Відео

warm fuzzies ❤

I wasn't looking for the laboratory maximum speed. I wanted a reasonable upper end I could design for in a specific use on a robot.

@@WhitmanTechnological Ok, nice video you had. I have been playing with steppers for years and I accelerate them by a totally digital ramp generator of my own design. It uses binary rate multipliers, up/down counters and comparators. You can program in the acceleration, initial (starting) speed and final speed. After that (keeping the accel constant) you just output new higher or lower speed and you get the stepping frequency for that.

@@MrJPI That sounds awesome.

I use FreeCAD and Blender, and they're both pretty lightweight to run. The need for a better computer really only comes when you start making massive and complex things. As a design gets more complex in FreeCAD it'll take longer to compute the updates - and a faster CPU will help. In Blender, as you get more sophisticated, increase the output quality, add more detailed textures and render at a higher resolution, it'll put your graphics card to work. I upgraded my GPU from a GTX960 to an RTX2070 Super and it cut the time to render video sequences in half. At a minimum, if it has an operating system and boots up and was manufactured in the last 15 years, there's a good chance it can run FreeCAD and at least an older version of Blender. I have a few old Core2 Duo computers I built in 2007 that still run and they'll happily run FreeCAD and Blender. It'll just take a few decades to render anything in HD! www.blender.org/download/requirements/ forum.freecad.org/viewtopic.php?t=16701

Well, let's see... I have some gears right next to me printed in nylon from a gearbox that I don't need anymore. I put some dabs of "Gorilla Super Glue Gel" on one and stuck another gear to it. It's the super smooth sides that were against the glass during printing, so it should give the least adhesion vs a different size. In the time it took me to write those last sentences, it set and I can pull with most of my strength and it doesn't come apart. Seems like it's safe to say that super glue will happily adhere to the Taulman 645 these gears are printed with. I applied some uneven force and the gel broke down the middle, so it sticks harder to the nylon than it's own structural ability. Should work like a charm for gluing to your axle (as long as it also sticks to the metal rod). You might do that baking soda trick if you want to add more structure to the joint, too. Most of my gearboxes are either compound planetary gears or the outputs are not uniform/have mounting holes. I haven't tried attaching directly to an axle.

Thanks for the tip on RS-485. I'll read through those specs. Also thanks for pushing those impedance numbers. I assumed because it's Ohms that it was synonymous with resistance. Since you took the time to correct me, I googled it "how to measure impedance of a ribbon cable" and now I'm up to speed. ☺ After uploading this video, I found that the USB hub supplying +5V power for the controller board has a 0.3V wobble at the same frequency as the wobbles in the green signals in this video. I'm going to try to smooth out the input voltage, and hopefully that'll reduce the amount of noise at the source of the signal, too.

I looked up CANbus. Interesting... I'd need to add some chips to the controller and sensor boards to support it. It does look like it'd be resilient to noise, since it signals both high and low. Thanks for the tip. :)

I'm using an A4988 driver in this video. I'm not driving the motor anywhere near it's theoretical maximum. I was testing for a use case with a relatively high torque load where I wanted to be able to reliably begin spinning at max speed with minimal ramping. Most stepper drivers can probably go much, much faster than I do in this video. As other commenters have pointed out, ramping up the speed instead of just starting from a dead stop will get the motor spinning a lot faster. Higher voltages can also help overcome the inductance of the motor coils as the speed increases, too. ~300 RPM is the fastest I want the motor to go when I designed the gearbox for a robot wheel system, but the motor/driver could easily double (maybe even triple?!) that if it was completely unloaded and I ramped up the speed more gradually.

It's been almost a year since my last video on this channel. In that time I started fresh with the fourth iteration of the robot. It's been a long adventure, mostly fun, with one fairly major roadblock that took me a long time to identify. It's a big step forward from the last Nephele video with the robot kinda moving on its own. I picked materials more carefully for the gears and structure. The roadblock that held me up for a long time was the ringing in the ribbon cables between the microcontroller and the analog-digital-converter chips embedded within each joint. I was getting a bunch of bad data from the ADC chips, and only through months of trial-and-error, research, and careful review of the oscilloscope did I eventually discover that the high-low signals are ringing in the ribbon cabled like a bell. The ADC interprets the fluctuation of the voltage from the ringing as extra high-low pulses, and sends back more bits of data than the microcontroller reads. It was a brutal but fascinating lesson to learn, and I think I have a fix for it. With any luck I'll have the hardware for Nephele 3 working in the next few months - and I can post a new videos. I have about 100 GB of raw footage from the last year, but three months of "I don't know why it's not working" and 20 hours of cutting nearly identical iterations of the same circuit board on my CNC router isn't enough substance to make an entertaining video. Lots of progress, but no functional milestone worth bundling into a video just yet. Thanks for asking :)

