Jump to content


  • Content Count

  • Joined

Community Reputation

126 Excellent


About Mr.RC-Cam

  • Rank
    RC-Cam Mentor

Contact Methods

  • Website URL
  • ICQ

Profile Information

  • Location
  • Interests
    R/C, FPV, Embedded Programming, Electronic Design.

Recent Profile Visitors

8,621 profile views
  1. Mr.RC-Cam


    Here we are four years later. The goflye drone's website is 404 and their Kickstarter campaign was vaporware. Some backers are still waiting for their refund. Yikes. https://www.kickstarter.com/projects/gofleye/fleye-your-personal-flying-robot/comments Too many cool sounding Kickstarter launches end up as product failures. Getting your money back is often painful, sometimes impossible. Glad I stay away from crowd funding platforms.
  2. An accurate schematic of the welding machine's PWM controller board could help us design alternate current control solutions. I don't have the free time to reverse-engineer the controller board, but perhaps someone else will volunteer to do that.
  3. I did not find schematics to this welder. Having them would have saved a lot of time developing the Sparky Project. The current setting POT is sourced by the PWM chip's feedback path. I originally planned to use the ESP32's DAC to control weld current but had to abandon that idea because of the welder's feedback configuration. The MCP45HV51 digital POT was a convenient solution.
  4. Surprisingly, this welder runs cool. As a test I burned a half-dozen 7018 rods uninterrupted at 85A and temperature rise was minimal. So I'd say that the IGBT transistors and oversized fan are helping out. In case of overheated silicon this welder has a thermal sensor to shut it down. And Sparky will voice announce the overheat alarm condition. But I have a feeling that this audio alert may never be needed. Oops, indeed I am guilty of a typo. This welder's PWM controller uses a SG3525A, not the SG325 nonsense I typed. - Thomas
  5. In this design, a SG325 PWM controller chip and the IGBT transistors are used to switch (at 30 to 50 KHz) the rectified line voltage for the step-down voltage conversion. These drive the primary of a large Toroidal transformer (seen in the image in my previous post). The transformer's low voltage secondary is DC rectified. The reference patent claims the toroid's ferrite material is a special microcrystalline alloy. The toroidal transformer has independent primary and secondary windings. So the AC input and DC output are galvanically isolated. BTW, the low voltage controller supply (15VDC) is also isolated. It uses a TNY275P series chip and bobbin type transformer. I too have watched YouTube videos where the welder was tripping breakers. But they were using 115VAC outlets (versus 230VAC). A 80A welding current will trip a North American 115VAC / 20A circuit (80A x 26VDC / 0.85% = ~2450W). A dedicated 230VAC 30A-50A outlet is is recommended for common welding machines. The market is flooded with very capable Stick, TIG, MIG machines. So I could have bought my way into welder ownership. But I'm also drawn to interesting challenges and I like to hack. So I took a chance on a insanely cheap Chinese Inverter machine. Despite all its shortcomings, it seems to satisfy my needs. And it has given me the chance to finally try my hand at stick welding, something I've been wanting to do for many years. Exactly! Creating Sparky was interesting and a fun challenge. That's a reward of hacking. Even my failed hacks are a win, since something is usually learned along the way.
  6. Project Update: I upgraded the bus bar that bridges the inverter's output to the positive Dinse style connector. The original aluminum bar stock piece was replaced with copper. Maximum short circuit current increased to 121 Amps (was 102A before mod). I also performed a test weld with some 1/8" (3.2mm) 7018 rod. Peak current increased to 115 Amps (was 96A before mod). The new copper busbar is larger than the original. It's 2mm thick and overall size is 24mm x 100mm. The copper surface that directly contacts the aluminum heatsink was tinned with tin-lead solder. A third mounting screw was added too. See photo above. I highly recommended doing the copper mod to any mini-sized welder that was factory assembled with cheap aluminum bus bars. It's a simple hack with a big reward. No doubt there are more opportunities to improve the welding current. It would be fantastic to find a way to push it to 150 Amps. That seems achievable, especially with some help from the community.
