Do I have to measure the phase of the current?

[[email protected]]

Hello,

I am working on a current regulating circuit where the phase
of my current is important.

To be more precise:
I have the following circuit:


- Vref
|
\ OPA227 +--------------+
R1 / | |
\ |\ | |
| | \ | |
+---------|- \ |++ |
| | >----|| 2SK3680-01 ---
\ +-----|+ / |++ - Battery
R2 / | | / | ---
\ | |/ | -
| | | |
| | (GND) | |
+--------------------+ |
| | |
| \ |
| / |
| R_Shunt \ |
| / |
| | |
+----------------+--------------+


My Vref voltage will be an AC signal up to 10 KHz with a
certain DC offset. As I have to know (but not to control)
the phase of my current at 3° precision, I would like to ask
you:

Do I have to measure the phase (which would impose another
circuitry) or do I know the phase with the wanted precision
as I am imposing it?

I fear that not only the Opamp, but also the Mosfet will
introduce a phase shift higher than 3° compared with the
Vref voltage.

Some important data: The Opamp OPA227 has a Unity Gainbandwith
of 8 MHz and a Slew Rate of about 2.3 V/us. The Mosfet 2SL3680-01
is a power Mosfet, but I don't know too much about what criterias
might be important. It seems (if I interpret the datasheet correctly)
that the Cgs is not more than 10 nF.

The Gain of the Opamp never exceeds 10. The AC-signal is supposed to
be never higher than 2 Vpp.

I would be glad if you could help me a little bit!

Yours,

Bernhard

 

[Winfield Hill]

[email protected]-nancy.fr wrote...

I am working on a current regulating circuit where the phase
of my current is important. To be more precise:
I have the following circuit:

- Vref
|
\ OPA227 +--------------+
R1 / | |
\ |\ | |
| | \ | |
+---------|- \ |++ |
| | >----|| 2SK3680-01 ---
\ +-----|+ / |++ - Battery
R2 / | | / | ---
\ | |/ | -
| | (GND) | |
+--------------------+ |
| | |
| / |
| R_Shunt \ |
| / |
| | |
+----------------+--------------+

My Vref voltage will be an AC signal up to 10 KHz with a
certain DC offset. As I have to know (but not to control)
the phase of my current at 3° precision, I would like to ask
you:

Do I have to measure the phase (which would impose another
circuitry) or do I know the phase with the wanted precision
as I am imposing it?

I fear that not only the Opamp, but also the Mosfet will
introduce a phase shift higher than 3° compared with the
Vref voltage.

Some important data: The Opamp OPA227 has a Unity Gainbandwith
of 8 MHz and a Slew Rate of about 2.3 V/us. The Mosfet 2SL3680-01
is a power Mosfet, but I don't know too much about what criterias
might be important. It seems (if I interpret the datasheet correctly)
that the Cgs is not more than 10 nF.

The Gain of the Opamp never exceeds 10. The AC-signal is supposed to
be never higher than 2 Vpp.

As you noticed, the MOSFET's high gate capacitance wil be trouble
for the opamp. I've edited your circuit to show the standard way
to deal with this problem, isolating the opamp from the FET with a
series resistor, and providing direct high-frequency feedback.

- Vref
| OPA227
\ Cf +--------------+
R1 / ,---||----, | |
\ | _ | | |
| | | \ | | |
+-----+---|- \ | Rg |-+ |
| | >-+--/\/\--|| 2SK3680-01 ---
\ +-----|+ / |-+ - Battery
R2 / | |_/ | ---
\ | | -
| | (GND) | |
+---------------------------+ |
| | |
| / |
| R_Shunt \ |
| / |
| | |
+-----------------------+--------------+

Although the Rg Cf components stabilize the opamp feedback loop,
they limit the ability of the circuit to work accurately at high
frequencies. One solution is to use an opamp that has a lower
open-loop Zout, or even better, use a power buffer between the
opamp and Rg, allowing you to reduce the value of both Rg and Cf.
E.g., http://focus.ti.com/docs/prod/folders/print/buf634.html

In this manner you should be able to achieve < 3 degrees at 10kHz.

