Hunka Hunka Burning Love Posted July 13, 2017 Share Posted July 13, 2017 I bought one of those, but it ended up dying after a couple of uses. Hopefully yours will last longer. I ended up getting a small borescope / endoscope that has a USB connection that you can use with your phone or laptop. I think mine might be 5.5 mm in diameter as it's pretty small. I got some soap stuck in my ear, and it was bothering me so I used my scope to see where it was located and scooped it out with the included small picks. Quite handy! http://www.ebay.com/itm/1-5M-5-5-7mm-Android-Endoscope-Waterproof-Borescope-Inspection-Camera-Lot-FE-/232241997854 3 Quote Link to comment Share on other sites More sharing options...
esaj Posted July 13, 2017 Author Share Posted July 13, 2017 5 minutes ago, Hunka Hunka Burning Love said: I bought one of those, but it ended up dying after a couple of uses. Hopefully yours will last longer. If it does, it's not that big of a loss At least it has survived several uses today, but who knows, might die on the next try 5 minutes ago, Hunka Hunka Burning Love said: I ended up getting a small borescope / endoscope that has a USB connection that you can use with your phone or laptop. I think mine might be 5.5 mm in diameter as it's pretty small. I got some soap stuck in my ear, and it was bothering me so I used my scope to see where it was located and scooped it out with the included small picks. Quite handy! http://www.ebay.com/itm/1-5M-5-5-7mm-Android-Endoscope-Waterproof-Borescope-Inspection-Camera-Lot-FE-/232241997854 Handy for that, but if I ever intend to use microscope for soldering, I need a "hands-free" -model, not to say that that one couldn't be rigged somehow... not so sure if I will use the microscope for soldering in the end, at least it will likely take some practice to learn to look at the screen instead of directly towards the board while manipulating the components and the iron. Professionals seem to prefer stereoscopic microscopes, I guess it helps that you have some sort of "depth-view", especially when dealing with really small parts (like 0201's or even smaller), but the high-end scopes start at several hundreds and probably go to thousands or more. Good thing that hobby projects haven't got deadlines, I seem to be having trouble getting started on anything bigger since the pedal-marathon, basically I'm just lazy The speedo-/odo-/tripmeter + battery voltage display-project's been stuck ever since I couldn't get all the 7-segments, connectors & wiring fitted in the encasing I had in mind, probably have to fall back to simpler means like an OLED or less displays on that one... Got some parts to power it straight from the wheel-battery, so that might also be an option now. I also want to build a small bench-drill press frame (maybe with a stepper-driven z-axis), as I noticed what a pain in the ass it is to drill a single hole with the CNC when I made the drilling file -mistake with the pedal, but 1.0mm or smaller drill-bits aren't good for working by hand/cordless drill (easy to snap the bit / miss). Even though the CNC can be controlled manually (well, manually through the software), if felt a bit much having to set everything up for a single hole. But I suck at mechanical design and haven't got anything like linear guides Also still got the spot-welder waiting for the mosfet-bank, although I've had the parts for a while, and of course a couple of other ideas on the back of my mind... 2 Quote Link to comment Share on other sites More sharing options...
Carlos E Rodriguez Posted July 13, 2017 Share Posted July 13, 2017 23 minutes ago, esaj said: If it does, it's not that big of a loss At least it has survived several uses today, but who knows, might die on the next try Handy for that, but if I ever intend to use microscope for soldering, I need a "hands-free" -model, not to say that that one couldn't be rigged somehow... not so sure if I will use the microscope for soldering in the end, at least it will likely take some practice to learn to look at the screen instead of directly towards the board while manipulating the components and the iron. Professionals seem to prefer stereoscopic microscopes, I guess it helps that you have some sort of "depth-view", especially when dealing with really small parts (like 0201's or even smaller), but the high-end scopes start at several hundreds and probably go to thousands or more. Good thing that hobby projects haven't got deadlines, I seem to be having trouble getting started on anything bigger since the pedal-marathon, basically I'm just lazy The speedo-/odo-/tripmeter + battery voltage display-project's been stuck ever since I couldn't get all the 7-segments, connectors & wiring fitted in the encasing I had in mind, probably have to fall back to simpler means like an OLED or less displays on that one... Got some parts to power it straight from the wheel-battery, so that might also be an option now. I also want to build a small bench-drill press frame (maybe with a stepper-driven z-axis), as I noticed what a pain in the ass it is to drill a single hole with the CNC when I made the drilling file -mistake with the pedal, but 1.0mm or smaller drill-bits aren't good for working by hand/cordless drill (easy to snap the bit / miss). Even though the CNC can be controlled manually (well, manually through the software), if felt a bit much having to set everything up for a single hole. But I suck at mechanical design and haven't got anything like linear guides Also still got the spot-welder waiting for the mosfet-bank, although I've had the parts for a while, and of course a couple of other ideas on the back of my mind... Maybe you you should do Lean-one-piece-flow and finish one at a time. It would possibly reduce your inventory also. Lol 1 Quote Link to comment Share on other sites More sharing options...
