Adjustable Ballast
Rough and Ready Ballast
First up is a V /I plot for the "rough and ready ballast" which consists of two MOT primaries on a single MOT core, primaries in series. The non-linearity in the curves is due to core saturation.
How linear does it need to be?
Next is a V/I plot for a 15/30 NST with the secondaries shorted compared to the "rough and ready" ballast using both one and two primaries at 120V (0-130). Note the linearity of the ballast is better than the NST when both primaries are in series keeping the core out of saturation. (more turns/volt)
Saturation Infatuation
Next is a saturation curve for the 3.5 square inch MOT core used in the experiment.
Next is a saturation curve for the "double wide" MOT core (two identical cores combined into a single wider one) with 7 square inch core area.
These curves were done with ten turns of #10 AWG THHN. Current is vertical, voltage horizontal. From this it can be seen how many turns per volt are needed to stay out of core saturation. The curve starts to turn up at 16V, (double wide core) so I need 1.6 turns/volt to keep the ballast linear. For operation at 240V that is 384 turns, more or less.
Double wide core Ballast
The next graph is the V/I plot for the finished "double wide core" ballast made from two identical MOT cores and wound with #11 AWG 200 C magnet wire. Forgive me, I did this last July and do not remember how many turns I was able to stuff into the winding window (not enough), but just enough to keep linearity up to 240V, saturation starts there. It is still more linear than a NST, so serves my purpose.
The last graph is just a plot of the inductive reactance vs gap thickness (plastic spacers, mostly thin sheets of GPO3) but you need to use your imagination to "smooth" the curve between the points. GNU Plot or whatever application I used to make the graphs seemed to lack a "smoothing" feature.
For the Image deprived
Here are some pictures of the disassembly of one of the NSTs and some pictures of the "testing stage" double-wide-core ballast. Of course it is very important to clamp the "I" part onto the "E" part very securely during testing. The finished ballast would of course have a suitable clamping structure built in, with the ability to "quick change" the shims.
Disassembly
Step 1: Install bolts to hold the "I" section together during removal.
The weld holding the "I" to the "E" needs to be cut.
Step 2: Cut the welds. I used a cutoff wheel in a die grinder.
Another view of the cut. It just needs to be as deep as the weld.
Step 3: Split the core into the "E" and "I" sections.
Step 4: Remove the shunts.
Step 5: Remove filament winding and loosen insulation around secondary.
Step 6: Remove the coils from the MOT core without damaging the primary. This part is fun. Have patience.
Missing are the pictures of the "Rough and Ready" Ballast. It was just a single MOT core with two primaries on it. It works surprisingly well for such a simple thing, but not linear enough for my purposes. Also missing are the pictures of the assembly of the double wide core. I needed to machine the faces of the "E" and "I" to get a smooth fit. Just a bit of a shave.
The Winding
Winding the new coil. This was a real pain. Suggest using caution buying magnet wire on Ebay. I got good dollar value on the wire, but it had been re-spooled at least twice and seemed to be work hardened making it a real fight to get the wire to lay in place. I resorted to "tapping" each wrap into place with a small ball peen hammer. I used 3M brand Scotch 27® thermosetting fiberglass cloth electrical tape to hold the windings together and to insulate from the core. The coil form was made of wood and too soft, so the inside corners came out a little too rounded and too narrow, so I had to grind round corners on the center leg of the "E" so the coil would fit without cutting the insulation on the sharp corners.
The Double Wide Core Ballast
Sitting around since July, this is April, yeah it got rusty. A little sandpaper will fix that. The double wide "E" with the new hand wound coil installed, the double wide "I" and some of the GPO3 shims are shown here.
The "I" sitting on the "E" with one shim.
Side view.
Side view with the "I" and shim removed.
My Powerstat has a 30A fuse, so I only tested to 30A or just a little more, but with the 200°C insulation and the Scotch 27® tape this ballast should be good for 40A with a 60% duty cycle, or better. This estimate is based on the measured temperature rise running at 30A 240V for 20 minutes, the mass and the surface area of a box with the dimensions of the core, and an ambient temerature of 25°C. Only time and real world testing will tell the true story.