nonsense

At best it would surely require a very specific kind of toroidal core to make it work. But there is no real information given about the toroidal core at all. On the circuit it just says L1 100μH.

"john larkin" snipped-for-privacy@htigct.com wrote in message news: snipped-for-privacy@4ax.com...

If you make R3 12.3k you can have a sinewave for the first 300ms

Version 4.1 SHEET 1 3020 2548 WIRE -1184 160 -1264 160 WIRE -1008 160 -1104 160 WIRE -1984 320 -1984 288 WIRE -1984 320 -2032 320 WIRE -1760 320 -1840 320 WIRE -1648 320 -1680 320 WIRE -1584 320 -1648 320 WIRE -1456 320 -1504 320 WIRE -1392 320 -1456 320 WIRE -1264 320 -1264 160 WIRE -1264 320 -1312 320 WIRE -1184 320 -1264 320 WIRE -2032 336 -2032 320 WIRE -1008 336 -1008 160 WIRE -1008 336 -1120 336 WIRE -1984 352 -1984 320 WIRE -1184 352 -1232 352 WIRE -1232 384 -1232 352 WIRE -1648 400 -1648 320 WIRE -1456 400 -1456 320 WIRE -1008 400 -1008 336 WIRE -928 400 -1008 400 WIRE -848 400 -928 400 WIRE -1648 496 -1648 464 WIRE -1456 496 -1456 464 WIRE -1840 608 -1840 320 WIRE -1008 608 -1008 400 WIRE -1008 608 -1840 608 FLAG -2032 336 0 FLAG -1984 208 vcc FLAG -1984 432 vee FLAG -1152 304 vcc FLAG -1152 368 vee FLAG -928 400 output FLAG -1232 384 0 FLAG -1456 496 0 FLAG -1648 496 0 SYMBOL voltage -1984 192 R0 WINDOW 123 0 0 Left 0 WINDOW 39 0 0 Left 0 SYMATTR InstName V3 SYMATTR Value 15 SYMBOL voltage -1984 336 R0 WINDOW 123 0 0 Left 0 WINDOW 39 0 0 Left 0 SYMATTR InstName V4 SYMATTR Value 15 SYMBOL OpAmps\\LT1357 -1152 272 R0 SYMATTR InstName U1 SYMBOL res -1088 144 R90 WINDOW 0 0 56 VBottom 2 WINDOW 3 32 56 VTop 2 SYMATTR InstName R3 SYMATTR Value 12.3k SYMBOL res -1296 304 R90 WINDOW 0 0 56 VBottom 2 WINDOW 3 32 56 VTop 2 SYMATTR InstName R2 SYMATTR Value 10k SYMBOL res -1664 304 R90 WINDOW 0 0 56 VBottom 2 WINDOW 3 32 56 VTop 2 SYMATTR InstName R1 SYMATTR Value 2.2k SYMBOL ind -1488 304 R90 WINDOW 0 5 56 VBottom 2 WINDOW 3 32 56 VTop 2 SYMATTR InstName L1 SYMATTR Value 100000n SYMATTR SpiceLine Rser=1m SYMBOL cap -1664 400 R0 SYMATTR InstName C1 SYMATTR Value 33n SYMBOL cap -1472 400 R0 SYMATTR InstName C2 SYMATTR Value 33n TEXT -1800 64 Left 2 !.tran 0 300m 0 1u startup TEXT -1800 8 Left 2 !.options plotwinsize=0 numdgt=15

A better inductor simulation along these lines may produce a sinewave untill the wind direction changes.

"Edward Rawde" snipped-for-privacy@invalid.invalid wrote in message news:1162v3c$28a0$ snipped-for-privacy@nnrp.usenet.blueworldhosting.com...

Like this, but only about 60dB down on harmonics.

