PI active damping

[mickstv]

Fair enough.

[Davor]

I've found the trick to make active damping work is a trasconductance amplifier. Notice one single stage can amplify well into the volts range with very low noise.

Would you care to share your solution?
I have a solution for devices such as Minipulse, where damping resistor is a part of a feedback network. It could perhaps be improved, but even this way it enables using a simple trimmer to adjust damping on the go, which seem advantageous when experimenting with various coils.

[green]

Teleno. I'm wondering how you damp the coil in reply #43 with no fly back voltage.

[Teleno]

Would you care to share your solution?

For the time being I'd rather stay cautious because I haven't seen this solution anywhere and it might be worth a patent.

I can share its characteristics:

- Very low input impedance through virtual ground (coil practically short circuited at end of transient), meaning very low input noise.
- Signal goes directly into a single stage current-to-voltage amplifier (high gain with a small resistor -> low noise).
- Output of single stage already usable without further amplification.
- No saturation states to recover from (fast response).
- Active damping without instability at end of transient.
- Single supply (except for active damping).

I've ordered some nice high-voltage n-channel depletion mode MOSFET's to to achieve active damping with a single supply.

IXTP08N50D2 500V 0.8 Ohm 800mA
DN3545 450V 20 Ohm 200mA

Available at Mouser electronics.

In short: the simplest PI front-end with low-noise, high gain and single supply. Active component count is 6: 2 x MOSFET, 2 x bipolar, 2 x schottky diodes.

Teleno. I'm wondering how you damp the coil in reply #43 with no fly back voltage.

The flyback voltage is there, but the signal I posted is the output of the current-to-voltage amplifier which cannot swing above +Vcc.

[green]

The target time constant looks to be about 1 usec. Could you share what the target is and your coil diameter. Looks like you have a good solution,hope it works out for you.

[Teleno]

The target is a gold chain 22kt. The coil for this experiment 7cm diameter, spider-web with 12 tuns, 13uH at 600mA. it's just a proof-of-concept table experiment.

My next coil will be a 50uH spider-web of Litz wire running at 2A, and we'll see what it brings.

I might also try a ferrite core just like the Falcon MD20. The intended depth for this detector is about 2 in. (5 cm.) for a gold particle the size of a pinhead.

Looks like you have a good solution,hope it works out for you.

Thank you. I've never seen a current-mode PI detector and the concept seems promising. Think of it like this:

- A 1 Ohm coil receiving a 1 mV signal will produce 1 mA if shorted.
- In order to get 1V from 1mV you need need a high-impedance amplifier with 1000x gain. If you use op-amps you can't get away with lees than 100K resistors - noisy! If you do it in stages it's also noisy.
- Now in order to get 1V from 1mA you just need a single 1K resistor. Simple an noise-free. What's more, it's a lot faster than an Op Amp.

You get the idea? And in both cases it's the same target!

[Davor]

Interesting idea. So active damping is used simply to quench the coil response, and you expect to obtain some transformation of target responses by means of switching between a shorted coil and high impedance.
I'm on a journey now and I have no idea how well it may work.
Regarding patenting, please take a great care about it. There are many modes of protecting your IP other than patent. Patents are probably the worst of all. A private person patenting is usually a sucker that gets squeezed financially and psychologically by the time a patent is granted, and then it "protects" your idea in a limited market, provided you sue everyone breaching your IP right. In other words - it sucks. If you are able to produce your idea, again, patents only suck your resources that you'd be better off investing.
Copyright is your friend. Whatever you publish becomes yours by the fact of creation. Unlike patents, copyright can not be taken away from you, and you can't be prevented to produce your invention.

[Teleno]

Interesting idea. So active damping is used simply to quench the coil response, and you expect to obtain some transformation of target responses by means of switching between a shorted coil and high impedance.

The Vds/Vgs characteristic of the MOSFET does a soft switching. When the Vds is high (coil transient) it works in the saturation region (high impedance), when it's low (end of transient) it enters the linear region (low impedance). Just connect the source to a virtual ground that can absorb the instantaneous peak current, this woud be the virtual ground at the input of a trasconductance amplifier. The target's current I enters virtual ground and a voltage R x I is available at the amplifier's output.

[Davor]

I'm still unable to simulate this. You are aware that you'd need to quench the coil's own response prior to taking a target sample? That means you'd have to use a third impedance that is equal to the critical damping while the coil's own response dissipates, then enter a short circuited stage, and lastly the high impedance stage. Otherwise you'd have weak targets masked by the coil's own response.

[Teleno]

I'm still unable to simulate this. You are aware that you'd need to quench the coil's own response prior to taking a target sample? That means you'd have to use a third impedance that is equal to the critical damping while the coil's own response dissipates, then enter a short circuited stage, and lastly the high impedance stage. Otherwise you'd have weak targets masked by the coil's own response.

The active damping is tuned to absorb exactly the amount of energy stored in the coil with no target. Any target alters the energy balance, so the residual energy after active damping belongs to the target.

Here's a simulation setup:



damping.zip

The resistance of the damping MOSFET (trace: (V(n004)-V(n013))/Id(M2)) peaks at 458 Ohms in the middle of the transient and goes down to 0.42 Ohms when the target signal (trace: Id(M4)-Id(M2)) is at a maximum. The difference target/no-target is 1.4mA after the transient. In practice you have an offset of 5.4mA (Id(M2) biasing) , so your target is the difference between 6.8mA and 5.4mA. Offset can be reduced using a constant current sink set around 5.4mA.

[green]

Finally got active damping with spice to maybe look better than critical damping. Have tried some in hardware that didn't work, haven't tried this circuit yet. Wondering if anyone is using active damping that would share there circuit.

[Teleno]

Remove R1, it slows the decay. Tune Vgs of M2 instead to achieve optimum decay rate.

[green]

Remove R1, it slows the decay. Tune Vgs of M2 instead to achieve optimum decay rate.

Lowered R1 to 65 ohms(limits current M2 when Tx on). Can adjust gate volts to get a good looking signal but gate volts setting is a lot more sensitive. Have you tried a similar circuit with hardware? Is temperature a problem? Everything I've tried so far in hardware isn't as good as critical damping.

[Teleno]

Lowered R1 to 65 ohms(limits current M2 when Tx on). Can adjust gate volts to get a good looking signal but gate volts setting is a lot more sensitive. Have you tried a similar circuit with hardware? Is temperature a problem? Everything I've tried so far in hardware isn't as good as critical damping.

The gate voltage is indeed very sensitive to temperature, you need a feedback loop to control it.

[green]

The gate voltage is indeed very sensitive to temperature, you need a feedback loop to control it.

What signal do you use for feedback?