No announcement yet.

Quiet1's SG build

  • Filter
  • Time
  • Show
Clear All
new posts

  • Quiet1's SG build

    I’ve been following John Bedini’s work for at least 20 years and wanted to build one of his “interesting” devices. However, the information available back then was, at best, unclear and comments on the various forums were terribly confusing. When I recently stumbled across the SG book series by Peter Lindemann and Aaron Murakami, things finally made sense and the desire to build a SG machine returned with renewed vigor. In this thread I will document my SG build, staying as close as practical to the design in the books. One exception is that I included the advanced upgrades such as tuning and generator mode into the initial build. This made “upgrading” quite easy.

    My main goal is to learn what this machine has to teach me. But it would be really nice if it manages to run “for an extended period of time!”

    Disclosure: As I write this, the project has completed the intermediate stage and has moved into the advanced material. I plan to describe the project chronologically from the beginning to help those who may follow. However, due to lax documentation, some photos may appear slightly out of order.

    Thanks for reading. Enjoy the ride!

  • #2

    If you are thinking of building a SG unit, I strongly suggest you start by getting a copy of the SG handbook series, available at These three volumes describe in detail how to build the machine and how it works. Everything you need is clearly explained. I read my books several times before starting on my project, and then a couple more while building. I’m still using them for reference. Great Stuff!

    The forum here is another excellent source of information. I’ve been lurking for a couple years and gleaned much information.

    Read and Learn.


    • #3
      Sourcing the Wheel

      After reading the SG books, and with a good idea of what I’m getting into, my first step for the project was to purchase a bicycle wheel. I had previously read on the forum that several people bought their wheels from Amazon, only to find that they are out of round and have terrible wobble. Straightening (“truing”) a wheel is a learned skill and requires a few special tools, as I observed when I had a bicycle rebuilt a few years ago. So, despite the attractive price on Amazon, I opted to purchase my wheel from a local bike shop. The results were fantastic.

      I stopped by the shop in mid-winter. A very nice older gentleman, the owner, was behind the counter. I described my project to him as a science fair type motor where there’s a bunch of magnets on a bike wheel and a pulsing coil under it making the wheel turn. (I really try to avoid talking about overunity and perpetual motion.) He accepted my story but I may have caught a very quick reaction of “that’s weird - but whatever.” We headed to the back room where he has hundreds of wheels hanging from the ceiling. He picked one of the proper size, put it in his truing stand, and gave it a spin. He checked for out of round and side to side wobble. With perhaps a half dozen minor adjustments to the spokes the wheel was running true. He then slightly loosened the hub bearings for less friction. In less than three minutes the wheel was perfect for use in my SG! The total cost was a little over $30.

      If you are looking for a bicycle wheel to use in your own build, please consider visiting your local bicycle stores. They will appreciate the business and the convenience is well worth the extra few dollars spent.


      • #4
        And then it was time to build the frame…

        To say my carpentry skills are terrible is being both generous and kind. John’s frame certainly looks easy enough, but it is made of wood! Fortunately there’s always a work around…

        Nearing retirement a few years ago I wanted to have a modest shop at home for projects like this SG. However, space was limited. At the time hobbyist 3D printers were getting popular and capable of surprisingly good results. I opted to go with the 3D printer instead of getting a mill, a lathe, and all the required tooling. It took a while to learn the printer, but now I can turn out a good part with little effort.

        So instead of cutting apart a 12”x48” board, I brought up the 3D mechanical CAD package. I made the base a bit taller to hold a standard size aluminum plate for the front panel (6” x 24” x 1/8” is a good size. I cut it twice to get three panels, each 6” x 8”). Reading ahead I also learned that I’d probably need to include some adjustment in the wheel to coil distance. I did this by making the uprights a bit longer and adding plastic shims (3d printed, of course) under the coil. Some shims are 1/8" thick, others are 0.10” thick. This gives flexibility in finding that “just right” distance for best performance. Every part has holes allowing the entire assembly to be bolted together. I also included mounting ears & holes so I can mount another aluminum plate on the back in case it is needed later. (Cap dump circuit, perhaps?) One glitch was that the uprights are longer than the printer can handle. So I designed them as multiple pieces and hold them together with brackets and screws.

        The 3D printing was generally uneventful, but it did take a good number of hours. I don’t mind having my robot laborer working on the SG project while I’m off doing something else - like having a nice cold beverage!

        Here is a picture of the 3D printed uprights and a coil winding jig:


        • #5
          Winding the Coil

          When it came time to wind the coil I followed the instructions almost exactly. My two exceptions are:
          1) My trigger wire is red instead of green, and
          2) My spool is 3D printed (of course)

          Otherwise I laid out the wires, twisted them together, and wrapped the resulting cable onto the spool, all as described in the beginner’s book and YouTube video. I printed a small fixture and handle to help with winding the coil, but it really wasn’t necessary.

