PEI is surprisingly strong. The only damage i was able to inflict was from the hot nozzle itself. Also my metal scraper was treated with angle grinder so it's de-sharpened and rounded over.
@@KaminarisTech ok I’ve heard this is a problem with GaN mosfets in motor drive topologies because there’s no parasitic diode across the drain to source will be interested to see how your drive goes
Thank you for this video, I really enjoyed it. One thing that perhaps can be expanded upon is the types of probes relative to the build plate size. Most probes work on the large beds you were demonstrating. No so good on a 120 x120 bed due to large offsets because certainly in the case of the Voron 0 or similar you can only get to probe two thirds of the bed. The new klicky and it`s cantilevered predecessor will score here. Also perhaps the new E3D PZ, I am waiting for one to arrive to test. For example I spent months developing a condensed tap for the Voron 0 to achieve full build plate cover. Anyway, many thanks again for an interesting video, well done
I do an adaptive heightmap for every print at fast speed after everything heats up. Takes very little time. Eddy has been great, my Bltouch was old and losing accuracy.
Very cool project. Loved the tip about using a ferrite on the gate. I was going to say please make the psu video, but it looks like you already did haha
in 15 min you cleared my all the doubts , for which i was dancing around for last 7 days very specificaaly , which firsm where i ca use aloneg with which controller thankk you very very much
Hey @KamiTech 😃Thanks for the Video! The Klipper Config Link doesnt work anymore. Can you refresh it please? Im working on an VzBot with Trippl Z and Octopus Pro myself 🙂
My Bambu P1S seems to know how to get the nozzle clean and the probe with the nozzle and some sensors under the bed. So far it has given me perfect first layer on every print. On my Sapphire Plus I have a home-made servo + optical sensor probe and on my Ender 3 I have a home-made modified unklicky style. Both are ok, but nowhere near what the P1S delivers. And despite lots of attempts tweaking my Klipper startup code, neither is getting its nozzle anywhere nearly as clean as the P1S before the actual print starts ...
I use Voron tap. 3D printed. Wiping the nozzle clean on the build plate like on the Bambu;ab printers, first layer is always perfect now. I use a nozzle wiper with bucket too. I used clicky before and hated it. Z offset was always a hit and miss.
If you think BTT eddy is good boy have I got news for you regarding Cartographer 😂 might wanna get one and try it and you'll have no opinions on BTT Eddy.
You could always use a pair of regular touch probes on either side, if they fit. And maybe touch probe accuracy could be improved if it gave analog readings over an interval of time with small timesteps, that could be processed by klipper? Bit of Machine learning maybe? Can't hurt
You can put your build plate on adjustable stilts. Ditch the springs, replace them with a nut that locks the bed to the top of the bolt. Then put two nuts around the bed carrier. You can adjust them once, then jam them against the carrier and never need to touch them again. I use nuts in flat large printed thumb wheels. Springs are bad because they respond with vibrations. Furthermore every time the spring oscillates, it nudges the bed adjustment nut just a little to unscrew itself. You still want a bed mesh to compensate for bed shape weirdness which might well be temperature dependent, but you might not even need a probe unless you have several print surfaces that you swap in and out. Wanna know what i use for a bed probe? For a time i had a microswitch button probe that i attached to the carriage magnetically manually; now i use a 9g servo which swings a leg, on the leg is a mouse microswitch. I used to use a 3.7g servo but i broke its gears. I currently use it mostly to balance out the Z axis dual independent motors, and to home because i no longer have a spot to mount the Z endstop. I didn't want to pay the price of a good touch probe, and i think the microswitch is better anyway. I actually don't want inductive since i use glass plates and will probably be using all sorts of clip-on bed type things.
Using bltouch/crtouch since they exist.. and even before that I made my own touch probe using a pin and optical sensor.. 0 problems, repeatability in range of 1 microstep. bltouch had minimum problems with heated chamber, but crtouch doesn't.
