Variable Coil Current Control

[Carl-NC]

Here is what I meant... normally we run R3=0 and the current is limited by the supply voltage divided by all the series resistances. With R3 (called a degeneration resistor) Q1 acts like a voltage-controlled current source. If VCTRL comes from a regulated source then the current does not vary as the battery drains.

Furthermore, you can 'shape' VCTRL during turn-on to get different effects. Bring it up with a pulse, ramp, or exponential. You might see some interesting effects. Mind the heat, though.

- Carl

[o.mag]

hi carl,
i understand, I=(vctrl-vbe)/R3
you are right Q1 may be hot

olivier

[Tinkerer]

Here is what I meant... normally we run R3=0 and the current is limited by the supply voltage divided by all the series resistances. With R3 (called a degeneration resistor) Q1 acts like a voltage-controlled current source. If VCTRL comes from a regulated source then the current does not vary as the battery drains.

Furthermore, you can 'shape' VCTRL during turn-on to get different effects. Bring it up with a pulse, ramp, or exponential. You might see some interesting effects. Mind the heat, though.

- Carl

Thanks Carl,

I will try that with some TO 3 and big heat sinks.

Tinkerer

[Max]

Here is what I meant... normally we run R3=0 and the current is limited by the supply voltage divided by all the series resistances. With R3 (called a degeneration resistor) Q1 acts like a voltage-controlled current source. If VCTRL comes from a regulated source then the current does not vary as the battery drains.

Furthermore, you can 'shape' VCTRL during turn-on to get different effects. Bring it up with a pulse, ramp, or exponential. You might see some interesting effects. Mind the heat, though.

- Carl

Hi,
I see... so it's a different thing than what I posted. Well, I think this way it's interesting cause shaping the tx phase (charging process) could have interesting effects during the charge phase.
That is if we are looking at e.g. side effects during the charging phase (like the perturbances due to a near target during that phase).

But my question is... what are the effects we look for in rx phase ?

I mean usually we are looking for the eddy currents effects (opposite magnetic field due to the conductivity of metallic target) that are due to the L*di/dt phase , so we look just for effects due to the abrupt switch-off no matter how the current peak was reached, if e.g. by exponential charge or ramp-like increase, what's important is the last value of current and relative fall time (the di/dt slope).

If we are interested just in effects that are generated by the switch-off why care about how the charging process happens ?

Or maybe I miss something along the way ?

Kind regards,
Max

[Max]

hi carl,
i understand, I=(vctrl-vbe)/R3
you are right Q1 may be hot

olivier

Hi,
that's for sure... if the difference is high the transistor will get a lot of heat to sink...

But infinite gain ?

That's pretty easy calculation (if we don't consider minor issues) : suppose you have 3A as current I=3 , then you need Vctrl = 13.5V
so... 3=(13.5-0.6)/4.3=3 ... ok...

that means also you have Vr3= 4.3*3=12.9V Vbe=0.6V then the current flowing in R1 will be Ir1= (13.5-0.6-12.9)/100 = 0

No... it's a small current of some milliamps : bc337 has dc current gain typ from 160 to 320... then can say we have something 9.4mA to 18.75mA depending on transistor... but nevermind!

Suppose the gain is 300... we have 10mA Ib , and Vr1=1V (much more realistic) means Vctrl will be instead 14.5V but change few I think... what's important is that you get 3A from a BC337!

The BC337 is a bad choice... can give you 500mA at limit...(some are rated 800mA in datasheet but don't trust them, my advice) then melt down! with big smoke!

No good there...I would use , instead, some array of good old 2N3055 with sized current driver required... or will use (much better) a mosfet as electronic rheostat (much less heat there).

The transistors aren't that good for current regulation in the several amp range... unless you use dedicated ones or arrays of them with appropriate fault safe network (cause they will never share exactly the same current... and one can burn easy with a cascade effect).

PS: typically VCEsat is in the 0.7V range... means you have about 0.7V*3A= 2.1W over the juntion! with 625mW absolute maximum rating... so like said above you are at 3 times the power it can run without damages! BIG SMOKE!

Kind regards,
Max

[Tinkerer]

Tinkerer,

I wonder what would happen if you switched low value resistors in series with the ground side of the TX. I.E. pulse1 = 0 Ohm, pulse 2 = 0.25 ohm, pulse 3 =0.5 ohm etc. You could switch the resistors after your rx samples, but before the next Tx pulse. This would give a stepped TC coil that could be digitally controlled. You would need to use non-inductive higher watt resistors. I think there has been some work on dynamically changing the damping resistors but not the TX series resistance?