 

[John Woodgate]

I read in sci.electronics.design that
"[email protected]-nancy.fr" <[email protected]>

As I have to know (but not to control)
the phase of my current at 3° precision, I would like to ask
you:

This is the phase of the DC into the battery referenced to the phase of
the moon?

 

[Winfield Hill]

Winfield Hill wrote...

[email protected]-nancy.fr wrote...

I am working on a current regulating circuit where the phase
of my current is important. To be more precise:
I have the following circuit: [ snip ]

My Vref voltage will be an AC signal up to 10 KHz with a
certain DC offset. I have to know (but not to control)
the phase of my current at 3° precision [ snip ]

Some important data: The Opamp OPA227 has a Unity Gainbandwith
of 8 MHz and a Slew Rate of about 2.3 V/us. The Mosfet 2SL3680-01
is a power Mosfet, but I don't know too much about what criterias
might be important. It seems (if I interpret the datasheet correctly)
that the Cgs is not more than 10 nF.

The Gain of the Opamp never exceeds 10. The AC-signal is supposed to
be never higher than 2 Vpp.

As you noticed, the MOSFET's high gate capacitance wil be trouble
for the opamp. I've edited your circuit to show the standard way
to deal with this problem, isolating the opamp from the FET with a
series resistor, and providing direct high-frequency feedback.

[ snip modified drawing ]

Although the Rg Cf components stabilize the opamp feedback loop,
they limit the ability of the circuit to work accurately at high
frequencies. One solution is to use an opamp that has a lower
open-loop Zout, or even better, use a power buffer between the
opamp and Rg, allowing you to reduce the value of both Rg and Cf.
E.g., http://focus.ti.com/docs/prod/folders/print/buf634.html

In this manner you should be able to achieve < 3 degrees at 10kHz.

I just looked up the specs on your elegant Fuji 2sk3680 600W FET
(595 in stock at Allied, $9.32 each). It's a 500V part, and you
should be aware that all high-voltage FETs are prone to go into RF
oscillation (10 - 50MHz) when used in linear, rather than switching
modes of operation (note, you haven't shown your load, which could
have some effect on this scene). Some remediation may be needed to
prevent RF oscillation, some kind of lossy RF element, such as a
resistor or ferrite bead in the FET's drain pathway, etc. Another
possibility is that you choose a low-voltage FET, if appropriate.

.. - Vref
.. | OPA227
.. \ Cf +--------------+
.. R1 / ,---||----, | |
.. \ | _ | O - RF loss |
.. | | | \ | | element |
.. +-----+---|- \ | Rg |-+ |
.. | | >-+--/\/\--|| 2SK3680-01 ---
.. \ +-----|+ / |-+ - Battery
.. R2 / | |_/ | ---
.. \ | | -
.. | | (GND) | |
.. +---------------------------+ |
.. | | |
.. | / |
.. | R_Shunt \ |
.. | / |
.. | | |
.. +-----------------------+--------------+

One other issue, be sure to use Kelvin connections to R_Shunt.

 

[John Woodgate]

I read in sci.electronics.design that Winfield Hill

In this manner you should be able to achieve < 3 degrees at 10kHz.

That suggest that you understand what the OP means by 'the phase of the
current'. Please explain; which current? Phase with respect to what?

 

[Bernhard Krämer]

Winfield Hill wrote...

[email protected]-nancy.fr wrote...

I am working on a current regulating circuit where the phase
of my current is important. To be more precise:
I have the following circuit: [ snip ]

My Vref voltage will be an AC signal up to 10 KHz with a
certain DC offset. I have to know (but not to control)
the phase of my current at 3° precision [ snip ]

Some important data: The Opamp OPA227 has a Unity Gainbandwith
of 8 MHz and a Slew Rate of about 2.3 V/us. The Mosfet 2SL3680-01
is a power Mosfet, but I don't know too much about what criterias
might be important. It seems (if I interpret the datasheet correctly)
that the Cgs is not more than 10 nF.

