I love the beacon. Its the best money I have spent on 3d printer gear for some time. I'm using it on my Voron 2.4 and its just as accurate as the TAP probe while being far faster, lighter and more consistent. I am getting the the best bed meshes I have ever had. One thing that the quick start guide does not cover is that you need to use the ls /dev/serial/by-id command on the pi to get the beacons device ID to add to the klipper config.
Cant use it on mine - Mandala Rose works bed has huge magnets in it, but I went with that as its flatter than a stick on magnet will ever be. I hope someone brings out a proper lidar probe at some point, that will be a game changer.
I know the video is a bit old, but now that they also have Eddy, which is the same thing but in a different board form factor. I will be getting one of these for my new Sovol Sv08 printer that is arriving next week! I love the speed and consistancy of the bed meshing making adhesion more percise! I cant wait to try it out! Thanks for your great video!
To be honest, the time saving isn't a huge selling point for me as I'm regularly running multi-day prints but I'll quite happily swallow the cost if the accuracy is considerably better. Slowly upgrading my machines to Klipper so reckon I'll give it a shot on one of them. Can't see it being too long before there's a repository of user-designed files for mounting it to at least the more common printers so that'll remove a major hurdle for most people.
On a 500mm ratrig the mesh leveling can take up to 10 minutes with the bed warmup included. That's a lot of time between prints. This would cut meshing down by 30% i bet. That adds up in a print farm environment.
I can't wait for the market to open up for competition and introduce more fine tuning and innovation in this style of ABL. Given a little time im certain things will get very interesting.
Beacon is quite intresting, some other quirks is that it can measure Backlash in the Z axis for adding some software compensation. wont fix it, but minimize it.
Another nice video, thank you for taking your time to share your knowledge for a newbie like me! You have been helping me through a lot with 3d printing.
OMG, I love this. It will eliminate the need to worry about nozzle ooze when using Klicky Probe or Tap. Ordering one to try. Great video Michael, thank you!
but it adds worry about thermal expansion of the nozzle or having to adjust the offset for different nozzle. I still personally like tap more, but this seems great nonetheless
Nozzle ooze is a non issue if you config for it. Use a gantry mount dock for the klicky/Euclid probe. Add a 5mm retract to the print end macro and a 5 minute pause after nozzle heat up followed by a nozzle scrub in print start. If your using a high flow hotend with a longer heat zone you may need to increase the pause time to allow all the oozing to finish before the scrub. Observe and adapt.
@@Daepilin The amount of thermal expansion is going to be tiny and pretty much not be a realistic effect. Literally the textured PEI sheet most people use is going to have a bigger variance than the thermal expansion. For example the linear thermal expansion of brass/steel is going to be under .03mm on a 10mm piece of steel/brass(the amount where you could possibly have any expansion. That is the difference between 70 and 260 as well. If your properly heating the bed/nozzle before probing then your talking about a .009 variance between 160 and 220 which is the general temperature range in what you would probe/print at. Most probes alone have a variance in accuracy of .01-.02mm alone. Hell your setting the z-height in .1mm accuracy. Now on something like SLS or resin that kind of thermal expansion actually would be an issue that needs to be factored in. On an FDM machine where your typically concerned about .1mm steps a .03mm difference is not going to be a major factor to worry about.
Beacon is epic. I use it on my machine, it's a million times better than the POS BL Touch I was enduring for the longest time. Beacon is currently the best Z probe and bed scanner by a longshot, all the other stuff is trash and obsolete. Beacon: Fast, accurate, small, lightweight, superior.
I removed the springs under my Anycubic Mega X print bed two years ago. Installed jam-nuts to prevent leveling knobs from rotating. Manually leveled it twice since then after repairs. No leveling mesh. I fear I am doing something wrong, for I never have leveling issues.
Really cool video, thank you for taking the time for this content, owning the Bambu X1 Carbon, and now the Prusa XL *I preordered before the Bambu, this kind of bed leveling will be game changing if there is adoption. As you mention, the resolution and the time to get to result will improve print times substantially. Now I print functional designs, but those running print farms, saving 3/5 min prep time will be and added benefit. I just ordered to put in on my RailCore 300ZL.
I like innovation, so I am pleased this product exists. I am impressed with its speed and accuracy. I get overwhelmed with the Code aspect of this hobby, but am taking it a step at a time. One day it will be more manageable and make sense. This probe is cool and I do think it will find a home, but likely on the higher end user bracket. Voron, Sekit, etc etc or motivated hobbyist users. I don't think its bed limitations will be a deal breaker.
I am running TAP on my Voron, wich has the absolute advantage, that i can change Nozzles AND Bed surfaces, even both at the same time and i never have to readjust for any of it, since the Nozzle is the probe and the offset always stays the same. If tuned correctly once, you have no more guesswork at all.
Our build platforms are flat and parallel within 0.05mm, so no need for waste of time additional processes. And yes, our oldest machine is 4 years old without any problem towards bed flatness.
I really love your channel, I watch almost every single video and I truly appreciate your work. I have checked out several products you've reviewed, and this one was one of the most exciting to me. Since I am very new to this 3d printing thing, I have very basic entry level products. I have an ender 3 s1, but also got a secondary sprite pro print head, because I intended on swapping out the pro head for hotter materials and using the standard sprite head for lower temp materials. In December, I think, I caught the product launch for the sonic pad. They REALLY sold me as TH-cam doesn't seem to have any really negative remarks about it and it has KLIPPER!!! Now, as a new person to this world, and a gadget nerd, I just had to know of the awesomeness of this entirely new program that I'd never heard of, and it is impressive!!! So I bit and, bought me a brand new shiny thing that was going to be fantastic!!! Move on to the current video on this AMAZING, AWESOME, Beacon ABL sensor. You had me at fast, high resolution, and non contact. I was fully ready to drop my $80 on this magnificently cool product. But as a lot of people are using it on ender c series and ender 5 products, I reached out to Beacon3D to ask a couple of pre purchase questions, what they told me was VERY disappointing. Though the sonic pad runs "Klipper", it isn't REALLY KLIPPER. Apparently, creality messed with the source code and didn't say how, to Klipper's open source code update log. This basically negates the license for klipper, and this was reinforced by the klipper official discord channel. They openly stated that they don't acknowledge the sonic pad. And that creality has violated the license, their words. Beacon3D said as much as well. So basically, if I wanted, which I actually do want the Beacon abl, I would have to dump simplicity and then learn how to use, set up, and run klipper in a raspberry pi, and the sonic pad would basically be a $159 paper weight. It's VERY frustrating to me that no one has brought this up. Creality is basically conning everyone who buys their sonic pad by NOT specifying that it isn't REALLY klipper, it is a half measure knock off of the fully functioning program. I truly hope that Beacon eventually works it out, or that Creality creates a compatibility with Beacon. Because THIS is exactly what I wanted, for a bed leveller. But if you weren't aware, potential buyers of Beacon BLS should be informed, as it is non compatible, at all, with Beacon, according to Beacon 3D.
Hall sensors work with eddy currents too. The Prusa mk4 uses a hall sensor for the filament detector, for example. I am not sure if this different approach uses exactly the same method of orthogonal magnetic fields and their interruption. Take a look at hall sensors.
Hall sensors detect charge deflection in a conductor due to the presence of a magnetic field, aka the hall effect. Eddy currents are reverse induced current in a secondary conductor. Two different things. Hall sensors don't detect eddy currents. Common industrial hall sensors such as the pinda etc have their own built-in permanent magnet to provide the field which then gets disturbed by the presence of a metal element such as a steel print bed, the hall sensor detects that disturbance. Again hall sensors don't detect eddy currents, they detect magnetic field variation, now an eddy current can produce a magnetic field and if that current were to change a hall sensor could detect that. I suspect this is exactly how this sensor works, I designed something similar for a different application (machining) about eight years ago, we used a small coil to generate a timed changing field, a comparator compared the resultant periodic output from a nearby hall sensor with our known periodic field change to detect target distance. It's just a spin on the traditional permanent magnet sensor I.E instead of having a permanent magnet producing a fixed field and varying the distance to target while measuring the hall output, you vary the field and calculate the instantaneous distance to target by measuring the hall output in synchronisation with your varying field.
