Induction Balance Stuff - Single/Multi Frequency Response, GB, Disc, Measurements, Ideas, Fun, etc.

[Aziz]

Hi all,

I have found another TX coil in my old box. No enamelled copper wire this time. It should withstand high voltages up to 600 V. Pity, the copper litz wire isn't tinned.
TX: L=440µH, R=0.3 Ohm.
I have reduced my shunt resistor to 250 mOhm due to power efficiency considerations. I don't have any low TCR shunt resistors (low TCR = low Temperature Coefficient Resistor). I have used 4x 1 Ohm parallel metal film resistors.

Now the transmitter is 20-30 times more quiet. No sparks between the interwire windings of the TX coil anymore.

Nevertheless, the bare enamelled copper wire can still be used for high voltage uses. As Spider-Web formed TX coil. Between the winding-to-winding sections, the voltage difference isn't much. Enamelled copper wire should withstand this voltage. Coil leadings and cables need thicker insulation however. The coil would weigh less.
I would like to use such a coil as I still have enough enamelled copper wire.

The RX coil isn't critical. Enamelled thin copper wire is enough. The RX coil won't see the high voltage due to inductance balance configuration.
Cheers

[Aziz]

Hi all,

OMG, this is the outcome of the big hammer. I have hit something.
Bzzzzzt, puff, smoke. No joke this time.

I spent at least two hours searching for the reason. I didn't find. The test board is alive. The mosfet is alive. Everythings seems to work. But still not transmitting. Something smells burnt and the mosfet is getting hot. It shouldn't get hot.

Well, Murphy's Law has struck again.

Finally, I checked the tune capacitor Ct1 and the TX coil.
There we have. My transmit coil has burned out. It has now something around 5 µH and 0.3 Ohm now (it should have 1.1 mH and 4 Ohms).

And why the hell was my transmitter so noisy?

I have finally disassembled my old TX coil. Fck! Enamelled copper wire (0.5 mm). Looking very strange! What happened to the enamel coating?

Look:


Every dark region had a fine and destructive spark over the time using high flyback voltages (200 .. 500 V). Dark regions had been corroded over time.
The last spark welded an interwire connection and shorted internally the coil finally. Somewhere.

There must be 1000' of little sparks. Even much much more.

And this is the reason for, why my transmitter was so noisy during the recent tests.
This was my next task to search for noisy sources. It's done now.

I need to wind a new TX coil. I need a wire, which can withstand the high voltage (600 V) and high current rate.
I am looking for coated tinned litz wire. AWG 16 or thicker (AWG 14).

There are more pictures in this post. Have a look and think of wires for coils.
Cheers

[Aziz]

Hi all,

I face with big problems with the double tuned transmitter. It doesn't make sense to follow this concept anymore.
There is a much much simpler solution.
Well, if it doesn't fit, use a bigger hammer.

We are reducing the transmitter back to the off-time tuned TEM and measuring the on-time current using a shunt resistor.
The voltage divider R2/R3:
R2 will be replaced by a large electrolytic low ESR capacitor (+ at Vc, - at R3). R3 becomes the shunt resistor (100 mOhm ..1 Ohm).
Leave Ct2, we don't need it anymore. Choke L1 remains with low ESR and 1 mH .. 4 mH and it isn't critical at all (take what you have).

We are measuring the low frequency on-time current instead of the voltage at node Vc. All effects on TX inductance change remains. We can still monitor the battery voltage and have a transmitter reference signal. The original TEM transmitter is much more flexible and robust to use. Does not create problems and is always working fine. It is very power efficient and it is a power beast too. TX power will be controlled by source voltage level and operating frequency.

Where is my big hammer?
Aziz

[Aziz]

Hi all,

the last experiment reminds me to the off-resonance principle. So the transmitter is suffering from energy. It is kept alive with minimum amount of energy to operate (for instance to oscillate). Any change of the energy caused by targets can be seen in the amplitude of the transmitter voltage.

Same principle can be achieved with this dual-tuned transmitter. But we have to limit the energy in the transmitter to detect the big changes of the energy balance. We have to limit it either, because the series resistor get quite hot and you can detect the temperature gradient of the hot resistor. And the power efficiency gets lost.
Limitting the TX-coil current will limit the detection depth either. But it is still possible to operate the TX-coil one decade lower as usual (instead of 1 A, 50 .. 100 mA peak). This won't cause much temperature drift of the current limitting resistor.