The Gain of the Opamp never exceeds 10. The AC-signal is supposed to
be never higher than 2 Vpp.

As you noticed, the MOSFET's high gate capacitance wil be trouble
for the opamp. I've edited your circuit to show the standard way
to deal with this problem, isolating the opamp from the FET with a
series resistor, and providing direct high-frequency feedback.

If I understand you modification correctly, then the Rg Cf components turn
the Opamp stage into some sort of low pass, preventing oscillations at high
frequencies due to the Gate-Source capacity of the MOSFET. If this is true,
I guess that a power buffer would allow reducing the Rg and Cf values
because the Gate-Source capacity could be charged faster which would reduce
the stage' sensibility to oscillate.

I just looked up the specs on your elegant Fuji 2sk3680 600W FET
(595 in stock at Allied, $9.32 each). It's a 500V part, and you
should be aware that all high-voltage FETs are prone to go into RF
oscillation (10 - 50MHz) when used in linear, rather than switching
modes of operation (note, you haven't shown your load, which could
have some effect on this scene).

How does this happen?

Some remediation may be needed to
prevent RF oscillation, some kind of lossy RF element, such as a
resistor or ferrite bead in the FET's drain pathway, etc. Another
possibility is that you choose a low-voltage FET, if appropriate.

. - Vref
. | OPA227
. \ Cf +--------------+
. R1 / ,---||----, | |
. \ | _ | O - RF loss |
. | | | \ | | element |
. +-----+---|- \ | Rg |-+ |
. | | >-+--/\/\--|| 2SK3680-01 ---
. \ +-----|+ / |-+ - Battery
. R2 / | |_/ | ---
. \ | | -
. | | (GND) | |
. +---------------------------+ |
. | | |
. | / |
. | R_Shunt \ |
. | / |
. | | |
. +-----------------------+--------------+

One other issue, be sure to use Kelvin connections to R_Shunt.

I have one more question: To prevent distortion of my AC-current signal,
shouldn't I integrate a resistance from Drain to Gate and an equal one from
Gate to Opamp-Output?

As it might help and/or might be interesting, I'll tell you the purpose of
my circuitry. Nearly nothing is very secret, so I can tell relatively free.

Another laboratory which is in collaboration with us, has just begun
research on fuel cells. Fuel cells might be the future in powering cars,
little plants, and more. In order to produce effective fuel cells,
optimizations still have to be made in nearly every part of the fuel cell.
One approach to learn more about the chemical reactions taking place in the
fuel cell is the so-called "electrochemical impedance spectroscopy".

That means, you measure the spectrum of the fuel cells' impedance, and by
fitting the impedance on existing models (see e.g. Randles Model), you
learn a lot about where you could optimize further.

I am going to measure the impedance of the fuel cell in the following way:
First, I conceive a variable charge able to set a DC current. You have to
know that the impedance of the fuel cell is non-linear: it is dependant of
the current that flows.

Then, I am going to superpose an AC-current on the DC-current. What will
happen? The voltage delivered by fuel cell will also oscillate at the same
frequency, and at an amplitude & phase dependent on its internal impedance
at this frequency. By sweeping through all the frequency range, you can
obtain the impedance spectrum of the fuel cell.

The current regulator you see in the schematics above is the heart of my
measurement circuit (Thanks, Mr. Hill, I took the circuit out of your
book . The same regulator will appear six times in parallel to stand all
the current which will pass in the most extreme case. With a few more
parts, a little lock-in amplifier comes available which will measure the
phase and amplitude of the relatively small current and voltage variations
(they have to be small due to the non-linearity of the impedance).

Meanwhile, I decided that I will measure amplitude and phase of both current
and voltage. First, and this is the reason why I initially posted this
message, I believed I could spare measuring the current phase and amplitude
as I am imposing their values. But now, I know it is much better to measure
it anyway.


Yours,

Bernhard