Hunka Hunka Burning Love Posted July 14, 2017 Share Posted July 14, 2017 Here's an alternative to using a microscope. They come in 2.5 and 3.5x and higher I think? http://www.ebay.ca/itm/Dentist-Dental-Surgical-Medical-Binocular-Loupes-3-5X-R-Optical-Glass-/391574917945 1 Quote Link to comment Share on other sites More sharing options...
esaj Posted July 22, 2017 Author Share Posted July 22, 2017 (edited) Finally learned to solder watching the microscope view from the screen instead of looking at the board and using 0805's, and it does help a lot. I've done a bunch of other stuff (a frequency meter & some PWM mosfet-drivers, motorized stand for the microscope from a broken DVD-drive, new power distribution board for the next rack-front panel and some small stuff) over the past week, but today I made a bunch of hastily designed & cut led-blinker boards, mostly for practice: The board's 18mm across (0.7 inches), the cap and the led on the board are 1206-size, the resistors are 0805's (so the next smaller size from the 1206's to which I'm used to), transistors are SOT-23, so although small, nothing really small. Although the other stuff (the microscope stand & the frequency meter) might be more interesting than a blinking led, I thought of writing about this (at least for now). This is a self-oscillating "micro-power" blinker (average current is in the microamperes-range, µA, millionth of an ampere), it's run by a 1.5F / 5.5V supercapacitor, should blink it a good while (15-20 hours, even more?) on one charge, recharges in about 20 seconds It blinks around twice per second, but taking a picture just when it blinks is of course next to impossible. The first one I made has been going for over 12 hours now, another with usb-charging & two boards in parallel and faster blink rate (so higher current usage) on a single cap has been running for over 7 hours... Not sure what real use these are though, maybe attach to clothes during winter time when it's dark? On a lithium (primary, non-rechargeable) coin-cell, which is about 150mAh, it could theoretically run for half a year, and some people have managed to make ones that use around 6µA on average, reaching over two and a half years of run-time on a coin-cell. Or if enough light is available often enough, a supercap (or rechargeable lithium cell or some other type of battery) can charge from a small solar-panel and run until it breaks down (10-20 years? I don't know how long supercaps last in use...). Shot a video of making one of the boards, but of course it didn't go "perfectly" (nothing ever does when shooting a video ), although not really bad either. Not sure if anyone wants to watch it though, there's no audio and it's long and pretty boring, so I sped it up to 4x to get it down to less than 4 minutes (original was about 15 minutes even with parts cut out): It's a bit tricky to solder a board this small, as it isn't attached to anything and is so light that it easily moves around by accident. Also need to check for short circuits all the time with milled boards, as it's easy to slip a bit of tin into the cuts as there are no solder-masks. The blink at the end doesn't look that bright, but it's easier to "fool" naked eye than camera, especially in darker conditions. Edited July 22, 2017 by esaj 3 Quote Link to comment Share on other sites More sharing options...
Hunka Hunka Burning Love Posted July 22, 2017 Share Posted July 22, 2017 That is really awesome!!! Loved the video! It's fun watching as these parts are so tiny! You really did a nice job on the placement and soldering. Was that on the first try? 1 Quote Link to comment Share on other sites More sharing options...
esaj Posted July 22, 2017 Author Share Posted July 22, 2017 5 hours ago, Hunka Hunka Burning Love said: That is really awesome!!! Loved the video! It's fun watching as these parts are so tiny! You really did a nice job on the placement and soldering. ' Thanks, although I've had a lot of practice since I started using SMDs late last year and I (still) don't work with the really small sizes like 0402's (but working my way there, slowly ). Good tweezers help a lot with handling the small parts, I'd say they're about the most important tool when working with SMDs by hand, you can easily get by with a cheapo soldering iron with a basic 0.8mm chisel-tip, but not with crappy tweezers. I did OK(ish) on an Aliexpress SMD-tweezer set, which wasn't that bad, but bought better (and expensive, for tweezers, around 15€ on sale) early this year, and have been using those almost exclusively: http://ideal-tek.com/scheda.php?m=0&f=8&l=3&c=Smd Tweezers&idp=856 There's a huge selection of different tweezers out there, TME was having -25% off from the normal prices for tweezers, and while I did look through the selection for a long