IGSL

[Davor]

Here is my input. Apparently when power supply is simulated correctly there is no problem with starting even with 470k.

The upper right part of the schematic is a simplified negative voltage source. There is an optional 10p capacitor for improving symmetry, which is not needed as the improvement is negligible.
You'll see an expected artefact that is present only with BC557, but not with BC327. It could also be a LTspice thing. Anyway, there is a little difference between the two.
FFT is done with flat top windowing to improve level reading, and it reflects your FFT-s as well.

The schematic in LTspice is attached here:

[Davor]

Very nice!

You can observe a direct link between the saturation level with harmonic content. In short, all "I" cases should be avoided in favour of the first cases in "II". Low saturation levels are not working in favour of class C operation, but we know it already.

It appears that 680k is still a bit conservative, as CRO-s reveal that it turns the oscillator just about to the edge of class C operation. Perhaps even a lower value will do better. In Spice the oscillation start is not assured with 470k, but reality is always the ultimate arbiter to any simulation, and it could prove beneficial. I'd try 470k as well. I'll try this configuration in Spice with BC547 and BC557 as well and see if simulation confirms this behaviour as well.

You can observe the stabilising effect of Tx amplitude in relation to saturation depth. It makes sure that regardless of the coil loading, and with a constant Vcc, the amplitude will remain constant, and that was a goal.

What is your negative voltage value?

[sini]

Hi,I made measurements with emitter . Note, that scale of emitter is changed ! Everything is measured with BC547 and BC557.Please, tell me optimal settings.Sini

[Davor]

It will be very interesting to see your measurements. I'm seldom at my shack lately, and I have some other things to fix first, but there is lots of time till spring.
If you fancy some light reading, seek A. Hajimiri and T. H. Lee - A general theory of phase noise in electrical oscillators and see what it says about AM modulation of oscillators. In fact it says very little, but that precious little is very important. The paper is mostly about phase noise that is of least concern to metal detectors. Short version on AM - every oscillator has lots of small signal gain, and there are always some effects that push oscillation to the rail limit - read saturation - and the amplitude purity is solely related to the rail voltage purity. Noise at rails will surely turn to AM noise, and will find it's way to the Rx in form of chatters. Furthermore, a coil is a part of it's surroundings, and with changing permeability the coil Q changes. Such changes will turn to AM in case there is any room for oscillation amplitude change.
Frequency/phase will also change, but it is not a problem at all. The reason to it is a very low sensitivity of the targets to frequency change. Even an octave will change a target phase response by less than 15° and the Rx to Tx phase problem does not exist at all because Rx is locked to Tx.
Hence the AM behaviour is of utmost importance to us.

[sini]

Thanks,

very well explained. As I underestand, focusing on AM troubles is more important -> slight saturation ( decreasing AM noise ) is more wanted than decreasing higher harmonics. I will buy BC327, BC 337 and I will do new measure.

[Davor]

@sini, as noted in my previous post, I believe slight saturation is beneficial for keeping AM at bay, so S_DSC02611.JPG (standard), with S_DSC02571.JPG (1M), and S_DSC02593.JPG (680k) would be it. You'll find a saturation spike resembling something close to this Before ABtoC Emitter.jpg revealing the saturation spike just a bit overshooting the rail. This spike should not go beyond the rail so this CRO shot is not the perfect example. The spike is influenced by emitter trimmer and a supply voltage, which is mostly constant.

My preferred transistors for Tx are BC327/337 because of superior switching ability and low noise.

As for the harmonics, I identified the 2nd as a source of offset drift, in effect similar to "earth field" effects in PI. The 2nd distorts the sinewave in a way that upper side becomes more blunt and the lower more spiky (or the other way around) and their equilibrium is not 50:50 split in time, but some ratio close to it, and also changing by the 2nd content. As a consequence the "air" signal" becomes modulated by this difference from 50% and may pass as a phantom target.

3rd harmonic passes through the Rx directly as a consequence of Nyquist mirroring, and in that sense the 3rd is a same level of trouble as the fundamental "air signal". It is more lively than a fundamental, in fact 3 times as lively, but still overshadowed by a fundamental. A short estimation:
- say the 3rd is attenuated 30dB against the fundamental
- 3rd amplitude modulation index is 3 times of a fundamental ... +10dB
- resulting attenuation + modulation index ... -30+10=-20dB ... so the 3rd's troubles are still 20dB below the fundamental's AM troubles.

@golfnut, yes, those harmonics are the reason 3rd, 5th and all other 2n-1 (th) harmonics have a free pass in any kind of communication receiver, Therefore we have to put some filter in front of a mixer. In software defined radio such filters are often called "anti aliasing" as all the frequency content that is found past 1/2 of the Nyquist frequency are called "aliases". In VLF-s the coil is often in some tank configuration doing exactly that.

[golfnut]

One thing I wonder about is with these switching style mixers - the LO drive is a square wave and full of harmonics anyway. S

[sini]

Hi Davor,
I use BC557 as PNP and BC547 as NPN transistor. I will explore emitter spike.
My questions are:
What is worse for synchronous detection? Second or third harmonic ?
I can adjust saturation by 220 Ohm trimmer. Please, see pictures and recomend me some saturation ( number of picture ). Few month ago you said, that it is good to have small saturation becouse of AM noise.
Many thanks, Sini

[Davor]

...errr ... my mistake ... 680k seem to be a better choice after all. I had all the pictures opened simultaneously, and I confused 680k with 1M.
What is important to compare are S_DSC02612.JPG (standard) with S_DSC02573.JPG (1M) and S_DSC02595.JPG (680k), all in slight saturation and producing some 15.8dB of carrier, but with 2nd harmonic to carrier ratio of 30.8dB, 33.2dB, and 36dB respectively, and 3rd harmonic roughly the same.

Of course, 680k uses the least current of them all.

Keeping a Tx in slight saturation is a good thing as it remains amplitude-locked to the rail voltage. Amplitude changes are reflected as false targets, so chasing signal purity by means of reducing Tx amplitude is not entirely a good idea. Stabilising Tx amplitude by some envelope integrating method is also a bad idea, because the stability requirements make such integrator very slow, and incapable of maintaining the amplitude quick enough so that the changes skip the Rx filters.

I'd say it will be beneficial to further explore the emitter spike as a straightforward indication of Tx saturation depth.

[Davor]

Nice work! What transistors are in your Tx?

So apparently 1M works better than 680k in this case. It is somewhat influenced by hfe of the transistors. The other thing to observe closely is the voltage across emitter resistor, in this case a trimmer. You'll see a spike that corresponds to the transistor saturation. It is off-centre as there are some delays in the process, but more important is to avoid the spike's overshooting the rail voltage.

Anyway, 1M is a keeper.

[golfnut]

Hey, real engineering - love it . S

[sini]

Hi all,

I made some experiments with Davors AB->C Tx oscillator.STANDARD is standard oscillator AB, 680KOHM is C oscillator with 680 kohm, 1MOHM is C oscillator with 1Mohm. I made some settings of 220 ohm trimmer ( Tx amplitude ).Here are my results:
Sini

[Silver Dollar]

Try this thread for Lobo info;

http://www.geotech1.com/forums/showthread.php?18196-Tesoro-Lobo-news/page4&highlight=lobo

[Davor]

Again, seek Lobo.

[parun14]

And
What the higher frequency one that have a plan pcb to build?