I'm gonna consider it when someone will do the mount for my ender 3 s1 as I'm not gonna learn CAD pretty soon. Anyway its a great idea and I think it will become mainstream in a couple of years
I want to use a tap with this so I don’t have to adjust my z offset regardless! I’m not a gram counter though. Ona Voron 1.8.just heard about on one of the forums so not my idea lol
This feels like the first usable version of a game changing advancement. Once this style becomes more mainstream and cheaper it will easily become the new standard.
I don’t at all disagree, but as time passes I find it harder to understand why non-Klipper users even exist. It’s just phenomenal compared to the alternatives.
@@dsnineteen kipper is a pain is the ass to learn when what you have works fine and you have to find an elusive and overpriced rasberry pi. Plus if you have a X1C this thing would pretty much be pointless.
It depends, if you have a removable build plate then you typically want to because the act of removing it and putting it back can throw off the previous level a slight amount. That is less important with a hard mounted bed on a core XY machine but on something with leveling screws you want to do it every time.
You can do the mesh scan far faster. Beacon actually samples at 1 kHz regardless of your toolhead speed and just throws away samples when you're moving slower. That mesh can be done identically in just a couple seconds.
It still pays to do it at a reasonable speed, just because if you go super fast, beacon will start to pick up issues with your motion system itself. I can scan my bed at 800mm/s but I get a little vibration and surface discrepancies at that speed, vs doing it at 500mm/s.
I'd like to see a price justification, what exactly are they charging 80$ for and why they can't go down. They could sell SO many if they went to the 50 mark and maybe below
Clones will no doubt be appearing shortly. Given the number of inductive bed sensors already on the market I would think prior art exemption would render any patent they hold useless not that that's much of a consideration for many producers.
These are using off the shelf Texas instruments chips that cost only a couple of dollars, they are even using the suggested layout that comes with the chips documentation. Some chinese companies will come along and properly open source it and sell it for $10 in a couple of months.
A TMAG5170 by any chance? I see it has a SAMD21 as it's mcu. That's about $7 total, throw in another generous 8 for the boards, passives and connector and it's all of 15 bucks retail to make the hardware. So yeah, highly likely.
This is pretty cool. I wonder if this amount of resolution is actually used in the firmware compensation though, or if it ends up collecting a ton more data than the machine/firmware can actually use. Very cool tool regardless!
@@lloydrmc he is saying that an old dirty rag has a ton of resolution and filth that you could model... but why? That's less smooth (more detail, texture) than a spring steel magnetic sheet, and to triangulate the tilt and compensate Marlin firmware even has a 3 point bed leveling option that suited needs for early 3d printing with bed tilt. Assuming the surface was very smooth and uniform, and not warped. But that's like a polygon from doom II trying to model the surface of a pumpkin . But the more point samples you can take, the more accurate your resolution. That's more like a point cloud with a ton of points sampled from a lidar or camera which can be turned into a detailed mesh. Much like bamboo labs and their microlidar approach. Just as with pixels, you get a more detailed picture of the surface deformations as well as tilt of the bed in the end with more samples. However, for Derek his prints get by with a triangulation then mesh creation in a 3 pts by 3 pts array, or 9 points. The bed is smooth enough but is warped slightly on either side and you can mathematically interpolate the smooth curves that arise from the samples given and get really close to the actual surface. Think about a statistics class were you take a bunch of samples and try to use linear regression or something to find out where the line or function actually lies. We are getting into bed leveling point cloud here 🤯
@@derekc423 Depends if you care about the bottom surface finish. Your sheet is not going to be perfect and will have some dimples in it. Most might not be large enough to affect adhesion but will make the surface look less uniform. Then it's fast and light, so suitable for higher spec printers. Putting it on an ender is wasteful though.
Confused why an 8bit mainboard would cause issues? You said the Beacon attaches to an SBC via USB and is implemented in Klipper as in independent mcu, so what aspect of the 8bit board was breaking it? This is a really interesting device and is definitely a novel solution (albeit klipper only, which could be seen as a drawback for some). Heres hoping for future innovation and refinement.
it's less to do with beacon itself and more to do with multi-mcu homing, the 8bit board is responsible for the motor movements but the probe MCU is responsible for triggering the, well, probing, and the code needed to have those two MCUs work in concert, whether it's beacon or a mouse switch, is a little much for an 8bit board to cope, especially multiple probes at speed.
@Allan Mitchell yh that makes sense. I'll admit, my knowledge of the internals of Kliper are much less than that of MCUs. Why does the beacon NEED to control the Z homing? I can see why it helps etc.. but its equally not really essential to actually generating the bed mesh.
@@Thats_Mr_Random_Person_to_you because both MCUs are involved, one for motion, one for sensors, it's the same if you're using a second MCU for quad Z motors but the end stops are on the main MCU, the code to get them to cooperate without communications lag is hefty resource intensive.
If this probe came out before the abundance of touch probes available now, then I could see wanting it. But I prefer sensing the bed surface, not the metal underneath it.
Neat. I wonder what the variance of thickness the PEI is on the covered metal sheets is, and if it could affect things. If its not much, this seems like promising tech.
This like all inductive probes measures the metal and the glue and pei sticker are indeed a factor but MOST of the time it's a tiny one. For those not concerned about speed things like the new Nextruder from Prusa or Voron TAP are technically more accurate, but with the speeds involved you're also more likely to do more high res meshes (11x11 or more) as opposed to the slower options where even 5x5 can seem to take forever.
@@urufushinjiro on my mantis tap I manually calibrated input shaper to 14k easily on mzv. If you calibrate with adxl and just go with that it introduces noise that isn’t accurate. So I can print fast on my trident.
That's pretty amazing, and really tempting... But the Voron 2.4 files still say they haven't actually been tested, so I think I'll stick with Klicky for now. It's definitely going to be on my radar in the future, though.
As much as I love the concept of the beacon, the speed and accuracy, the fact that I use FR4 (GR10) plates for all of my printbeds makes the beacon not usable for me. Also, I recently switched nearly all of my printers to klicky-probe. A beacon for each one would have been very expensive...
yup. i even have a timber print bed... glass, fiber, timber... they no worky! and i just dont bother with bed levelling, i can jog the head across and get it pretty well much spot on before the "autoprobe" has even got its first measurement...
Neither was I, but I replaced my klicky probe on my 350mm Voron V2 with this and not only is it comically fast (I sweep at 500 mm/s) but the sample standard deviation is way better than the physical probe. With such a large bed I was always having problems with first layer consistency but this thing makes everything perfect in a tiny fraction of the time of the physical probe. This is the best piece of 3d printer hardware I have ever tried by a wide margin
@@BarrettAnderies Thank you for your thoughtful reply. I'm concerned with the consistency of the nonconductive coatings on spring steel plates. Do you think that is an invalid concern? I buy cheap stuff where possible. I don't know that I'd be willing to stop to facilitate electromagnetic trickery.
@@RichardBronosky I've seen this mentioned a few times and I pretty much ignored it because I'm printing large parts on my voron with a first layer height of 0.3mm and width of 1mm, so small height variations in the coating don't matter much to me. Also, I'm using a textured bed which I imagine allows the plastic to move into lower pockets if there is an adjacent higher pocket from the coating application process. However, now that you bring it up I have a few thoughts: 1. If you're printing big parts with large first layer height/width I think it's a non-issue unless the variation in coating thickness is extremely large 2. Thought experiment: Suppose your BLtouch samples two points. The first is a very low point due to coating variation, the second is a very high point due to coating variation. Both points have a large error compared to the metal underneath due to coating variation (the cheaper the bed plate, the worse this scenario is in theory). The bed mesh is going to interpolate between these points. Most of the interpolation will not match the actual surface (except in the middle, assuming the rest of the coating was perfectly applied). Then imagine the same scenario with an inductive probe which is measuring the metal. Since the inductive probe can achieve a much higher mesh density, it suffers less from the potential worst case scenario physical probe measurements constructed in this thought experiment. It seems to me that the physical probe would actually be more prone to poor surface mesh collection due to the lower sampling density and therefore higher risk of sampling points with higher error due to coating thickness variation. I guess the question is which causes more true error, but tbh I have no idea. It would be pretty easy to connect a Beacon scanner and a BLtouch to the same print head and sample with both at the same time and compare the meshes. You could even raster the BLtouch (very slowly) to collect a very high density mesh for comparison. All I can say confidentially is that compared to my (homemade) klicky probe, the Beacon is significantly better. 3. Solid state components don't wear out as fast as mechanical components. Not only have I found it better from a mesh precision/accuracy perspective, I've also found it much more reliable mechanically. I print exclusively ASA which produces a ton of VOCs which coat everything in the chamber. Every 2-3 prints my klicky probe would fail because the magnets had collected so much VOC gunk that they were no longer electrically conductive. I realize this wouldn't be a problem for a permanently mounted probe like a BLtouch, but I still think that in general the Beacon would be reliable for longer periods without service/re-calibrating. I would be curious to see how much surface coating variation there actually is by comparing an inductive mesh to a really high density physical mesh, and I don't think it would be that hard to do with Klipper, but I've been so happy with my Beacon that I can't be bothered to do it myself. I will be putting Beacons on all my printers going forward
Cool as this is, I think we are approaching bed level the wrong way as a community. We need to demand that manufacturers give us straight beds that don't need these secondary steps that just trade one problem for another.
