PI active damping

[moodz]

I've been using I/T=E/L to calculate current decay. Your graph shows infinite change, wondering how that can be.

Not quite infinite ... :-) ... its about 35 ns in the last drop from about 500 ma down to 0 in the coil.
In the PIC below you can see the difference ... the green trace is a 500 damping resitor showing typical decay ... and the red trace is the damping resistor replaced with a active damping circuit using the same coil etc ...

[moodz]

The PIC below shows the currents in the respective damping circuits ... the blue trace is conventional damping resistor and the the green trace is the active damping ..... the active damping stays constant for the duration of the flyback voltage ... the damping resistor is somewhat slower. For active damping the current has to be EXACT or the coil will over / underdamp remembering that the active damping current is INDEPENDANT from the flyback voltage level. ie it does not vary with voltage across the coil.

[Teleno]

For active damping the current has to be EXACT or the coil will over / underdamp

I gues you use the Rx signal as the input of a feedback loop to control the current sink, the loop having a time constant longer than the detector's swing time (about 1 s.), making it a dynamic detector. The loop also provides ground balance in slow-varying magnetic ground.

For a static detector you'd need to measure L during Tx time and set the damping current of a temperature-compensated current sink accordingly.

[moodz]

I gues you use the Rx signal as the input of a feedback loop to control the current sink, the loop having a time constant longer than the detector's swing time (about 1 s.), making it a dynamic detector. The loop also provides ground balance in slow-varying magnetic ground.

For a static detector you'd need to measure L during Tx time and set the damping current of a temperature-compensated current sink accordingly.

The damping error is sampled from the rx signal however the loop has to be much faster than the detector swing time as the damping feedback signal is proportional to the inductance of the coil ... as ferrites will "modulate" the inductance as you swing across them ... if the loop cannot track these they will appear as false targets. So the loop must be quite fast ... ie above 10 Hz.

For a static detector the control processor opens the control loop leaving the damping control voltage constant. Generally the temperature wont change over several 10s of seconds enough to affect the damping however if you were going to use for long periods you would definitely need temp compensation.

[Teleno]

The damping error is sampled from the rx signal however the loop has to be much faster than the detector swing time as the damping feedback signal is proportional to the inductance of the coil ... as ferrites will "modulate" the inductance as you swing across them ... if the loop cannot track these they will appear as false targets. So the loop must be quite fast ... ie above 10 Hz.

I assume the damping error is the early Rx signal. Then how do you tell whether you're sampling the error or a target's response? A fast compensation loop would also eliminate the target, wouldn't it?

[moodz]

I assume the damping error is the early Rx signal. Then how do you tell whether you're sampling the error or a target's response? A fast compensation loop would also eliminate the target, wouldn't it?

Intuitively ... what you say makes sense ... the active damping would cancel the target signal ... however it does not because the active damping solves the damping equation such that dv/dt = 0 and di/dt = 0 at a specific sample time for the coil only ... the targets will have different time constants ... so any target responses not having a tc identical to the coil will not be affected as they are sampled at a different time according to the desired tc. Since ferrrous objects have the most affect on coil inductance ... then it can be shown that response to iron targets for example is reduced somewhat .... in comparison to non-ferrous ... for gold hunting this is good. ;-) though. The active damping if tuned correctly will completely eliminate a ferrite response if scraped or waved near the coil. Response to very small non-ferrous targets ( eg .0x gram gold targets ) is unaffected. I am talking mono coils here .... DD coils etc will benefit also.

If the ground contained no ferrous components then the active damping loop response would not matter so much. ( could be slow or fast ).

Another factor to deal with in these control loops is things like AC mains etc.

[moodz]

PIC below may explain better. There are three traces of coil current in an active damped coil. The blue trace is the normal optimal damping, red trace is a target trace and green trace is a ferrite near coil causing an increase in inductance. For the the normal(blue) and target(red) the peak coil current is unchanged ... however for the ferrite loaded coil ( green ) there is a decrease in peak current ( ie less energy stored in mag field ). Yet even though the ferrite and target are equivalent in their magnitude of effect on the coil ... the target signal is much larger. So yeh if you can measure the energy stored during TX ... you can calculate if its being affected by ferrite and can compensate. For example you can use a suitable fixed active damping value and the control loop will vary the Transmit duty cycle to achieve optimal damping. ( varying the RX timings also required. ) :-)

[green]

Not quite infinite ... :-) ... its about 35 ns in the last drop from about 500 ma down to 0 in the coil.
In the PIC below you can see the difference ... the green trace is a 500 damping resitor showing typical decay ... and the red trace is the damping resistor replaced with a active damping circuit using the same coil etc ...
[ATTACH]44138[/ATTACH]

Interesting PIC's, could you post one that includes coil volts, coil current and damping current?