I will do some more experiments with it. I still have not connected a receive coil yet. I am observing the TXref signal (just like detecting on a mono coil).
Cheers

[Aziz]

Hi all,

this is becoming really a funny thread.

The inductance of the choke L1 was not the real reason for unstable operation point of the transmitter. It was it's low series resistor value (too low).
I have accidentally taken a second series choke (L1 + another choke in series I have found in my box). The second choke had thin wires and therefore high series resistor. Well, this solved the issue yesterday, but I told you bull-sh1t yesterday.

Murphys law?
(Sh1t happens)
I have to put a series resistor between choke and node Vc to decouple the transmitter from the battery source.
The transmitter becomes now a critical energy balanced system.

Good news:
The targets have more effect at node Vc now:
- stealing inductance from the TX coil -> inductance change of TX-coil
- the critical energy balance changes -> resistive Targets steal energy from the energy balanced transmitter.
The ground has less effect on this.
I can use the information from the TXref to minimize the target losses caused by the lossy GB method and will have more sensitivy to large targets too.
This is what I didn't expect.
Cheers
Aziz

[Aziz]

Hi all,

funny and mysterious things did happen again.
I have finally connect the capacitors Ct1 (FKP1, 4.7 nF, 1250 VDC) and Ct2 (MKC4, 2.2µ+1µ+2.2µ=5.4µ, 100 VDC, all parallel) with a small TX coil (low Q coil, L=1.1 mH, Rs = 4 Ohms), which I have found in my old box .

Plugged the 9.6 V battery to the breadboard. Bzzzzt, puff, bang, smoke!
No, I'm kidding you.

It worked, but I couldn't reach the optimum (=critical) operating point (0 V at node Vc). The transmitter was stable down to Vc at 2 V. Below this point, wild oscillations of the transmitter happened. Above no problems (higher operating frequency, less TX-coil current).
WTF, I need a full voltage swing of the Vc node. The battery voltage Vb should be optimal translated into a stable AC signal (half sines), which should be between 0 V and sqrt(2)*Vbat V.

I have tried many things to solve this issue (seperate batteries for analog and transmitter parts, different mosfets, etc.).

I have found the issue: Make the choke L1 inductance higher (minimum 8 mH).
Have a look at the schematics above. The Choke L1 and Ct2 are forming a series LC resonant circuit. If my operating frequency comes near to the resonant frequency to set the optimum operating point, wild oscillations happen then. Even large targets nearby the TX coil made the transmitter quite unstable. The higher inductance of the choke shifts the resonant frequency to the low frequency region.

I have to replace my choke L1 yet. This transmitter is somehow quite interesting. I'm sure, we can realize the metal detector even with a mono TX coil (no RX coil). With reduced sensitivity of course. I will be testing the transmitter further. Pictures will be provided later.
Cheers

PS: I have made the transmitter incidentaly tripple tuned TEM by taking low inductance choke.

[Aziz]

Hi all,

funny thing happened on timing jitter issue.
#1 My old analog oscilloscope is generating the most of the jitter noise itself. It's own 1 kHz reference signal reveals how noisy the delayed and magnified time base is realised.
#2 I was looking for signal transitions way beyond 1 sigma (1 sigma = standard deviation), maybe 3 - 4 sigma for long period of time using min and max and calculating the timing difference for jitter. It's not correct. It is gaussian distributed.
So my measurements are not correct and timing jitter isn't really an issue.

[Aziz]

Hi all,

I have soldered the breadboard and checked the clock detectors and mosfet gate signal. They are working as expected and specified.

But the EMI noise picked up by the board and line output cables (+sound card generated noise on line output), will lead up to 400 ns timing jitter on the mosfet gate signal @3 kHz pulse frequency. This is absolutely no go. We are producing more phase noise in the transmitter.

A crystal controlled internal oscillator to generate the mosfet gate signal would minimize such noises however.
But hey, it doesn't matter. We have our TX reference signal (TXref) to compensate for timings jitter. The RX-channel will be synchronized to the TX-channel. The TX-channel will be infected for further processing however (limitted to battery voltage level and synchronization). Anyway, I will keep up finished the transmitter. We will see, what we can do.