Been thinking about building a voron, but I was surprised there isn't a design for trident nor 2.4 for a bed loadcell design like the way bambu does it.
Thanks for your interesting take on this tricky subject. Here's my input: I have a Voron 2.4 with CNC TAP. It's reliable, in fact it's pretty close to Fire and Forget when it comes to printing. What it isn't is fast!, but that's not a problem for me. I was going to change (I won't use the term upgrade because it might not be) to a Cartographer (see below for my Cartographer experience). You can change your PRINT_START so that the nozzle temp is set at 150 before bed meshing - there is a either a cool down time or a warm up time - time for me is not an issue. I haven't done this change, so mine probes the bed with the nozzle at print temp (260 normally). It has not marked the bed (over 500 hours of printing). It does leave little blobs/snots at the probe mesh points if you print PETG, but they can be scraped off. Cartographer - I fitted a Cartographer (USB) to my VzBot 330 AWD, it didn't play well! I think it was down to the magentic sheet on the build plate. On my Voron there is only one "field" or band of magnetism, on the 330 there are many bands of magnetism (N/S/N/S/N.....) (See my short - ru-vid.comzoDxIrudB4w). No matter what I did, I could not get a reliable first layer. So then I bought a Beacon H - this was working fine, good reliable first layers. But, last night after about 5 hours of printing the first layers all went to hell in a handbasket. I tried recalibrating the Z probe - that didn't work. I don't if the Beacon got "cooked" slowly over the 105c heat from the bed. I'm going to try it again tonight to see if the stability has returned. If it has, I'll subject it to another 5 hours of ABS printing and see if the problem arises again. My experience with the automatic calibration of Beacon did not end well, I had to go back to the beginning and do the manual calibration. For my 330 I'm probably going to spend a lot of time getting the bed parallel to the frame, this will mean ditching the silicone stand offs and replacing them with a mechnical "bottle jack" type of device which would give the ability to raise/lower the bed corners individually. The bed on the 330 is seriously thick (8mm) so it should be stabile enough.
Nice video, very explainable! I have a problem with the rods. I convert an x5sa pro to a vzbot330 awd. Did you cut shorter the rods?? i tried to install them and they are pretty longer than it should be.
I do not have a metal build plate, so probes like the Eddy don't work. I print on a plate of FR4. Why? - I only print high temp materials, mainly ASA, and ABS so I don't want to rely on a magnetic surface which will loose it's magnetic properties after a few weeks printing on a 100 degree or above bed. - It is dead flat and much cheaper than a machined alu slab. - Is does not expand or contract when heating and cooling down. - Because FR4 self releases the print completely after a bit of cooling down, I just take it of the bed. No need to touch it with my fingers.(and no glue stick required) - Because I never touch the bed and FR4 is such a stable material, I never have to reprobe a bed mesh, I'll be printing in a few minutes (yes, it heats up to 240/100 C in 3 minutes) I have used a servo deployed probe for many years and that worked fine most of the time. Only issue as the servo not deploying 100% sometimes. Since a few years I use a 3D touch probe which serves me well.. on *any* surface.
Indeed eddy probes won't work on non metallic surfaces but It never occurred to me that someone would try this without metallic liner. Because in case you don't know, as long as you have something ferromagnetic beneath FR4, eddy current probes will work. You do NOT need magnets anywhere in the bed, just ferromagnetic metal :) Unfortunately: - You do not need to rely on magnets with powder coated PEI spring steel either, I just used clips for quite some time. - FR4 being "fiberglass-reinforced epoxy laminate" expands and contracts similarly to aluminium, it's thermal expansion coefficient is somewhere around 14-17, its nowhere near granite or cast iron - PEI does not need glue stick either, never in my entire 3D printer career I have used glue on build surface - FR4 will eventually bend like a banana, it is only a question of how long will it take. This is a common problem with it if you take a look at old BGA electronics repair like PS3 PS4. At some point it's impossible to resolder GPU/CPU back because circuit board is permanently bent.
