Variable Coil Current Control

[Tinkerer]

With traditional PI, the coil current usually varies a little depending on the state of charge of the batteries.
The change of battery voltage has a little influence on the depth of reach.
The lower coil current due to the lower battery voltage, allows for a faster Flyback decay and earlier sampling.

Now, when we use an Induction Balanced coil (IB), things are very different. The varying TX voltage and coil current have a strong influence on the balance of the coil assembly.
Therefore the TX voltage and the Coil current need to be controlled.
Once we have to use voltage and current control, we might just as well get the best out of it and use these features to obtain more information about the target.

Let's use this thread to discuss how I do this and other possible methods how it could be done.

Tinkerer

[Max]

With traditional PI, the coil current usually varies a little depending on the state of charge of the batteries.
The change of battery voltage has a little influence on the depth of reach.
The lower coil current due to the lower battery voltage, allows for a faster Flyback decay and earlier sampling.

Now, when we use an Induction Balanced coil (IB), things are very different. The varying TX voltage and coil current have a strong influence on the balance of the coil assembly.
Therefore the TX voltage and the Coil current need to be controlled.
Once we have to use voltage and current control, we might just as well get the best out of it and use these features to obtain more information about the target.

Let's use this thread to discuss how I do this and other possible methods how it could be done.

Tinkerer

Hi,
TB electronics engineer (don't remember his name now... maybe thomas ?) made that some years ago; it's not completely new concept.

Referred as pulse to pulse current stability (or something like that) it's usually implemented by a direct sampling of current flowing in the coil: a small resistor of usually 0.1 ohm used for converting the current into a voltage and then the voltage is read by a comparator or something like that who will feedback the tx pulse generator (also an MCU if necessary... but in old stuff... old timers like 4538 etc) to terminate when a particular level of voltage (thus current flowing in the coil) is matched.
That's my preferred method also... I implemented some years ago with success.

In some Eric Foster and TB's guy discussion I remember they talk about methods for mantaining e.g. the same energy between pulses... that's about the same thing (the energy stored in the coil , if we don't consider losses, is totally stored in inductor as E=1/2(L*I^2) ... so if L doesn't vary (and that's acceptable in our model I think) the energy stability is a direct consequence of current stability between pulses...
The method proposed by the TB's guy was storing in a large capacitor the energy required then transferring it abruptly by the commutation of an SCR diode... or something... like happens in DC-DC converters... I think this method appears in a patent of TB electronics... but now I don't remember exactly.

Kind regards,
Max

[Tinkerer]

Hi,
TB electronics engineer (don't remember his name now... maybe thomas ?) made that some years ago; it's not completely new concept.

Referred as pulse to pulse current stability (or something like that) it's usually implemented by a direct sampling of current flowing in the coil: a small resistor of usually 0.1 ohm used for converting the current into a voltage and then the voltage is read by a comparator or something like that who will feedback the tx pulse generator (also an MCU if necessary... but in old stuff... old timers like 4538 etc) to terminate when a particular level of voltage (thus current flowing in the coil) is matched.
That's my preferred method also... I implemented some years ago with success.

In some Eric Foster and TB's guy discussion I remember they talk about methods for mantaining e.g. the same energy between pulses... that's about the same thing (the energy stored in the coil , if we don't consider losses, is totally stored in inductor as E=1/2(L*I^2) ... so if L doesn't vary (and that's acceptable in our model I think) the energy stability is a direct consequence of current stability between pulses...
The method proposed by the TB's guy was storing in a large capacitor the energy required then transferring it abruptly by the commutation of an SCR diode... or something... like happens in DC-DC converters... I think this method appears in a patent of TB electronics... but now I don't remember exactly.

Kind regards,
Max

Thank you for the feedback. The information is very interesting.
This type of coil current control will indeed compensate for diminishing battery power and some variation in coil parameters. this control is designed to stabilize the coil current.

I am talking about a kind of control TO CHANGE THE CURRENT, to use variable coil current to gather more information about the target.
If you look at the pictures in my post NEW TECH PHOTOS, you will notice "THE PIVOT" This gives an indication about the TC of the target and for FE.
It is the same thing as the phase shift with VLF.
It is produced by a certain relationship between the Tau of the TX pulse and the TC of the target.
Now, controlling the di/dt of the coil current also let's us control this relationship.

