ZP ACTIVE DAMPING and NOISE CANCELLATION

[moodz]

Decided to split this topic away from the MAGPI detector which is a project that uses ZP damping.

ZP aka Zero Point damping uses the non linear characteristic of mosfets where for a given gate - source potential the drain current will substantially remain constant even though there are large voltage changes at the Drain.

This "non linear resistance" can be used to damp a coil faster than any sort of fixed resistance.

The damping is achieved by a negative feedback loop that samples the "damping error" and sets the gate voltage appropriately. Once the optimal damping point is achieved the gate voltage remains substantially constant.

The amplifier used in the loop is functioning as a TIA or transimpedance amplifier ( current to voltage ) but is also the preamp for target samples ...

However if a single feedback loop is used the damping loop can be affected by Earth Magnetic field and Mains Hum .... so a second feedback loop is provide to sample this "noise" and remove it also.

A couple of variants and a sim file for LTSPICE.



ZPDAMPING.zip

moodz

[Mr.Jaick]

I really like the Idea of ZP
and I have some questions if you don't mind
first about the auto EF adjust loop
it's a control loop, the op amp will do anything on the output to make two inputs equal
non-inverting is ground so the output will change until the incoming samples to the inverting input are at ground potential thus eliminating any offset
but the question is; does it produce some noise voltage itself?
I ran same simulations on that and i can see some bad triangle noise




could that be the opamp is trying to keep the output at desired value thus hovering around it and making this noise?
op amps are set to "Boyle model" in the simulation so they are as close to real world as possible

[Carl-NC]

Might be a simulation artifact. The default integration method is trapezoidal, try setting it Gear instead. In Spice this is done with the command:

.options method=gear

I don't know how to do it in MicroCap.

[Mr.Jaick]

Might be a simulation artifact. The default integration method is trapezoidal, try setting it Gear instead. In Spice this is done with the command:

.options method=gear

I don't know how to do it in MicroCap.

it was exactly the same command in Micro Cap
it's real clean now and works very nice

you're a lifesaver man thank you again
now what could be the purpose of 100R resistor in series with integration cap? sets a minimum gain for high frequency noise maybe?
can't see any affect in droop or noise in my oversimplified simulation

[Carl-NC]

The output impedance of a bipolar opamp is generally inductive because of the beta roll-off of the output devices. So it's not a good idea to drive a purely capacitive load, some series R will improve stability.

[Mr.Jaick]

strange for me
any current through that capacitor has to go through input resistor (10K) since input of both op amps are ideally infinite
i was thinking that the input resistor will prevent the op Amp from any type of oscillation

[green]

Tried adding a 10us TC target and a Rx coil to simulation to compare with normal PI decay. Normal decay looks like I would expect, ZPDAMPING not. What am I doing wrong?

[green]

Tried adding a 10us TC target and a Rx coil to simulation to compare with normal PI decay. Normal decay looks like I would expect, ZPDAMPING not. What am I doing wrong?

I see one thing I did wrong. Coupled Tx to target and Tx to Rx instead of target to Rx. Changed simulation, still doesn't look right. Any suggestions to try?

[green]

Got Rx to make sense.

[Vadim18]

Heres my idea for active damping, tested last year in software.

[moodz]

Procastination is the biggest enemy of progress and so is too much time in front of the simulator .... so I had some boards made to test the ZP damping theory nearly 3 years ago now but only just built one of them up ( properly ... there was an earlier one but that is another - interesting story ).

Anyway to cut a long story short the assembly is complete and except for some minor tweaks to the source code for the PIC controller was able to obtain good results.

For a simple thruhole design with just 1 SMD transistor ..

Excellent damping of the flyback in under 1 microsecond from a 260 volt ( negative going peak ).

Excellent Ferrite , EF , mains EMI rejection. Rejects hot rocks that fool my GPZ7000.

Detects .05 gram nugget at 1 cm in air. Detects 10cm x 10 cm foil target in air at 1 meter. Using a minelab ( 10 inch ? ) monocoil.

No controls to tweak .... there are several settings to adjust for factory baseline in the optional LCD / rotary encoder but not required for operation - feedback loops handle everything.