Better tolerances and less heat warping would be great, though there are associated costs and tradeoffs. I suspect that for most cases ABL is a better use of time/money vs better tolerances.
Unless you demand manufacturers provide beds that have been cast rather than formed, then been heat aged/treated, then machined and then ground and are as a result prepared to pay the hefty price that comes with those secondary very costly processes you are not ever going to get a flat bed, and even then it won't be absolutely perfect, ever, perfect flatness doesn't exist in the real world, only acceptable deviation, measurement and compensation will always have a place, this is true in all engineering. But what you are really demanding is that they be cheap and widely available on cheap printers, because you can already buy precision, you just have to pay for it, this demand is simply unrealistic. Precision toolmaking has been around a good long time now, it's not like this is a new thing, just like people expecting to pay cheap tool prices for precision tools that are expensive to manufacture.
@@m3chanist Great writeup, it's even more lopsided than I thought. It is interesting though, how "good enough" precision has gotten cheaper recently. I'm talking about the advent of $10 steel digital calipers, ABL, and so on. I suspect that modern machine tools also make the really high end precision also a bit cheaper, but still very very impractical to put on a consumer grade printer; and of course, that may be balanced out by increasing labor costs for skilled machinists.
I came across beacon the other day and i'm designing a hotend. It's one of those parts where trying to implement it after the fact is a bit of a struggle when you start considering screw and extrusion locations. I'm not giving up on the idea, I like the concept, though im curious how it will fare once prusa's new leveling system is looked at and perhaps is adapted to the previous probes.
By new system do you mean the load cell leveling?? If so they don't seem comparable in terms of end user benefit. Like load cell leveling is probably better than having a bltouch hanging off the side but its not really any faster. It's more accurate but leveling accuracy is not a problem unless your printing with extremely small nozzles. Which means speed should be the only thing that we are looking at for real improvement. Especially because even if you are satisfied with the current time it takes to make a mesh, a faster system can allow for a more detailed and comprehensive mesh for in same amount of time.
@@nocare Yep, but more specifically the gcode running behind it that allows for bed leveling only in the print area. Realistically speaking, that little bit of code when adapted across the industry will be a pretty decent upgrade for 'traditional' probing methods imo. It really only makes a ton of sense in my opinion for the DIY CoreXY community. Most other machines regardless of brand will struggle to meet all of the prerequisits. For everything else, speeding up the traditional probing methods with an adaptation of prusa's gcode will be the way to go. Which is where I want to see how that code develops with point density.
Wait... if klipper pulls the load off the board so it processes the commands, and the beacon connects via USB to the PI. Why is running 8 bit boards an issue? The printer mainboard shouldnt have much load on it at all? Wasnt the point of klipper to get more use out of the older boards originally?
it's less to do with beacon itself and more to do with multi-mcu homing, the 8bit board is responsible for the motor movements but the probe MCU is responsible for triggering the, well, probing, and the code needed to have those two MCUs work in concert, whether it's beacon or a mouse switch, is a little much for an 8bit board to cope, especially multiple probes at speed.
The printer main board can monitor directly attached endstop without using processing power. But the printer mainboard has to do some processing to understand when Klipper communicates with it. So it actually puts more load on the printer mainboard if endstop is remote and coming via Klipper.
Oh wow those prerequisites, cant say that it is the future of 3d printing. Yes it is fast but only on certain printer with certain firmware with certain bed which i don't have anything of these because i am using old CR10S and Ender 3. It is obvious that this ABL is made for Vorons in mind ... but also for others if you go thought the pain to mount.
Everyone will be using klipper this time next year of course its the future of 3D printing. Lol the Ender 3 isn't the future of 3d printing its the past what are you talking about. Might not be in your future but you aren't 3D printing.
Why can't you run this on an Ender? You can get a double-sided, spring steel PEI sheet kit with bed magnet for ~$20 (a good upgrade anyway) and a 32-bit BTT board for $40-$60. Running Klipper requires a Raspberry Pi or similar device, but many of us already have one lying around (or are already running either Klipper or OctoPrint on one), so that isn't much of an issue. Prerequisites solved, yes?
Nice guide/review! Is there a probe that is an upgrade from a Bltouch that would work with the same port/connection? I'm modding an old Artillery Genius to Klipper and a Stealthburner toolhead, but I'm not really looking to upgrade the 8 bit mainboard (unless maybe a cheap direct drop-in exists, i really don't want to rewire).
The MK4 uses a strain gauge sensor with the hot end floating on the sensor. It likely uses an HX711 for analog amplification, and a small microcontroller in addition to send a trigger pulse once the rising edge reaches a certain threshold. It's a lot like the CR6-SE and other nozzle-as-probe systems.
@@ThantiK In the Prusa interview of Thomas Sanladerer, Prusa mentioned they were getting the full info on the pressure on the nozzle, so not just a trigger at a certain level. Whether that is correct or I misunderstood, I leave up to you - check the video from Tom yourself and let me know how you understood the statements ...
@@ThantiK josef prusa claimed they get full detailed pressure feedback from the hotend and talk about using it to precisely detect underextrusion or filament jams. this does not sound like trigger pulse to me. if they implement it the way you say, then they can't do what prusa is claiming. so maybe prusa is lying?
Eddie Currant, you say? I think he was in the year below me in school. 🤔😄 I keep coming across new _'ooooh, shiny_ things to add to my CR10s-Pro, only to find out that I'm, once again, hampered by the f$%king 8-bit board! 😭
I'd be interested to see this compared to the BDsensor for Marlin. The BDsensor is the same price as a BLTouch clone and seems to function similarly to this sensor.
It must unless there's some magical sensing of nozzle location I'm missing. Using a load cell to do nozzle taps (Bambu and Prusa XL/Mk4) seems like a slower, but also more fool proof method. I've ruined a number of beds before when probes got tweaked and I didn't realize it... a few minutes extra at the start of a print is a small price to pay to avoid that. Even that can be susceptible to ooze on the nozzle, of course, but as seen on existing printers it can be mitigated. This is very cool and I may decide to try it on my V-Core eventually, though I wish it allowed for more distance to the bed, like a LiDAR would... Catching on failed prints is usually how the previous probes got messed up.
@@TMS5100 I agree, but I think we've got a ways to go before hot ends with load cells become commonplace. It's less of an issue though on the one printer I have that can't do nozzle taps. That one has to preheat for a while anyway, so... What's another couple minutes doing the pinda probe at that point?
I"m definitely interested but I'll wait for someone else to publish a mount for my printer. I can design simple things, but that mount looks too complicated for me. (sigh)
It's all nice and shiny but in the age of force sensor bed leveling (where you don't need to calibrate anything when you change nozzle) I wonder if it is still relevant. Especially with the popularity of garolite beds on the rise, and strong magnets embedded into the bed. Not to mention the compatibility with Klipper only.
I hope someone might be able to help me on this. I've installed the probe its working great but when I run a mesh it skips over the 1st 60mm of the very left of bed and completely goes off the bed on the right side. I've adjusted the Mesh_min and the mesh_max and also the probe offset . I'm stumped on this..
Does klipper mesh adjustment code can actually handle these thousands of points? Or does klipper applies an average of points to arrive at handful of points to create the corrections? If it's interpolation then all these thousand point probe is just gimmick. Moreover this still require you to change the offset when you switch nozzle, and if you are printing with abrasive filament then print after print that offset changes and this sensor will not help. Probably prise or anker kind of load sensor in the nozzle makes sense.