          The welding rods for the coil’s core gave me pause. While I did find the rods in bulk at a good price, that left me with the problem of cutting them into accurate lengths. In the end I opted to purchase the cut rods from TeslaGenX. To me, the small extra cost was well worth the time and aggravation. And the quality of the purchased rods is much better anyway. Thanks, TeslaGenX!

          Gluing the rods into the coil’s core went easily and exactly as described. It was a great feeling when the coil was finished.


          • #6
            Mounting Magnets

            I purchased the suggested ceramic magnets (1” x 2” x .5”) from a low price online supplier. I ordered several extra magnets because I know they are brittle. I expected one or two to be damaged in shipping, and another one or two from poor handling on my part - and that’s about how things turned out.

            Spacing the magnets evenly on the rim turned out to be easy as the spokes of the wheel provided excellent guides. I cut some cardboard into a small “L” shape to help me center the magnets (left/right) on the rim. Initially I used hot glue because it set faster than the super glue. I wrapped the wheel and magnets with strapping (shipping) tape and was happy with the progress. However, the hot glue did not hold for more than a couple days. I followed up with super glue and all has been well since.

            Having a true running wheel did not guarantee a balanced wheel with the magnets mounted. The wheel definitely had a heavy point and always wanted to rotate that part down. I used a bolt and a few nuts to balance the wheel. Some I glued to the inside of the rim, some I placed in the hole where the tire valve stem fits. The wheel now rotates and stops at any location with no preference for up or down.

            Alas, static balancing must not be the same as dynamic balancing. When the wheel is running at high speed it has a fair amount of side to side wobble. Part of it is probably from using plastic instead of less flexible wood for the uprights. But there must also be some magnets that are slightly off center or perhaps heavier on one side. The wobble doesn’t appear to affect how the machine runs, but it’d be nice if it was running true. Maybe on the next machine - I plan to leave this one as it is.


            • #7
              Front Panel Assembly

              And now I had the task of taking a couple handfuls of parts and mounting them on the aluminum plate.

              As mentioned earlier, I decided to cheat and include the intermediate and advanced circuitry into the build right away. This included matched transistors (from TeslaGenX), matched 100 ohm resistor (sorted by me), a 1K potentiometer, and a switch to select radiant mode or generator mode. I placed the parts on the aluminum plate, found a good layout, and started drilling holes.

              The transistors were mounted to the plate with electrically insulating but thermally conductive insulator film and small amounts of heat sink grease. I placed a strip of electrical tape under the transistor leads to reduce the chance of unwanted shorts. Electrical tape also went around the edges of the aluminum plate so that any sharp edges might be less likely to damage insulation. Then I wired it up point-to-point style, frequently using the sturdy transistor leads as supports. (It will never pass military soldering standards, but then again, it will probably never leave my basement!) I frequently and carefully referred to the schematics in all three SG books to be sure everything got wired as desired.

              I like how it turned out. To my eye it has the 60s retro look of a secret government project, or a mad scientist’s lab / lair. (Can you hear it? In the background, far off... “It. Is. Alive!”)

              Here is a picture of the front panel after assembly to the frame and coil:


              • #8
                Assembly and First Run

                At last my collection of parts was starting to look like a SG machine. There’s the frame with a wheel and magnets, over there is the front panel, and over there is the coil. My next task is to assemble them together.

                The first step was to remove the wheel to have access to the area around the coil. With the welding rods extending below the spool, it wasn’t possible to slide the coil in from the side. Not a big deal, just unexpected. I added a random collection of shim spacers under the coil that looked “about right”, knowing that changes will be needed. With a few screws the coil mounting task was complete.

                The front panel mounted easily to the frame on the first try. (gotta love that CAD accuracy) Wiring the coil to the panel took some time but did not present any new challenges. As before I referred to the three manuals frequently to be sure everything was wired correctly. I knew that one wrong connection could bring hundreds of amps from a battery. That would not be a happy day.

                I used a lab power supply (30V, 3A) instead of a battery to do the first power-on test. I turned the current limit low so it was less likely to damage anything in case of a wiring error. Fortunately everything was wired properly and no magic smoke appeared. I shorted the output wires (no charge battery) and gave the wheel a spin. The power supply quickly went into over-current limit mode, even with the limit turned to the 3 amp max. That seemed odd. (I later measured the peak input current to be around 4.5 amps, explaining the power supply’s behavior.) I tried adding some large capacitors but that didn’t help. Finally I replaced the power supply with a battery (lawn and garden, U1 size). Now I heard the turn-on thump as described in the beginner’s book. I gave the wheel another spin and this time it started to pick up speed.



