field test unit no 001 "model T"

[Tinkerer]

Instead of a snubber, which is partly resistive and wastes some energy, we could go full monty and use TEM exciter instead. It would not need a charging period and it would conserve energy. Such TEM pulse would be duration limited by it's capacitor to the largest possible with a desired shortest target tau and sample delay.

Have a look at this simulation.
The green trace is the coil current.
The blue trace is the coin voltage.
The red trace is the RX coil current with IB configuration.

Adding more capacitance to C1, reduces the coil voltage and increases the Flyback time. You need the increase the OFF time.

The full cycle is 100us, for a 10,000 PPS

Try the target response with different target TC's

I add the LTSpice sim in zip

[Tinkerer]

Well, that's a good reason to compare the two states:
- with avalanche mode
- without avalanche mode
and the effects to the damping time variation and response.

Who is going to make the interesting comparison?
Ready to get ?
Aziz

The problem with the switching noise, is the coil, cable, shielding capacitance. This spreads or integrates the noise spikes so that the resulting effect is prolonged or persists some time. This is clearly visible when using an IB coil arrangement that allows observation of the RX waveform during the whole PI cycle.

[Davor]

The same accelerating effect happens with TEM exciter. Point is that a capacitor gets reverse polarised, and in fact a reverse voltage accelerates the transition. The main difference between a snubber and a bare capacitor is that a circuit with a snubber has a Q well below 0.5

Perhaps a simple snubber could be a solution for accelerating the simple rigs such as Surf or Baracuda?

[Ferric Toes]

Well, that's a good reason to compare the two states:
- with avalanche mode
- without avalanche mode
and the effects to the damping time variation and response.

Who is going to make the interesting comparison?
Ready to get ?
Aziz

Not me. I am satisfied that avalanche noise is not significant when using the delay + sample system. Of greater priority is better match between coil and preamp to reduce noise there.

Three years ago I designed a high power PI for underwater pipline detection. Peak pulse current was 20A in a 500uH coil. Power Mosfet on fan cooled heatsink. Temperature sensor on heatsink kicked fan on at 70C. Coil ran at 70C in air. Mosfet avalanche volts 250, on resistance 0.03 ohms. In the workshop I used a 20in square dual stacked differential coil to cancel external noise. No visible increase in noise at preamp o/p whether TX was on or off.

One feature of diode/capacitor snubber that might make a small difference is that the rate of field switch off speeds up as it approaches zero. 1/4 cycle of the LC frequency. This might result in a better signal from small objects where the switch off with an avalanche circuit is less than 5X object TC.

Eric.

[Aziz]

Methinks your conditions are not quite the same if your coil is tuned. You would see a burst of noise if the peak exceeds the avalanche voltage. A paper on Google dealing with Noise in Semiconductors says that low noise performance returns once you are out of the avalanche region.

Eric.

Well, that's a good reason to compare the two states:
- with avalanche mode
- without avalanche mode
and the effects to the damping time variation and response.

Who is going to make the interesting comparison?
Ready to get ?
Aziz

[Ferric Toes]

Methinks your conditions are not quite the same if your coil is tuned. You would see a burst of noise if the peak exceeds the avalanche voltage. A paper on Google dealing with Noise in Semiconductors says that low noise performance returns once you are out of the avalanche region.

Eric.

[Aziz]

OK, but in a practical PIB detector I would be sampling soon after the start of the TXon period when the Mosfet is fully turned on. On resistance is usually 0.5 ohms or less, which from the noise point of view would be in parallel with the coil resistance (say 1 ohm). Noise should be miniscule compared to input resistor and preamp noise. What happens during the avalanche period does not matter as no sampling is taking place until after a further delay. You could have noise superimposed on the switchoff ramp but I can't see that this amounts to anything significant. I have used snubbing circuits but found that the diode used to charge a capacitor introduced low level effects, while Mosfets gave a much cleaner linear ramp type of switchoff.

Eric.

Hi Eric,

the avalanche noise can be particularly observed on high current pulses. I have a tuned pulse transmitter (we call it TEM TX here), where this can be observed in an induction balanced configuration easier. As long as the flyback voltage isn't above the avalance voltage, the mosfet operates quiet. But if the mosfet avalances at higher voltage levels, a horrible noise can be clearly observed.

BTW, the avalanche mode heats the mosfet and it won't be very much amused by that too. Maybe the excessive heat is contributing to the noise much.
Aziz

[Ferric Toes]

OK, but in a practical PIB detector I would be sampling soon after the start of the TXon period when the Mosfet is fully turned on. On resistance is usually 0.5 ohms or less, which from the noise point of view would be in parallel with the coil resistance (say 1 ohm). Noise should be miniscule compared to input resistor and preamp noise. What happens during the avalanche period does not matter as no sampling is taking place until after a further delay. You could have noise superimposed on the switchoff ramp but I can't see that this amounts to anything significant. I have used snubbing circuits but found that the diode used to charge a capacitor introduced low level effects, while Mosfets gave a much cleaner linear ramp type of switchoff.

Eric.

[Aziz]

How would you observe this noise? Does not the very target act as an integrator?

Eric.

Well, if you put a secondary RX coil in induction balance condition, one can measure the avalanche noise. Some mosfets do have a very bad avalanche noise behavior. Some better.

This noise isn't much but if you want to beat the evil best, its getting very important to avoid it.

Aziz

[Davor]

Instead of a snubber, which is partly resistive and wastes some energy, we could go full monty and use TEM exciter instead. It would not need a charging period and it would conserve energy. Such TEM pulse would be duration limited by it's capacitor to the largest possible with a desired shortest target tau and sample delay.

[Ferric Toes]

No Eric, it is noisy during the avalanche mode (off-time flyback period) modulating the target response, timings (time constant modulation) and so on...
This is one of the secrets of ML (& the QED). Snubber circuits help to control the max. voltage to avoid the avalanche mode.
Aziz

How would you observe this noise? Does not the very target act as an integrator?

Eric.

[Ferric Toes]

If we drive it waaaay upstream to absurdity, we could produce infinitesimally short pulse spiking at few kV ... with very mild target excitation.

On the low side, we could slow down flyback pulse, at non-lethal voltages, to achieve more.

The "few kV" mode raises potential problems with breakdown of components, coil insulation etc. The 1N4148 diodes are bulletproof for spikes of 500mA for a few uS, but after that, who knows? I go for low side when designing PI for small objects. 1000's more pulses per sec. can be averaged in a given time to achieve a good S/N ratio. Thin wire for TX, 0.25mm PTFE or genuine Litz of same gauge for very fast coil. Doing away with the coil cable and having TX/RX close to coil helps enormously by reducing capacitance.

Eric.

[Aziz]

Hi Aziz,

Are you saying that they are noisy after the avalanche period is over, the device is off, and into the sampling period? I have looked for this but never seen it.

Eric.

No Eric, it is noisy during the avalanche mode (off-time flyback period) modulating the target response, timings (time constant modulation) and so on...
This is one of the secrets of ML (& the QED). Snubber circuits help to control the max. voltage to avoid the avalanche mode.
Aziz

[Ferric Toes]

Hi Aziz,

Are you saying that they are noisy after the avalanche period is over, the device is off, and into the sampling period? I have looked for this but never seen it.

Eric.

[Aziz]

BTW, the mosfet avalanche mode isn't the goal for us.
Has anyone seen the noise in this mode??? Horrible!!!!
Aziz