It adjusts for the interpolated mesh. But with such high resolution its more like real bed instead of interpolation. You can still use the auto z feature thats often used together with klicky, although i have found it a gimmick at best. Only certain probes benefit from it, mostly the omron inductive probe used on vorons since it has no temperature compensation, or when using a dual sided flex plate with a inductive or infrared sensor since both trigger off the metal and not the pei on top. When using a mechanical probe like klicky and change bed surface you are better of not having auto z as the beds are not always the same thickness. Same goes for textured vs smooth. On the topic of nozzle changes, such nozzle probing or auto z features are interesting for revo user as well as people who only use cheap knockoff nozzles. My collection of phaetus and e3d nozzles are all within 0,01mm of each other, so no need for adjustment there, the cheap brass ones that came with some of my printers deviate much further, up to 0,05 from the target. All of course measured unused
Interesting. With my experience using CRTouch and board native installation of Marlin variants, the firmware is the biggest part of the issue with accuracy of the mesh. One possible benefit of this sensor vs CRTouch is interface and cabling. The CRTouch is a raw on/off sensor which some say may get noise and false trigger sensing thus affecting mesh accuracy. Since this sensor uses USB as its interface, EMF noise affecting accuracy is eliminated. Still, you apparently need Klipper, likely on a Pi, and Pi's are hard to come by. But I haven't looked into Klipper at all, so my comment is probably naive.
I mean, for most creality owners the build plate restriction and actual cost of the device itself make it really hard to sell. The more you think about it, the more niche of a device it becomes. Sure, enthusiasts will buy it and do some interesting things, but the vast majority of people buying this probe will be running voron/ ratrig/ other DIY corexy machines. Imo, this probe only really makes sense on something like a CR-10 that you're willing to convert to klipper. At that size, the speed makes a ton of sense and you can get incredible fidelity. The Pi ordeal is definitely a PITA. And i heard the Pi 5 is on the way. So that's not getting better anytime soon.
I'd say that with the price it's not something many will add to $300 Creality printers but may be of value to those running Voron, RatRig V-Core, VZBot, etc. These already run on klipper. I agree, right now the cost of pi is ridiculous and prohibitive for many but a lot of people are already running printers using Klipper as firmware with Rpi already onboard. If it weren't for the price tag, I'd be ordering one of these to add to my custom coreXY but I just can't justify the expense right now.
Well, the official price of the Pi isn't bad. It's the scalping resellers that has the ridiculous price. In hush voice: "want a Pi now? Here's one for $200"
@@danman32 Not just the scalpers, the official sellers as well. When they get a batch they put them into overpriced kits, you can't just buy a pi for $35-$45 anymore, you have to buy a $150-$200 'kit' with a lot of stuff in it you don't even need but it's the only way to get the pi. I'm glad I had a few pi before the chip shortage, I won't buy another unless/until the prices get back to normal.
Would it be possible to run the skr mini v3 and a Microswiss NG for a ender 3 pro? The NG's aluminum chassis is quite wide. Just an idea for my future overhaul. 😶
That seems neat, but I'm not about to give up my glass bed - or my Marlin. Really the only thing preventing Marlin compatibility is the USB interface (I wonder if it speaks raw serial?) but the glass bed is sadly a matter of physics.
Why would the mainboard matter if it's connecting to the Pi or whatever other SBC you might be using? Isn't the secondary ECU doing all the hard work? What does the mainboard have to do with the process?
When printing, the system would have to add/subtract the interpolated Z axis correction many times per layer **. No 8 bit board has a high enough CPU clock to do this in a timely manner. ** makes me wonder why we're not doing Z axis correction in the slicer?
I'm just upgrading to a canbus board to reduce my wires run to the head, could this be run to the CANBUS board rather than a home run directly to the main board?
Their documentation only states that it needs to be plugged into a single board computer, which a CANBUS board is not? Or am I mistaken? The CANBUS is only an MCU board correct?
Found my answer, haha. The USB-C on the CANBUS is only an input, not capable of bidirectional usb transfer trying to couple with the 4 wire CANBUS 😞 disappointing as running a 4 wire, and a USB cable is not ideal, but will make that sacrifice for this scanner 🤷♂️
wondering what the limitations are to resolution? any degradation in consistency or accuracy if you set the resolution to something insane, like 200x200? edit: actually just realized that still requires drawing 200 lines across the bed lol. maybe 200x10 is actually what I was imagining when I wrote that.
I love the beacon. Its the best money I have spent on 3d printer gear for some time. I'm using it on my Voron 2.4 and its just as accurate as the TAP probe while being far faster, lighter and more consistent. I am getting the the best bed meshes I have ever had.
One thing that the quick start guide does not cover is that you need to use the ls /dev/serial/by-id command on the pi to get the beacons device ID to add to the klipper config.
😂😂😂😂
Acceleration values has to have a higher upper limit compared to the tap? Far lighter and less moving parts.
Cant use it on mine - Mandala Rose works bed has huge magnets in it, but I went with that as its flatter than a stick on magnet will ever be. I hope someone brings out a proper lidar probe at some point, that will be a game changer.
Do you also use the QGL? Could you perhaps share your macro/config?
I know the video is a bit old, but now that they also have Eddy, which is the same thing but in a different board form factor. I will be getting one of these for my new Sovol Sv08 printer that is arriving next week! I love the speed and consistancy of the bed meshing making adhesion more percise! I cant wait to try it out! Thanks for your great video!
Thanks again for your expertise .. I used you original videos to build my 3 ratrigs and now an upgrade to the Beacon ..
To be honest, the time saving isn't a huge selling point for me as I'm regularly running multi-day prints but I'll quite happily swallow the cost if the accuracy is considerably better. Slowly upgrading my machines to Klipper so reckon I'll give it a shot on one of them. Can't see it being too long before there's a repository of user-designed files for mounting it to at least the more common printers so that'll remove a major hurdle for most people.
Have you tried it? If so, how well did it work?
On a 500mm ratrig the mesh leveling can take up to 10 minutes with the bed warmup included. That's a lot of time between prints. This would cut meshing down by 30% i bet. That adds up in a print farm environment.
The newest version of The Beacon also allows you to use the nozzle as a probe, as well as it has an accelerometer built into it
I've been waiting for you to get a vid for this since I saw the posts. It looks fantastic and now I really wanna mess with this
I can't wait for the market to open up for competition and introduce more fine tuning and innovation in this style of ABL. Given a little time im certain things will get very interesting.
Thanks!
I already have one, but currently sat in the upgrades to do pile.
A great review, thanks. I ordered one for my new printer last week, I really must get it finished!
Beacon is quite intresting, some other quirks is that it can measure Backlash in the Z axis for adding some software compensation.
wont fix it, but minimize it.
i never thought about doing z tilit in the start g code, seems like a great idea
I've Voron Trident, it's mandatory z-tilt or it won't know the z axis position, the stock probe only gives homing position.
Another nice video, thank you for taking your time to share your knowledge for a newbie like me! You have been helping me through a lot with 3d printing.
Quick to scan and easy to setup. I like it.
OMG, I love this. It will eliminate the need to worry about nozzle ooze when using Klicky Probe or Tap. Ordering one to try. Great video Michael, thank you!
but it adds worry about thermal expansion of the nozzle or having to adjust the offset for different nozzle.
I still personally like tap more, but this seems great nonetheless
Nozzle ooze is a non issue if you config for it. Use a gantry mount dock for the klicky/Euclid probe. Add a 5mm retract to the print end macro and a 5 minute pause after nozzle heat up followed by a nozzle scrub in print start.
If your using a high flow hotend with a longer heat zone you may need to increase the pause time to allow all the oozing to finish before the scrub.
Observe and adapt.
@@Daepilin The amount of thermal expansion is going to be tiny and pretty much not be a realistic effect. Literally the textured PEI sheet most people use is going to have a bigger variance than the thermal expansion.
For example the linear thermal expansion of brass/steel is going to be under .03mm on a 10mm piece of steel/brass(the amount where you could possibly have any expansion. That is the difference between 70 and 260 as well. If your properly heating the bed/nozzle before probing then your talking about a .009 variance between 160 and 220 which is the general temperature range in what you would probe/print at. Most probes alone have a variance in accuracy of .01-.02mm alone. Hell your setting the z-height in .1mm accuracy.
