Taco bed - BTT Eddy - Belt Tension and all the things

Sovol SV08
,
1

Guys,

Let me begin by saying I’ve never been within 10 Feet of a SV08. I am, however, a retired (disabled) mechanical engineer. Having lots of time I read all the forums on here and a couple other places. I see lots of frustration with bed mesh issues causing first layer issues.

Many are looking to the BTT Eddy as a solution. Unfortunately the posts from people who have gotten their machines to run with the Eddy report that issues persist. It is my observation that the SV08 bed sensor is MUCH more accurate than people give it credit for BUT when the bed changes shape after the mesh is generated no amount of precision in the mesh is able to predict what the bed shape will be after it shifts.

This leads to the question “why is the SV08 bed so squirmy”. Asking myself that lead me to go to Github and download the step files. I then examined how the machine is put together. What I found surprised me in 2 ways. 1) How could Sovol bolt together 2 semi rigid aluminum structures with16.5 mm of injection molded plastic in the middle and expect it to NOT be squirmy? 2) How is it thar the community hasn’t come up with a good solution.

I have a proposed mod that:

  1. Eliminates 3 of the bed hold down bolts to prevent pre-stressing the system by forcing 2 almost flat objects to be parallel .
  2. Eliminates the plastic base as a load bearing spacer in the system.

Mods are low cost (less than $20 tools and hardware), fairly labor intense (need to separate the plastic base from the bottom frame) and non reversable (have to cut into the plastic base).

I don’t have a unit to test on so I’m looking for an Alpha tester (or 2).

3 Likes
2

Just an idea, not tried, please use your mechanical knowledge to give this idea your analysis:-

1: remove the magnetic sticker
2: fit a piece of custom cut to size silica glass about £70.00 on eBay. Stick it on with heat stink paste, this will transfare heat but not any distortions from below.
3: stick a new magnetic sheet to the glass add the build surface as now, inductice probe sensing.

silica glass is said to be very flat and should put up the heat changes

I

3

Maybe you can tell us little bit more about the solution that you suggest? I will be happy to test something but i will not make an irreversible change without design the rollback possibilities (i have only one printer so in case, i prefer to print some part to rollback it before the changes). If that is not possible, multiple people here received a spare bed, maybe that can use one of them to try.

BTW, ok that is a trouble but i am not sure if that is the key point. The design of the bed is bad, for sure… but for the price + for the size, that is difficult to have a perfect stability. Any fix that i found yet, works “about well” but the only one who look like “perfect” is to replace the bed by a vorron recommanded one for around 400€… why not, but in this case, we can rebuild the full vorron systems, that is a verry expensive part in the printer.

PS: I am not mechanical engineer (and i am probably bad to that) but i saw some people who said in the reverse that is missing a screw at the middle of the plate (on the plastic part, behind the metal plate). Maybe that is possible to do that without irreversible changes but that is exactly the reverse of that do you suggesting. If i well understood the vorron advices, that is also something more like of that you are suggest (only one screw strongly fixed, all other are just here to avoid a rotation).

4

Just let me know if you think it may fix up an down hills :grinning_face:

Yahoo Mail: Search, organise, conquer

5

I’m an electrical engineer but I sometimes impersonate a mechanical engineer. :smiley:

I’ve engaged in some musings over ways to inexpensively fix the SV08 taco bed warping problem. In the early days of consumer 3D printing, I did a lot of printing on borosilicate glass which is very flat and doesn’t warp when heated. My SV01 has a textured glass plate. It’s manually leveled but it doesn’t warp.

I’d definitely be interested in a glass or ceramic build plate for the SV08, although I’d prefer the print surface to have a magnetic sheet and PEI coated steel build plate. I thought of crazy schemes such as inductively heating the steel sheet or using infrared through the glass to heat the bottom of the magnetic sheet but, at least for a retrofit, the best option seems to be a simple adhesive resistive heater on the bottom of the glass plate and wait while the print surface reaches the commanded temperature with the low thermal conductivity of glass.

If someone sold a drop in replacement glass build plate for the SV08 for $150 or less, I think it would be very popular. I’d buy it! Regardless of the speed and accuracy of the bed probe, the SV08 still needs a 350x350 mm build plate that is flat within at least 0.2 mm across the build surface from 25 C to 100 C, and the original design is far from that.

6

First thing - 3 points define a plane. If you THINK 4 points are all in the same plane you are not measuring accurately.