time (I think they already carried something like 200 different models, probably something like Mouser or Digikey has much larger selection ), not having the possibility to try any almost made give up. I picked those (Ideal-Tek SM108.SA, also known as Lindstrom SM108.SA, as IdealTek is OEM for Lindstrom) simply based on a couple of threads in some electronics forums, and they seemed to be one of the most popular ones and good for general use, people have used those down to 0402 -sizes (so less than a quarter of those 0805-resistors I use in the video, as length is halved and width is slightly less than half), and somebody posted a picture of the heads after 10 years of use: https://www.eevblog.com/forum/reviews/best-tweezers-etc/msg1194495/#msg1194495 5 hours ago, Hunka Hunka Burning Love said: Was that on the first try? That was the second blinker-board I made (I didn't shoot a video of the first one), the board & the video was done in one go, but I did edit it down to make it shorter, cut the first few minutes, as there are lots of pauses and not much is happening, as well as cutting out the pauses between getting the next component from a sample book and such (5-20 seconds clips here and there). As the video was still 15 minutes, I then just sped it up in Youtube. A board like that can be put together in 15-20 minutes (the one in the video took 25 minutes total with all the pauses), maybe even faster if you have all the components lined up (and don't mix them ) so you can just pick and go all the time (and with a real PCB fabrication-house made board, there's be no need to check for shorts to surroundings), but that's barring any bigger mistakes, it can easily take double that if things go wrong (like a short-circuit under a component after all legs are soldered). One of the board also had bad cuts (there was some hard to see copper stuck at the bottom of the cut), so it took a lot longer as I was wondering where the short-circuit was for a good while. I made 9 of those boards in a single waste bakelite-board with fast feed & single-pass cut, the kind of settings that destroy the V-bit in seconds with FR4, so the cut quality is a bit iffy But these were mostly for practice anyway Btw, it's been 21.5 hours for the slower blinker and about 16 hours for the dual-board setup, and the blinkers are still happily going, the average current must be a few tens of microamps... 2 Quote Link to comment Share on other sites More sharing options...
Hunka Hunka Burning Love Posted July 22, 2017 Share Posted July 22, 2017 That's fantastic! Yeah it must help to have good quality tweezers. I got a set of tweezers, but they flex and aren't very good. I had no luck resoldering a small IC after hot-air removing it. It would accidentally fly off half the time, or I couldn't see the proper orientation (pin 1 or 6? Hmmm ) that I got frustrated and gave up. How you liking the el cheapo Chinese re-work station? I see you still have the mini-vacuum tube on it. Mine doesn't have that. 2 Quote Link to comment Share on other sites More sharing options...
Rehab1 Posted July 22, 2017 Share Posted July 22, 2017 (edited) On 7/13/2017 at 5:24 PM, esaj said: Good thing that hobby projects haven't got deadlines, I seem to be having trouble getting started on anything bigger since the pedal-marathon, basically I'm just lazy You lazy! No way! 10 hours ago, esaj said: It's a bit tricky to solder a board this small, as it isn't attached to anything and is so light that it easily moves around by accident. Have you ever tried removable Glue Dots to hold projects in place? Edited July 22, 2017 by Rehab1 2 Quote Link to comment Share on other sites More sharing options...
Marty Backe Posted July 22, 2017 Share Posted July 22, 2017 (edited) 2 hours ago, Rehab1 said: You lazy! No way! Have you ever tried removable Glue Dots to hold projects in place? Even better is something like "Blue Tack". It's a tacky putty like substance that doesn't leave a residue but is very effective at holding things even if you don't have a flat surface. Here's what I use: Elmer's Tac 'N Stik Reusable Adhesive Edited July 22, 2017 by Marty Backe 3 Quote Link to comment Share on other sites More sharing options...
Rehab1 Posted July 22, 2017 Share Posted July 22, 2017 5 hours ago, Marty Backe said: Even better is something like "Blue Tack". It's a tacky putty like substance that doesn't leave a residue but is very effective at holding things even if you don't have a flat surface. Here's what I use: Elmer's Tac 'N Stik Reusable Adhesive Heck I would have settled for a snot booger just to keep the tiny board from moving around! I actually burned off calories watching @esaj trying to corral it! Entertainment at it's best! 2 Quote Link to comment Share on other sites More sharing options...
