M
MooseFET
I agree. He may be happy with just a common base stage.
-
Categories
-
Platforms
-
Content
I agree. He may be happy with just a common base stage.
Thanks. actually trying to make a CT loaded with really low impedance
till 50MHz. The low impedance at DC is not requirement...I tried the
OP-CT..trying to improve it by having really small burden.....
till 50MHz. The low impedance at DC is not requirement...I tried the
OP-CT..trying to improve it by having really small burden.....
J
John Popelish
With a fast opamp, you might actually be able to load it
with a small negative impedance, to compensate for the
series resistance of the winding. But it probably would not
make a measurable difference from what you would get with a
perfect zero burden impedance.
The fast opamp I listed with a 50 ohm feedback resistor and
a 50 ohm output series resistor should produce a pretty low
burden, but produce an output equivalent to what you would
get with a perfect CT and a 50 ohm burden.
F
Fred Bartoli
John Popelish a écrit :
Negative resistance will give better LF response and plays no role at
50MHz. At this frequency the important thing is leakage inductance.
More precisely, from the burden voltage POV, the primary side referred
leakage inductance is what matters.
as you know, with CFB opamps you can't reduce the feedback impedance at
will. I didn't checked, but onsemi used 330R in their spec. sheet so it
must be somewhat optimal. And at unity gain, the curves show 4+dB of
peaking and reducing the FB resistor down to 50R seems hazardous.
Anyway, even at a low 1:10 ratio, just a tiny 1nH primary leakage
inductance will give 31R secondary transformed impedance, showing that
burden voltage won't be impacted at all by lowering the opamp input
impedance beyond a few 10s ohm (easily provided by a simple common base
stage).
Carefull primary side design is a must there.
Negative resistance will give better LF response and plays no role at
50MHz. At this frequency the important thing is leakage inductance.
More precisely, from the burden voltage POV, the primary side referred
leakage inductance is what matters.
as you know, with CFB opamps you can't reduce the feedback impedance at
will. I didn't checked, but onsemi used 330R in their spec. sheet so it
must be somewhat optimal. And at unity gain, the curves show 4+dB of
peaking and reducing the FB resistor down to 50R seems hazardous.
Anyway, even at a low 1:10 ratio, just a tiny 1nH primary leakage
inductance will give 31R secondary transformed impedance, showing that
burden voltage won't be impacted at all by lowering the opamp input
impedance beyond a few 10s ohm (easily provided by a simple common base
stage).
Carefull primary side design is a must there.
J
John Popelish
Fred said:
Yes, thank you for reminding me the point of a negative
resistance CT burden. The lase application I saw was
extending the low frequency roll off of an ELF coil antenna.
To get anything like best performance from an opamp burden,
the opamp would probably have to be mounted directly on the
coil. Coming up with a core material with high permeability
at 50 MHz would also help. Even the winding form would
matter. It might make a measurable difference if the
secondary came back around the core to the starting point,
rather than going all the way around the core, forming a
turn that was not coupled to the primary.
I have no experience with CFB opamps, but am looking for
circuits that allow them to shine. Do you have any feel for
how the optimum value of the feedback resistor depends on
the shunt to ground impedance at the inverting input?
Does that mean that the core window should be no larger than
necessary to contain the two windings?
F
Fred Bartoli
John Popelish a écrit :
Recently I had the opportunity of a design making use of this for a CT.
The first estimation was extending the LF cut off from a few Hz down to
under a 10th Hz.
One thing you have to take care of is that copper has a 0.4% tempco, and
a total negative resistance will obviously have a nasty behaviour so you
have to take margins so that the resistance stays positive at the lowest
temperature. If you want to get closer to perfect you might design in
the same tempco for the negative resistance part, but you still have to
ensure to stay positive.
That sure would help.
The inverting input impedance is essentially a 50/100R resistance (plus
bonding impedance) and what drives the output voltage is the input
current. All the interesting properties of CFB opamps come from this.
