Self running coil?

[gravityblock]

Gyula,
I think it's time to wind a new toroid. I hate winding toroids

A user at the Energetic forum asked me if we had 4 coil sections instead of 2 would the inductance be more than 2 coil sections?  I said I don't know   what do you think

Thanks for your help

Luc

This has crossed my mind also.  It's a really good question.

GB

[gyulasun]

Gyula,

I decided to reconnect the no magnet toroid and re-tune it to the best of my ability to use the less current but also keeping the RMS value to the same as the magnet dual coil toroid test above (see scope shot below to compare both). I was able to get the no magnet single coil toroid down to 1.8uA compared to the magnet one above at 0.8uA

There is a small gain using the magnet but by keeping the inductance on the dual coil to the same as the single coil we don't have much much of the PM flux participating in it. Just one 1/2" x 1/8" plus two paper spacers. We need to consider this also.

I think it's time to wind a new toroid. I hate winding toroids

A user at the Energetic forum asked me if we had 4 coil sections instead of 2 would the inductance be more than 2 coil sections?  I said I don't know   what do you think

Thanks for your help

Luc

Hi Luc,

Thank you for the test.  I asked it because normally at such low audio frequencies toroidal cores give higher Q at a few kHz higher frequency (6kHz --> 8kHz) and, besides, higher frequency involves higher inductive reactance too (in the inductive reactance formula the reactance linearly increases with the frequency, (X_L=2*pi*f*L), so does the resonant impedance, Z=Q*X_L.

And if the resonant impedance is higher at a higher frequency wrt a lower frequency, the MOSFET will draw less current at its drain side when switches ON, this is the same as if you were using a ,say, a 50 kOhm resistor in the drain (instead of the coil) at 6kHz and then you were using ,say, a 70 kOhm resistor at 8.3kHz: obviously the the current draw would be less in the latter case.

And your split wound coil definitely must have a lower self capacitance with respect to the normally wound coil, I estimated the 533pF as being a close match to reality: your split wound coil may have 533pF LESS self capacitance than the normally wound.
This means that if you split the coil into ,say, four parts (for four quarters) you surely will have a coil with even less self capacitance. However, I can only GUESS whether the inductance in this case increases four times instead of the double value, maybe yes.

I think it also would be useful to test the two coils (the ones you kindly tested for me yesterday) in the following way:  please see Diagram 4 in this link: http://www.st-andrews.ac.uk/~www_pa/Scots_Guide/experiment/lab/expt3/expt3.html

You could use the same C capacitor for both coils in parallel, its value could be anything between 1000 and 1200pF what you can find and combine, at 6kHz the 635mH coils should resonate with 1108pF.  The R resistor in the Diagram shows your coils DC resistance, no need for putting there anything.

It is possible you would have to use 470kOhm or even 1MegaOhm series resistor instead of the 100kOhm shown because the resonant impedances received from toroid coils should be in the several hundred kOHm range.

PLEASE use your Fluke voltage meter in AC to see the voltage V_out at the output , I do not know your Fluke AC input impedance (maybe 1MegaOhm with 20-50pF parallel capacitance, just use a 1-2pF series coupling capacitor at the V_out output to reduce any detuning and loading effect to a minimum.  USE the SINE wave setting on the gen (but if you curious, switch to square wave too  )

You would first search for the resonant maximum output voltage, starting with the normally wound, no magnet coil, paralled with the 10..pF cap. Try to adjust with the signal level control just 10V RMS at the resonant maxed output (input level actually not important now, only the output from the resonant circuit).  The reason why I ask the 10V RMS output at resonance is shown here:

http://www.allaboutcircuits.com/vol_2/chpt_6/6.html#22055.png

Now if you detune the signal gen first to a lower frequency so that the output voltage changes gradually from 10V to 7.07V RMS, you notice this frequency, ok?  It is possible you need to use a digital frequency meter to better read this lower frequency, if your dial on the generator is rather rough for finer tens or hundreds Hz details. Please tune also to an upper frequency where the output voltage also reduces gradually to 7.07 RMS, and notice this frequency too. 

Now if you substract the lower freq (say it is 5.8kHz) from the higher freq (say it is 6.2kHz), the difference is 6.2-5.8= .4kHz   The Q value is received if you divide the resonant frequency (say it was exactly 6kHz) with the difference:  Q=6/.4=15  it is much possible your toroidal coils will have a much narrower 3dB bandwidth, it means the Q values will be higher,  this is why a digital frequncy meter is a neccessity when you detune to the half power points on the resonace curve.

This way we could have a much better inside onto the two different coils, regarding their Q, resonant impedance and self capacitance. And this test setup creates a different circuit 'enviroment' for the coils, namely no nonlinear dynamic MOSFET capacitance.