esaj Posted July 23, 2017 Author Share Posted July 23, 2017 (edited) 18 hours ago, Hunka Hunka Burning Love said: That's fantastic! Yeah it must help to have good quality tweezers. I got a set of tweezers, but they flex and aren't very good. "Normal" household tweezers can be horrible to work with, if that's what you have and need to occasionally work with SMDs, get a basic SMD-tweezer set from eBay or similar, and/or a suction tool (see below) if you work with larger ICs. They're cheap and most likely the tweezers will be better than what you currently have. Quote I had no luck resoldering a small IC after hot-air removing it. It would accidentally fly off half the time, or I couldn't see the proper orientation (pin 1 or 6? Hmmm ) that I got frustrated and gave up. The chips usually have a small dot next to pin 1, or if not, the datasheets should show some way of identification, for example, some small ICs with pins on two sides have (also or only) a notch at the other end or a slant on one side: Hate the slanted ones, on small packages the slant can be hard to see, plus I'm not sure if it's always on the side with the pin 1 or could be the opposite, need to check the datasheets. Larger IC-packages can be hard to handle with tweezers, they're mostly good for the smaller "basic" components like resistors and capacitors in standard smaller SMD-packages (1206, 0805, 0603, SMA/SMB/SMC etc) and smaller ICs (like *SOP/SOIC and whatever that MPU-6050 one is above), a suction-cup style tool might be more useful for the larger packages (BGA, higher pin-count *QFPs etc) where you have enough surface area on top to grab onto: I have one of these "manual" ones, plus one that attaches to the fume removal -port of the station for continuous suction. Quote How you liking the el cheapo Chinese re-work station? I see you still have the mini-vacuum tube on it. Mine doesn't have that. So far I've liked it. The soldering iron's 75W, so it heats up really fast, and while I was annoyed a bit with the fume extractor tube getting in the way of wiping the tip on sponge, I've kept it there, since the extraction works pretty well. The iron uses a "normal" rotatable potentiometer knob on the station front panel for setting the heat (and shows the set heat also on a 7-segment, as well as the actual tip temperature when not adjusting), the pump speed for the hot air also has a rotatable pot, but the hot-air temperature uses separate up/down/reset -buttons, and is much slower to set to wanted heat. The hot-air always starts at 90 degrees Celsius after the machine is turned on and you need to keep the "up" -button pressed for a good while before it rolls up into the 300's which I usually use. Not really a big issue, but I hope they'd have used similar pot for that too. I've also used the tweezers that can be attached instead of the iron for replacing basic 2-pin components like resistors when trying stuff out (or have to remove a shorted component that's already soldered on both ends): So much easier just to grab the component from the ends with the tweezers and lift it out. Soldering new ones back in with the tweezers themselves isn't that easy, have to be fast as they're heating the component from the moment you pick it up and would destroy it before long, so I prefer to use the normal iron for that, but for removal the tweezers are really handy. Only minor annoyance with them is having to screw out the soldering iron and screw in the tweezers, then wait for the tweezers to heat up (they're much slower than the iron). Having two GX16-5 -ports and a switch to select between the tweezers and the normal iron would be handier. I still need to practice more with the hot-air, the largest IC I've removed so far was something like 208-pin QFP with small pin-pitch on the DVD-drive I dismantled for the sled, I just used the board for practice. I don't think the IC survived, it took several minutes to heat up the pins enough for the chip to come off, and at that point the surface already was "warped" a bit, meaning it must have gotten really hot. Not that I'd have had any use for the chip anyway, as it was some special purpose optical drive -control chip. Gotta dig up more broken stuff with SMD-parts for practice before I face a situation where I have to remove a large chip without destroying it, and maybe also get the special purpose hot-air nozzles for QFPs (quad flat package). Only problem is, you need a separate nozzle for each QFP-size... I tried a BGA (ball-grid array) -nozzle but it heats up the center too much, QFP-nozzles direct the airflow to the sides only: 18 hours ago, Rehab1 said: You lazy! No way! Well, maybe not lazy, but easily distracted by other projects, and then ending up with lots of unfinished stuff? Quote Have you ever tried removable Glue Dots to hold projects in place? 15 hours ago, Marty Backe said: Even better is something like "Blue Tack". It's a tacky putty like substance that doesn't leave a residue but is very effective at holding things even if you don't have a flat surface. Here's what I use: Elmer's Tac 'N Stik Reusable Adhesive 10 hours ago, Rehab1 said: Heck I would have settled for a snot booger just to keep the tiny board from moving around! Blu-tack might work, I have some in the drawers. Usually the board moving is not an issue, as they're at least slightly larger than the blinker-boards, the white surface seen in the microscope-video is my silicone soldering mat, which typically has enough friction to prevent the boards from moving around. For SMD-work, I need to rotate the board around to get better "angle of attack" at the sides of the components all the time, so "helping hands" or anything that sticks "too much" is not good either. Maybe I should do a rotatable & quick lockable stand next? Quote I actually burned off calories watching @esaj trying to corral it! Entertainment at it's best! Glad you liked it Edited July 23, 2017 by esaj 3 Quote Link to comment Share on other sites More sharing options...
litewave Posted July 23, 2017 Share Posted July 23, 2017 14 hours ago, esaj said: "Normal" household tweezers can be horrible to work with, if that's what you have and need to occasionally work with SMDs, get a basic SMD-tweezer set from eBay or similar, and/or a suction tool (see below) if you work with larger ICs. They're cheap and most likely the tweezers will be better than what you currently have. The chips usually have a small dot next to pin 1, or if not, the datasheets should show some way of identification, for example, some small ICs with pins on two sides have (also or only) a notch at the other end or a slant on one side: I thought you were messing with your followers when you included the chip diagram labelled "THX1138D" and "1337OHAI". I suspected you were just having some Photoshop fun at our expense, until I did a reverse image search and found it at The Evil Mad Scientist website. (For the unsuspecting, follow the links. Second term has two parts). I will be very concerned if these chips start showing up in electronics devices for real. 3 Quote Link to comment Share on other sites More sharing options...
