I actually made that little extension cord relay breakout many years ago for a heater, which (unlike my fridge pump) had no other points exposed. I think my reasoning was if one of those wires was to be touched accidentally, I'd rather it be neutral. When an appliance is connected to the output end of the cable, bridging the two pins underneath the relay (like I did with my fingers it seems) would be a similar shock regardless of which wire was broken out, and touching just one of the pins would be a 50/50 chance of shock, also regardless of which wire was broken out. When there is no appliance connected on the output, the chance of shock is higher if the live wire is used on the relay. It was a long time ago, but that's probably what I was thinking. But in the context of this video where the fridge pump has other points exposed, this arrangement causes one of the points on the pump to be permanently live, which is not cool. Perhaps this is why people are telling me that breaking out the live would be preferable? In any case, I will cover it all up so I can't accidentally touch anything, and maybe also add a second relay to switch the live as well. btw I know there is still a chance of shock from just the neutral wire only if things elsewhere have not been wired properly, so the above 'shock or no shock' deductions are only probabilities. If we assume that wiring is *usually* labelled correctly, I think the logic still is still valid.
Always switch the active wire.(Brown) This assumes whatever you are plugged into is wired correctly. If you are going to shock yourself again, can you post a few more vids in advance while the paramedics are shocking you back into rhythm :)
With the solenoid you can typically use PWM. Use a high duty to open it then you can drop down to a lower duty to maintain. This should greatly reduce the heating on the solenoid. In the past I have found 20% PWM can maintain a valve open.
There's also dedicated solenoid driver chips that do that automatically, taking the error chances out of the software side. Same PWM trick can be used in larger relays.
hmmm yeah that makes sense. With the resistor, the solenoid does warm up a bit. One time I accidentally left it on for a few hours and it was warm but not too bad, and this is probably the warmest weather we'll have all year so I think I'll leave it like that. Also, the rather low-spec F103C8 of the weenyPRU is running out of timers to do PWM. I'm already thinking I need a new PCB design so I will probably expand it and use the more capable F103RC, if so I will make sure that at least a few more of the outputs have timers. Even if I never implement this solenoid control, they might be useful as 'analog out' for other things.
With using the frig compressor as a vac pump........they are originally designed with refrigerant flow in mind. This is to say the low pressure return under normal application cools the motor and oil. When used as a vac pump, that is out of the equation and the motor will overheat if run for long periods. That plus one needs a breather pipe, reasonably long and high enough to allow any expelled oil to bleed back into the motor. As mentioned by others, NEVER switch the neutral (well, not on it's own like you did there) ....you can be almost certain that brown from a sealed cable is the active BUT electrical regs require ( apart from obvious dodgy bare connections) that portable (anything powered from a plug/lead) gear should be double pole switched. THAT ensures any possible wiring reversal (not so common these days as points are already internally wired) does not effect the outcome. i.e. turn off turns off the active FOR SURE.
Exactly my thoughts. Without reasonable recirculating lubricant solution (the oil should be on both sides of the compressor) the motor has its expectancy in hours...
@@medvidekkrupicka1404 I just have my experience with another oil based compressor - it came through the nozzle. The oil is distributing outside... happy breathing. I'd only go for an oil free compressor (I also built my own machine actually).
You should add a pull-down resistor to the base of transistor in those end stops. It is needed to properly close the transistor and to set its working point. The LED with its resistor is not enough, because there is just too much of a voltage drop on that LED (2-3V for green IIRC), but bipolar transistors often starts to open at about 0,5-0,6V. It will also decrease sensitivity a bit. Value can be in range of tens of k, you can try using same 20k as you already have. Or even use trimmer to tune sensitivity, just don't go too low because you can fry your optical sensor.
Thanks, that makes sense. For some reason I really struggle with this kind of thing, I just can't picture how things will interact. It doesn't help that mistakes or non-ideal conditions in the physical world can screw things up too. It took me four hours twiddling around with a breadboard to even come up with that very simple circuit, and it was somehow working nicely without the pullup above the transistor on the output side. So I made the order without it, only to find my delivered boards didn't work. Back on the breadboard, it now didn't work without the pullup (wtf?), which is why I had to solder the 10k there. Then after mounting them on the machine, for whatever reason (longer wiring?) they were then unreliable until I changed the 10k to 2k. Overall this kind of analog development really pisses me off, give me software any day!!