Version 4.1 SHEET 1 3020 2548 WIRE -1184 160 -1264 160 WIRE -1008 160 -1104 160 WIRE -1984 320 -1984 288 WIRE -1984 320 -2032 320 WIRE -1760 320 -1840 320 WIRE -1648 320 -1680 320 WIRE -1584 320 -1648 320 WIRE -1456 320 -1504 320 WIRE -1392 320 -1456 320 WIRE -1264 320 -1264 160 WIRE -1264 320 -1312 320 WIRE -1184 320 -1264 320 WIRE -2032 336 -2032 320 WIRE -1008 336 -1008 160 WIRE -1008 336 -1120 336 WIRE -1984 352 -1984 320 WIRE -1184 352 -1232 352 WIRE -1232 384 -1232 352 WIRE -1648 400 -1648 320 WIRE -1456 400 -1456 320 WIRE -1008 400 -1008 336 WIRE -928 400 -1008 400 WIRE -848 400 -928 400 WIRE -1648 496 -1648 464 WIRE -1456 496 -1456 464 WIRE -1840 608 -1840 320 WIRE -1008 608 -1008 400 WIRE -1008 608 -1840 608 FLAG -2032 336 0 FLAG -1984 208 vcc FLAG -1984 432 vee FLAG -1152 304 vcc FLAG -1152 368 vee FLAG -928 400 output FLAG -1232 384 0 FLAG -1456 496 0 FLAG -1648 496 0 SYMBOL voltage -1984 192 R0 WINDOW 123 0 0 Left 0 WINDOW 39 0 0 Left 0 SYMATTR InstName V3 SYMATTR Value 15 SYMBOL voltage -1984 336 R0 WINDOW 123 0 0 Left 0 WINDOW 39 0 0 Left 0 SYMATTR InstName V4 SYMATTR Value 15 SYMBOL OpAmps\\LT1357 -1152 272 R0 SYMATTR InstName U1 SYMBOL res -1088 144 R90 WINDOW 0 0 56 VBottom 2 WINDOW 3 32 56 VTop 2 SYMATTR InstName R3 SYMATTR Value 12.23k SYMBOL res -1296 304 R90 WINDOW 0 0 56 VBottom 2 WINDOW 3 32 56 VTop 2 SYMATTR InstName R2 SYMATTR Value 10k SYMBOL res -1664 304 R90 WINDOW 0 0 56 VBottom 2 WINDOW 3 32 56 VTop 2 SYMATTR InstName R1 SYMATTR Value 2.2k SYMBOL ind -1488 304 R90 WINDOW 0 5 56 VBottom 2 WINDOW 3 -47 56 VTop 2 SYMATTR InstName L1 SYMATTR Value Flux=100000n*tanh(5*x) SYMATTR SpiceLine Rser=1m SYMBOL cap -1664 400 R0 SYMATTR InstName C1 SYMATTR Value 33n SYMBOL cap -1472 400 R0 SYMATTR InstName C2 SYMATTR Value 33n TEXT -1800 64 Left 2 !.tran 0 2 0 1u startup TEXT -1800 8 Left 2 !.options plotwinsize=0 numdgt=15

I think R3 should have been 12.5k so perhaps calling it 12k5 is not a bad idea after all.

It's improbable that the opamp can output enough current to saturate that toroid. Q is pretty low.

I suspect that the author fine-tuned the opamp gain until he got a pretty stable sine wave. That tweaking process is a form of AGC.

The opamp slew rate might be the nonlinear gain-set element.

Ugly breadboard.

Those Aspencore sites have some really bad circuit examples.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

It is certainly nonsense. The amplitude control mechanism is the op amp running out of gain - running out of slew rate is just a manifestation that.

It's not a recipe for a low harmonic content sine wave, despite Edward Rawde's long held conviction that it is. Any number of people can be fond of bad ideas - religions depend on it - but it doesn't make them right.

That makes no sense to me Bill but don't bother replying.

Datasheet shows FPBW (remember those?) to be nearly 3 MHz at +/-15V supplies. And that's a guaranteed minimum, so no agc to be found there.

