on one or the
other half cycle. I’ve had it happen in self clocked switching supplys
where the 2 transistors weren’t well matched, leading to a 45/55% duty
cycle. That difference is dc current to the transformer.

The default PWM frequency of Klipper is 10Hz and will work fine with an
SSR although the control will be more granular than you would get with a
MOSFET controlling DC.

So klipper does set that. So where do I slow it down to 1 or 2 hz?

A 300mm square 3D heated 3D printer bed consumes between 250W to 300W
when heating. 64VAC at 60A is 2.7kW RMS - that’s roughly ten times as
much power. What kind of bed are you using that requires that much power?

The bed on this SP-5 is around 12 to 13 minutes from ambient to 80C for
PETG. The hot end is another short 3, and I kill another minute waiting
for the edge of the bed to get up to temp, so just heading for home to
start a print is around 16 minutes.

You’ll definitely see your electrical bill jump each time you run the
printer.

Electrical goes up with the square of the voltage ratio, so 200 watts at
24 volts=1575 watts at 63 volts. I stage the warmup, bed first, else it
trips off a so called 900 WA ups. This scheme will bypass the ups for
bed power. The ups is big enough to run the printer while the 20kw
kohler gets started, a little over 9 seconds to full power again.

Thank you Myke, take care & stay well.

Cheers, Gene Heskett.

#pwm_cycle_time: 0.100
#   Time in seconds for each software PWM cycle of the heater. It is
#   not recommended to set this unless there is an electrical
#   requirement to switch the heater faster than 10 times a second.
#   The default is 0.100 seconds.

FWIW, on my printers I use 6 to 8 mm MIC Aluminium plates as base for the heated bed. Heating is done with a silicone heating mat. The mat is powered by 220V with a heating power of around 0.9W per square centimeter. It is controlled via SSR.
It takes about 3 minutes to reach 80°C on the heat mat. On the actual printing surface there is a considerable delta of ~10°C

keywords are “when heating”. Once heated, the duty cycle of the PWM will drop to 10%, maybe 15%, because it only takes the design “average” wattage to maintain that chosen temp. IOW, if it needs 100 watts to maintain the temp once at temp, the applied voltage and current will be regulated by the PID to 100 watts Average. That short time however will exacerbate the potential dc thru the coils if it always fires on the leading edge of the + half cycle, and turns off at the end of the + half cycle. That extra + half cycle 10 times a second would be the dc component.

So the transformer sees a dc component if its on for 17 half cycles and off for 103 half cycles. Even numbers match, and there is no DC, but odd numbers are worrysome. I could design a clocked at 60hz shift register that would fix that. Clock the shift at the positive zero crossing. The PID might hunt a bit more but I doubt you would see the wiggle on octoprints temp screen.

Take care and stay well Myke.

Cheers, Gene.

And its individual, something NOT mentioned in Config_Reference.md, I find to slow the bed heat cycle, its has to be in the {bed} list. So I’ve slowed that to the max allowed of 0.300 seconds. Works fine.

Now, whats a good starter value for pressure_advance: ?

The max value times .5 might be a good starter, but just like the pwm_cycle_time: max of .300, you have to exceed it and read the log to find what that limit might be.

I note that the starter line’s end has a small blob, which if that is set correctly, should disappear, is that a good indicator?

Thank you guys, take care and stay well.
Cheer, Gene

Oh, Gene. You are kind of my personal challenge. Must be some sort of Karma trial…

To quote from the above mentioned document:

The heater_bed section describes a heated bed. It uses the same heater settings described in the “extruder” section.

For all what is sacred to you: Please, please, PLEASE stop to mess and guess around. Read Pressure advance - Klipper documentation and follow it

The html docs you linked to are buckets clearer than Config_Reference.md, thank you,
I’ll do that as I’m currently waiting on parts for the big ender 5 plus.

Take care and stay well.
Cheers, Gene.

I’ve just run up one of my home designed printers (220mm square standard 3mm plate with a stainless steel/PEI surface - 24VDC from an LRS-350-24 with Octopus control) and my Voron 2.4 (300mm square - 110VAC with an SSR). I’ve just run them both up to 80C from 23C with the following results:

  • Home grown: Time to 80C, 215 seconds (3.5 minutes). Measured temperature on surface one minute after reaching 80C: 78C. Heater Resistance 1.4 Ohms. PWM to maintain 80C after one minute - 46% meaning that power to maintain is 189Watts.
  • Voron 2.4: Time to 80C, 110 seconds (just under 2 minutes). Measured temperature on surface one minute after reaching 80C: 58C. As @sineos reports, it takes a while to get up to temperature on the 8mm aluminum bed - at 5 minutes it’s at 66C, at 10 minutes its 74C. Heating pad resistance 15 Ohms. PWM to maintain 80C after 10 minutes - 55% meaning that power to maintain is 426Watts RMS.

The Voron 2.4 actual heating time caused me a bit of grief when I was bringing up the printer because my first attempts at ABS were miserable with no adhesion to the build surface until I got the idea to actually measure the bed surface temperature and then wait for the temperature to come up and stabilize. That aluminum plate has a lot of thermal mass which has its own positives and negatives.

If you have a large aluminum bed, then your heating times make sense - if you are running a 3mm plate with an PCB heater and it takes 15+ minutes, then there is something that needs to be investigated. On my home grown printer, if I start the hot end heating before starting the print, I can go from cold to starting the print in just over four minutes - if it’s taking you 15+ on a similar configuration then there’s some other issue that you need to address before going for the industrial option.