Popular Post esaj Posted July 23, 2017 Author Popular Post Share Posted July 23, 2017 6 minutes ago, litewave said: I thought you were messing with your followers when you included the chip diagram labelled "THX1138D" and "1337OHAI". I suspected you were just having some Photoshop fun at our expense, until I did a reverse image search and found it at The Evil Mad Scientist website. (For the unsuspecting, follow the links. Second term has two parts). I will be very concerned if these chips start showing up in electronics devices for real. I have followers? I've become a cult leader! But how many circuits can you build with an operational hamsterfier? 4 Quote Link to comment Share on other sites More sharing options...
steve454 Posted July 23, 2017 Share Posted July 23, 2017 (edited) that looks like an Arkansas pig packer (carrier) Now I have to check Netflix for TH1138 and see if it is still available. Edited July 24, 2017 by steve454 3 Quote Link to comment Share on other sites More sharing options...
litewave Posted July 24, 2017 Share Posted July 24, 2017 (edited) 19 hours ago, esaj said: I have followers? I've become a cult leader! Ha, I linked to the wrong "OHAI" in my response (original link was to a tiny town at the bottom of New Zealand). I think it really refers to the Chef plug. 19 hours ago, esaj said: But how many circuits can you build with an operational hamsterfier? Exactly one at the moment But anything that uses a AA battery (~1-2w) should work until the poor fellow tires out. Edited July 24, 2017 by litewave 3 Quote Link to comment Share on other sites More sharing options...
esaj Posted July 24, 2017 Author Share Posted July 24, 2017 (edited) 3 hours ago, litewave said: Exactly one at the moment Cool video! And the way how he makes a pulley wheel is actually really good, might use that some day... In other news, I'm pissed. I finally kicked myself in the butt and got back to building the spot welder. I designed the board for mosfet-bank last night, milled it out today after work and set to populate the board with the components. I had planned to put 15V TVS-diodes (transient voltage suppressors, also known as "transil") on the mosfet gates to prevent any ringing from destroying them (the maximum Vgs is +-20V, and the gates are being driven with 12V). I figured using voltage above what the gate drive has would be safer, although transils usually only start to conduct a good 10% above the nominal voltage. Nothing too fancy, the idea of using diodes to speed up shutdown (and hope that the gate driver can take it ) and ferrite beads on the gates to mitigate ringing is pretty much from the Firewheel board design (and some app notes about paralleling mosfets using ferrites). The actual heavy lifting is handled by the MIC4452, a 12A (max) gate driver, and the current per gate is limited to around 1.75A max with the 6.8-ohm resistors (except when they shutdown, and a large portion of the current goes through the reversed diodes). Maybe should have something like a few kilo-ohm resistors between the sources and the gates too, to so they're not left floating at any point... I actually went through the trouble of milling out all excess copper around the gate circuitry (well, except near the driver chip), as supposedly the gates can get some inductive coupling if the traces go near the high current lines. Don't know if it matters, but I guess it won't hurt. For transils, my inventory spreadsheet lists the voltages, nominal maximum current/power dissipation, whether the transil's uni- or bidirectional, amount at last count and the casing. Not the name of the component. I use the spread sheet when designing circuits in KiCAD, and only dig up the components when I'm ready to solder them on the board. So, I dig through the bags of components (I keep the components from "real" distributors and stuff from Aliexpress separate) and locate the bag of 15V transils: Never noticed it before, but now that I did, I thought it was a bit weird, why name it "SMCJ9.0A", with the number 9 in the name, if it's a 15V trans... I check the datasheets. Multiple manufacturers make this part, but they all say it's a 9V transil. I turn on the linear PSU, set current limit to 20mA and put 9V on the transil. Not conducting. 10V. The current limit is hit immediately. Shit. How can a large scale component distributor fuck up the component descriptions that bad? I checked their pages, and it's now listed correctly, but the printed label and the manifest of my original order list that as 15V transil, so their pages must have said the same at the time... The bag is missing one diode. I know where it went. It's in the robot motor controller that I abandoned after the mysterious breakdowns. It's actually on the main 12V rail, coming from a LiPo pack, and supposed to protect the rail from possible overvoltage spikes coming from the motors. I was pretty sure at the time that my design was just faulty, maybe I should dig up the board from the electric trash-box and try removing the diode, replace what's broken (this time) and see if I can get it to work so it won't blow up random components (usually the ones connected to the 12V line) all the time. Maybe. Some day... In the end, I used 12V transils (I tested them with the PSU, they won't start conducting until slightly above 13V) for protections, but the cases were smaller (SMB's, the SMCJ9.0A's were SMC's, as the name suggests), luckily they could still reach across the gap: Coating the traces with solder might have been a bit excessive, considering that the spikes are short and nothing like what runs through the mosfets and the ground plane... There's also a reason why the mosfets have only two holes here (the middle one, drain, is left out), but I'll get back to that later, when I have built the whole damn thing. Edited July 24, 2017 by esaj 3 Quote Link to comment Share on other sites More sharing options...