Runs of two lithium 3.7 volt cells or 12 volts ext. ( @ 250 ma max )

First pic below ...
Yellow is TX control ( high is off )
Light Blu is preamp o/p ( gain = 1000 approx )
Dark Blu is flyback measured directly off coil -260 volts peak ( no snubbing or avalanche in normal operation )

[Mr.Jaick]

such a winner moodz.
I think there are a few things to consider:
1) the damping loop sample delay needs to be decreased as the battery voltage decreases, because less energy in the coil to be damped, but what if we don't decrease the delay when the battery gets discharged? is it gonna do any harm to the loop stability or performance? most probably not...

2) the damping FET dynamically discharges the coil energy in the shortest amount of time to the ground potential, doesn't that little FET get red hot I wonder? especially for higher-power TX pulses.
what is your tx pulse width in this example?

by the way, the forum feels Fresh, thank you Carl.

[moodz]

Hi mrJ

Thanks .. to answer your questions the on board voltages are all regulated and the CPU monitors the battery voltage. So for two lithium cells the voltage will range from about 8.4 volts fully charged to 7.4 discharged. The detector draws 250 ma …so two 18650 cells will provide more than 10 hours of operation.

The tx timing can be variable but for the sensitivity as stated the Tx is on for 40 microseconds and the Rx period is 40 microseconds. The Tx power is a bit over 1 watt continuous so the little FET handle this no problem.

moodz.

[Mr.Jaick]

Are you using a 5v linear reg as your TX regulator? I can't see any robust switcher on this board.
so yes the system requires constant current at TX turn-off, but does it oscillate or react funny if the peak current slowly decreases(also what if a large metal target is close to the coil and changes the current)?
like for example we set the optimal ZPD timing for a peak tx current of 4A, then after a few hours the current reduces to 3A peak(same PW) but we haven't changed the timing yet;
now obviously the optimal timing for a 3A peak is different than 4A peak so some loss of performance there is inevitable(if we don't do anything about it); but does it produce any type of noise or undesired effects? in theory, it shouldn't go bad for reduced energy but for increased coil energy it would definitely go wild if it's set to its critical point.

the other point is that it dissipates the coil energy which makes the damping FET hot( as hot as a damping resistor? ), originally I thought it would make the coil dissipate the power within itself (dynamic push-pull).
So with this being said, the damping system is not suitable for any type of serious power, because as the TX power goes high you need to select a heat-sinkable power MOSFET as the damping device and those MOSFETs have high Coss and defeat the sole purpose of the idea(enlighten me if I'm wrong).

ML also uses a SOT-223 package for damping(ADC -> DAC -> FET) in gpx5k, and that detector doesn't use a low-power TX; dissipating even 2W power as heat with that package will easily roast it, especially in the long run, so that's not how it works.

[moodz]

Are you using a 5v linear reg as your TX regulator? I can't see any robust switcher on this board.
For lowest noise a linear 5 volt regulator is used.
so yes the system requires constant current at TX turn-off,
A constant current at turn off is not required.
but does it oscillate or react funny if the peak current slowly decreases(also what if a large metal target is close to the coil and changes the current)?
A large metal object causes a slight decrease in inductance of the coil but there is no oscillations or large changes in current.
like for example we set the optimal ZPD timing for a peak tx current of 4A, then after a few hours the current reduces to 3A peak(same PW) but we haven't changed the timing yet;
now obviously the optimal timing for a 3A peak is different than 4A peak so some loss of performance there is inevitable(if we don't do anything about it); but does it produce any type of noise or undesired effects? in theory, it shouldn't go bad for reduced energy but for increased coil energy it would definitely go wild if it's set to its critical point.
A few hours ?? not sure what you mean here ... the TX voltage is regulated .. so the coil current will not change in a few hours.
The circuit uses feedback loops that can keep the optimal operating point stable. ( which is the purpose of feedback loops )

the other point is that it dissipates the coil energy which makes the damping FET hot( as hot as a damping resistor? ), originally I thought it would make the coil dissipate the power within itself (dynamic push-pull).
So with this being said, the damping system is not suitable for any type of serious power, because as the TX power goes high you need to select a heat-sinkable power MOSFET as the damping device and those MOSFETs have high Coss and defeat the sole purpose of the idea(enlighten me if I'm wrong).
You dont need high power to make a sensitive detector ... however if you are familiar with SIC mosfets its possible to obtain very low Coss and 100 watts of power dissapation.

ML also uses a SOT-223 package for damping(ADC -> DAC -> FET) in gpx5k, and that detector doesn't use a low-power TX; dissipating even 2W power as heat with that package will easily roast it, especially in the long run, so that's not how it works.

ML uses these small FETS as blocking switches they are not damping any power. ( As far as I know ).

moodz.