I saw someone else with a laser X on their router and thought it was cool. I designed my own router, and I wanted to know that I could play with the open source grbl software that was controlling it. The lasers make an X on the surface where the bit will plunge. I spliced my own code into grbl to toggle a pin on the microcontroller, and I wrote it to listen for M61 to turn the laser on, and M62 to turn it off. Just a little excuse to play with lasers and familiarize myself a little bit with modifying the router firmware. It's not even remotely useful, but it looks cool. XD

I wrote the height probing and Z coordinate correction software myself. I wrote my own program to open gcode files and stream them to grbl - the open-source firmware that runs my router. So as I worked through getting reliable boards, I streamlined the software that controls the router to formalize the various steps I needed to do. My toolchain is KiCad to spit out Gerber and drill files > FlatCAM to convert the Gerber into gcode > my custom software to probe the copper blanks for the area where the traces will be cut and then the gcode is modified to include a Z coordinate for all the cutting moves > this modified gcode is streamed to the router which runs grbl. If I didn't perform the z correction, I would break bits and have unusable boards.

@WhitmanTechnological awesome and thanks for taking the time to respond 😁 I have an industrial CNC but it's not in anyway worth trying to use for PCBs so planning to get a small hobby type one that runs GRBL but wasn't sure if the machine itself had to be able to do a height map or the software I.e. flatcam. I guess the machine is capable but I guess I wasn't sure if the map is done via the software so machine firmware etc is not a dependency

@@LucidFX. Nice dude! Grbl has a command to probe on an axis (G38.2 Z-6 F20 is what I use - probe up to 6mm in the negative z direction at a super slow 20mm/min), and when the probe triggers, grbl will output the machine coordinates at the point where the trigger happened (looks something like "[PRB:67.86,49.23,-46.875]"). I pull the Z coord from that, and then measure the difference in z coordinates across a grid to generate the heightmap.

​@WhitmanTechnological awesome. I think I'll pull the trigger on it then😅 thank you again 🙏

Yesdooood!!! There goes all your free time! If you're comfortable with C#, I can post my height probing code in the comment here as something to start from on your projects.

To get the router to make incremental passes that cut out the teeth nicely, you need to select a face that has the teeth edges as either the inner or outer edge. At 15:24 you can see I select the top of the gears and then use the Profile with the outside of that face to get a path that cleanly cuts the teeth and out. And then around 18:52 I select the face inside the rings and use that outer edge to profile the inner teeth. Profile will compute the path that will precisely cut out the inner or outer edge of whatever face(s) is(are) selected. Make sense?

Thanks,, funny I can try dozens of times without success and one minor change it it works easy and perfect. Have a great day.@@WhitmanTechnological

That just made my morning, lol.

I sure hope so. I've been making stuff with it for three years!

A younger me would have jumped at that. I've always loved helicopters. I'm mired in the robot project that most of my other videos are about and won't have time to make a video to do it justice, sorry. The swashplate is a pretty cool piece of hardware. Sleeved onto the rotor so it can slide up and down, and then ball jointed from the sleeve so it can pivot. A lower plate connects to an upper plate with a bearing to allow the top to rotate while the bottom is fixed and both plates describe the same plane. Upper plate has four ball joints that link to the leading edges of the rotors. Lower plate has two or four linkages on ball joints that connect to servos or cables to control the whole thing. An amazing design, and pretty incredible to see the rotors changing their pitch so heavily during a single rotation. I read the FAA Rotorcraft Flying Handbook once, and the part about the flex from flapping and feathering - where the rotors are curving up and down dramatically as they transition from the heavy-load-into-the-wind half of their rotation to the unloaded-away-from-the-direction-of-travel half of their rotation was something I hadn't really thought about before. Amazing those things can fly so smoothly at over a hundred knots considering the wild change in forces on the individual components over the span of a single rotation.

@@WhitmanTechnological sir it’ll be slow and steady you can create something like a course where people pay to learn all this complex design you’re good and I think it won’t take you long time to do because of your experience please think about it sir

That's awesome, thanks for sharing. Nothing beats the turnaround during prototyping of being able to make your own circuit boards. :)

Thank you, Dan! :)

I'd put in the notch in FreeCAD and then 3D print with the notch. Lately, I'm doing regular spur gears (not helical) so the gears all slide together easily. For a helical gearset, I'd make the notch either by subtracting a thin cube in the Part workbench, or pocketing a thing rectangle in the Part Design workbench. Congrats on your new printer!!

@@WhitmanTechnological Tally ho mate! Much obliged good sir!

Thanks for the info. If I do some manual controlling of steppers in the future, I'll write it up to trigger pulses instead of toggle the pin. Woot!

