Z Wobble, or is it?

I just swapped out my leadscrews and nothing has changed so I am not certain this is a Z wobble issue. The printer is a TronXY X5S using a single motor and two leadscrews with 4 smooth rods (two on each side) for stability.

What could be the problem to cause that? It looks like Z wobble to me but the new leadscrews are dead straight and they have never been constrained. Oh, and this issue just popped up a few days ago.

[i.imgur.com]

---

_______
I await Skynet and my last vision will be of a RepRap self replicating the robots that is destroying the human race.

bang-bang control of bed temperature?

4 guide rails is overconstraint and may cause the bed to walk up and down. You shouldn't need more than two rails.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

I am not sure if I explained it right but over the last 2.5-3 years I have had no issues. I replaced the nut, and the rods and I have been using bang-bang on all of my printers for the HB and the nozzle's PID I retuned yesterday (after the issue showed up).

Here is what I own and the rods/rails I mentioned: [content.toolots.net]

edit: Shortly after I purchased it, years ago, I removed the two motors and did a 1 motor 16t to two 40t using this mod [www.thingiverse.com]

Edited 2 time(s). Last edit at 02/12/2020 07:09PM by Dark Alchemist.

---

_______
I await Skynet and my last vision will be of a RepRap self replicating the robots that is destroying the human race.

Looks to me like z-ribbing, since the whole layers are narrower and wider; rather than z-wobble, where each layer would be the same width but just displaced from one side to the other. This normally is caused by some inconsistency in Z height per layer, typically with periodicity as you are showing.

If the pitch of the bands you see matches lead screw pitch, it's likely related somehow to the lead screws that you just replaced. It looks like maybe 10 layers per cycle; if those are .2mm layers, that would match the common 2mm pitch. Maybe you didn't get the lead screws mounted perfectly square to the frame. The mounts in the mod you used seem a bit weak; I used pillow blocks mounted on solid aluminum risers, instead of 3d-printed mounts, when I did a similar mod.

Did you change pitch or lead when you bought your new lead screws? (Otherwise, why did you replace them?)

Does the banding layer pitch stay the same if you print a larger object, or an object with layers of different sizes (say, a cone or pyramid)? If so, I'd be shocked if it's related to bang-bang heating. It's unlikely to be a highly-periodic difference in filament thickness; I think bad filament is usually not regular in that way. If you have the same number of layers per vertical band when printing a cone or pyramid, that would definitely point to your lead screws.

You can get something like this effect from layer heights that result in non-integer step per layer thickness resulting in aliasing, but I don't think that's your problem. With your 16:40 (2:5) ratio, your presumably 200 steps per rev stepper motor, and your 2, 4, or 8 mm lead (travel per turn) lead screws, do you end up with an even number of steps per layer? Worst case there for .2 mm or .3 mm layers looks like an integer full step count at least every four layers if I'm doing the math right, which wouldn't match what looks like a 10-layer cycle.

(Incidentaly, note that specifying the layers as relatively simple fractions of mm is related to using metric hardware; if you were using imperial lead screws denominated in inches, you would use related fractional inches for layer thickness too. For example, if you had a 20TPI lead screw, you might use 1/80 inch layers, or .3175mm, to avoid z-banding from aliasing errors. Ugh.)

I am using 2mm pitch leadscrews (T8-2) but the distance between the bulges seems to mach the teeth of a GT2 belt as close as I could eyeball them. I have printed an object 6mm tall and 300mmx220mm and sure enough there it was as well. The layer height used was .24 and the very large object I printed used 0.2mm layer height (not shown)

They are straight as no binding and everything lines up but I do agree I wish I had the mounts in metal instead. I suspect this is something I will have to live with after all of my testing.

Did you change pitch or lead when you bought your new lead screws? (Otherwise, why did you replace them?) I always buy spares.

This is why I buy 1mm and 2mm lead leadscrews so I can use whatever layer height I wish to without fear. What should I print next to narrow this down?

Edited 2 time(s). Last edit at 02/12/2020 09:22PM by Dark Alchemist.

---

_______
I await Skynet and my last vision will be of a RepRap self replicating the robots that is destroying the human race.