  7. It's now Nov 2019, four years since the last post. Market growth has blown out those 2015 projections. The 2018 commercial market was valued at $14 billion USD. Projections say it will be $43 billion in 2024. All this was started by model aircraft hobbyists that wanted to have fun with technology infused flying machines.
  8. Time to wrap things up. The welding controller was an interesting project. However it took more time to complete than I dare admit. But I learned some new tricks, so that is all the reward I need. Epilogue: ======= Sparky introduced me to client side Bluetooth Low Energy (BLE) communication. This has a lot of potential for my battery operated IoT projects. The BLE library isn't perfect but the developers have obviously put enormous effort into creating it. I learned how to play wav files using the ESP32's DAC pin. All of Sparky's tones and voices are handled by the ESP32 XT DAC Audio 4.1.0 library. This lib was created by Steve at www.xtronical.com. His work deserves to be officially integrated into the Arduino library ecosystem. I found that a INA219 "high side" sensor module can be used as a low side current sensor. And measuring 200 Amps is possible from this $2 module by swapping out its 3A rated shunt resistor. Tricks like this are a hacker's delight. And best of all, I finally have a DC stick welder with custom features. Cost was about $200 US for everything, including helmet and leather protection. My setup isn't something that a serious metalworker would dare own. But it's a good match for my hobby/household projects. Of course I need more practice before I try to build something. So please pardon the sparks and smoke while I burn some rods. Project files are available on Github: https://github.com/thomastech/Sparky
  9. Sparky's real-time current measurements confirmed that the actual welding current is 45A to 100A. It's certainly not the advertised 20A to 200A range, but the findings are not a huge surprise. That's because the measured min/max currents match the information reported in the unboxing video shown earlier. I noticed the DC-DC inverter circuit's primary side filter has insufficient capacitance for a "200A" welder. It could be a partial explanation for the under-performing welding current. While searching for IGBT inverter welder schematics I found the patent to what may be the original Chinese designed ZX7 IGBT inverter stick welder. The reference schematic has four 470uF/400V caps, whereas my welder has a single 680uF. There's not enough room in the primary circuit region to install enough caps (high voltage electrolytics are big beasts). But I was able to squeeze in a 330uF/450V next to the factory installed cap. Unfortunately this small increase in capacitance was wasted effort; There was no noticeable improvement. So welding current is half what it should be. How does this impact Sparky's design? Software has been updated so that the available current settings match the welder's actual capabilities. Looking forward, it would be ideal to investigate the inverter power supply to see if the current can be increased. But digging deeper into this can wait since the welder has been useful to me as-is. And since this is an open source design, perhaps another project builder will tackle this hardware challenge and share their improvements. EDIT: See Nov-01-2019 Post, Bus Bar upgrade.
  10. Sparky's remote Amps Adjust feature requires a Bluetooth Low Energy "lost key finder." This device is officially named iTAG, but I refer to it as a FOB since it can attach to a key chain. It was intended to be used to find lost keys and has a button for taking smart phone selfie photos. The FOB device is sold by online Chinese retailers for about $1. That's about the same price as the CR2032 coin cell that powers it. Although it is a low energy device, I've burned through three batteries playing with it. OK, not all of button presses were fun and games; Hundreds of presses occurred during code development. Besides automatic low-battery shutdown, the FOB has a manual On/Off function. To turn it On, press and hold the button until you hear two short beeps. To turn it Off, press the button until a single long beep is heard. If it is paired with a smart phone the pairing must be disabled before it can be used with Sparky. During power-up (boot) Sparky will scan-search for the FOB button. If it is found a pairing will be performed. If not found during the boot scan the Bluetooth function will be disabled and no further scans will be attempted. But you can use Sparky's Settings menu to perform a manual scan. The Bluetooth connection can become unpaired (disconnected) during use. Especially if the FOB is moved too far away from the Welder. When this occurs several periodic connection retries will be