[moodz]

Interesting PIC's, could you post one that includes coil volts, coil current and damping current?

Coil current : BLUE
Coil voltage : GREEN
Damping Current in Current sink : RED



For one pulse the energy dissipated by the damping circuit is .... RMS vpeak x Idamp x time
Energy in damping cct = 0.7071 x 276 volts x 0.284 amps x 1.08 microseconds = 59.85 microjoule


... and the energy stored in the coil is 0.5 x L x Ifinal x Ifinal
Energy in coil = 0.629 amps x 0.629 amps x 300uH x 10E-6 x 0.5 = 59.34 microjoule

This shows that the damping circuit is correctly dissipating all the power stored during TX On because Energy stored in coil magnetic field = 59.34 uJ AND energy damped = 59.85 uJ showing very good correspondence.

This PI is running at 10 Khz ... so the power = joules per pulse x pulses per second = 59.5 uJ x 10000 = 0.595 watt.

The coil is damped in approx 1.08us.

[green]

Coil current : BLUE
Coil voltage : GREEN
Damping Current in Current sink : RED

[ATTACH]44159[/ATTACH]

For one pulse the energy dissipated by the damping circuit is .... RMS vpeak x Idamp x time
Energy in damping cct = 0.7071 x 276 volts x 0.284 amps x 1.08 microseconds = 59.85 microjoule


... and the energy stored in the coil is 0.5 x L x Ifinal x Ifinal
Energy in coil = 0.629 amps x 0.629 amps x 300uH x 10E-6 x 0.5 = 59.34 microjoule

This shows that the damping circuit is correctly dissipating all the power stored during TX On because Energy stored in coil magnetic field = 59.34 uJ AND energy damped = 59.85 uJ showing very good correspondence.

This PI is running at 10 Khz ... so the power = joules per pulse x pulses per second = 59.5 uJ x 10000 = 0.595 watt.

The coil is damped in approx 1.08us.

Thanks. Now I have to figure why I/T=E/L doesn't work.

[moodz]

Thanks. Now I have to figure why I/T=E/L doesn't work.

... to be fair ... here is a circuit I prepared earlier. This is the circuit that provided the values for the calculations above.
Note : a 500 ohm "damping" resistor still appears in the circuit as the diode in the current sink is an imperfect switch. This resistor "mops" up the residual decay.

M2, V3 and D2 form the constant current sink.

I have cut the feed back loop out so that a steady state is maintained to do the calculations ( ie fixed control voltage )



Enjoy ... no laws of physics were violated in the making of this circuit. LOL.


.. and the LTSPICE file of same ..

[green]

... to be fair ... here is a circuit I prepared earlier. This is the circuit that provided the values for the calculations above.
Note : a 500 ohm "damping" resistor still appears in the circuit as the diode in the current sink is an imperfect switch. This resistor "mops" up the residual decay.

M2, V3 and D2 form the constant current sink.

I have cut the feed back loop out so that a steady state is maintained to do the calculations ( ie fixed control voltage )

[ATTACH]44160[/ATTACH]

Enjoy ... no laws of physics were violated in the making of this circuit. LOL.


.. and the LTSPICE file of same ..

Thanks for the circuit. Tried decay into inverting amp. Don't see an advantage with active damping, maybe I'm missing something.

[green]

I see I changed the BSS84 gate voltage when I was playing. Looks a little better with 5.3114 volt gate volts reply #86.

[moodz]

I see I changed the BSS84 gate voltage when I was playing. Looks a little better with 5.3114 volt gate volts reply #86.

Ideally you want to implement a feedback loop to control the gate voltage.

Here is a simple charge pump loop that gets the gate voltage within 1% of the required voltage however as mentioned in the patents a sampling loop with some gain is more optimal. However this circuit works great on surf PI's and pinpointers.

The loop provides negative feedback as it "inverts" the +ve overshoot ( damping error ) from the flyback voltage ( -ve ) damping.

Note also the patents talk about a switch .... not a diode D2 ( though a diode can be used ) ... much better results can be obtained with a switch ( eg mosfet ). The switch needs to be ON during active flyback and OFF during receive. ( eg 1 or a few microseconds )

C2 looks counterintuitive however it is isolated by diodes D4 and D3 so does not affect once you need to sample receive.

R2 is a "leak" resistor to prevent overcharge conditions in the charge pump storage cap. ( adjust for best loop response )

M3 prevents excessive +ve voltage being presented to the input amplifier.

Note this is a "crude" but easy to test loop. ( no microprocessor sampling control LOL ).