Im so glad that i found this channel. Its a treasure trove of information that the average guy actually needs without spending hours on research on the internet 🤙
For the shock you got at the relay: from a voltage potential point of view, one of those exposed pins is at ground potential, one is at mains potential (since closing them closes the circuit). Which pin is what will depend on how your country does mains plugs and the meaning of the mains cable colors for your jurisdiction. If the plug orientation is keyed like in the US, so blue could alyways be neutral and brown would always be pahse (again, colors differ by country). If it's unkeyed, like in the EU, either could be phase or neutral. If your shock came from the pump side, it would explain why it was light(ish), since the pump (or the relay before it) acts as a resistor limiting current. Later when you simulate a pick and place program (55:00 ish), while replenishing the small reservoir: this is likely rather slow because of the crushed line connecting them, having very limited throughput for air (essentially only in the tiny un-crushed parts at the sides of 'flat strip' that remains). This probably does't matter a lot (or at all) for the actual pick and place head, since that only cares about the pressure difference at the nozzle, not actual air flowing through, but it matters a whole lot more for the transfer line between the tanks. A stirdier line would likely accelerate that refilling of the vacuum in the small buffer tank a lot (obviously technically the air flows from the buffer tank to the main tank, but whatever).
Nice! I actually thought about the ambient light being a problem when I saw the gap in the sensor. However, the fact that the diode on the sensor reacted as expected must have been super confusing!
48:50 You tried to power the compressor through YOU. Although the left blue is neutral (presumably), the right blue is hot! It is returning from the compressor, which has quite low impedance so that you can enjoy a decent shock by touching the two blue wires. Your two fingers probably had 230V AC across them for a moment.
Terrific watching as usual. Reminds me of an accident I had long back when I tried using those PVC pipes as a accumulator for storing air pressure.I has used the same fridge compressor too. Would not have worked of course, unless I had a diaphragm or spring piston inside. All I ended up doing was blast the pipe with an almighty bang , luckily none of the shrapnel hit me.
Lots of exciting progress made here. Soon chips and components will be flying. FYI: An aquarium air pump can be converted into a vacuum pump. This can make for a fairly compact low-cost vacuum source. I suspect the fridge compressor is consuming more energy, and is definitely more bulky. The sound at the end is neat, more so when replayed at 2x. It's got a bit of a steam punk rhythm.
You could just use a piece of tube and a peristaltic pump rotating back and forth to get negative and positive air pressure, no vacuum pump or solenoids
Do you have experience with that? If it worked reliably in this situation, it would have some really appealing features: Cost Simplicity Completely portable "one-piece" machine Fully software configurable and controllable Able to supply a pulse of positive pressure, rather than just ambient pressure, to release the component You've got to love the idea! 👍
might be worth testing if your vacuum actually holds for a second or so if the picker solenoid gets closed. then you would be able to just pulse it every now and then to refresh until you are ready to place the part and open the second valve to ambient pressure
In theory, this machine will be so fast that the very fabric of space-time is in danger of being torn apart, so parts should only be on the nozzle very briefly. The worst case would be picking from one side of the table, going to camera and then back to the far side again, which as shown at 1:45 should take slightly over 1 second.... assuming the flyover camera works properly. But in practice, components will be placed much closer the PCB and camera so the vacuum on-time will be similar to the example shown right at the end of the video, seems like about 0.5 seconds for each pick.
PVC for a (negative) pressure sensor is a poor choice It will crack after a while. especially is you cycle it a lot. For vacuum that is not dangerous but don't think of using it for pressure. There's multiple examples of exploding pipes to be found on the internet...
pvc cement is just a solvant, it welds pvc (technicly its not a glue, thats why your threads were melting on that abs part), also did you consider a air compressor venturi effect vaccum?
I wanted a system that would be quiet, I find those venturi things so annoying I can hardly even watch a minute or two of someone else's RU-vid video using them, and then you need a compressor as well. It seems like overkill for lifting and placing parts that weigh a tiny fraction of a gram. I will say though, the venturi sound is preferable to having an aquarium pump turning on and off for every pick, those videos I'm muting after about 5 seconds :)
Warning Warning, danger Will Robertson. The way you have the relay breaking the mains supply to the pump is VERY dangerous. Blue is neutral meaning the live line is never switched off. At a minimum break the line ( brown wire) or better yet use a double pole relay and simultaneously switch neutral and line for maximum safety since it is not unkown for DIY wiring jobs to have the line and neutral reversed, particularly when fitting a plug. in regards the way the solenoid valve works better one way I suggest you look into "pilot operated valves". It is common for solenoid operated valves to be pilot operated and are not bi directional.