@@KaminarisTech Thanks for your reply. I don't know about Inductive probes reaching through a thick FR4 sheet? Coul work, but just touching it with a probe means you could switch out different plate thicknesses without recalibration (but I don't really do that) Of course you can also use clips on a PEI sheet. But these sheets are so thin that ABS could probably pull it up when there is no magnetism. FR4 might expand a bit when heated up, and that is probably why parts release when the plate cools down, but it does stay totally flat. (I believe I use 2mm thickness) I did not say that PEI needs glue stick, I only said that I never use glue stick on FR4. About the bending of FR4; PCB are thinner thus less glass fiber layers. They're also bonded with copper traces which have a different expansion rate. In an electric device, the components at the top heat up a lot, so that could all contribute to warping. None of that with an evenly heated FR4 sheet. I'll defenitely stick with printing on FR4, works like a charm for me
@@winandd8649 For inductive probes FR4 simply does not exist. Unless its one with copper layer. Yes it is not immune to nozzle change. As for FR4 thermal expansion i suggest you check out that phrase in google. It is eventually going to be a banana, normal electronics don't heat the "PCB itself" to 80deg.
I just ordered an eddy after having a horrible experience trying to set up my 2024k1 max which dug holes into 5 bed plates and knocked 2 nozzles loose, and bent the x rod trying to get through the initial calibration. One of the belts was loose and another was rubbing out of the box. Creality sent me some replacement strain gauges, a belt a bed plate and some boards but I am traumatized by the physical probing. I'm hoping eddy can solve this issue!
Beacon contact puts it into s tier, but doesnt work well with g10 plates due to variations in surface versus the underlying spring steel. Contact allows you to switch to nozzle touch when neeeded. All you need to do is run at 140c nozzle and you never have to clean the nozzle. BTT should implement contact also and I'll be a customer.
Unfortunately imo it is not S-Tier, has the same drawbacks as strain gauge. With added benefit of quick height maps. S-Tier solution would be something that is truly hands free and works on all surfaces.
My favorite combo so far was a v2.4 with a properly flat bed and physical probe like klicky, quickdraw, or bltouch. It's incredibly reliable compared to anything else I've used. I've worked with 7 machines on similar setups, and they're incredibly reliable. Currently running a cartographer on one machine and omron inductive probe on the other (no kinematic bed/gantry mounting), and they're nowhere near as reliable.
Carto works great for mesh leveling, but z offset can be iffy. I'm going to swap to a nozzle endstop like on the v2 for now, but I'm hoping the nozzle tap feature works properly, like on the beacon.
Thermal drift kills most non physical probes. Even with a heat soak macro that read the carto coil temp to make sure it meshed and homed within a 5c range and running the temp compensation macro, it still drifts a bit.
Indeed, kinematic bed + physical probe seems to be best accuracy and reliability wise as z-endstop. Unfortunately they suffer greatly from axis twist and they are very slow for heightmaps. I wish to try cast iron bed one day :D
For Nozzle cleaning, I looked at how Bambu Lab printers do it. They perform some strange movement, but the more interesting part. They Heat the Nozzle to 250ish no matter the material and just let it droop out the filament remaining in the nozzle. I applied this to my print start and the results are amazing. It takes 2-3 minutes of oozing, then the nozzle is clean and can be used for homing.
Here is the thing. Bambu have super short melt zone compared to like the Goliath hotend which i have and it drips for a LOOOOOOOOOOOONG time :) Still, 2-3 minutes is pretty long time to wait just for z-homing.
I don’t care about how long it takes. It has to be accurate. I use tap and a Bambu hotend on my Voron. I clean the nozzle with a wiper and then clean and cool down the nozzle on the build plate while going back and forth 10mm with z-0.2mm to scrape over the build plate. No z-offset issues anymore.. I tried the inductive probe then clicky and tap is the way to go. I use 3 measurements for each point.
