Pierre's 170W in 1600W out Looped Very impressive Build continued & moderated

[T-1000]

@L192 Hi
It is nice you still keep it going.And when looking all way back where essence is you might actually try to make coil arrangement where magnetic field could smoothly shift from one driving coil to another for simulating moving magnet. Like back in old TPU days... The rule of the thumb for induction is the rate of change and the drag for current strength.

P.S> Personally I paused a project due other priority things in life but will continue on it at some point as well.
Good luck!

[listener192]

Hi T-1000,
Thanks for your encouragement.
The geometry of the DZ does not support energized stator coils to achieve flux cutting in the rotor, as there is no perpendicular path.
I have looked at architectures that would support flux cutting utilizing a stator and the only one I can describe, uses a circular rotor with teeth around the periphery.

If you energize coil on such a rotor, the flux will cross into the stator flow left and right around the stator.

Consider a small flux level.. the flux will take the shortest path around the  inner edge of the stator. As the flux level is increased, the flux lines compress and will move outward to increase the flux density in the mid portion of the stator.

Add even more flux and the outer portion of the stator becomes full with flux lines. This process progressively cuts the coil windings on the stator teeth at 90 degrees, so this becomes the direction of movement using right hand rule and because the flux will flow through both sides of the stator coil in registration, the direction of electron flow (looking up at the coil from the middle of the stator) is anti-clockwise on both coil edges. 

Physical movement is not necessary to achieve the above however, what benefit this derives I am unsure as both flux linkage and flux cutting are subject to lenz.In the above example, as a current is induced in the stator coil due to a connected load, lenz law will cause a counter flux to be setup that will oppose the flux coupled from the rotor.

In the case of the DZ, my current thought is on being able to make the stepped waveform behave like a sine current in a flux linkage configuration. Once that has been achieved i.e. normal transformer operation, then the magic would be in the recovery of the energy stored in the magnetic field, given up every time a coil turns off. So 100W input yeilds  say 85W output but 80% of this is recovered ..68WInput =100W-68W =32W for 85W output.

At the moment the coils behave as DC solenoids, once current has risen in the coil, in particular when they are on for several step periods. This wasted energy appears as heat due to the coil IR losses. At the moment, the only way I can see to counter this, is to keep inductance present, to limit the current by keeping the current changing during the coil on periods beyond the initial current rise. This is the idea of the sine current application. Using half sines as the supply rail for the switches may also achieve the desired result. Initially I dismissed this due to an experiment conducted using the bridge boards however, there were two many other problems that stopped that attempted working properly.
The impediment to making this work is the synchronization of the stepped waveform with the zero cross of the half sines. Before I used an interrupt structure with case statements and a zero cross detector. This allow the interrupt to call the start of the stepped waveform routine. There was a problem however described below that was never resolved.

My current scheme uses delays in the code for the HSS delays, required as the opto couplers are slow to switch the MOSFETS. The use of delay in an interrupt driven routine is not possible as they use internal timers with interrupts, and these get turned off. Perhaps polling a pin for zero cross would be better, as delays could be kept for timing the steps.

If you model the 150uF cap that Pierre has across his FWBR, you see well developed half sines.  This is what he is feeding his relays with, so I believe this is significant.


Regards
L192

[FixedSys]

The geometry of the DZ does not support energized stator coils to achieve flux cutting in the rotor, as there is no perpendicular path.

Is it possible that in addition to linking; flux cutting is achieved through the distortion of flux from armature reaction and / or other fields?
It would appear that flux distortion is the only way possible to properly move flux to achieve cutting. Perhaps there's distortion occurring in sympathy with the 60 Hz or harmonic, and perhaps the OU is found in a combination of linking and cutting?

[listener192]

Flux cutting occurs in a generator with a physically rotating rotor, irrespective of armature reaction.
The rotor flux flows into the stator and distributes left and right around the stator. As the rotor  moves across a given coil the flux level increase and cuts both edges of the coil however, it is the increase in flux density across the coil that causes induction, as this is the effective movement into the coil (right hand rule applies). If the rotor was static, flux cutting could still occur if the flux level in the rotor was changed by a rise and fall in rotor coil current.


In the DZ, the flux from a stator coil enters the rotor but there is no path at 90 degrees to the rotor coil, so the flux passes through the center of the coil, changing the flux density hence producing induction by flux linkage.

Flux cutting offers no advantage over flux linkage.

L192

[listener192]

Line Synchronization:
I tried polling a digital pin to look for the rising edge of a zero cross pulse but this was way too slow.
I have achieved locking the stepped sine waveform to an interrupt. I moved the stepping code from the loop and placed it in a named ISR. The loop remains empty.I replaced any delay commands with delayMicroseconds, which don't use interrupt timers.
Just using a square wave from a function generator (purple waveform), I locked the stepped sine to 50Hz.
There is a bit of phase shift between the two signals which is because my original code did not start the waveform at zero. The pot input delay setting was used to get the frequency in the ball park.

I have a zero cross detector to connect next.

L192