If it matches the 2mm pitch of your lead screw, it's kind of guaranteed to match the 2mm pitch of your GT2 belt, and vice-versa.

You could design replacement mounts that are sturdier. Metal risers and pillow blocks were trivial though. I just drilled holes in aluminum stock. TDD has written many times about how to design plastic parts for high load; model a solid block and remove from the block only what you must remove for function. You could follow his advice there for the lead screw mounts (and, if you like, the motor mount). Or just buy some pillow blocks and mount them on something solid as I did. But the belts are pulling the Z screws both towards the center, which I am guessing is a likely cause for your problem, by presenting a slightly changing bearing surface with periodicity matching the pitch of the screws.

So, I'm not telling you what to do, but if I were in your shoes, the next thing I would print to narrow it down would be solid riser blocks to mount under pillow blocks for the Z screws so that they are more solid. No guarantees, just my best guess.

Not sure what you meant on that last part about riser as the leadscrews would eat them. I am currently printing a cone 30mm high and not is vase mode, of course, to see what it comes out like. I have not changed any perimeters so apples to apples. If I see the banding/ribbing I will stop it and try to grab a picture but neither of my phones do well up close (macro).

edit: Oh, it is over 3mm now and I don't see any issues.

6mm tall - [i.imgur.com]

Edited 3 time(s). Last edit at 02/12/2020 10:09PM by Dark Alchemist.

---

_______
I await Skynet and my last vision will be of a RepRap self replicating the robots that is destroying the human race.

It is almost 12mm now and no bulges BUT I can see those dark bands that you always see with ribbing (like in my first photo).
13mm and the bands are there but not at the same distances. [i.imgur.com] See those dark lines? Not sure how, if at all, the cone is affecting it but the bands are not uniform spacing now.

Edited 2 time(s). Last edit at 02/12/2020 10:07PM by Dark Alchemist.

---

_______
I await Skynet and my last vision will be of a RepRap self replicating the robots that is destroying the human race.

TDD == the_digital_dentist

The pillow blocks I'm talking about have set screws (grub screws) to hold the lead screw, and the lead screw would never touch the riser. I have no recollection where I bought them, but [www.amazon.com] is an example of the kind of pillow block I'm talking about. Wouldn't damage a riser block on which it was mounted in any way.

I put pillow blocks like that at the top of the lead screw as well; theoretically overconstrained, but the 500mm T8 would whip when the bed was near the bottom so in practice it helped. YMMV.

Quote
mcdanlj
TDD == the_digital_dentist

The pillow blocks I'm talking about have set screws (grub screws) to hold the lead screw, and the lead screw would never touch the riser. I have no recollection where I bought them, but [www.amazon.com] is an example of the kind of pillow block I'm talking about. Wouldn't damage a riser block on which it was mounted in any way.

I put pillow blocks like that at the top of the lead screw as well; theoretically overconstrained, but the 500mm T8 would whip when the bed was near the bottom so in practice it helped. YMMV.

I am using these [cdn3.volusion.com] Yeah, I realized what TDD was finally.

edit: Same thing and there are two setscrews in each. Now I removed mine at top as TronXY originally had as it made banding. Since I am using what you are I guess there is no simple way to fix this.

edit: This is my mounting - [i.imgur.com]

Edited 2 time(s). Last edit at 02/12/2020 10:31PM by Dark Alchemist.

---

_______
I await Skynet and my last vision will be of a RepRap self replicating the robots that is destroying the human race.

Looks like a shrinkage effect to me.
If it was it would be affected by temperature variation - you think nothing has changed there. Note that cooling fan influence may be more relevant to bang - bang so changing the cooling fan settings may have an influence.
Layer print time. It is on the small cube but may not be the same on larger objects. Hmmm - cooling fan on for small layers ?
Infill effects - if you had different infill layers.

Can't really see the sides of the cone so can't judge it. If it is a layer size effect then it would be more pronounced near the cone top.

Quote
MCcarman
Looks like a shrinkage effect to me.
If it was it would be affected by temperature variation - you think nothing has changed there. Note that cooling fan influence may be more relevant to bang - bang so changing the cooling fan settings may have an influence.
Layer print time. It is on the small cube but may not be the same on larger objects. Hmmm - cooling fan on for small layers ?
Infill effects - if you had different infill layers.