Keeping the above in mind the goal is to provide the bed module with exactly 3 points of contact so that no EXTERNAL thermal effects can warp the module.
(a pre mod module stability test just occurred to me, see next post)

Tool list:

  1. Standard printer tool kit
  2. 5/16 or 8mm drill bit
  3. Largish 90° countersink Like these

Supply list:

  1. Standoffs, 16mm (3 each) - These or similar
  2. Hard steel washers, 12mm OD (3 each) - These or similar
  3. JB weld (not the quick) or similar FILLED epoxy (unfilled transparent epoxies shrink a lot)
  4. Flathead M4x16 machine screws (optional) These or similar

Dismantle printer and glue a washer to the frame at 3 points where the bed hold down bolts attach, Clamp down with a standoff. Glue on the threads is not a bad thing.

image
image801×667 101 KB

Place bed module on standoffs and secure with screws from standoff kit. Press, pull and tug to determine if this configuration provides suitable structure. Rap sideways on the bed module and note if it produces a ringing note (bad) or a dull thunk (good).

OPTIONAL (collect baseline data to compare with final measurements)
Plug everything together with all the compliments on a non-conductive surface. Note the points where exposed mains voltage could be an issue. Tape up where you feel necessary. Positioning the touch pad may require some ingenuity, For reference the model shows the touch surface at 1.3mm above the plane of the build surface. Perform mesh measurements. Be sure to rename each mesh and save for later analysis. I’d do 3 temperatures, 40, 60 80. For each temperature I’d do 3 as rapidly as possible then at 10 minutes and 30 minutes. Utterly boring… You too can have an exciting and rewarding career as a test engineer.

Rubicon (point of no return)

  1. Use a 16mm countersink and countersink all 6 bed hold down bolt locations
  2. Use a 5/16 or 8mm drill bit and enlarge 3 holes where the stand offs pass through
  3. Use a 19mm countersink and cut away 2 -3 mm of the “pedestal” so the cover cannot contact the washers under the standoffs.
  4. Reassemble printer. Install optional flat head bolts at 3 unused bed hold down locations before installing the bed module.

Bed is now 0.5mm above stock location and should not contact the plastic base at any point.
Print a “ringing” tower. It is possible putting the bed on stilts created a resonance problem that the accelerometer in the toolhead can’t see. If so dampers can be added between the bed and base but that could result in the bed again being able to influence the mesh.

7

Bed mesh experiment:

Obtain 3 6mm (or 1/4") bearing balls. Remove and set aside the 6 bed hold down bolts. Raise bed and place a ball on the hole for the left middle, right front, and right rear locations. Carefully lower the bed onto the balls. Do some meshes. Is the “after heat soak” mesh closer to the initial than with the bed bolted down?

It wouldn’t be hard to figure a system of rubber bands (I’d use Viton O rings) to keep the bed on the balls during “normal” printing. The nozzle hooks the print and things could get out of place.

8

True, but it is flexible enough that it tends to conform to the shape of whatever it is laying on.

I considered using UV cure adhesive like the put behind the glass on cell phones. Float your glass plate, heat soak to a temperature of your choice and hit it with the UV.

2 issues.
All modern printers assume a conductive build surface and use some form of eddy current device when probing.
Adds too much mass for a bed slinger.

2 Likes
9

I do not understood all sorry. That is possible to have additionals pictures ?
Also, even if i do not understood all, i saw some differents approach, maybe that can help too :

10

Heatsink compound is an interesting concept, potential for a large mess but I don’t see a functional problem. It might be prudent to design and print some brackets to keep the glass centered on the bed in case of nozzle hooking a print.

No need for a custom cut plate though:

310 x 310 is $20.00

1 Like
11

Your first link is to repair a broken mount for the power supply, nothing to do with bed mounting. The second does raise the bed but adds additional layers of plastic. I’m trying to eliminate plastic as a layer

I guess I wasn’t clear, the metal frame in the pic in the first post is buried inside the printer base. In order to get to it you’ll need to:

  1. Raise the gantry to the top.
  2. Put the printer on it’s side
  3. Remove bottom cover.
  4. Remove the plate over the power supply
  5. Remove the power supply
  6. Remove the bed heater module
  7. Remove the main board
  8. Remove all the screws that attach the blue base to the frame
  9. Remove the touch screen
  10. Remove the pressure pad and nozzle brush
  11. Put the printer on it’s feet
  12. Remove the bed module
  13. Lift up the cover and tip it to remove (you might have to remove the gantry and top frame)

Install the washers and standoffs (red) so it looks like this

image
image1417×1006 144 KB

Drill away the plastic (refer to original post) and put it all back

Here is a shot of the bed on stilts with the cover off.

image
image1920×970 111 KB

3 Likes
12

Thanks that is very clear like that :smiling_face_with_three_hearts:
That is lot of work to do it… but that make sense and it’s the recommanded build for a vorron system. I would like to do it… but out of my capacity for now. I keep that in mind, maybe… i need more free time ^^

13

Finally a real engineer, not the garage level ‘designers’ that created this mess.

No amount of shimming is ever going to fix the TACO…unless you plan to run at one heat soaked temperature in perpetuity.