I have no idea why you are using two SSRs when really only one is required. I would have thought that having one side of the transformer to mains power and the other to a single SSR that passes power to the heating pad - basically how Voron (and others) do it.

What is a “self clocked switching supplys” as to relates to this discussion? I’m not aware of any switching power supply topologies that use AC. There are switching supplies that take AC input but they rectify the current to DC and may invert it back to AC, but the actual switching supply is working with DC currents. Your experience with mismatched transistors in a switch mode power supply have absolutely no bearing on what is being discussed here (SSRs don’t use transistors for switching - they generally use TRIACS unless you’re in the 4KVA range and above at which point they go with thyristors in a back to back configuration).

I’ve repeatedly asked you to put down a concise description of your system so I can get an idea what’s happening here because what you’re seeing honestly doesn’t make sense and you’re jumping to a conclusion without demonstrating the analysis that shows the extreme “solution” you repeatedly turn to is appropriate.

1 Like

Damn I’d love to be able to quote the most recent msg.

re your question about multiple SSR’s:
There will be, when I’m done, at least 3 printers maybe 4 if and when I get this voron built, tied to this kit, the 2nd SSR in each case is used to send the 63 volts back to the printer that requested it. I’m well aware I may have to goto the service and bring back an 8 gauge supply line for this tranny. Its a toroid, weighs around 50 lbs. I’ve put it in a box with castors and its originally wired for two printers and 4 SSR’s rated at 360 volts and 60 amps each. These SSR’s are much like the one switching the bed in your voron as I have the short, trident version of onee of those half built from a box that arrived badly damaged and missing wholesale qty’s of plastic parts. Now waiting, ack messages from the auto-responder, for the Chinese new years holiday before anything to replace the missing stuff is picked off the shelves. Scheduled for the 29th.

I haven’t yet hooked it up to any printer because getting maximum voltage ratings out of the makers is hell when the folks on the other end of the support email addy do not speak English as a first language and pretend to not understand the problem I’m trying to describe to them.

I consider 15 to 18 minutes startup time before it lays down a priming line as quite excessive, and yes, 24 volts is applied to the build plate heater as it warms, continuously until its target temp is reached… On the SP-5, its about 4 or 5mm of alu topped with a 3 or 4mm glass plate whose top coat is already missing in the middle of the bed.

That will at some point be replaced with a magnetic sheet and a smooth pei steel sheet, when and if I can find one that size, 300x300. Textured doesn’t fly, need smooth.

At that point I can also fit an inductive bed sensor which in my experience is far more accurate than a bltouch and hundreds of times less damageable by being bumped. Not to mention less than a tenner to install. Beats a micro-switch like a white mouthed mule.

Your pressure_advance calibrator is running now. Correction, finished, and absolutely no difference, bottom to top.
dbl chked command line as its way too long for octoprints input display, word “PARAMETER=” missing ack the log. fixed, new log made, running again. That is an octoprint bug, selecting “terminal” SHOULD focus on command line but doesn’t, and you have to click on it 3 to 5 times to get its attention so you can type something. PITA. Not your problem of course.

I’ve now ran it twice, no effect on the corners can be seen. Corner expansion in both directions remains at 60.39mm, about the size of the corner bulge. Bottom or top within .03 of identical. New, total restart of klipper, on a new log, w/o a bpi5 reboot, verified both preparatory command lines, invoked them again and started a 3rd print. Is there some other pre-condition I’m not setting? The test line I was going to use is #commented out of printer.cfg.

Where is the klippy.log?

klippy.log (2.5 MB)

Its longer, I took a nap while it was running the third time.

I see by the log you now have that it was at 1.0xxx at the end, but the bowden tube is over 2 feet of capricorn, I purposely made it a bit longer, because its not removable from those shark style fittings, so it has to be cut, the fitting unscrewed and the stub that was cut off, pushed on thru the fitting, on both ends of it.

If I imagine real hard, there might be some effect I can see on this 3rd run, but it is a long ways from enough at 1.00xxx.

So I took the 0 out from in front of the .02 for the increment, multiplying its effect by 10, and its running again. I just took a look, stopped it, pizza slices missing out of the corners by 6mm up. Next run, I’ll start at 1.00 and increment by .03.

5th run started.

I do see that one direction is very smooth, while 90 degrees away, there is a low level ripple the full width of the move.
Likely because the right side fan has a bent fin. Ripple might be .005mm, certainly not a problem for the parts it’s making for the ender 5 plus. Very slow fans on PETG, else it warps.

That will be a while being completed, the size of idler I need for the X belt turnaround is odd to maintain the geometry and ETA time is Mar 1st. Grrrrrrrrr. I might make a plastic idler for temp progress.

So I may work on the voron trident if I get the missing parts in time or the tronxy-400-pro (its anything but pro) in the meantime. Gotta have something to keep me out the bars. ;o)> That tronxy is the worst case of BBLB I’ve seen in years.

Take care & stay well Sineos.

Cheers, Gene.

So I started the 7nth run at . with a .02 increment. It got better right a the top with a PA of 1.08. The 6th run puked, wore out the filament, cut it in 2 and spit out 30" of it on the floor. I vacuumed the trash out of the extruder w/o cleaning the gears. At 1.08, 3 corners still stick out a bit and one is rounded inward. And the ossets in the inner wall look a lot better.

So I’ll run with this unless the extruder turns argmentative. At that point I put an orbiter on top of the hotend.

This is an improvement, thank you Sineos.

You can edit your previous reply to add a quote by highlighting text and selecting the quote reply button that appears.

Where?

Cheers, Gene