Can't really see the sides of the cone so can't judge it. If it is a layer size effect then it would be more pronounced near the cone top.

Nope, it is a repeating pattern typical of Z Banding and there is just nothing I can do, short of having a CNC machine at my disposal which I don't have, to fix this. Just something I will have to live with and sand off if needed.

---

_______
I await Skynet and my last vision will be of a RepRap self replicating the robots that is destroying the human race.

I suggest you try PID control of bed temperature.

There are two theories on the cause of the problem related to bang-bang temperature control. There is a deadband where the temperature gets to a specific value and then the heater switches off. Then the temperature drops to a specific value and the heat turns back on. Theory 1: the bed warps slightly up and down as the heat switches on and then off. Theory 2: the heater drops the supply voltage because the wiring is bad and or the power supply is underspec'd for the printer/heater. The varying supply voltage affects the nozzle temperature, etc.

You should be able to measure the voltage variation on the supply with the heater switched on and off, or just try PID control and see if the performance improves. It definitely won't hurt.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

TDD, I can't do PID to the bed (never could) as my external mosfet gets so hot it did this to me about a year and a half, or so, back and burned up the FET. [i.imgur.com]

---

_______
I await Skynet and my last vision will be of a RepRap self replicating the robots that is destroying the human race.

That shouldn't happen. You probably need to reduce the PWM frequency for PID operation.

Even though the FET has very low on-resistance, it probably gets hot because the signal that is switching the FET does not have sharp rise/fall transitions. That means the FET spends enough time in the transition between on and off to cause it to heat up, especially if you try to run it at a high frequency. With low frequency operation, the transitions happen less often so less power gets burned up in the FET. OTOH, its possible the FET isn't turning completely on or off, in which case it's going to burn up no matter what gate frequency you use.

The bed is massive and takes time for heat to propagate, so you can set a low PID frequency (try 10 or even 1 Hz) and it will still work fine. There's no need to run a high frequency gate signal to the FET.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

TDD, it was one like this [images-na.ssl-images-amazon.com] and I moved to one like this [images-na.ssl-images-amazon.com] and when I use PID it gets bloody hot as well. I figure it was the frequency that causes it since the new one (the second picture) doesn't even get warm with bang bang on doing 480w.

Do you happen to know what I change for the frequency for just the bed pwm frequency in Marlin 2.0? I only found for the fans when I looked and low frequency on the fans makes a bad buzzing sound. Oh, and I am now on a LPC1768 100mhz ARM and the burn was on a AT1268 16mhz.

Edited 2 time(s). Last edit at 02/13/2020 08:44PM by Dark Alchemist.

---

_______
I await Skynet and my last vision will be of a RepRap self replicating the robots that is destroying the human race.

Went back and reread your earlier posts. I didn't realis you were using bang - bang only on the bed.
I still think this is a temperature effect so it may be worth looking at the PID settings.
What temperature are you printing at?
I don't know the details on how the firmware runs the PID test but anything that interacts achieving temperature near the target can screw the PID calculation. For instance, if the heater performance is limited by the PSU then it is not controlled by the PID settings so the calculation is screwed. The same applies if the thermistor has an issue. What i don't know is if the temperature limit in firmware is applied during PID calculation. There may also be current limits from the electronics. Im no expert but i think this temperature oscillation is indicated by too high a P or I value.

Marlin 2 Configuration.h says:

* The PID frequency will be the same as the extruder PWM.
 * If PID_dT is the default, and correct for the hardware/configuration, that means 7.689Hz,

7.689 Hz is approximately 10 Hz for the purposes of this discussion. If an approximately 10Hz signal caused the MOSFET to burn up, something else is wrong. Using the board's integrated hotbed MOSFET as a driver for the external MOSFET should work just fine. It shouldn't matter which board you are using.

I ran a Tronxy X5S bed on an external MOSFET with 24V supply and PWM with no MOSFET-related problems, until I decided to use the frame as a skeleton on which to build a new printer because the X5S had so many faults it wasn't worth addressing them serially.

