Only people with that affliction of 'golden ears' can hear the difference then.
It works the other way, the ringing actually returns power to the supply wich therefore increases efficiency.
Well in my experience its certainly worth trying ordinary wire becuase Litz showed dissapointing results, the extra DC resistance seemed to be much more dominant than AC resistance due to skin effect calculations would suggest.
You might want to try winding it so as to avoid intrawinding capacitance, winding it not in layers horozontally but vertically, like discs.
Well, in that case I would measure a lower power supply current (note: I measures this DC current before the 8x1000uf capacitors bank), but unfortunately as I wrote you, with the air core inductor the current increases. The result is that the amplifier need more power and this additional power is dissipated into heat by the air core inductor.
As a matter of fact my air core inductor is similar to a disc because is made of 36 turns wound in 9 layers with 4 turns per layer. It is a sort of 6 mm thick disc with a diameter of 20 mm. There is no reel (the turns are glued together) and the diameter of the central hole is about 5 mm.
It can get hot only if it has resistive losses. Maybe the glue you used is lossy at the switching frequency.
--
OOO - Own Opinions Only. Try www.jmwa.demon.co.uk and www.isce.org.uk
2006 is YMMVI- Your mileage may vary immensely.
John Woodgate, J M Woodgate and Associates, Rayleigh, Essex UK
Of course; that's part of my excuse for existing. But what I wrote in this case is serious. The assumption is that the air-cored inductor must be less lossy than the ferrite-cored one, and indeed its DC resistance is lower (which it might not have been). So the only way that electrical power can be converted into heat is for some extra high-frequency loss to occur, and since there can't be a core loss [1], it must be dielectric loss.
Do you have another explanation?
[1] Unless, of course, the OP has put a big brass bolt through the centre hole and not told us! Especially if both ends of the bolt are earthed!
--
OOO - Own Opinions Only. Try www.jmwa.demon.co.uk and www.isce.org.uk
2006 is YMMVI- Your mileage may vary immensely.
John Woodgate, J M Woodgate and Associates, Rayleigh, Essex UK
That might be your problem then. Your air cored inductor rope has a DC resistance of 0.0298 ohms per meter at 650KHz and 40C. With 9 layers its AC resistance is 5.42 ohms per meter, proximity effect. Your single wire cored inductor wire has a DC resistance of 0.056 ohms per meter and an AC resistance of 0.36133 ohms per meter at 650KHz and 40C.
Those might be approximate figures but it looks like you have something like
1.6m on your air cored inductor and 1m on your cored inductor...... 45turns as a single layer?
--
OOO - Own Opinions Only. Try www.jmwa.demon.co.uk and www.isce.org.uk
2006 is YMMVI- Your mileage may vary immensely.
John Woodgate, J M Woodgate and Associates, Rayleigh, Essex UK
In message , dated Sun, 27 Aug 2006, Genome writes
That seems a HUGE increase in resistance. These 'pancake' coils used to be regarded as very good for HF work. Your figures would suggest that they are quite useless.
--
OOO - Own Opinions Only. Try www.jmwa.demon.co.uk and www.isce.org.uk
2006 is YMMVI- Your mileage may vary immensely.
John Woodgate, J M Woodgate and Associates, Rayleigh, Essex UK
As you say...... your mileage may vary immensly. The particular geometry he is using may well make the sums innacurate but those are the numbers the sums come up with.
formatting link
Weren't 'pancake' coils wound in a funny progressive way with the wire wandering from inside to out and back again so they looked like a yarn bobbin with lots of air in the structure? Perhaps that helps things?
The reactance of the 11.5uH at 1MHz is 6.28E6 x 11.5E-6=72 ohm so that at 16App in a series RLC, this implies a voltage of 16x72=1,150Vpp. I really don't see this happening in a 5V 74AC00 driver circuit. The 470nf has comparatively negligible reactance at 1/(470E-9 x 6.28E6)=0.3 ohm. Then there is the admitted observation that the currents do not add up. Your problem is most certainly related to your scope probe where the unconcentrated magnetic field of the air core coil is coupling with the geometry of the probe cable in such a way as to induce an EMF across the
1Mohm scope input impedance that looks like a 16App measurement across the 0.22R. I take 16App to mean a voltage of 16x0.22=3.5Vpp apparent across the 0.22R and this requires only 3.5Vpp/1MR=3.5uApp induced current in the probe- quite a minuscule amount. You have the same problem with T68 core but to a much lesser degree. Your measurements are
*all* off, and therefore nothing can be concluded about your seemingly amazing circuit.
funny, I thought your conclusions were somewhat more than nothing. Although they could be more concisely put as: "look harder"
In general I always assume measurement errors first.
in addition to all the other comments, beware "swinging" inductors. At low currents, permeability (and hence inductance) can be much, much higher; steel is a bugger for this, as are most powdered iron cores. -2 is pretty low mu (10), so its not as bad as some, but the phenomenon still exists.
Are the strands in the copper plait well isolated from eachother? If not, it is better to use a single wire. There should be some spacing between layers to reduce parasitic capacitance. Use glue sparcely. The bigger the coil is made, the easier it is to get good performance. Beware of metals and other absorbers in its vicinity.
Fred, your claims are quite reasonable. In another post of this thread I had already recognized the parasitic inductance of the 0.22 ohm shunt resistor as the main cause of the error. However that doesn't modify the original question; my oscillator requires a different supply current according to the inductor used as load (air core rather than toroidal core) and here we are talking about DC current!
No John, the hole is filled with air only! With regards to the glue, I tried with a similar inductor with no glue at all and I got the same behaviour.
I have made further tests in order to find answers to some other interesting questions:
1) Does the Litz wire play an important role here ?