49:15 you can get a shock from the blue wire if something else in the house is turned on that completes the circuit back to the positive side so its not full 240v maybe 30% etc....or if someone has wired some outlets the wrong way where red is now blue etc....thats the way I beleave I got a shock also from my mucking around with my electric blanket as a kid.....Brett
Hell... that's doing it the hard way! Three year old me just poked his pudgy little finger into the gap between the plug and the socket... worked exactly as advertised! 😵💫 😱 😭
For precision parts that you need to flip consider geting a ER collet like ER40. Lathe has morse taper (mt3 or mt5) under chuck so you could get good collet and get very low runout even if you flip part in collet.
Best option of all is to leave the part mounted in the chuck (or collet), and don't disturb it until all concentric operations are complete. Even if you use square stock in a three-jaw chuck you can get accurate measurements provided you are only making cylindrical cuts, and you don't remove the workpiece from the chuck. If I can't avoid removing it, I remove the chuck from the lathe with the workpiece in it. (Not always practical) It then depends on how accurately the chuck can be refitted to the lathe.
Ideally I would have brought the stepper motor to the lathe for fit checking, to avoid unscrewing the part from the nozzle at all. But the motor was already soldered into wiring on the machine and I was too lazy to undo it.
@@PiefacePete46 i presumed that this mini lathes dont have screw on plate or pin(like in D ) to index a chuck. But has screws from behind plate to mount a chuck and backplate is intregal part of spindle. Seems most easy to adapt such collet to get something more that is easy to put and remove.
Think you might need to add an option to pause the placing when the vacuum is too low, allowing time to recharge the small reservoir? The squishy tubes might be wasting some vacuum? and come on Chris switch the active not neutral wire from the mains!
I watched the video, I looked at the wiring "hack", I left a reply confirming New Zealands house wiring colour standard... and I didn't even spot the switching on the neutral! I must pay closer attention I must pay closer attention I must pay closer attention I must pay closer attention ... 😜
I actually made that little extension cord relay breakout many years ago for a heater, which (unlike my fridge pump) had no other points exposed. I think my reasoning was if one of those wires was to be touched accidentally, I'd rather it be neutral. When an appliance is connected to the output end of the cable, bridging the two pins underneath the relay (like I did with my fingers it seems) would be a similar shock regardless of which wire was broken out, and touching just one of the pins would be a 50/50 chance of shock, also regardless of which wire was broken out. When there is no appliance connected on the output, the chance of shock is higher if the live wire is used on the relay. It was a long time ago, but that's probably what I was thinking. But in the context of this video where the fridge pump has other points exposed, this arrangement causes one of the points on the pump to be permanently live, which is not cool. Perhaps this is why people are telling me that breaking out the live would be preferable? In any case, I will cover it all up so I can't accidentally touch anything, and maybe also add a second relay to switch the live as well. btw I know there is still a chance of shock from just the neutral wire only if things elsewhere have not been wired properly, so the above 'shock or no shock' deductions are only probabilities. If we assume that wiring is *usually* labelled correctly, I think the logic still is still valid.
Yes, the controller will be able to see that vacuum is insufficient and wait. At least the information is available to do that, not sure whether future Chris will implement it though.
Ha, your lathe comment at 15:30 ish reminds me of all the times I forget about certain tools or techniques for my projects 😂 Just one question, why the optical limit switched instead of physical switches? My laser cutter has these tiny spriny little button switches that seem to work quite well.
I actually can't remember why I didn't go for regular switches... size or weight maybe, or repeatability, or because the activating lever needs to be rigid and well attached instead of just cardboard and double sided tape. It was quite a while ago.
Don't you hate it when you admire your masterpiece, made with the most complex series of operations, and then realise there was actually a really simple way you could have done it! 🤔 🤬 😡
Originally 3d printers used optical sensors for limit switches. Pretty quickly we found out there are impacted by lights outside the switch. Immediately I thought that green led will cause issues
I was wondering about the green LED while laying out the board, but was trying to keep it as compact as possible so that's where it ended up. But the sensor is infrared so I think it's ok. At least, I wasn't having any problems until that sunny afternoon.