                • #9
                  Setting the Gap

                  Now that the wheel was running, albeit not well, I enjoyed a few hours of just playing with it. I’d bring it up to speed, shut down, and repeat with some minor change to the setup. I soon noticed that the machine responded quite differently when the output wires were shorted vs. when a charge battery was attached. Since then a charge battery has been connected for every run.

                  I connected an oscilloscope to the trigger wire (between the pot and 100 ohm resistors) and spent quite some time getting familiar with the waveform and better learning how the machine operates. The scope clearly showed multiple switching at low speed, and how to “shift gears” with the pot to get into single switching.

                  A few days later it was time to adjust the coil-to-magnet distance, or “air gap” as I sometimes call it. And then I noticed a big goof in my mechanical design. Because of the welding rods extending below the coil, I couldn’t just slide the shim spacers in and out. Instead, I’d need to remove the wheel to lift the coil enough to remove or add the shims. Oops. My workaround was to make new shim spacers that have a slot to fit around the welding rods. The new shims allowed me to make adjustments while keeping the wheel in place.

                  The intermediate book recommends adjusting the coil to magnet distance for maximum speed at minimum input current. I obtained an analog ammeter for measuring input current and an optical tachometer for measuring the wheel’s speed. Over several runs I measured the input current and wheel speed at various air gaps. Calculating “RPMs per Amp” showed that smaller gaps made things worse, and that I needed to increase the coil-magnet distance. With larger distances the wheel speed increased until it became difficult to achieve single switching, and the machine just felt unstable. (hard to explain) I decreased the air gap about .025 inches to where the machine ran stable and called it good. The coil to magnet distance now measures about 0.145 inches.

                  If you look closely at the picture in post #7 above you can see the white and black shim spacers under the coil.


                  • #10
                    Arduino data logger and controller

                    It was about this point in the project when I went off syllabus for a while. While it would be really nice to have a good data logger to measure battery voltage during a run, I didn’t want to spend that kind of money. My answer was to use an Arduino, a small microcontroller originally designed for teaching coding and electronics in schools. Details at

                    I paired an Arduino UNO with a data logger “shield” (in the old days we’d call it an expansion board) from Details here:

                    This shield includes an onboard real time clock, SD card interface, and prototype area for making a small custom circuit. I built an input voltage divider and noise filter here since the Arduino analog input is limited to 5V max. I use the Arduino UNO and data logger to measure and record battery voltages to the SD card, typically at 10 second intervals.

                    Adafruit has an excellent article/tutorial on this data logger shield here:
                    Their whimsical project detects if the light in the refrigerator really goes off when the door is closed - Because sometime you just gotta know! My software is based on their example code, with relatively minor modifications where needed.

                    Then I considered that I could have the Arduino also control the SG machine. To do this I added a relay shield from Details here:
                    Now the Arduino can switch relay contacts that I use to perform various functions. I wired one relay in series with the SG power input / run battery. This lets me stop the SG when the charge battery is fully charged, or reaches a set voltage, or meets any other desired criteria. I wired a second relay to a light bulb connected to the SG charge battery. The light bulb is used to discharge the battery at about a C/20 rate. A picture of this setup is shown below.

                    The Arduino is designed to stack shields on top of the processor board. In the picture, the UNO is on the bottom, the data logger is in the middle, and the relay board is at the top. The pushbutton is there to start a run sequence. The pushbutton isn’t needed; the project just evolved that way.

                    In a typical run, I have the Arduino programmed to run the SG until the charge battery is fully charged. (The run battery on the left has a small battery charger attached to keep it up.) When done, power to the SG is switched off. A four hour rest period follows. Then the light bulb is turned on until the battery’s lower voltage limit is reached. Another four hour rest period is performed before finishing the cycle. At that point I read the data on the SD card and typically start another cycle. Sweet!

                    But the best part of this setup may be the data it produces. The data logger shield writes to the SD card in .CSV format. I load this into a spreadsheet for study and further processing. At this point it is easy to make very nice charts such as the one below. In this typical run it is easy to see where the battery was charging, resting, discharging, and resting again. (Labels added for your convenience.)

                    This setup has served me very well. While I know the accuracy isn’t as good as an expensive instrument, the price and controller feature more than makes up for it.


                    • #11
                      Thanks for posting the Data Logger Shield that will be a great help for ppl experimenting with Bedini Tech of all kinds..... We Needed this a LONG Time Ago......


                      • #12
                        Hi RS_ Thanks for the kind words.

                        Yes, the Arduino has been very handy. Watch for more charts to come.

                        I have a nearly identical Arduino setup doing battery cycling per the Battery Secrets and Battery Rejuvenation videos. I’m getting great data and results on that project also.