If you do not understand anything wrt to this test, please ask.

rgds,  Gyula

[gyulasun]

It's clear that it does more. Imagine that the magnet brings the core close to saturation. If you are feeding a sine A into your coil, without the magnet it would create a magnetic flux which changes its direction as the current changes its direction. Let's say, you get a megnetic flux inside the core which changes between -100 mT to +100 mT. If the core has a saturation flux of 1 T, and with the magnet you create a flux of 950 mT inside the core, you get a bias from the magnet, and if you apply your sine wave the flux is centered at 950 mT with +/- 100 mT, but because the saturation flux is 1000 mT it will be clipped. So you have some kind of asymmetric behavior here.

This could, as i already explained some days ago, also have the effect of 'magnetic rectification' which might explain some of the strange effects.

Hi Skywatcher,

I also say the saturation is surely involved and the nonlinear behavior of the toroidal core with the magnet attached makes understanding much harder.

Traditional toroidal ferrit cores normally has a 0.4-0.5T saturation limit, modern cores obviously better with 1-1.5T limits.

However, I do not think everything is to be blamed by saturation. See Luc first picture in his reply here: 
http://www.overunity.com/index.php?topic=8892.msg235414#msg235414

The voltage should show a nice sinewave but its upper positive half is  narrower than the lower one. And this core did not have any magnets attached: the distortion must have come from the lack of the optimal bias for the MOSFET to work on its linear transfer line, it acted as a switch.

But the distortion is more pronounced for the core with the magnet in his bottom picture there, using the same switching condition and I suspect the resonant impedance of the coil is also higher-- another reason for getting distortion in the output signal. This may turn out when Luc can test the coils without the MOSFET.

rgds,  Gyula

[Magluvin]

I hope this doesn't seem unrelated, but I'm working with Luc's setup and trying variations to which this applies.

I worked out a circuit to get an AC square wave pulse to drive my parallel resonant circuit for the coil, however, it ends up with:

555 timer x 1
p-mos x 1
n-mos x 5

I spent quite some time foraging through google before I came up with this solution - however it seems ungainly (a lot of components!  I'm new to electronics in general) - does anyone know of an easier way to turn a 12V source into an AC square wave?

You can view the circuit at

http://www.falstad.com/circuit/

And then import the schematic I'm attaching to this post.

Also, I appreciate all of the work that's being shared here, I know I haven't posted much, but I try to refrain unless I have something new and or original to say.  Most of what I'm seeing others are seeing and posting in more than sufficient detail to warrant a post from myself   Thanks all

Hey Void
Good job.  I had talked about this effect a while back with this sim. I wonder if it is an artifact with the sim. But it is exciting to see it go.

Thanks

Mags

[gyulasun]

Hi Gyula,

Well, the standard formula to calculate L,C or F cannot predict C on it's own.

My point is, that it must be somehow possible to get a quite close estimation by using data from the data sheet of the particular mosfet, to -on beforehand- determine what it's 'final' Capacitance 'result' would be when being plugged into Luc's A.

The 1578.54 pF is derived from the 5 to 50 Khz scope shots Luc provided... If you put his main coil mH and the belonging frequency for each scope shot in the online LC calculator, you will see that all scope shot frequency's/mH's come to between 1533 and 1580 pF
So if we can get a good estimation of this C value,  we might in the first place get a better understanding between Ciss, Coss, Crss, Rds(on), Vgs(th) and Vgs, wrt the A, and by such we would be able to find the 'perfecft' match mosfet.

In the 2nd place, it *might* be such, that for optimizing the A to 100 percent [well as close to it as possible], we need to know the mosfets C value to begin with, meaning the A would be fine tune designed around that value.

My values by the way might differ slightly from online type calculators, as I do calculations the old fashion way .

--
NextGen67

@NextGen

I put in bold your text above and I ask you WHY do you use the resonance formula as you described for calculating something which is NOT in resonance?

Because at the drain side there is NO resonance ok? There IS resonance in  the gate source circuit where the sinusoidal voltage is shown.

Question is why Luc had to tune the main toroid core/coil with the magnet?

I think that for each individual frequency for the oscillator to work properly (and on 'properly' now I mean Luc's quest for the lowest uA draw), the drain circuit's reactive impedance has to be trimmed so that for each frequency the phase shift and the voltage amplitude BE optimal for feedback to control the gate. And the 'optimal' means the positive gate voltage peaks in time fall just between the drain voltage pulses, not earlier or not later AND then the fedback amplitude is such that it matches the MOSFET's threshold voltage level.  In this respect the Q of the input side coil that is tuned by a ferrite rod is also to be watched for, maybe the coil Luc used there with its high DC resistance had a just optimum value for the job, it would also be good to test with a better Q coil.

At the gate source side the MOSFET input capacitance C_iss is surely can be calculated from the measured frequency and ferrite rod coil inductances. I did some calculation in this respect, using an online LC resonance calculator. Let the coil between the gate source is Lg.

f[kHz]             5.58    8     12     16     20     25    30    40     50
Lg[mH]            310   147   65.3  38.7  25.3  16.3  11.7  7.3    4.9
C_iss[pF]           2624 2692 2693  2556  2502 2486 2405 2168  2067

I am not sure that with all these info as I see this oscillator there is something useful to be drawn into a formula but who knows?

rgds, Gyula