Now on something like SLS or resin that kind of thermal expansion actually would be an issue that needs to be factored in. On an FDM machine where your typically concerned about .1mm steps a .03mm difference is not going to be a major factor to worry about.
Pretty interesting indeed! Thanks, Michael! 😊
Stay safe there with your family! 🖖😊
Beacon is epic. I use it on my machine, it's a million times better than the POS BL Touch I was enduring for the longest time. Beacon is currently the best Z probe and bed scanner by a longshot, all the other stuff is trash and obsolete. Beacon: Fast, accurate, small, lightweight, superior.
Great to see this review, I was very interested when I saw this was available.
I removed the springs under my Anycubic Mega X print bed two years ago.
Installed jam-nuts to prevent leveling knobs from rotating.
Manually leveled it twice since then after repairs. No leveling mesh.
I fear I am doing something wrong, for I never have leveling issues.
Really cool video, thank you for taking the time for this content, owning the Bambu X1 Carbon, and now the Prusa XL *I preordered before the Bambu, this kind of bed leveling will be game changing if there is adoption. As you mention, the resolution and the time to get to result will improve print times substantially. Now I print functional designs, but those running print farms, saving 3/5 min prep time will be and added benefit. I just ordered to put in on my RailCore 300ZL.
I like innovation, so I am pleased this product exists. I am impressed with its speed and accuracy. I get overwhelmed with the Code aspect of this hobby, but am taking it a step at a time. One day it will be more manageable and make sense. This probe is cool and I do think it will find a home, but likely on the higher end user bracket. Voron, Sekit, etc etc or motivated hobbyist users. I don't think its bed limitations will be a deal breaker.
I am running TAP on my Voron, wich has the absolute advantage, that i can change Nozzles AND Bed surfaces, even both at the same time and i never have to readjust for any of it, since the Nozzle is the probe and the offset always stays the same. If tuned correctly once, you have no more guesswork at all.
Our build platforms are flat and parallel within 0.05mm, so no need for waste of time additional processes.
And yes, our oldest machine is 4 years old without any problem towards bed flatness.
Within 0.05mm across what distance? That number without the other is meaningless. And when it's heated? How is it attached to the rest of the printer?
@@m3chanist across build length and width. The build envelope is 400x400x650. 100% machined
I really love your channel, I watch almost every single video and I truly appreciate your work.
I have checked out several products you've reviewed, and this one was one of the most exciting to me.
Since I am very new to this 3d printing thing, I have very basic entry level products. I have an ender 3 s1, but also got a secondary sprite pro print head, because I intended on swapping out the pro head for hotter materials and using the standard sprite head for lower temp materials.
In December, I think, I caught the product launch for the sonic pad. They REALLY sold me as TH-cam doesn't seem to have any really negative remarks about it and it has KLIPPER!!!
Now, as a new person to this world, and a gadget nerd, I just had to know of the awesomeness of this entirely new program that I'd never heard of, and it is impressive!!! So I bit and, bought me a brand new shiny thing that was going to be fantastic!!!
Move on to the current video on this AMAZING, AWESOME, Beacon ABL sensor. You had me at fast, high resolution, and non contact. I was fully ready to drop my $80 on this magnificently cool product.
But as a lot of people are using it on ender c series and ender 5 products, I reached out to Beacon3D to ask a couple of pre purchase questions, what they told me was VERY disappointing.
Though the sonic pad runs "Klipper", it isn't REALLY KLIPPER. Apparently, creality messed with the source code and didn't say how, to Klipper's open source code update log. This basically negates the license for klipper, and this was reinforced by the klipper official discord channel. They openly stated that they don't acknowledge the sonic pad. And that creality has violated the license, their words.
Beacon3D said as much as well. So basically, if I wanted, which I actually do want the Beacon abl, I would have to dump simplicity and then learn how to use, set up, and run klipper in a raspberry pi, and the sonic pad would basically be a $159 paper weight.
It's VERY frustrating to me that no one has brought this up.
Creality is basically conning everyone who buys their sonic pad by NOT specifying that it isn't REALLY klipper, it is a half measure knock off of the fully functioning program.
I truly hope that Beacon eventually works it out, or that Creality creates a compatibility with Beacon. Because THIS is exactly what I wanted, for a bed leveller.
But if you weren't aware, potential buyers of Beacon BLS should be informed, as it is non compatible, at all, with Beacon, according to Beacon 3D.
Looks very promising, I may replace my Euclid with this. Thanks!
Thats a cool ruler bro, you create great videos.
Ooh, looks wonderful! I really need to build a fully enclosed Voron 0 some day...
Hall sensors work with eddy currents too. The Prusa mk4 uses a hall sensor for the filament detector, for example.
I am not sure if this different approach uses exactly the same method of orthogonal magnetic fields and their interruption. Take a look at hall sensors.
Hall sensors detect charge deflection in a conductor due to the presence of a magnetic field, aka the hall effect. Eddy currents are reverse induced current in a secondary conductor. Two different things. Hall sensors don't detect eddy currents. Common industrial hall sensors such as the pinda etc have their own built-in permanent magnet to provide the field which then gets disturbed by the presence of a metal element such as a steel print bed, the hall sensor detects that disturbance. Again hall sensors don't detect eddy currents, they detect magnetic field variation, now an eddy current can produce a magnetic field and if that current were to change a hall sensor could detect that. I suspect this is exactly how this sensor works, I designed something similar for a different application (machining) about eight years ago, we used a small coil to generate a timed changing field, a comparator compared the resultant periodic output from a nearby hall sensor with our known periodic field change to detect target distance. It's just a spin on the traditional permanent magnet sensor I.E instead of having a permanent magnet producing a fixed field and varying the distance to target while measuring the hall output, you vary the field and calculate the instantaneous distance to target by measuring the hall output in synchronisation with your varying field.
I'm gonna consider it when someone will do the mount for my ender 3 s1 as I'm not gonna learn CAD pretty soon. Anyway its a great idea and I think it will become mainstream in a couple of years
Awesome showcase and guide, well done
Seems like quite a neat sensor... but I'm not ready to give up the bltouch cause I have so many different steel sheets
I want to use a tap with this so I don’t have to adjust my z offset regardless! I’m not a gram counter though. Ona Voron 1.8.just heard about on one of the forums so not my idea lol
This feels like the first usable version of a game changing advancement. Once this style becomes more mainstream and cheaper it will easily become the new standard.
Nice, i'll wait for the Aliexpress version
Interesting but certainly not for everyone, not for glass beds, thick PEI sheet beds and non Klipper users
I don’t at all disagree, but as time passes I find it harder to understand why non-Klipper users even exist. It’s just phenomenal compared to the alternatives.
@@dsnineteen kipper is a pain is the ass to learn when what you have works fine and you have to find an elusive and overpriced rasberry pi. Plus if you have a X1C this thing would pretty much be pointless.
@@TheRattleSnake3145 there are cheaper boards that work fine for klipper. But it's still an added cost for sure.
@@TheRattleSnake3145 orangepi from aliexpress is the answer
@@TheRattleSnake3145 😂😂😂
1:55 Why would you need to probe the bed before every print?
You don't need to, but it's become "because you can"
It depends, if you have a removable build plate then you typically want to because the act of removing it and putting it back can throw off the previous level a slight amount. That is less important with a hard mounted bed on a core XY machine but on something with leveling screws you want to do it every time.
You can do the mesh scan far faster. Beacon actually samples at 1 kHz regardless of your toolhead speed and just throws away samples when you're moving slower. That mesh can be done identically in just a couple seconds.
It still pays to do it at a reasonable speed, just because if you go super fast, beacon will start to pick up issues with your motion system itself. I can scan my bed at 800mm/s but I get a little vibration and surface discrepancies at that speed, vs doing it at 500mm/s.
While I really like my BL Touch, this is faster, with no moving parts and nothing to break off. I'd say it's a winner.
Thank you very much
The price is a major factor. Get it down to $15-$20 then it will be awesome, but the current meta of what we have does just fine.
I'd like to see a price justification, what exactly are they charging 80$ for and why they can't go down. They could sell SO many if they went to the 50 mark and maybe below
Clones will no doubt be appearing shortly. Given the number of inductive bed sensors already on the market I would think prior art exemption would render any patent they hold useless not that that's much of a consideration for many producers.
I was wondering how VZBot 330 did bed leveling and found he started using a Beacon V1, worked great and didn't add the weight of a 'TAP'.