Rehab1 Posted July 25, 2017 Share Posted July 25, 2017 On 7/23/2017 at 4:21 AM, esaj said: I've also used the tweezers that can be attached instead of the iron for replacing basic 2-pin components like resistors when trying stuff out (or have to remove a shorted component that's already soldered on both ends): I am impressed with the micro SMD tweezers..... along with just about everything else your topic discusses! Always an education in progress. I see the SMD tweezers can get quite elaborate and expensive! 2 Quote Link to comment Share on other sites More sharing options...
esaj Posted July 25, 2017 Author Share Posted July 25, 2017 (edited) Cable going to the negative plate of the capacitor bank soldered to the board: Probably would have been a better idea to solder that first and then the components, things got pretty hot, but at least it seems the mosfets didn't die Another angle: Pretty messy... I later on "hot air leveled" the solder... actual HASL, Hot Air Solder Leveling, is a bit different process, but it worked out pretty good here: Excess flux is dripping all over the place, I couldn't even get it all cleaned off, looks like the hot air burned some of it straight onto the board, but most. Heat sink drilled & threaded. It's some old CPU heat sink, just a big hunk of aluminum. I'm not even sure how much the heatsink will help, although it will probably make it possible to dissipate much more heat, the question is will the heat actually flow out from inside the mosfet junctions fast enough for a sink to matter, or if it will just heat up so fast that the insides turn into mush before the heat can flow off. I hinted before that there's a reason why there are no holes drilled for the drain-legs in the big board. The reason is that I use another piece of board on top and solder the drain-legs there, along with the big cable that comes from the "low-side" welding cable: The mosfet legs look tiny in comparison... we'll see if they hold up. Basically it would be possible to use the heatsink itself as conductor too, as the mosfet tabs are directly connected to the drains. Still need to add diodes between the upper & lower board in the mosfet-bank, the idea is to bypass the body-diodes of the mosfets themselves, so in case the stray inductances pull the drain-side of the mosfets to negative in comparison to the source-side (the common ground of the entire thing), the mosfets diodes won't need to take the hit. Might work just as well without them, but can't hurt (I think...). After that, I "just" need to solder the actual caps to the bank (I have the other bank-plate with the cable done already), add the heavy-duty fast recovery diodes between the cables to free-wheel current if/when stray inductance causes high voltage spikes, add the tungsten bits as welding prongs and then it's ready for a test run with something simple giving a short (1ms) pulse to the gate-driver. I'll add the actual charging/discharging controller later on if it survives the test blasts. It might be that this whole contraption fails more or less spectacularly. If it happens, I'll have to see what needs to be changed, in hindsight, I should have just bought some heavy duty copper flats (busbars) and build most of the high current circuitry on those. There are so many ways this can fail, and I probably haven't even thought of all of the possibilities The capacitors have too high ESR to deliver enough current for welds The capacitors cannot handle the spike and die/explode, as too much power is dissipated in the internal resistance I haven't found an official rating, but somewhere there was a mention it could be around 160 milliohms per cap for the ones I'm using (cheapo 47000uF / 16V from Samwha). With 10-12 caps in parallel, it will be "only" 13.33.. - 16 milliohms... still, if the mosfets in parallel have about < 0.5 milliohms in full conductions, and add a good few milliohms for connections plus whatever the resistance at the welding points is, it might still be relatively high portion of the total resistance Mosfets overheat and get destroyed Could be just because the current is too high for them, or that the gate driver doesn't switch them fully on fast enough, or myriad of other reasons, gate ringing etc. Mosfet legs blow off Stray inductance causes high enough voltage to destroy the mosfets and/or other parts Solder connections dissipate too much heat and melt/vaporize off The 35µm copper layer (105µm on the capacitor banks) on the boards fries Ground plane shifts enough during the pulse to destroy the gates, gate drive and/or other components That's about it for now... Not sure how long it will take me to get to test this thing, but it's getting close. Edited July 26, 2017 by esaj 3 Quote Link to comment Share on other sites More sharing options...
esaj Posted July 30, 2017 Author Share Posted July 30, 2017 (edited) Sneak peek: Compilation of slowed down videos from where the above frames are taken: I'll get back to this later, it's pretty late here... Edited July 30, 2017 by esaj 3 Quote Link to comment Share on other sites More sharing options...
