Tweet
Silicon Carbide FET Experiments with SSG
Hi All, 
With all the recent discussions concerning Silicon Carbide FET’s, I decided to experiment with replacing the MJL21194 BJT’s on my oldest Simplified School Girl motor with them. The specs for this SSG are as follows: 5 strands of AWG18 by 155' litzed together for the power windings, no trigger winding, hall switch triggering with adjustable timing and pulse width, 26" bike wheel with 18 magnets.
Before the modifications, I ran a COP test as a base for comparison. Even though the wheel is well balanced, it does have some run out and the magnets are not exactly evenly spaced or tested for equal gauss. And the MJL21194's are not perfectly matched. So it’s not as well constructed as my newer SSG. Using 12v - 12AH SLA batteries in both the supply and charge positions, and with the hall’s set to deliver 19% duty cycle and turning off at 5 degrees BTDC. I got the following results running in Generator (common ground) Mode: 1.25 Amps average current draw for 51.75 minutes (.8625 Hr) reaching 16.0 Volts to replace the 1AH previously removed. The wheel speed started at 252 RPM and finished at 262 RPM. The COP was 1AH/( 1.25A x .8625 Hr) = 1AH/1.078AH= .927 COP
I then replaced the five MJL21194's with a single SiC FET and also replaced the neon bulbs with a TVS diode, and the 1N4007 Silicon diodes with Silicon Carbide Schottky diodes. Only minor changes were needed in the hall switch triggering circuit to go from base current triggering to gate voltage triggering. First test all five windings were powered by only one FET. It pegged the ammeter (3 Amp) and smoked the SiC Fet in about 5 to 10 seconds.
Second test I replaced the FET and changed the five windings from parallel to series configuration. Each winding is exactly 1 ohm. So in parallel they are .2 ohm and in series they are 5 ohms. This yielded about 2.4 amps draw at start up and dropped to about .2 amps at 108 RPM with very little charging of the secondary battery.
Third test I switched to 24 volts at both positions and repeated test 2 at the higher voltage. Initial current draw was about 4.8 amps and dropped to .4 amps at 120 RPM with again, very little charging.
Fourth test I installed five SiC FET’s with one FET and one TVS diode for each winding just like with the five MJL’s. This effectively places all the windings in parallel. (All five in parallel gives 1/5=.2 ohms total resistance.) I accidentally hooked this back up to 24 Volts by mistake instead of the 12 volts I intended. On start up it smoked three of the five FET’s, the ammeter, and the primary input wire insulation. (It was trying to deliver 24/.2=120 Amps at start up which is zero RPM.) ............ BUMMER
It finally dawned on me that when these SiC FET’s turn on they have almost no internal resistance and become almost like a mechanical switch. And with hall triggering they are either on or off unlike with trigger winding switching of a BJT which involves current gain and self servoing.
Fifth test I replaced all the damaged components and placed a 35 watt 12 volt headlamp bulb in series with the positive lead and started it up on 12 volts. The bulb lit up brightly flashing with each pulse and getting dimmer as the machine sped up and the reactive impedance of the coil increased. Next, I removed the headlamp bulb and gave the wheel a fast spin before I turned on the trigger circuit. This worked well with initial current draw of 6 amps quickly dropping to 2.5 amps. Now I had a procedure that worked!!!
Sixth test I charged both batteries on a 10A12 Bedini charger and then took 1AH out of one after resting and letting it recover to a standing voltage overnight. Performed a COP test with the following results: starting RPM of 282 and current draw of 2.5 amps, ending current draw of 1.6 amps and running 289RPM, reached 16 volts in 41 minutes, 2.05 amps ave for .683 hours to replace the 1 AH previously drawn, COP=1/(2.05A x .683H)=1/1.4=.714 This is well below the base test of .927COP
Seventh test changed the duty cycle to 15% and repeated test 6. RPM and current draw both dropped and charge time increased for a calculated .856COP. Better, but still less than the base run with the MJL’s.
Eighth test I changed the duty cycle to 16.7% and the turn off to 3 degrees BTDC and repeated the test procedure. This gave results very similar to the base test. 2.0 amp draw @ 220 RPM at start and 1.1 amps @ 262 RPM finishing at 16.0 volts in 42 minutes with an average amp draw of 1.55 amps. This calculates out at .92 COP.
Conclusions: These circuits are very responsive to duty cycle and discharge timing especially when using SiC FET’s or any FET for that matter. Triggering can’t start at zero RPM without excessive current draw or without buffering with an extra load to limit starting current. My goal is to use this type circuit in an attraction motor coupled to a flux gate generator. Just getting started with this but will get put on hold for three months at the beginning of 2024.