Tinkerer

[Max]

Thank you for the feedback. The information is very interesting.
This type of coil current control will indeed compensate for diminishing battery power and some variation in coil parameters. this control is designed to stabilize the coil current.

I am talking about a kind of control TO CHANGE THE CURRENT, to use variable coil current to gather more information about the target.
If you look at the pictures in my post NEW TECH PHOTOS, you will notice "THE PIVOT" This gives an indication about the TC of the target and for FE.
It is the same thing as the phase shift with VLF.
It is produced by a certain relationship between the Tau of the TX pulse and the TC of the target.
Now, controlling the di/dt of the coil current also let's us control this relationship.

Tinkerer

Hi,
I see... but it's not big issue.

If you use a flexible approach, e.g. an MCU timebase you can program the MCU for otherthings... suppose you have an DAC onboard then you could set the DAC output to feed e.g. 2 or 3 comparators reference inputs.

That way you could replicate the approach of above (single comparator I wrote above) with 2 or 3 comparators instead... and giving them each a different voltage reference. Do you understand what I mean ?

The MCU will issue e.g. 3 levels, as example, 1A, 2A and 3 amperes of current in 3 different tx pulses just setting the appropriate value at comparators, then the comparators output will be sent to an I/O line of MCU to signal where the desired current level is matched to terminate the tx pulses.

In reality you'll probably could use just one comparator and then will set each time the threshold... but I exposed like above cause if wanna save time... you performs DAC setting just one time for a cycle of e.g. 3 pulses (using 3 different comparators)... to save time.

So you made trains of 3 tx pulses, after each you do RX function and calculations: that approach will require after any tx pulse you made a separate RX cycle...

Then after collecting 3 rx cycles and data some correlation will be made to attempt disc!

It's not only possible but even easy to do with right approach.

Kind regards,
Max

[Aziz]

Hi Tinkerer,

here is an example for a coil voltage controller (see below). On the right side, the coil voltage at C5 is monitored with a comparator (with reference voltage band gap diode LM336-2.5). Once a defined reference voltage is arrived, a hold signal will be generated and the charging of the C5 is disabled. The stored energy (voltage) in C5 is hold and ready to fire the transmit coil.
The VP3 controls in the cycle period, when C5 is enabled again for charging. The speed of charging can be limited by R12. On non-critical periods, you can enable the charging to minimize noise. So VP3 can also be seen as a start coil voltage control (falling edge triggered). You can trigger as often as you want. Once the coil voltage level is arrived, it will switch off the charging of C5.

It shows the basic principle. You have to adapt into your requirements.


Aziz

[Carl-NC]

Why not just use a degenerated bipolar transistor?

[moodz]

With traditional PI, the coil current usually varies a little depending on the state of charge of the batteries.
The change of battery voltage has a little influence on the depth of reach.
The lower coil current due to the lower battery voltage, allows for a faster Flyback decay and earlier sampling.

Now, when we use an Induction Balanced coil (IB), things are very different. The varying TX voltage and coil current have a strong influence on the balance of the coil assembly.
Therefore the TX voltage and the Coil current need to be controlled.
Once we have to use voltage and current control, we might just as well get the best out of it and use these features to obtain more information about the target.

Let's use this thread to discuss how I do this and other possible methods how it could be done.

Tinkerer

hmmm nice idea ... but I think it is already covered in US 7474102 by BC himself. He uses switches .. you are proposing variable ... the switched schema could easily be regarded as a 'variable' methodology.
BTW not trying to squash the discussion ... came across this patent whilst researching prior art for a patent app I am applying for.

moodz.

[Altra]

Hi Carl,

Could you explain what "degenerated bipolar transisitor" means?

Does it refer to a semiconductor process for making the transistor?
Or is it a circuit configuration such as current sink/source etc?

Thanks
Mark

Quote "Why not just use a degenerated bipolar transistor? "

[Carl-NC]

A bipolar transistor with a resistor in the emitter, basically a switched current source. Switch the base off a regulated voltage, and you get a current that is constant vs battery.

[Max]

A bipolar transistor with a resistor in the emitter, basically a switched current source. Switch the base off a regulated voltage, and you get a current that is constant vs battery.