For the mechanics of planetary gears, there's an awesome UA-cam channel - youtube.com/@geardownforwhat - where they make all kinds of gearsets and lately a bunch of other cool stuff. That's where I got introduced to planetary gears. There's a few really good FreeCAD UA-cam channels with tutorials. I've seen youtube.com/@JokoEngineeringhelp mentioned a lot on Reddit as a channel with lots of tutorials to learn.

Check the pinned comment at the top. I have a couple versions of code to pulse a stepper driver that you can copy/paste.

So the first one is the pocket where it's carving out all the material, and you select the bottom of the pocket for that operation. After bulk removal, I think you want to do a Profile operation instead of a second 3D Pocket. You might be able to select just the bottom face, or you might need to select the vertical walls. I'm pretty sure there's an option for outside or inside - you'd probably want inside for the inner walls of a pocket. You can set a tiny step down to get a nice surface finish. I'm pretty sure that's the one you want. I'm travelling so I can't test it. Let us know how it goes!

@@WhitmanTechnological Dear Sir, Thank you for getting back to me. Here is what I'm trying to machine. Picture half of a basket ball sitting on it's apex with the open end up. The walls are about 10mm thick. I need to rough out the inside material, and then come back and smooth off the inside walls. I don't see where I can send you my FCStd file, or I would. I'm in no hurry for a solution to this problem. Thank you again.

@@kb3cxe Back at home now, I can hop into FreeCAD and set up something like you describe. You do a 3D Pocket first to carve out all the material. And then you do a 3D Surface operation for your finishing pass. Whereas the pocket expects there to be material to remove in multiple passes, the surface assumes all excess material is already removed. Does that work?

@@WhitmanTechnological Thank you for all your help. I'll give it a go, and see how it turns out. Thank you again. Roland

@@WhitmanTechnological Dear Sir, Thank you for helping me out on this. Where do I find the 3D Surface function? Thank you again. I'm using a 2022 MacBook Pro, OS 13.4, FreeCad 0.20.2

Picking up Moneylender like a chad! You just create a loan like you would if you are the lender, and for the borrower you put yourself, and for the lender you put your investor. Everything else is the same. Either click the New Loan button or from the main menu choose Loan > New Loan > Loan. When you run reports, you can choose a specific lender when you only want to see loans payable to you or loans payable to an investor instead. When you pay your investor, you record payments on the Payments tab, and Moneylender will do all the math for you.

I feel your pain. The button is on the right side, 7th button down. "Filled Zone" it's called in PcbNew. Or just Zone on the Place menu. Congrats on making your first board in new software - that's always where I hit the most hurdles.

A good solution. I bought some aluminum blocks and I want to try an undercut bit to make a special clamp just for the little 70x100 copper blanks. Hopefully, with a nicely undercut ridge that covers the majority of the perimeter of the board I can clamp in and out really easily without having to worry about bending anymore. I have some big sheets of 2-sided pcb blanks (like 14"x14" or something). I'll have to try your method when I attempt something on those.

I rewatched the part of the video with the acrylic to see what the context is for your comment. Seeing the old 2x4 router in action made me chuckle. I learned so much by making it. Only recently have I been brave enough to run my router at less than full speed, so I'll take your advice next time I'm doing acrylic and cut the router RPM and feedrate. I have a little air compressor that blows off the chips now, so that'll probably help keep things cool and clear of the cutting surfaces, too. I appreciate you taking the time to share your experience!

Sorry for taking so long to reply, I didn't notice this comment when it came in three days ago. I saw a marketing video for a company that sells a mini CNC router that's specifically for endmills (edit: for PCBs). Their process has you milling the traces first, and then your apply an epoxy solder mask on top of that. Then the router delicately mills just the epoxy off of the top of the foil (edit: where the solder pads should go). Given how warped the blank boards are and how very thin the copper cladding is, I haven't attempted it myself. I'm using mine for prototyping - to figure out how the circuits need to work, but I'm going to be ordering a small batch of boards from JLCPCB once I know the wiring is correct. I priced out my little "smart bone" boards, and it was like $6 with two day turnaround. Would take me almost as long to mill six of them myself and I'm just doing single layer boards. So the rumor is that you can gently carve off a solder mask with an endmill. I don't think I'd want to do it, though. With a layer on top of the copper, I couldn't height probe successfully. If the cladding isn't at least 0.2mm thick, it's be pretty dang hard to clear the pads without erasing the 0.035mm foil unexpectedly. A photo-cure epoxy might work with a stencil, though. So that's why I haven't explored doing it personally - it seems unlikely to succeed with my equipment.

