A question for experts in inductors

There may be inductive coupling between the coil and the probe (or the ground connection of the probe). R=0.22 ohm is a low value. Try to vary the path of the ground connection.

--
Sven Wilhelmsson
http://home.swipnet.se/swi

Yes, I agree that the air core inductor generates more EMI than the toroidal, and that could be picked up from the probe. In any case how could be explained the higher current supply required by the oscillator with the air core inductor? (Note that in this case the probe is not connected at all).

Icc=4.55 mA to 6.58 mA is not a big change. The loss (Q-value) of the inductors may differ substantially. Small shifts in resonance frequency could give big changes in supply current, especially if the oscillators frequency is determined by other things than the LC itself.

--
Sven Wilhelmsson
http://home.swipnet.se/swi

If you could first explain what you would expect the currents to be with the two inductors (and how you calculated these currents) I am sure it can readily be explained to you

If you could first explain what you would expect the currents to be with the two inductors (and how you calculated these currents) I am sure it can readily be explained to you

If you could first explain what you would expect the currents to be with the two inductors (and how you calculated these currents) I am sure it can readily be explained to you

That's basically impossible. A cmos gate has a certain intrinsic minimum output impedance, around 100 ohms, so it can only put out about 50 milliamps peak into a short circuit on a good day.

You're probably gettting some ground-loop or high-side coupling. Not to mention your .22 ohm resistor and its leads have anywhere from 0.050 to 0.500 microhenries of inductance right there. That makes your current measurements with it mighty inaccurate.

Regards,

grg

The fact that its a square wave probably means that its cuased by the inevitable extra capacitance of the air coil, due to it having more turns.

Colin =^.^=

You are right, I can't expect such currents from a cmos gate. However I'm sure that the measure is not affected by ground-loop or coupling (if I move the tip of the probe from one leg of the resistor to the other, that is shorting it to the reference ground, the spike disappears).

As you mention, I think that the spike as to do with the parasitic inductance of the 0.22 ohm resistor. I tried to hook the probe very near to the body of the resistor and the spike was a little lower (a sort of dumping burst with f= about 70MHz and Vpp= about 1.8V). In any case I'm not interested in making a correct measure of the current flowing into the inductor, I'm rather interested in knowing why using two different inductors (but with basically the same inductance) I got a very different current behaviour.

The oscillator frequency doesn't vary with the LC load. I tried also with differents frequencies (in the range 0.5...1MHz) but the circuit behaviour remain the same; the oscillator with air core inductor still draw more current. Your hint about Q value is quite intriguing, I suspect that the air core inductor should have a Q value considerably greater than the toroidal one. Could you suggest me a pratical test setup to estabilish the Q of the two inductors?

The Icc curent is just the DC current drawn by the active oscillator circuit. I would expect the same Icc current with both inductors. I got instead with the air core inductor an Icc current that is about

80% greater of the Icc current with the toroidal inductor (the addition current from no load condition to RLC loaded oscillator, is about 2.6 mA with toroidal and 4.6 mA with air core).

Yes, I think the extra parasitic capacitance of the air inductor could play a rule here. However in order to rise the Icc with the toroidal inductor to values similar to the air core setup I need to shunt the toroidal inductor with a capacitor not lower than 1000pf. I doubt that the my air core would have such order of parasitic capacitance.

Well I'm not an expert (far from it), but from my understanding of inductors in general - an air core inductor which has no core can not store the flux like the toroidal with a core, and hence their is less counter magnetic field. So maybe the air core uses full field collapse and must use more current to re-build the field while the toroidal is buffered by the core storing the flux field. Just my 2 cents worth. JTT

My calculations show 400pf would cuase a current of 2ma at 5vp-p

The Q of the air coil is probably a lot higher, it also probably has a lower resonant frequency, It could mean that there is a lot of ringing going at the edges wich could cuase the output stage to draw more current from the supply, or affect other gates and cuase them to draw more current.

Is the oscillator and output in the same chip? Are all the other gates inputs grounded ?

Without knowing what you are aiming to acheive its hard to suggest a solution.

Colin =^.^=

Also note that a ferrite-core toroid is going to have a very contained magnetic field, while an air-core coil is the exact opposite-- it's going to have magnetic loops flying all over the place.

Find an old CRT shield or other mu-metal and put your air-core coil in that. You may see quite a drop in the measured spike.

inductors

True there is less magnetic flux without the magnetic core, but this is why there are a lot more turns needed for an eqv air coil.