The flag doesn't touch anything so it doesn't matter if it's fragile. The microswitch will still need some kind of adjustable lump, that will need to be securely mounted.
@@iforce2d : I took one look at what was happening, and said "Induction... something in that little unit is acting as an induction sensor!". I would have wasted hours trying to nut it out, and finally at about 8:30pm, I would have said "Yay!... Cracked it!". (8:30 sunset for non-New Zealanders.)
If it's only used to find home then it doesn't even need to be adjustable as long as it's a fixed reference point. I would most likely break a flag off when working on the machine.
I saw that relay and imagined picture of you being zapped. But its ok I was thinking, and then you said you where zapped haha. I am working on your code for weeny to convert it to CH32V307 which has eth phy. I was supprised with code that you used one timer for all stepers. I am going trought code and convert it to that riscvchip and code style suprising similar.
Yeah I was quite pleased with how that turned out, to have it all running on such a cheap and low-spec chip, and at 50kHz too. It's quite amazing what these things can do if you know how to use them. That method also means that all steppers get the high pulse at exactly the same time - probably not important in practice but that is the ideal signal to have for synchronization. My wiring is looking very messy again, so pretty soon I might start designing a new board to include the solenoid mosfets, I2C pullups, and allow 8 steppers using a F103RC with 64 pins. Or maybe even a G431 that has a 170MHz clock, which would allow faster step rates.
I was always chasing speed of pulses but this 50khz seems good enough by your example. This chip has 4 advanced timers for motors and I was thinking I will have to use all of them (they can sync) but your code showed me different approach. I will try to mimic the same 4 channels per timer. I never did this before so its great excesise.
Yeah I think 50kHz is enough for most practical purposes. I calculated the fastest movement of this machine will be only 29kHz, which is faster than software stepping (via parallel port) could do, but still easily within the capability of the STM32. Having a faster clock helps in general though, especially when you keep piling on new features.
@@iforce2d stm32f103 is very old chip. 5v tolerance makes it neat but it has some shortcomings. It cant do i2c over dma if I remember correctly(perhaps together with pulse counter). Today there is so many new chips. Even clones of that series can clock more and perhaps have some things like this improved. Even for this ch32v code is so close to stm32. If i didnt have issue with nerve in my leg I thonk i could covert it quick. So your code is great example.
I actually can't remember why I didn't go for regular switches... size or weight maybe, or repeatability, or because the activating lever needs to be rigid and well attached instead of just cardboard and double sided tape. It was quite a while ago.
I wonder why you just didn't use common microswitches like from a mouse instead of the optical ones.. did you mention somewhere why ? ..Got to watch part one again , excuse my ads ^^
I actually can't remember why I didn't go for regular switches... size or weight maybe, or repeatability, or because the activating lever needs to be rigid and well attached instead of just cardboard and double sided tape. It was quite a while ago.
@@iforce2dThe only reason I could imagine is if you want to measure two (or more) positions on one axis.. have two alu strips pass one optical switch.. (or the other way around.. two sensors one alustrip) I actually cannot imagine how to mount that though .. like with a bridge ? but what would it be good for ? .. I wish you rainy days for prototyping then ^^
java Well ok, not just that. I'm interested to try making something myself. OpenPNP wants to micro-manage every movement, which is somewhat antithetical to how LinuxCNC would be best used by calling higher level subroutines. It also does not support the remote camera servers that I'm planning to use.
Well, the actual making of software is terrible for video content though. Just look how boring this one turned out :) On the other hand, the result can be fun to watch. Here are some pure bezier spline (constant jerk) movements, no LinuxCNC or anything, just my own program running with realtime priority. ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-nj7-fPy4DGg.html Since I understand 100% of everything happening there, it's tempting to go my own way entirely, even knowing it will be a lot of work. LinuxCNC brings a lot to the table but it wasn't really made for this use case, and learning the code base will still be quite some work. I'm really struggling to decide...
How did you get shocked? The motor is an inductor, therefore, all the wires in the motor completed the circuit and if you but your finger on both ends of the blue wires then you would get shocked.
Can't believe I'm still watching the Video after 27 minutes... Must be interesting.... 27:00 You would be better off using an Inductive or Hall sensor for limit detection instead of being required to be in a low ambient light room. On Ali look for "Mini Inductive Sensor NPN 3-Wire Square Proximity Switch". Cost About $4 each delivered. Single Hole Mounting 1:03:50.... Almost....😉