Gary Hammond,
With all the recent discussions concerning Silicon Carbide FET’s, I decided to experiment with replacing the MJL21194 BJT’s on my oldest Simplified School Girl motor with them. The specs for this SSG are as follows: 5 strands of AWG18 by 155' litzed together for the power windings, no trigger winding, hall switch triggering with adjustable timing and pulse width, 26" bike wheel with 18 magnets.
Before the modifications, I ran a COP test as a base for comparison. Even though the wheel is well balanced, it does have some run out and the magnets are not exactly evenly spaced or tested for equal gauss. And the MJL21194's are not perfectly matched. So it’s not as well constructed as my newer SSG. Using 12v - 12AH SLA batteries in both the supply and charge positions, and with the hall’s set to deliver 19% duty cycle and turning off at 5 degrees BTDC. I got the following results running in Generator (common ground) Mode: 1.25 Amps average current draw for 51.75 minutes (.8625 Hr) reaching 16.0 Volts to replace the 1AH previously removed. The wheel speed started at 252 RPM and finished at 262 RPM. The COP was 1AH/( 1.25A x .8625 Hr) = 1AH/1.078AH= .927 COP
I then replaced the five MJL21194's with a single SiC FET and also replaced the neon bulbs with a TVS diode, and the 1N4007 Silicon diodes with Silicon Carbide Schottky diodes. Only minor changes were needed in the hall switch triggering circuit to go from base current triggering to gate voltage triggering. First test all five windings were powered by only one FET. It pegged the ammeter (3 Amp) and smoked the SiC Fet in about 5 to 10 seconds.
Second test I replaced the FET and changed the five windings from parallel to series configuration. Each winding is exactly 1 ohm. So in parallel they are .2 ohm and in series they are 5 ohms. This yielded about 2.4 amps draw at start up and dropped to about .2 amps at 108 RPM with very little charging of the secondary battery.
Third test I switched to 24 volts at both positions and repeated test 2 at the higher voltage. Initial current draw was about 4.8 amps and dropped to .4 amps at 120 RPM with again, very little charging.
Fourth test I installed five SiC FET’s with one FET and one TVS diode for each winding just like with the five MJL’s. This effectively places all the windings in parallel. (All five in parallel gives 1/5=.2 ohms total resistance.) I accidentally hooked this back up to 24 Volts by mistake instead of the 12 volts I intended. On start up it smoked three of the five FET’s, the ammeter, and the primary input wire insulation. (It was trying to deliver 24/.2=120 Amps at start up which is zero RPM.) ............ BUMMER
It finally dawned on me that when these SiC FET’s turn on they have almost no internal resistance and become almost like a mechanical switch. And with hall triggering they are either on or off unlike with trigger winding switching of a BJT which involves current gain and self servoing.
Fifth test I replaced all the damaged components and placed a 35 watt 12 volt headlamp bulb in series with the positive lead and started it up on 12 volts. The bulb lit up brightly flashing with each pulse and getting dimmer as the machine sped up and the reactive impedance of the coil increased. Next, I removed the headlamp bulb and gave the wheel a fast spin before I turned on the trigger circuit. This worked well with initial current draw of 6 amps quickly dropping to 2.5 amps. Now I had a procedure that worked!!!
Sixth test I charged both batteries on a 10A12 Bedini charger and then took 1AH out of one after resting and letting it recover to a standing voltage overnight. Performed a COP test with the following results: starting RPM of 282 and current draw of 2.5 amps, ending current draw of 1.6 amps and running 289RPM, reached 16 volts in 41 minutes, 2.05 amps ave for .683 hours to replace the 1 AH previously drawn, COP=1/(2.05A x .683H)=1/1.4=.714 This is well below the base test of .927COP
Seventh test changed the duty cycle to 15% and repeated test 6. RPM and current draw both dropped and charge time increased for a calculated .856COP. Better, but still less than the base run with the MJL’s.
Eighth test I changed the duty cycle to 16.7% and the turn off to 3 degrees BTDC and repeated the test procedure. This gave results very similar to the base test. 2.0 amp draw @ 220 RPM at start and 1.1 amps @ 262 RPM finishing at 16.0 volts in 42 minutes with an average amp draw of 1.55 amps. This calculates out at .92 COP.
Conclusions: These circuits are very responsive to duty cycle and discharge timing especially when using SiC FET’s or any FET for that matter. Triggering can’t start at zero RPM without excessive current draw or without buffering with an extra load to limit starting current. My goal is to use this type circuit in an attraction motor coupled to a flux gate generator. Just getting started with this but will get put on hold for three months at the beginning of 2024.
Gary Hammond,
Comment