This project worked out pretty well. You said it, the temperature makes a massive difference in etching time and the quality of the finished board. When the basement was colder my results were soo much worse. Warmer days gave really clean traces. After I did this video I built a CNC router and have been trying to get a foolproof toolchain for making circuit boards with "mechanical etching" - routing out the traces with engraving bits or tiny endmills. I blew about $400 on endmills as I broke them over and over again. Happily, I finally have the whole process down to a science. Just in the last week I've made three different circuit boards for different things, each one came out flawlessly on the first run. I can start with a blank KiCad project and be holding the circuit board ready to solder in about 40 minutes. It's super awesome. My latest video chronicles some of the lessons I learned, and I'm about to start editing a new video with a few more of the problems I uncovered and resolved subsequently. It's been a process, lol 😅 Thanks so much for taking the time to share your ideas and I really appreciate the positive feedback. It's good to meet you 😊

@@WhitmanTechnological Your videos are definitely going to save me $$$ in broken bits!!! Keep up the great work please!

You could have the ring gear on the second stage link into the sun gear on a third stage. Hold the third stage ring fixed with the first stage ring, and your output would be on the 4th stage ring. But you can get gear ratios like 300:1 from just two stages if you choose the numbers of teeth appropriately. I used multiples of four (since I have four planets) for all the tooth counts so that the planets are identical. If you use multiples of two for tooth counts, you'll need two kinds of planets (and they'll have to be aligned carefully when the gearbox is assembled). If you use arbitrary numbers of teeth, each planet might have to be different, and every planet, sun and ring will need to be oriented properly when the gearbox is assembled. It makes the planets and assembly harder, but you can get astronomical gear ratios from just a two-stage assembly if you can tolerate the extra complexity.

You'd count the number of pulses, and stop sending pulses when you hit the desired amount of turning. Adding a target step count, counting the steps, and stopping the pulses when we've hit the target. We'd wrap that into example #1 something like this: int stepPin = 5; int dirPin = 6; long stepsPerRevolution = 200; //how many steps to turn the motor one time. long targetRotation = 30; //Number of revolutions to turn long totalSteps = targetRotation * stepsPreRevolution; //How many steps we need to take to rotate the desired amount long stepsCounted = 0; //Track the number of steps we've rotated the motor unsigned long currentTime; unsigned long nextStamp; bool isOn = false; //pulseDelay = RPM * steps/rev / sec/min * micros/sec and divided by two because each step requires going high and then going low (two toggles). unsigned long pulseDelay = 100 * stepsPerRevolution / 60 * 1000000 / 2; void setup() { pinMode(stepPin, OUTPUT); pinMode(dirPin, OUTPUT); digitalWrite(stepPin, LOW); analogWrite(dirPin, LOW); currentTime = micros(); nextStamp = currentTime; } void loop() { currentTime = micros(); //Add a check to stop rotating once we hit the desired amount of rotation. if (currentTime - nextStamp >= pulseDelay && stepsCounted < totalSteps) { //The pulseDelay has elapsed and it's time to toggle the pin. isOn = !isOn; digitalWrite(stepPin, isOn); //Toggle the pin. nextStamp += pulseDelay; if (!isOn) stepsCounted++; //When the pulse turns off, we count it as a step. } }

@@WhitmanTechnological Thanks for your time again. The second code worked for me. But still I could not control the number of revolutions using the same code. Can you help me do it?

@@sarahnasser6603 Did it stop after 30 turns? long targetRotation = 30; //Number of revolutions to turn That should be telling it to turn 30 times and then stop. You could make that a 2, if 30 rotations is too long to wait. I don't have a stepper driver and stepper handy to test the code, but my reply above looks like it should work properly.

@@sarahnasser6603 there is a typo in there: long totalSteps = targetRotation * stepsPreRevolution; //How many steps we need to take to rotate the desired amount stepsPreRevolution would need to be stepsPerRevolution for it to compile. I'm sure you fixed that already.

@@WhitmanTechnological No, the first code haven’t actually worked even after fixing the bug, the motor didn’t start at all. It makes sound as if it’s working but not rotating not even with a push. However, the second code worked well, the motor rotates. In my project, i need it to rotate 30 times and then have a 2 seconds delay. If you don’t mind, can you help me with that or just further explain the code and what is considered a revolution. Thank you so much for your help so far

This part here: //pulseDelay = RPM / sec/min * micros/sec and divided by two because each step requires going high and then going low (two toggles). unsigned long pulseDelay = 100 / 60 * 1000000 / 2; The RPM is set to 100. If you wanted 40 rpm, you'd change it like this: unsigned long pulseDelay = 40 / 60 * 1000000 / 2;

I just noticed this is missing a piece, the steps-per-revolution of the stepper motor. unsigned long pulseDelay = 100 / 60 * 1000000 / 2; It'd be more like: //pulseDelay = RPM / (sec/min) * (micros/sec) / (steps/rev) and divided by two because each step requires going high and then going low (two toggles). unsigned long pulseDelay = 100 / 60 * 1000000 / 200 / 2;

A quick search will show how to run the wires for an stepper driver like this: www.google.com/search?tbm=isch&q=stepper+motor+wiring