Interestingly for gaped ferrite cores, eg for energy storage in switch mode supplies, just about all the energy is stored in the flux in the air gap, very little energy is able to be stored in the ferrite itself.

Colin =^.^=

Hello There,

You might be suffering from a Beaconsfield.

Are you driving this network directly from the output of a single gate in a

74AC00 connected as an oscillator? Like what other people have hinted at.... it's not going to be happy with its load.

There is also this stuff about gain things and the load will affect them. It is possible that the parasitics of one load pulls the gain of your oscillator enough so that, while it is still oscillating it is oscillating in a more 'linear' manner and you are getting more cross conduction in its output stage.

Try driving the other three from the output of the oscillating one and use their outputs in parallel to drive the load.

DNA

The oscillator is made by one 74HC14 gate, the signal is then buffered through a 74AC00 gate. All the inputs of unused gates are grounded.

Yes, I understand your point of view. My oscillator setup was made only to reproduce and possibly to better understand what is happening in a real device I have just developed. The device is a class D audio power amplifier. In this case the output is a square wave (with very sharp fronts), the frequency is about

650KHz and it swings between 0 and 24 V. At the output of the amplifier is placed a low pass filter made of an inductor (L=10uH) and a capacitor (470nf).

I want to avoid using ferite inductors so to get rid of the core nonlinearity problems (lots of people claim a better audio quality with air core inductors). So I built a special inductor, the a) type I mentioned in my first post. It is wound on air with a copper plait so to also reduce the skin effect.

What I noticed is that this inductor will rise the quescient DC current of the amplifier of about 25 mA (compared with the amplifier using the toroidal core b) ). As the power supply is 25V that means about 0.6W of more dissipated power. The bothering drawback is that that power seems to be dissipated by the inductor itself because it becames quickly very hot (the toroidal core inductor keeps itself just warm). Considering that each amplifier channel has two inductors (it uses a bridge configuration) and there are 7 channel overall, the heat removal could be a serious issue.

As there is a relevant power dissipated over the air core inductor, I suspect that its parasitic capacitance shouldn't play a rule in this case. I mean, that capacitance should increase the power dissipated from the amplifier final stage instead of from the resistive component of the inductor, shouldn't it ? I begin to suspect that the Q value would be the main responsible factor of this behaviour. Can you confirm this impression ? Have you any hints to relieve this drawback ?

Aha, this makes it easier, I supose the non linearities of ferrite could cuase distortion here, not sure how exactly, but I think you need to measure it directly to see if its worth going to an air coil. An air coil probably is going to get hotter unless it is very much larger, simply becuase it needs a lot more turns, and so has more resistance.

What is going on is that there will be ringing at the switch output, this ringing will exceed the supply voltage and so will actually go back into the supply through the antiparallel diodes, with more loss in the coil there is less ringing so there is less current going back into the supply.

An alternative is to have a ferrite core but with a much larger air gap, the ferrite then works over a more linear region, or use a core wich is made of particles with a large space between them wich amounts to the same thing, of course you then need more turns.

The radiated effects from an open air coil might cuase a problem too, you might consider a toroidal air core.

Have you tried another but smaller filter right after the switch ? you might find you can then use air core or ferrite for one or both.

Colin =^.^=

On this narrow point, I recall the OP writing:

I read that the R is measured as less on the air core, here.

Jon

larger,

Good point wich I missed, however the AC copper loss is probably higher, the intrawinding capacitance wich will undoubtedly be higher cuases higher internal current too.

I tried Litz wire in an SMPS once and found it got hotter.

Colin =^.^=

The toroidal core used for inductor b) have very low permeability and I suppose it is realized, as you suggested, with particles with large air gap between them. The distorsion should be very low, nonetheless there is an heated argument between audiophiles about what kind of audio degradation this low distorsion could introduce (that recalls in some way the matter of which is the best speakers cables...).

Yes, that could be reasonable. The power supply is by-passed with a parallel of 8x1000uf low ESR capacitors (the total ESR is about

0.004 ohm), it appears quite clean and the spikes caused by the ringing can't be traced. Anyway it is a little odd the fact that increasing a loss factor (degrading the Q of the inductor) causes a better efficiency of the system. That seems quite like a paradox...

Yes, I'm aware of that.

Unfortunately it is quite expensive to double all the 14 filters, moreover there could be problems due to the eccessive phase lag not to mention the worse damping factor (as there is no output feedback, the output impedence is mainly due to the inductor impedence).

Colin, do you think that a Litz wire would show an AC copper loss higher than a single wire of equivalent section ?