Regards Mark

Hi Mark,

thanks for the feedback.
I have been considering something similar, but with different implementation.
Add a second, low R-ON Mosfet in series between the TX Mosfet and the coil. Use this Mosfet as a variable resistor and control it's gate with an 8 bit digital resistor that is controlled by the MCU.
this would make it possible to change the TC of the coil, make it very variable and therefore make it possible to tune the response to select specific targets.
However, this would need a significant amount of DSP.
I am hoping somebody with DSP knowledge joins in.

Tinkerer

[Tinkerer]

Hi,
I see... so it's a different thing than what I posted. Well, I think this way it's interesting cause shaping the tx phase (charging process) could have interesting effects during the charge phase.
That is if we are looking at e.g. side effects during the charging phase (like the perturbances due to a near target during that phase).

But my question is... what are the effects we look for in rx phase ?

I mean usually we are looking for the eddy currents effects (opposite magnetic field due to the conductivity of metallic target) that are due to the L*di/dt phase , so we look just for effects due to the abrupt switch-off no matter how the current peak was reached, if e.g. by exponential charge or ramp-like increase, what's important is the last value of current and relative fall time (the di/dt slope).

If we are interested just in effects that are generated by the switch-off why care about how the charging process happens ?

Or maybe I miss something along the way ?

Kind regards,
Max

Hi Max,

I much appreciate you help. But, yes, you are missing something.
Please look at the pictures at post: http://www.geotech1.com/forums/showthread.php?t=15441
These pictures are taken during TX, that is the excitation of the target. There is a wealth of information available during TX, that is lost after the Flyback has decayed.
Manipulating the TX ramp gives even more information.

I compare the traditional PI sampling to a black and white picture.
TX ON sampling to a Grey scale picture.
And sampling during the manipulated TX ON ramp, to a color picture.

I do not believe that discrimination is possible with traditional PI.
However, I am convinced that discrimination is possible with the enhanced information that the TX ON sampling provides.

Tinkerer

[Max]

Hi Max,

I much appreciate you help. But, yes, you are missing something.
Please look at the pictures at post: http://www.geotech1.com/forums/showthread.php?t=15441
These pictures are taken during TX, that is the excitation of the target. There is a wealth of information available during TX, that is lost after the Flyback has decayed.
Manipulating the TX ramp gives even more information.

I compare the traditional PI sampling to a black and white picture.
TX ON sampling to a Grey scale picture.
And sampling during the manipulated TX ON ramp, to a color picture.

I do not believe that discrimination is possible with traditional PI.
However, I am convinced that discrimination is possible with the enhanced information that the TX ON sampling provides.

Tinkerer

Hi,
now I understand, I was trying to explain myself why someone need that approach if just peak-current matter!

Now it's clear... you're looking for small variations in signal due to charging process , when current in the coil rise. I understand that the informations you could get that way could be useful cause we know traditional approach to disc PI has very big drawbacks.

Sure, I would say, that approach will add informations on targets.

Seems very good idea...

Kind regards,
Max

[Tinkerer]

Hi,
that's for sure... if the difference is high the transistor will get a lot of heat to sink...

But infinite gain ?

That's pretty easy calculation (if we don't consider minor issues) : suppose you have 3A as current I=3 , then you need Vctrl = 13.5V
so... 3=(13.5-0.6)/4.3=3 ... ok...

that means also you have Vr3= 4.3*3=12.9V Vbe=0.6V then the current flowing in R1 will be Ir1= (13.5-0.6-12.9)/100 = 0

No... it's a small current of some milliamps : bc337 has dc current gain typ from 160 to 320... then can say we have something 9.4mA to 18.75mA depending on transistor... but nevermind!

Suppose the gain is 300... we have 10mA Ib , and Vr1=1V (much more realistic) means Vctrl will be instead 14.5V but change few I think... what's important is that you get 3A from a BC337!

The BC337 is a bad choice... can give you 500mA at limit...(some are rated 800mA in datasheet but don't trust them, my advice) then melt down! with big smoke!

No good there...I would use , instead, some array of good old 2N3055 with sized current driver required... or will use (much better) a mosfet as electronic rheostat (much less heat there).