The AGC results from the Steinmetz relation applied to the core. Namely, core loss, Pcore=k x V^a. The wiki article states 2.4 < a < 2.6, but I recall for most of the low frequency ferrite toroids a=3. And the core can stay well out of saturation as this happens. The loss is due to wide range flux swing due to V. Peak flux is in direct proportion to V, and the loss is associated with the Raleigh loss of the hysteretic region of the resulting operating B-H curve.

Integrating this into the resonant pi, you get Pcore=V^2/Rp=kV^3, making Rp(equiv loss || R). Then Rp=1/(kV). Since loop feedback beta is ~ beta, the total loop gain Aol(OA) x beta becomes proportional to 1/Vout. There is your AGC.

Since L remains mostly linear independent of circuit operating state, the frequency selectivity is passable, but not great. The phase shift across the pi with frequency is just -0.046 degrees/Hz which is not terrifically good (kind of horrendous actually). And the low starting Q, which may be around 50, makes the feedback loop too broadband. And L probably runs 4,000 ppm/°C minimum, something else not so great.

And why does he need so much amplitude for an OOK reference? No reason, other than to make the AGC thing work with practical component values.

Peak flux may be in direct proportion to V in this situation - fixed frequency - but voltage is proportional to the rate of change of flux, and when flux peaks because the core has saturated you don't get any voltage out of the inductor at all.

You seem to be claiming that the EDN circuit works because the core starts saturating, which would mean that output waveform would be distorted. How much and how is the interesting question, but the EDN write-up doesn't go into that at all.

A remarkably incoherent explanation, even here, where clear explanations are thin on the ground.

To the extent that it does work.

Sloman isn't interested in electronics. For him this is just a forum for generating lame insults.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

That's why I'm not going to ask him to show the circuit I posted which "didn't include any controllable non-linear gain adjustment element". If he means the one with the inductor then it's clear how that works but it is not a circuit I'd use in reality except for experimental inductor behaviour demonstration. There are two types of circuit I'd use in reality. One has a control loop using an analog multiplier. You can use an MC1496 for this and they are available for next to nothing at Digikey. The other type uses an amplitude limiter which does generate harmonics. In this case you can filter the harmonics to end up with a clean sinewave. The best solution for any given application is probably a combination of both techniques. Use a control loop to generate the signal and a filter to remove any harmonics which are still above 140dB (usually the 3rd). In all cases, manual adjustment of the circuit will be required to push all harmonics below 140dB, assuming you have measuring equipment capable of telling you. You should also make sure that everything remains at a constant room temperature and be prepared to wait at least a few minutes after power is applied for everything to settle. So I'll leave it to Sloman to design the following oscillator.

Frequency: 1kHz +/- 0.01% Output level: 0dB into 600 Ohms. Output level stability: within 1% after 200 ms. Harmonic distortion: All harmonics to be below -140dB. Power supply: 9V batteries. Design for either 2 or 4 batteries. Temperature range: -55C to + 155C

How would you measure the distortion?

Peak flux is proportional to V and inversely proportional to F. It's hard to saturate a core at 100kHz. This circuit doesn't rely on saturation; it relies on the much stronger sensitivity to core loss. My guess is the ferrite saturates at 2,000 Gauss.

Your confusion probably stems from not realizing the feedback error is a current summing junction at virtual ground. So that decreasing the feedback impedance via a shunt equivalent core-loss resistor across the resonant pi results in more current feedback per unit of Vout, thereby lowering the gain.

What equipment?

I'd ask here what equipment I need to buy or build to do the measurement.

This claims to be able to measure it but I haven't read it in detail.

Depends on the frequency. If it’s reachable with air core coils and NP0 capacitors, it shouldn’t be too bad. You make a highpass filter with two notches right at the fundamental, which will knock it down by 60 dB or more, and then use a spectrum analyzer.

Cheers

Phil Hobbs

The fast Fourier transform does make it easy to extract specific harmonics from a digitised waveform, but you do need a good quality notch filter to keep the fundamental down to the level where the non-linearities in reasonably linear components don't generate their own harmonics.