VLF MD with digital signal processing : Bee-Buzz 1

[Aziz]

Hi all,

what should I say?
Is it better to lose one channel on the TX side (2-channel = stereo output) and get only half power to the TX coil and beeing able on the other side to null out the RX coil perfectly (perfect induction balance) and push up the pre-amp gain?

Now the good news:
I have injected through a series resistor (10 kOhm) from the second output channel an IB nulling correction signal directly to the left input channel (RX-signal). I have adjusted the amplitude and phase and have achieved the perfect IB. I mean really perfect as it can be done up to the noise floor level.
Instead of having mV residual RX IB mismatch voltage, I'm in the µV region.

This can be easily done for dual frequency too. The active IB nulling gets maybe complex for chirp TX signals. I haven't tried it out yet.
Aziz

[Aziz]

...except for the 20 db of amplitude modulation clearly visible on the fundamental.

Hi Paul,

I don't understand, to which post you are referring to.
If you refer to the TX-reference voltage in the lastest post, no this is coming from digital aliasing. The internal wave-form I do send to the sound card looks like the same on my display. On the oscilloscope, the TX wave-form is a pure sine of course.

Below is the interal TX wave-form:

[Aziz]

So what is the actual peak-to-peak voltage across the TX coil?

TX coil voltage measured on the oscilloscope: aprox. 130 V peak-to-peak.
Well, I can actually go up to 180 V as I haven't fully set to maximum. My reference TX-voltage will be clipped otherwise. I have to change my capacitive voltage divider.

[Olly]

So what is the actual peak-to-peak voltage across the TX coil?

[moodz]

...except for the 20 db of amplitude modulation clearly visible on the fundamental.

[Aziz]

This is it.

It got even 5 times better. I could probably get much better with active IB-nulling. Then pre-amp gain of 500 or 1000?
Cheers

[Aziz]

Hi all,

well I have connected a TX and a RX coil to my sound card. Pushed the transmit power to almost maximum power and tried to balance the RX coil. Only single frequency at the moment. TX coil is tuned with a capacitor at 27375 Hz resonant frequency. RX coil is not tuned. No pre-amp connected.
I could balance the TX and RX in first attempt into the 1/1000 of the ADC input range. This is something in mV region depending on the line-input level ruler. I could probably get it better and lower.

RX Phase detection is so much sensitive compared to magnitude. Which is the normal case if you look at the frequency response spectrum at resonance frequency. The phase change is much steeper than the magnitude. If you have a high Q TX coil system (low TX resistance), this is even much more sensitive. But we can do more. Much more.
As the RX coil won't be tuned at all with a capacitor in the multi-frequency mode, we really need an amplifier to get more depth and sensitivity. A gain of 50 to 100 is in game again.
And guess what? We need an Ultimate-Ultra-Low-Noise-KISS-Amplifier (UULNKA).

Without the pre-amp it would be a nice detector. But with the pre-amp ...
I will show you some pics if I have more time.
Cheers,
Aziz

[Aziz]

my problem is not a noise but EMI 60 Hz

Hi pito,

you have to go out field testing. There is no mains hum.

We should not have any low frequency mains noise as we are decoding at the much much higher transmit frequencies. Even the multiple harmonics of the mains hum shouldn't be a real problem. You must be doing something wrong. The noise could be other noise source however.
Aziz

[pito]

my problem is not a noise but EMI 60 Hz

[Aziz]

Hi all,

I have my purchased parts now.
The choke L1 in my TX circuit is getting difficult to realise. I haven't found a good coil former yet. I have several plastic coil formers, but they are either too big or too small or I don't get enough number of turns on it. Or I have to use 0.7 mm magnet wire, which increases it's resistance up to 1 Ohm.

For the choke L1 it applies, as tight and compact as possible to
- achive high inductance with less copper (less turns)
- low series resistance to increase power efficiency (by using thick magnet wire)
- high mechanical stability to avoid microphonic effect, which will cause electrical and audible noise (in low frequency audible range).

Anyway, I will use the single frequency TX for measurements until the choke L1 issue has been solved.
Yeah, I will make the Richard Lee's pre-amp soon. It is so nice.
Aziz

[Aziz]

Hi all,

now the real EMI noise at home. Fasten your seat belts please.

I have connected an RX coil (L=2.1 mH) to an ultra-low-noise pre-amp of gain 100. The pre-amp does not produce any significant noise. Even with gain of 100. But the sound card does.
Let's look at the shorted input of the pre-amp first:



And now with the RX-coil at the pre-amp input:



You see, the FFT spectrum rised on almost all regions up to 70 kHz. Above 70 kHz, the ADC of the Sound BlasterX G6 produces more noise so the EMI isn't visible, because the EMI noise is below the noise floor level of the ADC system. There are still regions with low EMI noise (around -110 dB level regions). And regions with bad, really very bad behaviour (all the peaks).

You see, that pre-amp gain of 100 is large enough. It should even be lower. Maybe 6 dB lower. This is factor 2. So pre-amp gain up to 50 is ok I think.
If we have an IB-mismatch, which produces 10 mV residual signal on the RX coil, this would be at gain 50 500 mV output. Ok, maybe too much. We should take lower gain. Or lowering our IB-mismatch residual RX voltage.
Maybe an active cancelling with a cancelling coil (one turn loop over RX coil)? Yes, this could be done very easily. I have done it 15-20 years ago. This is working perfect. Even on a dual frequency detector. But I would lose one output channel and would have half TX power at the end.

Anyway, you see the EMI noise (at home), which is causing a lot of headache.
Cheers,
Aziz

PS:
Next time, let's look the RX coil signal without pre-amp. But with working TX-circuit. I'm sure, we won't need any pre-amp with residual RX voltage up to 500 µV due to IB-mismatch.

[ivconic]

[Aziz]

Common-base is generally the better approach to low noise. Here is a CB preamp I was messing with several years ago. It does not require a floating voltage source. I don't recall where I found this but probably an audio application. The values shown are what was used in my last sim.

It's the same topology. A bit modified however. You have a buffer (amplifier) and the end of the output to get defined gain regarding output impedance to the next stages.
It's a (old and forgotton) moving coil (MC) amplifier.

Also very interesting.
Thank you Carl.

Aziz

[Carl-NC]

Common-base is generally the better approach to low noise. Here is a CB preamp I was messing with several years ago. It does not require a floating voltage source. I don't recall where I found this but probably an audio application. The values shown are what was used in my last sim.

[Aziz]

Hi all,

I have put a working LTspice simulation files for your convenience to play with. You can do transient, AC (frequency response) and amplifier noise analysis. Just make the desired command line active and mark the others as comment.

This common base transistor pair amplifier is acting like a TIA (transimpedance amplifier - current to voltage amplifier). This is what it makes it very very interesting. Watch the current through the signal source. It's voltage drops due to source resistance Rs.
It's a best finding gadget for decades!

Fortunately, I have purchased enough transistors ZTX851 and ZTX951 and I will get them very likely tomorrow.

Cheers,
Aziz