Jetijs,

Yes, I agree with you. The silicon-steel should have been able to do better. This result tells us two things. First, that your magnetic core material is returning energy at less that 50% of the silicon-steel. And second, that there may be a problem with the switching speed of your 2N3055's. I always use the MJL21194 by ON-Semiconductor. But I don't know why your 2N3055's aren't working better. Is it possible for you to try this test with a different transistor?

This was a great test to run. We were expecting a result that suggested that your core material was slow, but that the transistors were fast enough. Perhaps they are not doing as well as we thought either.

High speed BPT's are still the device of choice. MOSFET's are difficult to turn OFF quickly, so let's stay with BPT's for now. (Sorry, Elias).

We need to solve (or explore) one issue at a time, so maybe trying a different transistor on your energy return test set-up is a good place to start.

What do you think?

Peter

Peter,
I just browsed on the local electronic web shop and they seem to have the MJL21194 transistors. They are three times more expensive than the 2n3055, but I will buy one or two to make the tests, also I will have a use for them later. But I doubt that the transistors are the problem, I can't see any fall time on my scope, only rise times and off periods, this tells me that the transistor switches off fast enough, right? Also I used these transistors on some SSG setups and could get some very good results.
I just found some silicon steel open core material form an old contactor relay, I will also do some tests on that material, since this core is open and not closed like the induction motor startor core, maybe this will give some different results. After these tests I will post my results and then we will decide what to do next. I am not gonna quit on this!
Thanks,
Jetijs

Jetijs,

I agree that the scope shots look like the 2N3055 is turning off fast enough. But in a situation like this, everything must be looked at carefully, and all assumptions questioned. The MJL21194 has a history of working well in circuits like this, so let's see how they behave in your test.

Glad to hear you plan to stick with it.

Peter

Yes thanks for pointing that out. I was thinking of logic circuits instead of power circuits. Finding a way to discharge the base of the bipolar transistor faster would turn it off more quickly and increase the output even more.

Hi Elias,
does this mean that this circuit:



is producing more energy than was stored as 0.5 x C x V^2 in the beginning in C1 ?
If yes, you would have already creazed a circuit with COP >1 !

Please let us know more about this experiment.
Many thanks.
Regards, Stefan.

Hi Jetijs,
is your stator iron core solid iron or
laminated iron ?
Maybe eddy currents play a major role here,
that you can only get 24% back out ?
Also , did you see, how much you already are in saturation
in your BH curve of the iron core ?

Peter, is it better to drive the iron into saturation
or not ?
Also it probably depends a lot what kind of
core material you are using.
A low hysteresis metgals alloy would probably be much
better than just iron...
Maybe you also have too much hysteresis losses at the moment.

Regards, Stefan.

Stefan,


Welcome to this forum. This is unrelated to the topic of this thread so lets move the discussion to the corresponding thread:
Bedini's SSG Secrets
This circuit is an abstract form which yields to more charge, not more energy. But in my opinion it has the potential to produce more energy if we use better coils + better core materials + battery instead of a capacitor + correct timing of the pulses.

You are right.
I answered you over here and asked the energy question:

Bedini's SSG Secrets

Please answer this.

Many thanks.

Stefan,

Since we are trying to recover the maximum energy from the collapse of the magnetic field, keeping the iron below saturation is best.

Also, you are probably right that low hysteresis materials like MetGlas would work the best in a motor like this. But right now, we are just exploring the limits of using low cost materials like cast iron or silicon steel laminations.

Thanks for bringing these ideas up, since they had not been discussed very much yet.

Peter

Hi Jetijs,

I did an experiment to see how much output I could get out of pulsing a coil. The picture of the coil used is attached. It is pretty small (about 1cm in length). Probably it is about 5-10mH or less. Its core is about 1mm in diameter and is apparently made of ferrite.

For this purpose I used a 10,000uF capacitor for the input and a 10,000uF capacitor for the output, across the diode. I charged C1 to 12 volts and by pulsing it with an approximately 20% duty cycle at 5KHz I got the output capacitor charged to about 9.15 volts, while some charge was remained on the input capacitor at the end (about 0.5 volts). So I concluded that the efficiency of my pulsing system is about (9.15/12)^2 = 58%. Note that as the charge on the capacitor decreases the spike gets lower in amplitude and reduces the charging effect. I used a 900v FS7SM MOSFET and a square-wave signal generator for this purpose.

Hope this helps ...