Aziz

[ivconic]

Considering what could be seen in the Lavry documents; that card of yours is for home conditions and cheap trials; more than decent.
...
Very good PDF, thanks!

[Aziz]

Hi all,

I have finaly found a similar white paper of the LCR measurement, containing obviously the same formulas, which I use for the LCR implementation. Now I can check my LCR implementation by reverse engineering the source code and check the calibration procedures.
Link: https://download.tek.com/document/75..._MR_Letter.pdf

The calibration determines the parasitic impedances like the series complex impedance (Zs) caused by cables and parallel complex impedance (Zp) caused by the input impedance of the sound card. But I have not found something about the calibration correction math.

If a second alternate different coding gives the same results, I can trust the LCR measurement outputs.
Aziz

[Aziz]

Ok, I get it. But the cost is the loss of resolution. If I calculated correctly; 5-6 bits less than declared. By the way; how much is your audio card declared?
(I apologize if I missed that information in previous posts)

Sound BlasterX G6:
Line input: 24 or 32 bits @ 192 kHz sample rate (max)
Line output: 24 or 32 bits @ 384 kHz sample rate (max)

I'm using only 24 bits @ 192 kHz sample rate for both.
ENOB is 19 - 20 bits.

[ivconic]

Hi ivica,

thanks for your infos.
Nevertheless, we have to live with that, what we have.

Let's look at frequency response at around 90 kHz input signal on my G6 sound card.
Noise contribution: approx. 25 dB (mean noise floor at -94 dB)
Input filter signal attenuation: approx 7 dB
Total: 25+7 = 32 dB
Factor f = 10^(32/20) = 39.8 (almost 40 x)

We can rise the input signal with an amplifier with gain of 40 to compensate the high frequency signal loss. With additional amplifier noise contribution of course. The low frequency part will be rised from approx. -120 dB to -80 dB. I will leave a spare area for the RX amplifier on my test board.

I know, there are better external USB sound cards available. But the public specs are not reveailing the real performance (cheating, cheating, everwhere cheating). I have to test them by myself.

Cheers,
Aziz

Ok, I get it. But the cost is the loss of resolution. If I calculated correctly; 5-6 bits less than declared. By the way; how much is your audio card declared?
(I apologize if I missed that information in previous posts)

[Aziz]

Hi all,

I have been looking at the specs of the WIMA capacitors.

The high voltage capacitor Ct1 must withstand high voltage changes and currents. The 600 V voltage rate shown in the schematic isn't really enough. We must use at least 1000 VDC. Even higher is better (1250 VDC). As the pulse is very short (high frequency) and changing, the high frequency AC voltage ratings apply for it. If we have 500 V flyback voltage, we have to use the 3-4x DC voltage ratings. Otherwise, the capacitor Ct1 might get hot and can be damaged.

Best capacitors for Ct1: Pulse optimized foil capacitors with less loss.
Sorted from best to next best:
WIMA FKP1 (foil, best)
WIMA FKP4 (foil, next best)
WIMA MKP10 (metalized, next best)
WIMA FKPx (foil, next best other types)
WIMA MKPx (metalized, next best other types)

If your Ct1 gets hot or gives a hearable high frequency sound, look for a better capacitor.

The capacitor Ct2 isn't critical. It should have less loss, DC voltage rating of up to 50 - 100 V is enough. It will be operated at low frequencies (around pulse frequency).

Cheers,
Aziz

[Aziz]

Hi ivica,

thanks for your infos.
Nevertheless, we have to live with that, what we have.

Let's look at frequency response at around 90 kHz input signal on my G6 sound card.
Noise contribution: approx. 25 dB (mean noise floor at -94 dB)
Input filter signal attenuation: approx 7 dB
Total: 25+7 = 32 dB
Factor f = 10^(32/20) = 39.8 (almost 40 x)

We can rise the input signal with an amplifier with gain of 40 to compensate the high frequency signal loss. With additional amplifier noise contribution of course. The low frequency part will be rised from approx. -120 dB to -80 dB. I will leave a spare area for the RX amplifier on my test board.

I know, there are better external USB sound cards available. But the public specs are not reveailing the real performance (cheating, cheating, everwhere cheating). I have to test them by myself.

Cheers,
Aziz

[ivconic]

The real truth is always buried in the details.
If I take top model and its specifications as a reference; the first thing that catches my eye is the latency data, ha!