It's been years since I used Marlin, but IRIC there are variables somewhere that set the PWM frequencies for the bed and hot-end independently, and they default to hundreds or thousands of Hz. I was using an SSR to switch AC power into a 24V transformer that drove a 450W heater in MegaMax in its early days, and I turned the PWM frequency down to about 10 Hz. I even made an audio recording of the sound the transformer made when the PWM was switching. But, it has been years... I may not remember correctly and Marlin may have changed a lot.

Z wobble due to things like bent lead screws would cause the layers to shift -i.e. one side of the print would bulge outward and the opposite side would have a concavity. The photos look like the layers are bulging uniformly. That's what makes me think this is a thermal issue. If the bed warps upward when the heat is on, layers will get squashed thinner and create an outward bulge in the print. It could go the other way, to- the bed could warp downward and the layers would get thicker, and create a concavity instead of a bulge in the print surface. The mass of the bed and the time it takes to heat and cool would be responsible for the relatively slow changes in the bulging in time as measured by the Z height.

Edited 1 time(s). Last edit at 02/14/2020 12:43PM by the_digital_dentist.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

Quote

It's been years since I used Marlin, but IRIC there are variables somewhere that set the PWM frequencies for the bed and hot-end independently, and they default to hundreds or thousands of Hz.

At least in recent Marlin this isn't true - the default bed and hot end PWM frequency is around 7Hz, unless you change the setting of SOFT_PWM_SCALE.

Quote
MMcLure

Quote

It's been years since I used Marlin, but IRIC there are variables somewhere that set the PWM frequencies for the bed and hot-end independently, and they default to hundreds or thousands of Hz.

At least in recent Marlin this isn't true - the default bed and hot end PWM frequency is around 7Hz, unless you change the setting of SOFT_PWM_SCALE.

Yes, BUT is that 7hz for the LPC 1768 as well? Something tells me it isn't due to it running at 100mhz vs 16mhz.

---

_______
I await Skynet and my last vision will be of a RepRap self replicating the robots that is destroying the human race.

Quote
Dark Alchemist
Yes, BUT is that 7hz for the LPC 1768 as well? Something tells me it isn't due to it running at 100mhz vs 16mhz.

Timer frequencies are specified independent of CPU frequencies. I think that it happens that it's derived from the base CPU frequency on AVR, but if I'm reading the LPC1768 HAL headers right it's just an independent timer there. In both cases you should expect around 7Hz on your LPC1768.

You probably have a signal LED on your external MOSFET board, and 7Hz flickering should be obvious when the board reaches set temperature.

I understand your reluctance to trust the MOSFET with PID PWM control (once literally burned, twice literally shy?) but it really ought to work, has worked for others, and something is wrong if it overheated for you. You can avoid being burned again if you use a TCO attached to the heat sink on your external MOSFET board inline with the input positive supply so that if it gets too hot, the thermal fuse turns off all power through the MOSFET board. They can be a useful safety feature!

All I know is that I tried PID (using the AT1284p on the Melzi 2.0) on the red one and it got hot almost instantly but due to the beefier heatsink, and this time I was prepared, I was able to pull the plug before it burnt up. I switched back to bang bang and it never even gets warm which tells me it isn't an amperage dealio it is the switching frequency since bang bang is a steady state.

edit: Oh, the red led never flickers it is full on or full off and it doesn't flicker even on a smart phone camera.

Edited 2 time(s). Last edit at 02/15/2020 01:38AM by Dark Alchemist.

---

_______
I await Skynet and my last vision will be of a RepRap self replicating the robots that is destroying the human race.

Quote
Dark Alchemist
edit: Oh, the red led never flickers it is full on or full off and it doesn't flicker even on a smart phone camera.

That's what you'll see if the bed is set to bang-bang (i.e. PIDTEMPBED is not enabled). If you enable PIDTEMPBED the light will be solid until the temperature is within a few degrees of the set point and then start flickering.

Well, these external mosfets do not like 7hz but are fine with the steady state bang bang.

---

_______
I await Skynet and my last vision will be of a RepRap self replicating the robots that is destroying the human race.