These are using off the shelf Texas instruments chips that cost only a couple of dollars, they are even using the suggested layout that comes with the chips documentation. Some chinese companies will come along and properly open source it and sell it for $10 in a couple of months.
A TMAG5170 by any chance? I see it has a SAMD21 as it's mcu. That's about $7 total, throw in another generous 8 for the boards, passives and connector and it's all of 15 bucks retail to make the hardware. So yeah, highly likely.
A comparison between the Beacon and the much cheaper BigTreeTech Eddy probe would be interesting.
This is pretty cool. I wonder if this amount of resolution is actually used in the firmware compensation though, or if it ends up collecting a ton more data than the machine/firmware can actually use. Very cool tool regardless!
Klipper will handle mesh up to 100x100
@@urufushinjiro seems like you would need to be printing on an old dirty rag or something for much more than 3x3 mesh to make much difference.
@@derekc423?
@@lloydrmc he is saying that an old dirty rag has a ton of resolution and filth that you could model... but why?
That's less smooth (more detail, texture) than a spring steel magnetic sheet, and to triangulate the tilt and compensate Marlin firmware even has a 3 point bed leveling option that suited needs for early 3d printing with bed tilt. Assuming the surface was very smooth and uniform, and not warped.
But that's like a polygon from doom II trying to model the surface of a pumpkin .
But the more point samples you can take, the more accurate your resolution. That's more like a point cloud with a ton of points sampled from a lidar or camera which can be turned into a detailed mesh. Much like bamboo labs and their microlidar approach.
Just as with pixels, you get a more detailed picture of the surface deformations as well as tilt of the bed in the end with more samples. However, for Derek his prints get by with a triangulation then mesh creation in a 3 pts by 3 pts array, or 9 points. The bed is smooth enough but is warped slightly on either side and you can mathematically interpolate the smooth curves that arise from the samples given and get really close to the actual surface.
Think about a statistics class were you take a bunch of samples and try to use linear regression or something to find out where the line or function actually lies.
We are getting into bed leveling point cloud here 🤯
@@derekc423 Depends if you care about the bottom surface finish. Your sheet is not going to be perfect and will have some dimples in it. Most might not be large enough to affect adhesion but will make the surface look less uniform. Then it's fast and light, so suitable for higher spec printers. Putting it on an ender is wasteful though.
Confused why an 8bit mainboard would cause issues? You said the Beacon attaches to an SBC via USB and is implemented in Klipper as in independent mcu, so what aspect of the 8bit board was breaking it?
This is a really interesting device and is definitely a novel solution (albeit klipper only, which could be seen as a drawback for some). Heres hoping for future innovation and refinement.
I have no idea, i know that 8bit boards cant have as high of a step rate as 32 bit, but this shouldn't affect multi mcu homing.
it's less to do with beacon itself and more to do with multi-mcu homing, the 8bit board is responsible for the motor movements but the probe MCU is responsible for triggering the, well, probing, and the code needed to have those two MCUs work in concert, whether it's beacon or a mouse switch, is a little much for an 8bit board to cope, especially multiple probes at speed.
@Allan Mitchell yh that makes sense. I'll admit, my knowledge of the internals of Kliper are much less than that of MCUs.
Why does the beacon NEED to control the Z homing? I can see why it helps etc.. but its equally not really essential to actually generating the bed mesh.
@@Thats_Mr_Random_Person_to_you because both MCUs are involved, one for motion, one for sensors, it's the same if you're using a second MCU for quad Z motors but the end stops are on the main MCU, the code to get them to cooperate without communications lag is hefty resource intensive.
If this probe came out before the abundance of touch probes available now, then I could see wanting it. But I prefer sensing the bed surface, not the metal underneath it.
Neat. I wonder what the variance of thickness the PEI is on the covered metal sheets is, and if it could affect things. If its not much, this seems like promising tech.
This like all inductive probes measures the metal and the glue and pei sticker are indeed a factor but MOST of the time it's a tiny one. For those not concerned about speed things like the new Nextruder from Prusa or Voron TAP are technically more accurate, but with the speeds involved you're also more likely to do more high res meshes (11x11 or more) as opposed to the slower options where even 5x5 can seem to take forever.
This wouldn’t matter as it’s taken care of in the z offset
@@urufushinjiro on my mantis tap I manually calibrated input shaper to 14k easily on mzv. If you calibrate with adxl and just go with that it introduces noise that isn’t accurate. So I can print fast on my trident.
@@joe_duck inconsistencies in the material could though
@@IrocZIV I did put a new pei on and the 3m decided to bunch up so that makes sense. I fixed it somewhat but now some glue is missing lol
That's pretty amazing, and really tempting... But the Voron 2.4 files still say they haven't actually been tested, so I think I'll stick with Klicky for now. It's definitely going to be on my radar in the future, though.
As much as I love the concept of the beacon, the speed and accuracy, the fact that I use FR4 (GR10) plates for all of my printbeds makes the beacon not usable for me.
Also, I recently switched nearly all of my printers to klicky-probe. A beacon for each one would have been very expensive...
yup. i even have a timber print bed... glass, fiber, timber... they no worky!
and i just dont bother with bed levelling, i can jog the head across and get it pretty well much spot on before the "autoprobe" has even got its first measurement...
Even on pei on metal sheets, pei and the glue used to apply it can have uneven thickness. This is only measures the variation over the metal
excellent video... I was wondering can you share me the link for the stl of the fisrt layer test "X' design @13:10
Not a fan of induction for leveling. I'm waiting for a machinist grade "indicator" solution before I replace my BLTouch.
Neither was I, but I replaced my klicky probe on my 350mm Voron V2 with this and not only is it comically fast (I sweep at 500 mm/s) but the sample standard deviation is way better than the physical probe. With such a large bed I was always having problems with first layer consistency but this thing makes everything perfect in a tiny fraction of the time of the physical probe. This is the best piece of 3d printer hardware I have ever tried by a wide margin
@@BarrettAnderies Thank you for your thoughtful reply. I'm concerned with the consistency of the nonconductive coatings on spring steel plates. Do you think that is an invalid concern? I buy cheap stuff where possible. I don't know that I'd be willing to stop to facilitate electromagnetic trickery.
@@RichardBronosky it is a valid concern but unless you get especially unlucky with some exceptionally low quality sheet it's has a negligible effect
@@jakoob874 have you ever heard of Amazon? #PoesLaw
@@RichardBronosky I've seen this mentioned a few times and I pretty much ignored it because I'm printing large parts on my voron with a first layer height of 0.3mm and width of 1mm, so small height variations in the coating don't matter much to me. Also, I'm using a textured bed which I imagine allows the plastic to move into lower pockets if there is an adjacent higher pocket from the coating application process. However, now that you bring it up I have a few thoughts:
1. If you're printing big parts with large first layer height/width I think it's a non-issue unless the variation in coating thickness is extremely large
2. Thought experiment: Suppose your BLtouch samples two points. The first is a very low point due to coating variation, the second is a very high point due to coating variation. Both points have a large error compared to the metal underneath due to coating variation (the cheaper the bed plate, the worse this scenario is in theory). The bed mesh is going to interpolate between these points. Most of the interpolation will not match the actual surface (except in the middle, assuming the rest of the coating was perfectly applied). Then imagine the same scenario with an inductive probe which is measuring the metal. Since the inductive probe can achieve a much higher mesh density, it suffers less from the potential worst case scenario physical probe measurements constructed in this thought experiment. It seems to me that the physical probe would actually be more prone to poor surface mesh collection due to the lower sampling density and therefore higher risk of sampling points with higher error due to coating thickness variation. I guess the question is which causes more true error, but tbh I have no idea. It would be pretty easy to connect a Beacon scanner and a BLtouch to the same print head and sample with both at the same time and compare the meshes. You could even raster the BLtouch (very slowly) to collect a very high density mesh for comparison. All I can say confidentially is that compared to my (homemade) klicky probe, the Beacon is significantly better.
3. Solid state components don't wear out as fast as mechanical components. Not only have I found it better from a mesh precision/accuracy perspective, I've also found it much more reliable mechanically. I print exclusively ASA which produces a ton of VOCs which coat everything in the chamber. Every 2-3 prints my klicky probe would fail because the magnets had collected so much VOC gunk that they were no longer electrically conductive. I realize this wouldn't be a problem for a permanently mounted probe like a BLtouch, but I still think that in general the Beacon would be reliable for longer periods without service/re-calibrating.
