PI active damping

[Teleno]

Also known as Constant Current damping as disclosed in this patent by P. G. Moody, whose nick in this forum was "moodz" (he no longer participates).

Normally a "critical damping" resistor is used, calculated as:

At the end of the flyback the voltage induded by the target is measurable across resistor R.

Active damping replaces R by a constant current sink. This way the resistance seen by the coil higher than R at the beginning of the flyback and then much lower at the end. As a result the flyback period is shortened allowing for earlier sampling. The dI/dt is also faster inducing a higher signal in the target. Sounds like it's all benefits, doesn't it?

But... at the end of the flyback (Icoil near 0) the R of the current sink is very small. The coil is practically shorted and we cannot measure the voltage induced by the target.

Possible approaches to measure the target signal with this arrangement:

1. placing a diode between current source and the reference voltage (GND or VDD depending on the coil driver being P-MOS or N-MOS)

This classical solution increases the equivalent R as the voltage at the coil goes below 0.7 volts, causing ringing on top of the target's signal.

2. Using a separate receiving coil.

3. Measuring the end current in the coil instead of the voltage.

Let's see if we can materialize a circuit to do this.

[Teleno]

These are simulations of the flyback using "critical damping" (blue) and "C.C. damping" (red)

Coil current during flyback:




Signal at target (tau = 1us)

[Davor]

I can't see the images at the moment (proxy quirk) but I'm kinda sure you forgot about the linear resistance feature of a MOSFET that is commonly used for active loads. The best of all is that you can steer it by voltage applied to the gate, which may help with adjusting damping for different coils.

[Teleno]

I can't see the images at the moment (proxy quirk) but I'm kinda sure you forgot about the linear resistance feature of a MOSFET that is commonly used for active loads. The best of all is that you can steer it by voltage applied to the gate, which may help with adjusting damping for different coils.

I used a different current source, but yes, the usual high voltage MOSFET ends up as a resistor at flyback end, however, it comes with a parallel recirculation diode that has to be blocked by another diode in series. This causes ringing.

[Davor]

Yes, it must be replaced by active element as well.

[Teleno]

Yes, it must be replaced by active element as well.

I use a Zener as a floating ground so that he current at L1 never actually reaches zero, just a very small value.



Instability disappears but the target signal is no longer measurable as a voltage, it turns into a current signal.

[green]

Looks like a challenge. Wondering if we can't pick some circuit parameters, so we are all working on the same thing. Suggest: 300 uh, 935 pf, 2 ohms for the coil. Gives a circuit resonance of 300 khz. should be able to critical damp to less than 1 mv in a little over 10 usec.

[Teleno]

should be able to critical damp to less than 1 mv in a little over 10 usec.

From what peak current?

At 1.7A the flyback looks like this:



3us to zero.

and the target signal (current)

[green]

From what peak current?

At 1.7A the flyback looks like this:

[ATTACH]31230[/ATTACH]



3us to zero.

and the target signal (current)

[ATTACH]31231[/ATTACH]

1.7 amps is good. What would the preamp circuit look like? Mono or IB coil? I was thinking the current would have to decay to a lower value, maybe not.

[Teleno]

What would the preamp circuit look like? Mono or IB coil?

That is the question. The Tx coil ends up shorted rather than seeing a resistor in series. I guess an Rx coil is the solution.

I was thinking the current would have to decay to a lower value, maybe not.

The lower the value, the more unstable. A compromise must be found.

[green]

Could someone tell me what I'm doing wrong. I can't get the coil current to decay to near zero current with a CC sink. I used 220pf instead of 935pf I suggested in another reply because M2 adds capacitance. Don't think it's critical.

[Teleno]

Could someone tell me what I'm doing wrong. I can't get the coil current to decay to near zero current with a CC sink. I used 220pf instead of 935pf I suggested in another reply because M2 adds capacitance. Don't think it's critical.

Your zener is 4.7V, but V4 is 8v. You need a zener as close as possible to V4 yet a little under it. Then adjust V2 to get the fastest decay without undershoot.

[green]

Your zener is 4.7V, but V4 is 8v. You need a zener as close as possible to V4 yet a little under it. Then adjust V2 to get the fastest decay without undershoot.

Thanks

[Davor]

You could also move a Zener trainset to MUR460 cathode and go for a P-channel mosfet across a coil, and replace a zener in its source with resistor. That should do.
A small problem is with various leakage currents related to all devices that have any offset voltage across them. Simulations are helpful there only if all the leakage currents are accounted for in models.
I'll place my contender tomorrow.

[Teleno]

It would be useful to determine the impedance of the coil signal after the transient. If it's less than the critical damping resistor (R) we could adjust the current source to have the same minimum resistance. This way we'd have two improvements: faster decay and better impedance matching.

In my simulations, by manually adjusting the current source and its serial resistor I got 12mv for CC damping versus 2mv for critical damping. I used 100ohm in the CC circuit and the coil' s R is 500. This could mean the signal impedance is indeed less than R, but I'm not sure about this.