The transistors aren't that good for current regulation in the several amp range... unless you use dedicated ones or arrays of them with appropriate fault safe network (cause they will never share exactly the same current... and one can burn easy with a cascade effect).

PS: typically VCEsat is in the 0.7V range... means you have about 0.7V*3A= 2.1W over the juntion! with 625mW absolute maximum rating... so like said above you are at 3 times the power it can run without damages! BIG SMOKE!

Kind regards,
Max

I have a 2N3550 and big heat sinks at hand so I want to try that. I am not good at calculating the base drive current needed for bipolar transistors.
Could you help me with that?
How much current do I need to drive the 2N3550?

I have the same problem with the Darlington transistors of my capacitor charge pump of my power supply. I feel they should be able to supply more current.
Your help is much appreciated.

All the best

Tinkerer

[Max]

I have a 2N3550 and big heat sinks at hand so I want to try that. I am not good at calculating the base drive current needed for bipolar transistors.
Could you help me with that?
How much current do I need to drive the 2N3550?

I have the same problem with the Darlington transistors of my capacitor charge pump of my power supply. I feel they should be able to supply more current.
Your help is much appreciated.

All the best

Tinkerer

Hi,
the current required depends on dc gain of pass transistor, in our case it's 2N3055, that has typ gain of 70. If you use many 2N3055 to share the load current you'll must polarize each base accordingly and then figure out the drive current for each of them.

E.g. suppose you need 3A at output and use 3 * 2N3055 to carry each 1A, then you know each transistor have IC=1A and gain A=70 so you'll need IB=IC/A=1/70=0.014 A or 14mA

If you need something specific I can make calculations... but I think you could like the schematic below....

Uses LM317 as driver for transistor bases... you have both can use LM317 as regulator and 2N3055 as pass... now you see 4 of 2N3055 but you need just a couple or 3 I think... you can scale it as you need!

Component bill is at : http://www.elecfree.com/circuit/powe...17-and-2n3055/

Kind regards,
Max

[Altra]

Tinkerer,

I thought of using a fet also as variable resisitor. But I don't have enough knowledge about hexfets to understand the dynamics when it comes to switching inductive loads. How would the RDSon react under a changing load? We know the results if the fet is fully on or off, but what about when biased to result in RDSon of 2ohms? Does the RDS remain the same as the DS current and voltage change? Going in a full circle we are back to plain old current limiting. I had another idea tap your tx coil and switch between a high and low TC?

I think Moodz mentioned he uses DSP processors maybe you could get him to chime in.

Mark

[Tinkerer]

Thanks for the circuit, Max

All the best

Tinkerer

[Tinkerer]

Tinkerer,

I thought of using a fet also as variable resisitor. But I don't have enough knowledge about hexfets to understand the dynamics when it comes to switching inductive loads. How would the RDSon react under a changing load? We know the results if the fet is fully on or off, but what about when biased to result in RDSon of 2ohms? Does the RDS remain the same as the DS current and voltage change? Going in a full circle we are back to plain old current limiting. I had another idea tap your tx coil and switch between a high and low TC?

I think Moodz mentioned he uses DSP processors maybe you could get him to chime in.

Mark

Hi Mark,

when I use a manual pot to control the Mosfet in its pinch off region, I do not notice linearity problems. It may be different when using a digital pot.
I suspect that there are certain Mosfets that are better suited for that purpose than others.
The good thing is that the Mosfet are built to withstand heat and that a heat sink can be added.

Tinkerer

[Carl-NC]

The BC337 is a bad choice...

I clipped this circuit from a PI probe I designed. It is running only 50mA of 'on' current. For a high-power detector, you obviously need something like a 3055.

With heat, keep in mind that we are dealing with time-averaged power and a short duty cycle. Still, you could easily see a watt both on Q1 and R3.

- Carl

[Tinkerer]

I clipped this circuit from a PI probe I designed. It is running only 50mA of 'on' current. For a high-power detector, you obviously need something like a 3055.

With heat, keep in mind that we are dealing with time-averaged power and a short duty cycle. Still, you could easily see a watt both on Q1 and R3.

- Carl

I understood that. I am going to reduce the PPS to 600 and limit the coil current to 3 Amps with a maximum of 100uS pulse, so I will come out with under 2 Watt.

Thanks

Tinkerer