Elias

Attached Images

coil.JPG (12.4 KB, 28 views)

Hi

I played with the function generator a bit and found something interesting while connecting a lamp to the output along with a capacitor. See the attached circuit. There is a sweet spot for the minimum current drawn from the battery depending on the coil used. it is somewhat a bell-shaped curve. I experimented with the same coil and the result was interesting as it provided minimum current (29mA) to the inductor at about 28KHz and it increased when either increasing the frequency or decreasing it. In my setup I had:
Vin = 12V, Iin = 29mA
Vout = 10.20V, Iout = 27.5mA

The data above yields a recovery of about 80% of the input and it seems promising.
But when using capacitors only as the previous experiment I did at 5KHz, but now at 28KHz 50% duty cycle, I did not get more than 6 volts on the output capacitor. It seems also that lower frequencies with less duty cycle are better than higher frequencies with more duty cycle.

When the input voltage was 18 volts the minimum current drawn from the battery was at 31.5KHz:
Vin = 18.2V, Iin= 38mA
Vout = 16.36V, Iout = 35mA
In this case the recovered energy was about 81% of the input.

I experimented with a larger coil and that one gave me 76% efficiency at its sweet spot which was 2KHz.
In all of these experiments I used a duty cycle of 50%.

In conclusion I have to say that the motor seems to be able to recover the most in a certain RPM, or certain amount of pulses per second. Maybe we can design a system which can stabilize the frequency of the pulses independently from the RPM of the motor. increasing the input voltage, increases the recovered energy a bit.

Regards
Elias

Attached Images

recovery-circuit.JPG (14.0 KB, 38 views)

Are you talking about the motor I am working on? If so, then you have missed, that we already have an independent frequency of the pulses regardless of the RPM's of the motor. The switching is done by a 555 timer.

Ok, I have done some tests today. I used my motor circuit to pulse the coil with a welding rod core. I made these tests with 5 different frequencies. I used two 12v batteries in series (24V) on the input. In the first test I used a 24V battery in the output and pulsed the circuit with one 2n3055 transistor to get the basic results so that I can compare the with other tests. The switching circuit with the 555 timer and the optotrigger consumed 0.09A, that must be taken into account. For the second test I used the MJL21194 transistor soldered on the same base plate parallel to the 2n3055 transistors. This time the switching current was 0.11A, because the PNP transistor now had to switch four transistors instead of three. Basically I got a little bit worse results compared to the 2n3055 transistor, bat I think that this is because now the circuit had to open and close four transistors in parallel thus consuming more current. After all both transistors are working just as good and we can safely say, that the problem is not th the transistors used. In test number 3 I used a light bulb in the output and the 2n3055 for switching. This time I got lower voltages and overall efficiency. And in the last test I attached a 10000uF cap in parallel of the bulb in the output, this increased the efficiency a little bit. Here are the results:
Test1

Test2

Test3

Test4


Any comments, suggestions?
Thanks,
Jetijs

Yes and that's nice, but you didn't mention the frequency at which you pulsed your coil? My test data shows that there is a frequency for any coil that the efficiency gets maximized. Have you attempted to adjust the frequency of your pulsing? I have been able to get about 80% back with those coils, but maybe larger coils don't put back as smaller ones do. I'll do more experiments with different coils and see what happens.

Elias,

Thanks for running these tests. I have been saying that I thought 80% recovery was both possible and the goal of this project. If the motor can recover 80% of its input electrical energy AND produce 80% mechanical energy at the shaft by careful timing and small air-gap, then the motor should be able to operate with a COP=4.

Your tests show that 80% electrical recovery is possible with proper switching.

Thanks again,

Peter

elias,
yes I tried various frequencies with the 555 timer. I think, that the optimum frequency sweetspot is somewhere around where the pulse lenght is eaqual to the current rise time in the coil and that varies with different coils. I should get a function generator for these tests, because it is hard to adjust the frequrncy correctly with the 555 timer.

Peter,
are there any more tests you want me to do before I move on with a different aporach? Because with all the tests I have made, it does not look like that the problem with a low recovery is in the transistors, timing or frequency, it must be in the core material. If there is nothing I can do about this design anymore, then I have some other ideas
Thanks,
Jetijs

Lost of good experiments going on there lately!
But it might be time to move onto the more 'big boys' capture techniques.

You need to pulse both the pos rail and neg rail together.
There is a lot to this but to put it simply you have much limitations in how you are going to collect this energy.

Quick example.
Imagine this energy is a compressed spring, it is supported at one end by a red bracket and supported at the other end by a green bracket. Well if you are going to suddenly release its compression only at the green end, you have still got pressure at the red end as the spring keeps expanding. Ok, now lets suddenly remove both the green and red end together, there is no more pressure in our supporting frame which means the system is completely free and isolated from what the spring is doing. Mechanical and electrical all have the same effects.

(Side note: The purposed technique is not mine but from a well respected engineer I work with)
Steven

Attached Images

TwoSided.jpg (107.1 KB, 59 views)

Jetijs,

We started this investigation with the assumption that the problem w