Rehab1 Posted July 30, 2017 Share Posted July 30, 2017 Strong stuff!! I'm blown away again my your electrical/technical expertise and fabrication capabilities. The hot air leveling process worked great! Beatutiful job fanning out the cable strands on the copper plate! I'm learning so much from your topic but unfortunately much of it still over my head just like but rocket science. I'm not sure if this was a 'scared reaction' to your first test or the switch was not fastened to the table as you depressed it. 2 Quote Link to comment Share on other sites More sharing options...
esaj Posted July 30, 2017 Author Share Posted July 30, 2017 (edited) 4 hours ago, Rehab1 said: Strong stuff!! I'm blown away again my your electrical/technical expertise and fabrication capabilities. The hot air leveling process worked great! Thanks, mostly I just make up "fabrication techniques" as I go along, based on prior knowledge and what I happen to have at hand But yeah, leveling thick solder joints with hot air seems to work pretty great, real HASL uses "air knives", which create a thin, steady "curtain" of hot air through which the board is moved to remove excess solder after wave or dip soldering in PCB manufacturing (the pads of "real" PCBs you can order from PCB-fabricators are pre-tinned by this method): Quote Beatutiful job fanning out the cable strands on the copper plate! I'm learning so much from your topic but unfortunately much of it still over my head just like but rocket science. The strands are rather stiff, so fanning out the cables just required to bend the strands by hand and then press them against the copper clad. Quote I'm not sure if this was a 'scared reaction' to your first test or the switch was not fastened to the table as you depressed it. Both, the breadboard with the circuit was not attached to the table in any way, and I was wary of the sparks / possibly blowing components flying. What's not seen in the videos is that I stand further away leaning in to press the button/connect wires and hold a piece of plexiglass in front of my face with another hand Edited July 30, 2017 by esaj 2 Quote Link to comment Share on other sites More sharing options...
esaj Posted July 30, 2017 Author Share Posted July 30, 2017 (edited) After the initial pictures of the mosfet board, I added a MBR40100 (40A / 100V Schottky -diode) and a 30V transil between the drain and the source to handle reverse current and possible Vds -overvoltage. Soldering the bank wasn't that hard, but required a bit unusual ways to keep things in place and turning the board back and forth: I typically checked to see where the cap legs would hit, spread solder on the board with the hot iron first on the leg positions and then held the caps by one hand while heating the solder with the other. Once a row of caps was attached well enough, I turned the board around, soldered the legs on the other side properly, and then turned it around again to properly solder the initial blobs that were helding the caps in place. After all the legs were soldered, I still hot-air leveled the legs a bit and added charging cables on both sides: Didn't bother to clean up the flux, maybe I should have But at least in this case, it's just an aesthetic thing. For the electrodes, I used couple of broken/dulled tungsten carbide V-bits, and added a heavy duty fast recovery diode in TO-247 -case to suppress possible back-EMF (although, it probably should be near the other end of cables to be really useful): The jig holding the electrodes is just a piece of plywood with drilled holes and screws to hold the cables in place. Soldering the V-bits turned out to require a lot of heat, and the end result is messy. I'll probably cut those off at some point and make something better later on. I made the test setup in the garage, in case something blows up or catches fire Since I didn't want to move power supplies, I had to make do with a battery charger (charging the cap bank, I used LiPo-settings for 1-3 series packs, so the voltages were 4.2, 8.4 and 12.6V). Also I didn't find a 12V wall-vart, so I used a 9V one and a separate boost-module to supply 12V for the gate driver. I first tried with holding the electrodes down with a clamp, but the results weren't really impressive. I couldn't get enough pressure, so usually the current was stopped short as the contact was lost early in the pulse. I also just hit the gates full on instead of pulsing to get maximum discharge to see that things hold up. Later in I clamped the steel-piece directly on the electrodes, that's where the big spark videos are from. After testing a few dozen times with the clamping, and seeing that the sparks weren't that big and that nothing blows up, I held the electrode jig by hand to apply pressure. A much larger spark and sound ensued, and the power was enough to soften/melt the ends of the carbide bits: Well, actually that might not be a bad thing, they make better contact now I didn't get any proper welds done, as most of the resistance seems to be at the ends of the electrodes instead of the interface between the two metals. On the copper-clad the copper mostly blew away around the electrode (too high current) cutting the pulse early, but with better electrodes/jig and finding more proper voltages and pulse lengths, it could work much better. The pieces I used for testing were 1mm thick steel (I think it's steel?) ribbon and left-over 35µm (1oz) copper clad. Not ideal, but they did give some idea of the capabilities, close ups on steel: On the copper clad, high currents tended to blow away the thin copper, that's probably where most of the sparks came from (hot / molten copper bits flying): Today I tested with oscilloscope attached to get some