attempted before giving up; If pairing is not successful a manual scan will be required to reconnect the FOB. The Lolin D32 Pro ESP32 microcontroller includes all necessary Bluetooth Low Energy (BLE) hardware. The "ESP32 BLE for Arduino" software library is used in the firmware. I have customized this library to improve reliability (the updated lib files are included in the source code package). Be aware that the ESP32 may occasionally lockup (hang forever) during a BLE device scan / connect. Sparky's heart-beat icon is located in the lower right corner of the screen. When the FOB is connected it will appear as a blinking Bluetooth icon. Otherwise it will be a blinking red heart. It will stop blinking when the CPU is locked-up. Do NOT attempt to weld if the Bluetooth or heart icon is frozen (not blinking). A power cycle reboot is required to restore full operation. Before we end our FOB discussion I have a word of warning on choosing a CR2032 replacement battery. I made the mistake of ordering a 50 piece pack from Amazon. At $10 it seemed like a bargain. But these Chinese made coin cells have high internal resistance. So they are unable to supply the peak current needed when the FOB's button is pressed. This results in it randomly shutting off during use. A partial workaround requires hacking the FOB. I added a tiny 33uF cap across the battery supply (see photos below). The cap provides an energy boost during the button press. This simple trick keeps the FOB from shutting off due to low power. But the cheap coin cells still have short life compared to name brands.
  11. IMPORTANT: The provided STL files must be re-scaled to account for filament shrinkage; 101% works well for typical ABS filaments. Sparky is housed in two 3D printed parts. The Front Chassis holds the TFT, ESP32, and Perfboard assembly. Seven M2x6 self-tapping screws are used to hold these in place. The Rear Chassis holds the two speakers and PAM8403 audio amplifier. Hot Melt glue is used to mount them. Don't forget to add the two resistors and two capacitors to the audio amp (see schematic). The Front and Rear Chassis are held together by two self-tapping M2x10 screws. A 3D printed Clip slides into the slot at the far-end of the Rear Chassis. It locks the plastic housing to the metal cabinet. A single M2x12 screw holds it in place. The 3D printed strain relief can be used if the power cord is upgraded to 12/3 (12AWG) SJTO cable. See image below. Apply rubber adhesive (shoe glue), plus cinch zip tie wraps around the cable on the front and back sides of the strain relief.
  12. This is a good place to stop and talk about the digital POT chip that is controlled by the ESP32. In case it wasn't obvious, it replaces the front panel mounted Welding Current POT that was removed earlier. But this I2C Bus controlled component is no ordinary digital POT. Unlike common digital POTs, Microchip Technology's MCP45HV51 can handle the higher voltages used in the Welder's closed-loop PWM current control circuitry. That is to say, the feedback signal voltages are high enough to damage a typical digital POT. So don't attempt to go rogue and substitute it. And you may have noticed that the original front panel mounted POT was 1K ohms and the digital POT is 5K. In a perfect world we would use a 1K digital POT. But that value is not offered by the chip maker. Fortunately there's no harm in using the 5K substitute. BTW, during design validation I installed a "mechanical" 5K POT on the Welder. Testing confirmed that the wiper voltages were correct and welding operation was not affected. I rarely experience a lucky break like this one, so please join me in thanking the silicon gods for their mercy.
  13. The MCP45HV51 digital pot IC is a tiny TSSOP14 SMD component. A 14-Pin DIP adapter board is used to make installation a bit easier. The INA219 module, MCP45HV51 (with DIP adapter), and PS2501 optocoupler are installed on a piece of Perfboard. See images below. IMPORTANT: A previous post mentioned that the INA219 module requires shunt resistor removal and the installation of a filter network. So be sure to do that BEFORE soldering the module to the Perfboard.
  14. The ESP32 microcontroller is a Lolin D32 Pro. It has 16MB FLASH and 4MB PSRAM. It saves us from some tedious wiring by including a TFT port connector. Attaching the TFT color touchscreen is as simple as plugging in a 10-Pin cable. But there's some soldering needed on the ESP32. A 0.47uF cap (C1) is added to Pin-3 and a 10K ohm resistor (R3) is installed on Pin-5. Caps C2, C3, C4 are omitted (not needed). Then all the I/O wiring is added, per the schematic. See photos below.
  • Create New...