I agree with your assessment and have gone down this path over the last many months. I don’t have a lot of time to fix this brick on my bench but I finally got my mods back together yesterday and ran a gantry level - that failed. My fault. I had the opportunity to lower the bed several mm due to getting rid of the plastic bed support. Unfortunately the stepper motors now crash on the plastic floor as the limit. If I had put the whole thing in SolidWorks it would have found the problem - but who has time for all of that?

The basic problem is that we have several sub units all rigidly attached to one another with very different thermal expansion coefficients. This is the same thing as a bi-metalic bar designed to deform as used in old school thermostats. And we have 3 layers in the SV08, 4 if you consider the steel build plate. Plastic vs aluminum expansion coefficients are about 3/1.

My mods, so far, are to convert the bed to a floating 3 point system. This decouples the first layer. There is one point that is stationary and used as a datum to keep the bed in place. A second mount is a linear rail that keeps the bed in orientation. The third mount is a ball with a single point of contact that is allowed to rotate.

I printed up some new parts yesterday to move the bed up so the steppers won’t crash again. I - may - get it all back together today to attempt another gantry level. That should prove to solve half of the TACO problem. The other half is likely from the magnetic pad being glued to the aluminum plate. Again, the difference in thermal expansion coefficients between the aluminum plate and and the plastic composite magnet.

I expect to find this to still be a source of deformation and have a couple of ideas on how to decouple the magnet and the heat plate. The trick is going to be to decouple the 2 parts but still keep the build surface oriented throughout the print.

My mods are also low cost, but rather labor and modification intensive.

Edit:
Most of the above is directed at the average reader here, not Cardoc - who understands all of this.

I converted my ‘primitive’ Ender 3 to a 3 point mount long ago. I level it - ONCE - and print. The only time I’ve had problems is when I started printing the carbon PC parts for this project. The added bed temperature was enough to mess up the Z offset. The bed level, and mesh, remained unchanged.

Felt bed
Felt bed1920×2560 236 KB

Edit 2:
7 minute read!?! Really? Maybe for a third grader :roll_eyes:
Must be for the whole thread?

1 Like
14

Gantry level worked this time. First mesh is a huge improvement over the original with 2mm deviation. Yes, 2.0mm not 0.2mm. Still pretty poor.
The mesh results were confirmed with a straight edge and piece of paper. Still more work to do. Given the bed is now suspended on 3 points and free to float indicates the bed is either physically warped (will check manually when cold) or??? I’d expect a bowl shape if the disparity between the aluminum bed and magnet was the the cause.

Taco1
Taco11545×725 107 KB

1 Like
15

A couple layers of Kapton tape on top of the magnet sheet (under the build plate) will flatten out your bowl. If the bed stops “shape shifting” shimming the build plate is a robust solution. Hell of the shape is stable then let the software do what it was designed to do and follow the mesh as it prints.

Unless you are printing models that require top and bottom be parallel within .2mm. But if that is your use case NO consumer grade printer will reliably meet your demands.

My question is, with your mods how does the initial (no heat soak) mesh compare to the 30 (or 60) minute heat soak. Does the bed still shape shift?

To be clear - Shape change from ambient to print temperature is expected. Shape change as the heat soaks into the printer frame is our gremlin.

I don’t think the heater being glued to the bed is a significant source of “shape shift”. The silicone wrapper around the heater is elastic enough to allow the plate to grow. I’d be much more concerned about the die cast tray. When you squirt molten metal into a cool mold all kind of havoc ensues… resulting in trapped stresses.

1 Like
16

My primitive Ender 3 meets my demands (with the exception of size and input shaping). Maybe I’m just spoiled? Most of what I print is functional rather than art. I use 2 layers as a transition from first layer just to be safe. If I see more than a tenth of a mm variation then there is something wrong. I use an ohm meter and a metal shim to check the bed level/mesh when changing nozzles. Paper at other times has proven accurate - if you have the “feel” and listen for the stepper to start vibrating.

Did some more investigating. The Sovol sensor does look to be pretty accurate. My TACO stays the same (manual check) hot or cold. So I’m feeling kind of stupid in that I didn’t check for the TACO cold to start. That also means that my work to 3 point may have been wasted?

I did try shimming with paper. The Sovol sensor picks up the paper (.005mm) under the steel sheet. I’d not have expected the steel build plate to be so compliant.

Taco 3
Taco 31773×821 138 KB

Conclusions so far:
The Sovol Z sensor works. Yes, it is slow but a lot faster than my manual mesh on the Ender. Who can’t wait a minute to let the Sovol software to do the mesh?

With a 3 point mount there is no heat induced TACO. The reason for the cold TACO? Maybe the process of applying the heat grid? Or the deformation from the grain of the aluminum sheet after the first heat?

The stock Sovol aluminum/magnet bed is junk. - IMHO - but might be usable after a mesh with several layers selected in the slicer for transition to flat.

Careful shimming of the magnet surface with Kapton tape may work as a Caveman fix.