I would be curious to see how much surface coating variation there actually is by comparing an inductive mesh to a really high density physical mesh, and I don't think it would be that hard to do with Klipper, but I've been so happy with my Beacon that I can't be bothered to do it myself. I will be putting Beacons on all my printers going forward
It's funny how you design the mount, like the keepout zone also applies to plastic
video about upgrading the cr10 mainboard easily coming soon?
Cool as this is, I think we are approaching bed level the wrong way as a community. We need to demand that manufacturers give us straight beds that don't need these secondary steps that just trade one problem for another.
Better tolerances and less heat warping would be great, though there are associated costs and tradeoffs. I suspect that for most cases ABL is a better use of time/money vs better tolerances.
Unless you demand manufacturers provide beds that have been cast rather than formed, then been heat aged/treated, then machined and then ground and are as a result prepared to pay the hefty price that comes with those secondary very costly processes you are not ever going to get a flat bed, and even then it won't be absolutely perfect, ever, perfect flatness doesn't exist in the real world, only acceptable deviation, measurement and compensation will always have a place, this is true in all engineering. But what you are really demanding is that they be cheap and widely available on cheap printers, because you can already buy precision, you just have to pay for it, this demand is simply unrealistic. Precision toolmaking has been around a good long time now, it's not like this is a new thing, just like people expecting to pay cheap tool prices for precision tools that are expensive to manufacture.
@@szurketaltos2693 Quite right, ABL is very cheap compared to the cost of upping the precision of a large component.
@@m3chanist Great writeup, it's even more lopsided than I thought. It is interesting though, how "good enough" precision has gotten cheaper recently. I'm talking about the advent of $10 steel digital calipers, ABL, and so on. I suspect that modern machine tools also make the really high end precision also a bit cheaper, but still very very impractical to put on a consumer grade printer; and of course, that may be balanced out by increasing labor costs for skilled machinists.
I came across beacon the other day and i'm designing a hotend. It's one of those parts where trying to implement it after the fact is a bit of a struggle when you start considering screw and extrusion locations. I'm not giving up on the idea, I like the concept, though im curious how it will fare once prusa's new leveling system is looked at and perhaps is adapted to the previous probes.
By new system do you mean the load cell leveling??
If so they don't seem comparable in terms of end user benefit. Like load cell leveling is probably better than having a bltouch hanging off the side but its not really any faster. It's more accurate but leveling accuracy is not a problem unless your printing with extremely small nozzles. Which means speed should be the only thing that we are looking at for real improvement.
Especially because even if you are satisfied with the current time it takes to make a mesh, a faster system can allow for a more detailed and comprehensive mesh for in same amount of time.
@@nocare Yep, but more specifically the gcode running behind it that allows for bed leveling only in the print area.
Realistically speaking, that little bit of code when adapted across the industry will be a pretty decent upgrade for 'traditional' probing methods imo.
It really only makes a ton of sense in my opinion for the DIY CoreXY community. Most other machines regardless of brand will struggle to meet all of the prerequisits.
For everything else, speeding up the traditional probing methods with an adaptation of prusa's gcode will be the way to go.
Which is where I want to see how that code develops with point density.
@@Mytagforhalo I see. That makes sense.
Wait... if klipper pulls the load off the board so it processes the commands, and the beacon connects via USB to the PI. Why is running 8 bit boards an issue? The printer mainboard shouldnt have much load on it at all? Wasnt the point of klipper to get more use out of the older boards originally?
it's less to do with beacon itself and more to do with multi-mcu homing, the 8bit board is responsible for the motor movements but the probe MCU is responsible for triggering the, well, probing, and the code needed to have those two MCUs work in concert, whether it's beacon or a mouse switch, is a little much for an 8bit board to cope, especially multiple probes at speed.
It's to do with the multi MCU homing
The printer main board can monitor directly attached endstop without using processing power. But the printer mainboard has to do some processing to understand when Klipper communicates with it. So it actually puts more load on the printer mainboard if endstop is remote and coming via Klipper.
I might have to keep an eye on this to see if they knock out a non klipper version.
It's pretty klipper dependent because it connects directly to the Raspi to do the calculations on the linux PC
Why run worse probes and worse firmware? Just upgrade to klipper
..
I think the likelyhood that Marlin will just slowly die off is bigger. And a better outcome for everyone in the end.
Oh wow those prerequisites, cant say that it is the future of 3d printing. Yes it is fast but only on certain printer with certain firmware with certain bed which i don't have anything of these because i am using old CR10S and Ender 3. It is obvious that this ABL is made for Vorons in mind ... but also for others if you go thought the pain to mount.
It's actually made with the Annex printer (K3 etc) in mind.
Everyone will be using klipper this time next year of course its the future of 3D printing. Lol the Ender 3 isn't the future of 3d printing its the past what are you talking about. Might not be in your future but you aren't 3D printing.
@@backgammonbaconEveryone will still be unable to get a RPi to run klipper on next year.
@@maplobats BTT CB1 works great for running Klipper and it's about $40 USD. I've been running it for 4 months and I have had no complaints.
Why can't you run this on an Ender? You can get a double-sided, spring steel PEI sheet kit with bed magnet for ~$20 (a good upgrade anyway) and a 32-bit BTT board for $40-$60. Running Klipper requires a Raspberry Pi or similar device, but many of us already have one lying around (or are already running either Klipper or OctoPrint on one), so that isn't much of an issue. Prerequisites solved, yes?
Nice guide/review!
Is there a probe that is an upgrade from a Bltouch that would work with the same port/connection?
I'm modding an old Artillery Genius to Klipper and a Stealthburner toolhead, but I'm not really looking to upgrade the 8 bit mainboard (unless maybe a cheap direct drop-in exists, i really don't want to rewire).
Since Prusa MK 4 is open source. Can you cover how the MK4 uses the nozzle to bed level. I use glass so this device wouldn't work.
the future of bed probing is nozzle based probing like MK4 and bambu. not this device.
The MK4 uses a strain gauge sensor with the hot end floating on the sensor. It likely uses an HX711 for analog amplification, and a small microcontroller in addition to send a trigger pulse once the rising edge reaches a certain threshold. It's a lot like the CR6-SE and other nozzle-as-probe systems.
@@ThantiK In the Prusa interview of Thomas Sanladerer, Prusa mentioned they were getting the full info on the pressure on the nozzle, so not just a trigger at a certain level. Whether that is correct or I misunderstood, I leave up to you - check the video from Tom yourself and let me know how you understood the statements ...
@@ThantiK josef prusa claimed they get full detailed pressure feedback from the hotend and talk about using it to precisely detect underextrusion or filament jams. this does not sound like trigger pulse to me. if they implement it the way you say, then they can't do what prusa is claiming. so maybe prusa is lying?
I only print on Fr4 g10 garolite. I don't probe every print, why would you. Can't remember the last time I leveled my bed.
🤣
Leveling with the nozzle so close to the bed would I expect cause it to scratch a concave bed.
Eddie Currant, you say? I think he was in the year below me in school. 🤔😄
I keep coming across new _'ooooh, shiny_ things to add to my CR10s-Pro, only to find out that I'm, once again, hampered by the f$%king 8-bit board! 😭
Excellent always great videos, thank you for making it!
Great tut :) Wish someone would figure out how to install on the darn Ender 3 KE .... tried everything and nothing works :(
I’m not too great at 3D modeling. I’ll pick one up if I can find a mount for the Creality Sprite Pro extruder
I'd be interested to see this compared to the BDsensor for Marlin. The BDsensor is the same price as a BLTouch clone and seems to function similarly to this sensor.
BD sensor doesn't have temperature compensation among other things. It does support marlin though
@@Tedlasman temperature compensation is software.
@@backgammonbacon it's hardware too.
11:42 The vertical alignment of the belts is very poor! Print dimensions will be incorrect
Does it require a Z Offset? If so, i would rather have what the Bambu does and wait a few extra minutes.
It must unless there's some magical sensing of nozzle location I'm missing. Using a load cell to do nozzle taps (Bambu and Prusa XL/Mk4) seems like a slower, but also more fool proof method. I've ruined a number of beds before when probes got tweaked and I didn't realize it... a few minutes extra at the start of a print is a small price to pay to avoid that. Even that can be susceptible to ooze on the nozzle, of course, but as seen on existing printers it can be mitigated.