measurements, using the same pieces and proper power sources, holding the electrodes down by hand (and inside a small cardboard box to catch any stray sparks). I changed the pulse to 10ms and charged to full 12V: Gate rise at turn-on: Not bad, although not superb either. But since the mosfets haven't broken and the heatsink doesn't get even warm, good enough No overshoot or ringing at the gates (I also tried with the drain unpowered, the rise times are then much faster, <100ns, and it won't ring there either) . Gate fall at turn-off: The shutdown is faster, probably thanks to the diodes in parallel with the gate resistors. Slight undershoot (around 1V), but it seems the gate-source -diodes catch that. Here's a measurement of full 10ms pulse, channel 1 (yellow) is the gate voltage, and channel 2 (cyan?) is the capacitor bank voltage: The pulse is cut short after around 3ms as the copper is burned away, but the bank voltage drops by about 5V during that time (2V per grid square). Not sure if my math is entirely correct, but the average current could then be calculated something along like this: Capacitor bank capacity is 0.47F (10 x 47000uF caps), farad equals ampere-seconds over volt => (As)/V . Charged to 12V, the total charge is 12V * 0.47F = 5,64As (ampere-seconds). So, at a constant current of 5.64A, the bank would empty in 1 second. Solving for current from the farad-equation using units F = (As)/V gives A = (F*V) / s Amperes equals farads times voltage (drop) over time Plugging in the known values gives (0.47F*5V) / 0.003s (0.47 farads times the 5V voltage drop divided by 3 milliseconds) = 783.3333... A on average I'm not sure if this is the proper way to calculate the amperage, also this is the average over the 3ms period, if you look at the scope shot, the voltage drops like a rock for the first volt or so, which would indicate much higher current for a very brief period of time. With 783A average current, and average voltage (against ground) of 12V - (0.5 * 5V) = 9.5V would indicate a resistance of about R = U/I => 9.5V / 783A = 0.012133ohms or about 12 milliohms. Plausible? Well, I guess, the parallel mosfet resistance in full conduction should be about 0.4 milliohms or below, and the cables & solder contacts in the circuit probably have negligible resistance. That leaves the capacitor ESRs (of which I don't know the real values) and the actual welding points. Last scope shots with more measurement points, all on 2V per grid square: Again, channel 1 (yellow) is the gate voltage used for triggering, channel 2 (cyan) is the capacitor bank voltage, channel 3 (purple) is at the mosfet drains and the darker blue or whatever it's called (I'm colorblind, btw) is the math operation of channel 2 (capacitor bank voltage) - channel 3 (drain voltage), basically the voltage drop over the load: As can be seen, most of the voltage drop occurs over the load, as expected, so most of the power is dissipated where it should be. Close up: This time the bank discharged more properly, the voltage goes from 12V to about 3.5V over the 10ms period. Using the same equations as before, this gives and average current of (0.47F*8.5V) / 0.01s = 399.5A The average current is smaller than before, because the current drops the lower the capacitor bank voltage becomes. The main thing is that the device as a whole seems to work as supposed to. The heatsink on the mosfets is likely unnecessary (or at least unnecessarily big ), as it hasn't ever became even warm. The caps seem to be capable of discharging high enough currents, 0.47F is still relatively small bank for spot-welding, but for thin materials, it seems it could work. I still need to figure out if I need to use something else for electrodes, how to jig them so I can put enough pressure on the contacts, proper pulse widths (and pulse pattern, many welders use patterns like 1ms pre-heat pulse + 2ms delay + 4.5 actual welding pulse) and voltages and so on, of course these are highly dependent on material also (welding 0.15-0.3mm thick nickel strip on cells is different than welding 1mm thick steel onto copper etc). Also need a proper controller for the discharge & charge (so I can select the charging voltages, discharge pulse lengths etc) and get the thing encased. Edited July 30, 2017 by esaj 3 Quote Link to comment Share on other sites More sharing options...
Rehab1 Posted July 30, 2017 Share Posted July 30, 2017 2 hours ago, esaj said: I still need to figure out if I need to use something else for electrodes, how to jig them so I can put enough pressure on the contacts, proper pulse widths (and pulse pattern, many welders use patterns like 1ms pre-heat pulse + 2ms delay + 4.5 actual welding pulse) and voltages and so on, of course these are highly dependent on material also (welding 0.15-0.3mm thick nickel strip on cells is different than welding 1mm thick steel onto copper etc). Also need a proper controller for the discharge & charge (so I can select the charging voltages, discharge pulse lengths etc) and get the thing encased. I love your work buddy! There are 1000's of pieces of equipment that I will never possess the knowledge to construct and this project is one. 3 Quote Link to comment Share on other sites More sharing options...
litewave Posted July 31, 2017 Share Posted July 31, 2017 19 hours ago, Rehab1 said: I love your work buddy! There are 1000's of pieces of equipment that I will never possess the knowledge to construct and this project is one. @esaj videos inspire me to relearn and catch up. Since I'm still transient with split households on opposite sides of the planet (without many tools), I can't actively participate, but I am definitely paying attention. 2 Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.