                        • #13
                          Battery Conditioning

                          At this point in the project things were moving along nicely. The machine was running well, with speed and current draw about the same as the measurements given in the books and videos. My next step was to add a capacitor discharge circuit. But I knew that would take some time to build. It’s a shame to let the SG sit idle that whole time - there must be something to learn from it. Ah, how about running some battery conditioning tests!

                          Chapter 6 of the Intermediate book describes in detail observations made by John Bedini and Peter Lindemann that a battery charged with pulses needs significantly less time to charge after several daily cycles. I wonder if the SG can duplicate their results…

                          I verified that the wiring on my SG, Arduino, and batteries was correct. As described in an earlier post, the Arduino measured and recorded the charge battery voltage at 10 second intervals. Relays controlled power to the SG and also turned on the load light bulb for discharging the charge battery. The run battery was backed up with a small 3 amp mains powered battery charger. All good.

                          I gave the wheel a spin and everything started up fine. Over the next 20 hours (approx) the system charged the battery, let it rest, discharged it, and let it rest again. The charge and discharge rates were about C/20 but the cycle took significantly less because I was discharging to only 12.5 volts. I had already hurt this starting battery, and didn’t want to do more damage by discharging too deeply. A side benefit was that the test required attention only once per day.

                          The next day the Arduino was blinking its LEDs (located on the data logger shield) to indicate a complete cycle. I downloaded the data and started a new run, leaving all setpoints and batteries as they were. I continued this for six consecutive days. At the end of the series I loaded all the battery voltage data into one spreadsheet to graph and print. The result is the multi-line chart below.

                          The six lines are the battery voltages for each charge cycle. The first run is the brown line on the right. Following runs display from right to left, with the last run (light blue) on the far left. All start times are lined up at zero hours. It is clear that with each run the battery charged in less time that it took in the previous cycle. This agrees well with John and Peter’s results with their golf cart batteries.

                          Another item that John and Peter observed was that this fast charging effect went away after about 100 hours. I let my SG and batteries rest for 4 days. Afterwards I ran a cycle just as done previously. The resulting data was clearly different, as shown in the two line graph below. Here, the line on the left - the faster charging one - is data from the final run in the earlier series. The line on the right is the battery voltage after a 4 day rest. Clearly the fast charging effect went away, just as it did in the golf cart batteries.

                          I have no idea what causes this temporary fast charging effect, or “battery conditioning.” But I can see how it can be a very helpful “trick” when working to improve the efficiency of the SG machine.

                          This work was done in radiant mode with no capacitor discharge circuit. I hope to someday repeat this test with capacitor discharge in the circuit.


                          • #14
                            Capacitor Discharge Circuit

                            The more I thought about using the 555 timer based cap dump circuit described in the intermediate handbook, the less I liked it. Sure, it could work great IF the wheel speed and battery voltages are constant. But in real life? I may be constantly tweaking it to keep things running in the sweet spot. I’d rather let the machine do as much of the work as possible.

                            I decided to go with the comparator circuit posted by RS. See post #27 here:
                   Thanks RS! I did not include the modifications by Nityesh. My own modifications were to:

                            * Delete diode bridge D1 because the diodes in the SG perform the same function,
                            * Add one LED in series with the input to the opto-isolator, and
                            * Change R1 from 1.2K to 1.0K to somewhat compensate for the added LED.

                            Designing my circuit board involved finding available parts to purchase. The big capacitors offered a challenge. The SG books frequently talk about “photoflash” capacitors. Apparently that term is no longer used in the industry. Instead, I searched for “low ESR” (equivalent series resistance) capacitors. Also, I couldn’t find the suggested 15,000 uF, 80 volt caps. I settled for 63 volts and so far things have gone well.

                            I’ve been wanting to get a table top CNC router to make prototype circuit boards. When Amazon had a great deal on one, I jumped on it.
                            Assembly went well. I found and learned the necessary software to run it. Looking good.

                            And then I tried to route a board. I’m sure I’ll be able to do it someday, but it will take some time to figure out the best cutter, speed, feed, and depth of cut. But for now I needed a work-around.

                            And then I remembered seeing some wiring in old 1960s era military equipment. Hmmm.

                            I purchased some 0.063” phenolic from here:
                            I had the CNC router drill the holes in the proper locations and sizes, and then cut the board perimeter. The components (all thru-hole) fit perfectly. On the bottom side I did point to point wiring. One big advantage to this was that I could use large size (12 awg) wire in the cap discharge path. See pics of the assembled board below.

                            For mechanical mounting, I 3D printed a frame that fits on the back of the SG. The frame holds the circuit board and also stiffens the SG base. A picture of the mounted board is below.

                            The circuit operates just as desired. The caps discharge when they reach 24 volts. The discharge stops early, as intended, and leaves the caps at 18 volts. The discharge frequency varies automatically with wheel speed and single/multiple coil switching.


                            • #15
                              Thanks for posting all this. I am really enjoying reading about this build.