This is very cool and I may decide to try it on my V-Core eventually, though I wish it allowed for more distance to the bed, like a LiDAR would... Catching on failed prints is usually how the previous probes got messed up.
@@drdoakcom nozzle based probing is the future. it's the conclusion cnc machining learned decades ago and 3d printers are only just learning.
@@TMS5100 I agree, but I think we've got a ways to go before hot ends with load cells become commonplace. It's less of an issue though on the one printer I have that can't do nozzle taps. That one has to preheat for a while anyway, so... What's another couple minutes doing the pinda probe at that point?
@@drdoakcom bambu does it with piezo sensors under the bed. lots of ways to do nozzle probing.
@@TMS5100 yeah the probe on tool head + probe on bed setup on CNC machines is similar to Auto Z for Klipper. Good stuff
Nice Video.. Thx.. Does this works with my Qidi X-Max 3?
I"m definitely interested but I'll wait for someone else to publish a mount for my printer. I can design simple things, but that mount looks too complicated for me. (sigh)
It's all nice and shiny but in the age of force sensor bed leveling (where you don't need to calibrate anything when you change nozzle) I wonder if it is still relevant. Especially with the popularity of garolite beds on the rise, and strong magnets embedded into the bed. Not to mention the compatibility with Klipper only.
Please let me know how to use Lidar bed leveling
Are you going to do a presentation on Beacon Contact?
Bacon is delicious.
managing all the start commands in the slicer seems...like a lot. use print_start and just tweak it there?
I hope someone might be able to help me on this. I've installed the probe its working great but when I run a mesh it skips over the 1st 60mm of the very left of bed and completely goes off the bed on the right side. I've adjusted the Mesh_min and the mesh_max and also the probe offset . I'm stumped on this..
Does klipper mesh adjustment code can actually handle these thousands of points? Or does klipper applies an average of points to arrive at handful of points to create the corrections? If it's interpolation then all these thousand point probe is just gimmick. Moreover this still require you to change the offset when you switch nozzle, and if you are printing with abrasive filament then print after print that offset changes and this sensor will not help. Probably prise or anker kind of load sensor in the nozzle makes sense.
It adjusts for the interpolated mesh. But with such high resolution its more like real bed instead of interpolation.
You can still use the auto z feature thats often used together with klicky, although i have found it a gimmick at best. Only certain probes benefit from it, mostly the omron inductive probe used on vorons since it has no temperature compensation, or when using a dual sided flex plate with a inductive or infrared sensor since both trigger off the metal and not the pei on top.
When using a mechanical probe like klicky and change bed surface you are better of not having auto z as the beds are not always the same thickness. Same goes for textured vs smooth.
On the topic of nozzle changes, such nozzle probing or auto z features are interesting for revo user as well as people who only use cheap knockoff nozzles. My collection of phaetus and e3d nozzles are all within 0,01mm of each other, so no need for adjustment there, the cheap brass ones that came with some of my printers deviate much further, up to 0,05 from the target. All of course measured unused
Klipper seems to handle up to 100x100 meshes ime
Something is broken with your model for the SK-Tank adapter. Every time I try to slice it, it says no layers were detected.
Whe need a beacon mount for the EVA 2.4.2 Horn duct
I want it badly since I first heard of it
Is it possible to get this to work with a can bus and u2c board?
Interesting.
With my experience using CRTouch and board native installation of Marlin variants, the firmware is the biggest part of the issue with accuracy of the mesh.
One possible benefit of this sensor vs CRTouch is interface and cabling. The CRTouch is a raw on/off sensor which some say may get noise and false trigger sensing thus affecting mesh accuracy. Since this sensor uses USB as its interface, EMF noise affecting accuracy is eliminated.
Still, you apparently need Klipper, likely on a Pi, and Pi's are hard to come by.
But I haven't looked into Klipper at all, so my comment is probably naive.
I mean, for most creality owners the build plate restriction and actual cost of the device itself make it really hard to sell. The more you think about it, the more niche of a device it becomes.
Sure, enthusiasts will buy it and do some interesting things, but the vast majority of people buying this probe will be running voron/ ratrig/ other DIY corexy machines.
Imo, this probe only really makes sense on something like a CR-10 that you're willing to convert to klipper. At that size, the speed makes a ton of sense and you can get incredible fidelity.
The Pi ordeal is definitely a PITA. And i heard the Pi 5 is on the way. So that's not getting better anytime soon.
I'd say that with the price it's not something many will add to $300 Creality printers but may be of value to those running Voron, RatRig V-Core, VZBot, etc. These already run on klipper. I agree, right now the cost of pi is ridiculous and prohibitive for many but a lot of people are already running printers using Klipper as firmware with Rpi already onboard. If it weren't for the price tag, I'd be ordering one of these to add to my custom coreXY but I just can't justify the expense right now.
Well, the official price of the Pi isn't bad. It's the scalping resellers that has the ridiculous price.
In hush voice: "want a Pi now? Here's one for $200"
@@danman32 Not just the scalpers, the official sellers as well. When they get a batch they put them into overpriced kits, you can't just buy a pi for $35-$45 anymore, you have to buy a $150-$200 'kit' with a lot of stuff in it you don't even need but it's the only way to get the pi. I'm glad I had a few pi before the chip shortage, I won't buy another unless/until the prices get back to normal.
There are lots of Pi replacements that would run this just fine.
very nice i've seen this before and it was super cool.. BUT does it work with the Sonic Pad? if so buying it today lol!
4 grams and then the heavy cable, likely heavier then a regular bl touch for example. But neat though
"It did say it could take up to 10 minutes for this first section, but I found it was over in a matter of seconds"
So how did that underground workshop[ work out?
Would it be possible to run the skr mini v3 and a Microswiss NG for a ender 3 pro? The NG's aluminum chassis is quite wide. Just an idea for my future overhaul. 😶
That seems neat, but I'm not about to give up my glass bed - or my Marlin. Really the only thing preventing Marlin compatibility is the USB interface (I wonder if it speaks raw serial?) but the glass bed is sadly a matter of physics.
How do you compare this to the solution Prusa came up recently? the load cell?
Will this work with sonic pad?
will this work with kamp in klipper?
I bought the Tap mod a month ago, printer is still in progress and now I'm sad.
Why cant something be launched on Marlin first? 😢
Why would the mainboard matter if it's connecting to the Pi or whatever other SBC you might be using? Isn't the secondary ECU doing all the hard work? What does the mainboard have to do with the process?
When printing, the system would have to add/subtract the interpolated Z axis correction many times per layer **. No 8 bit board has a high enough CPU clock to do this in a timely manner.
** makes me wonder why we're not doing Z axis correction in the slicer?
I wonder if some sort of LiDAR scan is possible in the future
Of course. Isn't this what bambu is doing?
@@travistucker7317 Oh, ha, did not realize :) The future is now! Thanks
@@travistucker7317 no, Bambu lidar isn't real lidar and it doesn't actually do much.
@@urufushinjiro the idea is out there now, so it's a matter of time. Not sure how cheap you could do it
Very cool. But I’m still trying to learn how Klipper doesn’t recognize my M420 command and how to get it to load a mesh before every print.
BED_MESH_PROFILE LOAD = default
@@Autobot-j8k Hmm, thanks. Not really sure where I type this in though.
@@c0mputer in your START_PRINT macro
@@urufushinjiro thank you, I’m learning about macros now.
I'm just upgrading to a canbus board to reduce my wires run to the head, could this be run to the CANBUS board rather than a home run directly to the main board?
Their documentation only states that it needs to be plugged into a single board computer, which a CANBUS board is not? Or am I mistaken? The CANBUS is only an MCU board correct?
Found my answer, haha. The USB-C on the CANBUS is only an input, not capable of bidirectional usb transfer trying to couple with the 4 wire CANBUS 😞 disappointing as running a 4 wire, and a USB cable is not ideal, but will make that sacrifice for this scanner 🤷♂️
wondering what the limitations are to resolution? any degradation in consistency or accuracy if you set the resolution to something insane, like 200x200?
edit: actually just realized that still requires drawing 200 lines across the bed lol. maybe 200x10 is actually what I was imagining when I wrote that.
Klipper current limit is 100x100 but this does it just fine
This resolution looks high enough to compensate and print OVER an existing layer...