Beacon is pretty good :) especially now that it has contact. i've never had better first layers. Not to mention if you go full contact strong magnets don't matter :)
I was shocked when I used Beacon Contact the first time. I was like 'There's no way this just works... holy crap. I didn't have to set an offset. I didn't have to do ANYTHING. It just... worked.
При таком напряжении питания следовало бы посмотреть в сторону интегральных преобразователей сетевых работающих в диапазоне 110-220В, например Viper. TOPswich или кучи аналогичных. Есть уже готовые китайские модули, а можно было вообще разобрать китайский сетевой адаптер .Лучше применять трансформатор, а не дроссель, ибо в случае пробоя мосфета преобразователя можно сжечь сигнальную часть контроллера, плавкий предохранитель не успеет отработать. Плохо, что в вашей схеме с TL494 нет токового шунта, это опасно.
Yes I am aware that flyback with transformer would be better but again, not enough space. I did made version with OC protection with 10mOhm shunt and INA213
Second this thought. Halfway through the video I decided to subscribe and queue a bunch other videos. Dude didn't only share very insightful thoughts, but I had fun listening to them. Cracked a number of laughs as I identified with a lot of the issues listed.
Interesting comments, smiled at the brass brush nozzle cleaning, as been there done that years ago, and it really sucked, because the ooze set on the “bristles” and caused mis-steps. Shame you couldn’t get a Beacon, as I was waiting for that one. Good work on the BTT Eddy 👌
In my case brush also didn't do anything with petg, nozzle just grabbed chunks of it anyways and stringed away, dragging giga blob across build surface 😂
I have tried different inductive probe. Kicky and TAP. So far for me the TAP is the most precise one. But I had to change the X rail and the mini TAP rail for medium preload rail to get consistent. But yes it is bulky and heavy.
really cool! could you also make a video on how to configure the software for correct voltage and current readout? also on how to adjust the deadtime on the bride?
I'm surprised that tl494 can properly regulate the output at these high in/out ratios, without any tricks to increase the duty cycle like a tapped inductor. Also a bit of brute forcing took place - high ratios are not kind to components.
The IR2181 in combo with the MMBTA92 is an interesting PUSH-PULL design, although a bit wasteful of the IR2181. Something like a FAN7171 might have been a better choice. The IR2181 is very common, the FAN7171 less so, but often cheaper.
That is the thing, there are probably numerous better drivers but if they are not common i'd rather take more expensive but widely available part. I'm tired of parts that has to be constantly switched as they are not available.
@@KaminarisTech I understand, but IR2181 is ONE large manufacturer ONLY (Infineon), and so is FAN73711 (FAN7171 was a TYPE, sorry) from OnSemi. Both are parts around for a long time, it's just that FAN73711 is used far less often.
Very cool! With the startup circuit and the mosfet driver this design might be adapted to any regulator, not just the TL494. But why do you need 3 amperes? It seems excessive. As an alternative, i would suggest you again look into high-side buck regulators, which are technically AC/DC regulators, but all AC/DC SMPS use DC inputs in practice (rectified mains AC). For example LinkSwitch-TN by Power Integrations, MP157 by Monolithic Power Systems, or AP3928 by Diodes Inc. They are more common, simpler, and some of them work with low input voltages. But why is your input voltage as low as 32 volt anyway? Regarding SG3525 mentioned, it does not support 200 volt, nor can it drive a high side N-fet, i.e. some kind of mosfet driver is still needed. Anyway, good luck!
I do not need 3A, I mentioned it in video i was aiming at 1A. The absolute minimum for vesc is 500mA which is a bit too close for comfort if using built in bluetooth. I have not seen LinkSwitch-TN but i did see AP3928 and the problem i see with all those AC-DC switch controllers is that none of them can, at least on paper, go lower than ~80V AC. I have also not seen a full schematic of DCDC using one of those chips but if you have it, please do share. 32V was not a hard requirement either. I wanted to fully support at least 16S batteries which are 40-48V fully discharged. So it had to work a bit lower than that. 16S configuration is commonly seen in high-powered electric bikes and scooters