Hi Carl,
I think you maybe misunderstund his needs...(or maybe I do not understand what he needs ???)
I think you mean something like this in the picture. The bipolar transistor will act as a variable resistor to mantain a constant current in the output end of circuit... collector part I mean.

But I think, though useful in many cases, this simple circuit is not what he needs. That's cause the inductor (coil) will charge with a current that rise at exponential rate not that is constant over time.

The use of this approach will make the transistor act like a variable resistance (kinda of solid state automatic rheostat) in series with coil...having a low resistance value at start of tx pulse... then higher resistance when the current in the coil will rise to higher levels.

But this will make the charge of inductor a bit "stranger" than usual.

In usual RL circuit we have the I=(V0 /R)* (1-e^(-t*(R/L)))

Now... if R stays fixed as in a normal RL circuit the inductor current rise from 0 following the above equation (exponentially) and thus having an asymptotic increase to infinite after about 5 times the ratio R/L.

Now the problem is... when we made usual PIs we have an active device, usually a mosfet, that switch-on and so the current flowing will charge the inductor with active device exposing an apparent resistance (e.g. in the IRF740 the Rds on is about 0.5ohm) that stays about "fixed" during the charge... so things go like described by equation... is nothing unlike that charging the coil by a line, a switch and a fixed resistor... just we make by mosfet that's nothing more than an electronic switch for us.
In the case of transistor constant current source, instead, we have something that will change resistance to mantain the current stable, it's a completely different thing.

The approach is wrong... cause he needs just to control the peak-current value just before switchoff, that's why I proposed him the use of e.g. a sensing resistor + comparators for that: the charge will be exponential as usual but he could control and set a preferred value of peak-current before the switch-off take place for each of the tx pulses.

Kind regards,
Max

[Tinkerer]

Thanks for all the feedback.

Let me try to explain what I meant by "voltage and current control to obtain more information about the target".

At present I am using two different transients to excite the targets. The one transient has a di/dt that is about 20 times as the other one.
The "slow Tau" is good for long TC targets, the "fast Tau is better for short TC targets. The pictures I am showing in the thread below, are the responses to the transient with the high rate of change. After the end of this series, I will show the responses of the same targets with the slow transient.

http://www.geotech1.com/forums/showthread.php?t=15441

It is a bit like having a VLF that has simultaneous low frequency and high frequency.
When use different voltages for the TX, say 8V, 10V, 12V, 15V I would have TX pulses of very different rate of change. At present I use a 317 voltage regulator for that, but this restricts me to relatively low power.
There is a relationship between several parameters of the target, like the surface area, conductivity, mass etc., and the rate of change of the transient.
Therefore, the capability to use Tx pulses with different di/dt make it possible to extract much more information about the targets.

Tinkerer

[Max]

Thanks for all the feedback.

Let me try to explain what I meant by "voltage and current control to obtain more information about the target".

At present I am using two different transients to excite the targets. The one transient has a di/dt that is about 20 times as the other one.
The "slow Tau" is good for long TC targets, the "fast Tau is better for short TC targets. The pictures I am showing in the thread below, are the responses to the transient with the high rate of change. After the end of this series, I will show the responses of the same targets with the slow transient.

http://www.geotech1.com/forums/showthread.php?t=15441

It is a bit like having a VLF that has simultaneous low frequency and high frequency.
When use different voltages for the TX, say 8V, 10V, 12V, 15V I would have TX pulses of very different rate of change. At present I use a 317 voltage regulator for that, but this restricts me to relatively low power.
There is a relationship between several parameters of the target, like the surface area, conductivity, mass etc., and the rate of change of the transient.
Therefore, the capability to use Tx pulses with different di/dt make it possible to extract much more information about the targets.

Tinkerer

Hi,
LM317T can deliver max 1A, LM317K can go (hopefully) for 3A.

I think you need something more... cause even staying in the limits of above you'll get them run VERY hot... in a few time unless you use a large heat-sink stuff.

I suggest you'll dig some LM338K... it's a powerful linear regulator for 5A range, TO-3 package.

I suggest if find more than one... to put a couple in parallel and thus will stay safe in regulation till 10A.

The big advantage is that if you use the right voltage input level you can run them COLD at several amps, and that with small heatsinks.