I have built another air inductor, using this time a single copper wire with the following specs:
c) L=11.1 uH R=0.028 ohm Single copper wire diam. 1.2 mm
With this inductor the oscillator Icc rises to 9.4 mA (it was about
6.6 mA with the Litz-inductor, the type formerly denominated a)). Is the higher current due to the lower DC resistance of the c) inductor? However, going back to the initial question, it seems that the phenomenon is still more noticeable. The current drawn by the oscillator is higher with an air core bobbin and that is not imputable to the Litz wire.
2) How is the sensitivity of the Icc to the inductance value ?
Removing from the b) toroidal inductor (11.5uH) few turns, the inductance decreases to 9.8 uH and the Icc rises from 4.6 to 5.6 mA. Adding some more turns so to have a value of 12.6 uH shows that the Icc decreases to 4.3 mA. (note: in the last case the toroidal inductor has exactly the same inductance of the a) air core inductor)
3) How much is the parasitic capacitance of the inductors ?
Some tests I've made shows for the air core inductor a) a capacitance of about 5 pf. The capacitance of the b) inductor is quite lower, it should be below 1 pf (it is not an easy job to be more accurate). In any case the magnitude of these capacitances seems to be too small to have a part in the Icc behaviour.
Now I'm focussing my attention toward this two inductors:
a) L=12.6 uH R=0.046 ohm Copper plait made of 8 wires diam. 0.315 mm wound on air.
b1) L=12.6 uH R=0.055 ohm Single copper wire diam. 0.64 mm wound on a toroidal core (Micrometals T68-2).
Yes, you are correct in your supposition about the wires length.
I'm astonished that my rope could have an AC resistance @f=650KHz about 180 times greater than DC resistance. I would appreciate if Genome or Terry could give me some references or links about how to estimate the effective AC resistance in inductors. In the meantime I thought about a way to measure it. Here is what I've done:
I connected the inductor with a R=10 ohm so to implement a low pass filter. Then I feeded the filter with a sisusoidal signal with
4000mVpp @f=670KHz. With my double traces oscilloscope I have monitored both the input and the output signal and again...here are new unexpected results:
with toroidal b1) the output signal has an amplitude of 737mVpp. The amplitude becomes 836mVpp with the air core inductor a). Should I conclude that the AC resistance of the air core inductor is lower than the toroidal one?
Mind you, it gets turned off when I go to bed so......
I did scan and post the paper on an old site that is dead now, perhaps someone else still has a copy. I did it on a Lexmark which was a replacement for one that broke itself and the replacement broke itself. Anyway, the reference to the paper is on that page along with the software that does the sums for you. Nice purple as well.
It's been 'close' in the past....
I'm amazed too but like Terry said...... layers are really bad things when it comes to proximity effect, something to do with the integral around a current loop having to sum to zero so a wire on top of another one has to, errr carry its own current 'and' a current that cancels the effects of the one next to it and then things add up....... I don't understand it either but there is more there than just skin depth.
skin effect is when flux lines generated by the current flowing thru a conductor cut that conductor. The net result is an impedance that varies with depth, but in effect the magnetic field is forcing the current to the outside of the wire. This reduces the available area as current is only flowing in a thin "skin", so up goes the resistance. The net result is to make resistance increase with sqrt(f) above some corner frequency.
In practice you can usually assume your conductor is hollow, and one skin depth thick, giving a total area of perimeter*skin depth. Because of that, increasing conductor size doesnt help very much.
If the available winding width is constrained, adding more copper means increasing the number of layers.
Proximity effect is the same thing, but is caused by current flowing in nearby layers. This takes the area already limited by skin effect, and reduces it even further; more layers make the problem even worse. The net result is that AC resistance goes up much faster than sqrt(f).
I have seen a 1500W transformer running at 100kHz using 12 layers of
0.6mm copper foil on an ETD54 bobbin catch fire. skin depth is about
0.2mm at 100kHz, so the winding is 3 skin depths thick. The windings were interleaved (1/2P-S-1/2P), which means the effective number of layers was 6, due to the line of symmetry thru the centre of the secondary winding. This gave an AC/DC resistance ratio Fr = 100, so although the DC resistance was 1mOhm, the AC resistance was 100mOhms. Thus the temperature rise was about 100x that expected from the DC values.
I changed the winding from 0.6mm to 0.1mm, about 1/2 a skin depth, which gave an Fr of about 1.3. But the DC resistance went up 6-fold, so the overall AC resistance was 6x1.3 ~ 8x the original DC resistance. But
12.5x lower than the original AC resistance. The new transformer worked just fine, with about a 40C internal temperature rise IIRC.
You're quite the smug troll but not very convincing.
Magnetic nonlinearity does not apply to the aircore inductor. The OP now seems to be operating under the assumption that the inductance of the
0.22R is his problem. Since we know the reactance of his main L to be 72 ohms, and that his apparent 16App measurement results from a 3.5Vpp across the scope input, this would make the 0.22R inductance nearly 170 ohms for an L=27uH. This is not likely. Also, issues of proximity effect only pertain to the inductor design and not to the measurement. Even his VOM measurements of DC current are suspect because of the field effects. The proximity effect does explain larger than expected loss and lower than expected Q due to the resulting asymmetry in the current distribution in the wire.
not at all. that paragraph is perhaps the most useful in the thread. Its a nice application of "Sesame Street Theory" - one of these things is not like the other. In particular its a nice, clean BOTE debunking of a hypothesis assumed to be valid, using 1st year physics.
thats true, but it does to the other one. And hes interested in why they are different. Although with ur = 10, the effects will be fairly small.
The OP now
absolutely. the measurements are in all likelihood misleading, so its hard to draw any quantitative conclusions.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.