For example (for a given opamp):
- the loop gain almost depends on the FB resistor (that is without
extrem FB networks, ie while the lower FB network impedance is
sufficiently higher than the input impedance)
- one interesting consequence is that you can achieve much higher BW
with a neg input node parasitics capacitance than with VFB opamps.
- another interesting point is that the summing node impedance is upper
bounded by the neg input impedance.
Not necessarily (think of a bifilar wound toroid for ex.)
The basic rule, but that's so obvious that I don't dare mention this, is
to provide means of reducing unlinked flux paths.
Recently I had the opportunity of a design making use of this for a CT.
The first estimation was extending the LF cut off from a few Hz down to
under a 10th Hz.
One thing you have to take care of is that copper has a 0.4% tempco, and
a total negative resistance will obviously have a nasty behaviour so you
have to take margins so that the resistance stays positive at the lowest
temperature. If you want to get closer to perfect you might design in
the same tempco for the negative resistance part, but you still have to
ensure to stay positive.
That sure would help.
The inverting input impedance is essentially a 50/100R resistance (plus
bonding impedance) and what drives the output voltage is the input
current. All the interesting properties of CFB opamps come from this.
For example (for a given opamp):
- the loop gain almost depends on the FB resistor (that is without
extrem FB networks, ie while the lower FB network impedance is
sufficiently higher than the input impedance)
- one interesting consequence is that you can achieve much higher BW
with a neg input node parasitics capacitance than with VFB opamps.
- another interesting point is that the summing node impedance is upper
bounded by the neg input impedance.
Not necessarily (think of a bifilar wound toroid for ex.)
The basic rule, but that's so obvious that I don't dare mention this, is
to provide means of reducing unlinked flux paths.
I am picturing the primary as a sinlge pass through (as is common with
CTs), rather than a bifilar winding. In that case, it seems to me
that the optimum coupling would involve having the window area as
small as possible.
I guess you could fill the unused window area with silver (or super
conductor) to exclude flux from that space.
Could you add an additional resistance between the feedback resistor-
current transformer node and the inverting input, to lower the total
loop gain enough to stabilize the amp with a low value of feedback
resistor? I have also seen some series RC from inverting input ot
ground, to phase compensate CFB amplifiers operated with low feedback
resistances.
(snip)
John Popelish
F
Fred Bartoli
F
Fred Bartoli
[email protected] a écrit :
Sure, that would stabilize the loop, the limit case being having this
resistor value RFB, nulling the old feedback resistor and having the CT
connected to the output. Stable but not very efficient.
More seriously, it can be done, but what would be the purpose? Reducing
RFB to reduce the closed loop input impedance? I've never run the
figures so I can't say for sure but introducing the resitance is
lowering the opamp transimpedance on one side while you reduce the FB
impedance on the other. There's probably a sweet spot but I suspect it
to be with a null additionnal resistor.
Another thing to take with care is layout, especially since CFB is
always synonim to high BW. The dual of node capacitance parasitics for
VFB is series inductance to the neg input. Some interesting subtleties
can slip in there, as I discovered many years ago with my first
encounter with CFB, which probably triggers some of the 'hot rod' comments.
Sure, that would stabilize the loop, the limit case being having this
resistor value RFB, nulling the old feedback resistor and having the CT
connected to the output. Stable but not very efficient.
More seriously, it can be done, but what would be the purpose? Reducing
RFB to reduce the closed loop input impedance? I've never run the
figures so I can't say for sure but introducing the resitance is
lowering the opamp transimpedance on one side while you reduce the FB
impedance on the other. There's probably a sweet spot but I suspect it
to be with a null additionnal resistor.
Another thing to take with care is layout, especially since CFB is
always synonim to high BW. The dual of node capacitance parasitics for
VFB is series inductance to the neg input. Some interesting subtleties
can slip in there, as I discovered many years ago with my first
encounter with CFB, which probably triggers some of the 'hot rod' comments.
Similar threads