You could also make a quick and dirty regulator using parallels of e.g. (old nanny) 2N3055 things...bipolar , 75W stuff each...TO-3. But in this case you'll need small resistors... usually in the 0.3ohm range to prevent damages to the transistors.

Another solution is an array of LM317T (say you need MUCH of them)... teh advantage is that are about cheap parts... can find anywhere still and if you burn them , well... no regrets I think ! Or can use pass transistor... why not ?

There are other solutions like SMPS ics with e.g. low frequency transformer or inductor.... but they add bedlam of noise to the circuits they will supply.

Kind regards,
Max

[Tinkerer]

Hi Carl,
I think you maybe misunderstund his needs...(or maybe I do not understand what he needs ???)
I think you mean something like this in the picture. The bipolar transistor will act as a variable resistor to mantain a constant current in the output end of circuit... collector part I mean.

But I think, though useful in many cases, this simple circuit is not what he needs. That's cause the inductor (coil) will charge with a current that rise at exponential rate not that is constant over time.

The use of this approach will make the transistor act like a variable resistance (kinda of solid state automatic rheostat) in series with coil...having a low resistance value at start of tx pulse... then higher resistance when the current in the coil will rise to higher levels.

But this will make the charge of inductor a bit "stranger" than usual.

In usual RL circuit we have the I=(V0 /R)* (1-e^(-t*(R/L)))

Now... if R stays fixed as in a normal RL circuit the inductor current rise from 0 following the above equation (exponentially) and thus having an asymptotic increase to infinite after about 5 times the ratio R/L.

Now the problem is... when we made usual PIs we have an active device, usually a mosfet, that switch-on and so the current flowing will charge the inductor with active device exposing an apparent resistance (e.g. in the IRF740 the Rds on is about 0.5ohm) that stays about "fixed" during the charge... so things go like described by equation... is nothing unlike that charging the coil by a line, a switch and a fixed resistor... just we make by mosfet that's nothing more than an electronic switch for us.
In the case of transistor constant current source, instead, we have something that will change resistance to mantain the current stable, it's a completely different thing.

The approach is wrong... cause he needs just to control the peak-current value just before switchoff, that's why I proposed him the use of e.g. a sensing resistor + comparators for that: the charge will be exponential as usual but he could control and set a preferred value of peak-current before the switch-off take place for each of the tx pulses.

Kind regards,
Max

It might be an interesting experiment to use a constant current ramp for the TX pulse. I wonder how the targets will respond.

A still different way to compare responses, would be to change the TC of the coil in different steps. It also would change the di/dt, but it seems that the best type of response is for a tx pulse of around 1 coil TC.

Tinkerer

[Altra]

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

[Tinkerer]

Hi,
LM317T can deliver max 1A, LM317K can go (hopefully) for 3A.

I think you need something more... cause even staying in the limits of above you'll get them run VERY hot... in a few time unless you use a large heat-sink stuff.

I suggest you'll dig some LM338K... it's a powerful linear regulator for 5A range, TO-3 package.

I suggest if find more than one... to put a couple in parallel and thus will stay safe in regulation till 10A.

The big advantage is that if you use the right voltage input level you can run them COLD at several amps, and that with small heatsinks.

You could also make a quick and dirty regulator using parallels of e.g. (old nanny) 2N3055 things...bipolar , 75W stuff each...TO-3. But in this case you'll need small resistors... usually in the 0.3ohm range to prevent damages to the transistors.

Another solution is an array of LM317T (say you need MUCH of them)... teh advantage is that are about cheap parts... can find anywhere still and if you burn them , well... no regrets I think ! Or can use pass transistor... why not ?

There are other solutions like SMPS ics with e.g. low frequency transformer or inductor.... but they add bedlam of noise to the circuits they will supply.

Kind regards,
Max

Many thanks Max,

here is my present power supply. I can use up to maximum 2A coil current with 900 PPS, with a heat sink.
It needs improvement.
I suspect the biasing of the Darlington transistors is not very good.
The frequency of the sync pulse is about 10k which is the limit of the ICL7662 (not ICL7660)
Withe the sync pulse I have little noise during the sampling times.
The sampling times not to be confused with the sampling window.

all the best and thanks for the help

Tinkerer