Floating current source

Tony Williams wrote:
In article <c8n40j$q5d$1@news.Stanford.EDU>,
alan <no-longer-valid@yahoo.com> wrote:


I am looking to design a voltage controlled current source. I
would like the current source to be bipolar and fully floating,
i.e. both the source and sink are totally floating. Or in other
words, the current that goes out the source HAS to come back
through the sink. The input is an AC voltage source (not really
floating). This circuit is used in a four-point measurement,
but some other part of the circuit "pushes the voltages around",
thus the need for the current source to be floating.


I'll stick my neck out to get the discussion going.
Assuming the OP needs an AC sinewave current stimulus.

+---|>|---+---+---------+
| D1 | | |
| | )+ \
| | )sec1 /R1
| | )- \
| +| | D2 |
| C1=== +---|>|---+
| | | |
| | | |/e
| | '-------|PNP Q1
| | |\c
+ | | |
)|| | | +------
)|| | +sec3- | | Iout
pri)|| +-//////--+-----/--------------
)|| | | ======CT |
)|| | <--------+////+ |
+ | CT | | |
| <-------------' |/c
| | ,-------|NPN Q2
| | | |\e
| +| | D3 |
| C2=== +---|<|---+
| | | |
| | )- \
| | )sec2 /R2
| | )+ \
| D4 | | |
+--|<|----+---+---------+

It's a transformer with a pri and 3x secs. Two of the
secs (1 and 2) each drive a common base transistor,
PNP and NPN to provide alternating-polarity half-sine
output currents. The pri is voltage-driven by some
unspecified amplifier.

The main sec (sec3) delivers the total output power
needed for the PNP or NPN stages, plus the Vpk/Ipk
required by the unknown load. sec3 output is phased
with the sec1 and sec2 so that each stage is powered
up during each respective half-sine.

D1 and D4 prevent reverse currents flowing when
either transistor stage has a reverse voltage
across it. C1/C2 provide a small amount of dc
bias across a transistor stage at the start of
each half-sine.

Because it is a relatively open loop circuit it
would be prudent to measure the actual ac output
current and factor it's value (and actual shape)
into the unknown resistance calculation. A small
current-transformer (CT) provides this signal.
Okay- well I'll throw in my 10lbs of confusion over just exactly what
he's doing. This is STP or Scanning Tunneling Potentiometry which is a
variant on ST-Microscopy used to acquire "new insight into the local
electronic transport properties on the nanometer scale." The
conventional STM works by servoing a z-axis piezo positioner so as to
maintain a constant tip-sample tunneling current- which then ensures the
tip-sample separation, d, is a constant- and this gives incredibly high
resolution of a surface topography on the pico-meter scale through
measurement of z-axis variation. The tip-sample bias is also maintained
during this scan . A second variation is to vary this bias and acquire
the small perturbations in tip tunneling current to infer certain things
like "electronic density of states" and a bunch of other atomic
characteristics. So the thing looks like this:
View in a fixed-width font such as Courier.

STP acquires surface topography and potential distribution
of current flow through sample....


precise determination of
atomic topography dimension z
as tip scans x-y dimensions

feedback piezo positioner
tunneling CCS regulating I constant
+-----------+ tip
+----+ I tip--> |-----+ maintains constant d
| +-----------+ | between tip and sample
| |
| | I ~ exp(-Kappa x d)
| | tip
| | Kappa= decay constant
| z | function electron state energy
| | y TIP levels
| | / ^
| | / | Itip also =I ( Vtip-sample)
| +-----x d tip
| | <--on order several atomic diams
| v
| +---------------------+ <-----
| | | Isample
| +---| S A M P L E |----------+
| | | | |
| | +---------------------+ |
| | <------V (Isample)--- |
| | x |
| | |
| | |
| | ADJ SAMPLE |
| | BIAS FOR Isample |
| | _ |
| | | /| |
| +-------------| |/ ------------------+
| | /| |
| | / |
| | |
| | |
| +----------/\/\/\/\/\/\--------------+
| /|\
| | set reference for V(x)
| | |
+---------||------------+
| |
| Vtip,bias
---
/// tip bias voltage applied through resistor bridge

bridge adjusted for zero local potential between

tip and sample - then variation of Itip vs V (Isample)
x,y
can be acquired.
 
Fred Bloggs wrote:

A Google-proof diagram:
View in a fixed-width font such as Courier.

..
..
..
..
.. STP acquires surface topography and potential distribution
.. of current flow through sample....
..
..
.. precise determination of
.. atomic topography dimension z
.. as tip scans x-y dimensions
..
.. feedback piezo positioner
.. tunneling CCS regulating I constant
.. +-----------+ tip
.. +----+ I tip--> |-----+ maintains constant d
.. | +-----------+ | between tip and sample
.. | |
.. | | I ~ exp(-Kappa x d)
.. | | tip
.. | | Kappa= decay constant
.. | z | function electron state energy
.. | | y TIP levels
.. | | / ^
.. | | / | Itip also =I ( Vtip-sample)
.. | +-----x d tip
.. | | <--on order several atomic diams
.. | v
.. | +---------------------+ <-----
.. | | | Isample
.. | +---| S A M P L E |----------+
.. | | | | |
.. | | +---------------------+ |
.. | | <------V (Isample)--- |
.. | | x |
.. | | |
.. | | |
.. | | ADJ SAMPLE |
.. | | BIAS FOR Isample |
.. | | _ |
.. | | | /| |
.. | +-------------| |/ ------------------+
.. | | /| |
.. | | / |
.. | | |
.. | | |
.. | +----------/\/\/\/\/\/\--------------+
.. | /|\
.. | | set reference for V(x)
.. | | |
.. +---------||------------+
.. | |
.. | Vtip,bias
.. ---
.. /// tip bias voltage applied through resistor bridge
..
.. bridge adjusted for zero local potential between
..
.. tip and sample - then variation of Itip vs V (Isample)
.. x,y
.. can be acquired.
..
..
..
..
 
Fred Bloggs wrote:
Fred Bloggs wrote:

A Google-proof diagram:
View in a fixed-width font such as Courier.

.
.
.
.
. STP acquires surface topography and potential distribution
. of current flow through sample....
.
.
. precise determination of
. atomic topography dimension z
. as tip scans x-y dimensions
.
. feedback piezo positioner
. tunneling CCS regulating I constant
. +-----------+ tip
. +----+ I tip--> |-----+ maintains constant d
. | +-----------+ | between tip and sample
. | |
. | | I ~ exp(-Kappa x d)
. | | tip
. | | Kappa= decay constant
. | z | function electron state energy
. | | y TIP levels
. | | / ^
. | | / | Itip also =I ( Vtip-sample)
. | +-----x d tip
. | | <--on order several atomic diams
. | v
. | +---------------------+ <-----
. | | | Isample
. | +---| S A M P L E |----------+
. | | | | |
. | | +---------------------+ |
. | | <------V (Isample)--- |
. | | x |
. | | |
. | | |
. | | ADJ SAMPLE |
. | | BIAS FOR Isample |
. | | _ |
. | | | /| |
. | +-------------| |/ ------------------+
. | | /| |
. | | / |
. | | |
. | | |
. | +----------/\/\/\/\/\/\--------------+
. | /|\
. | | set reference for V(x)
. | | |
. +---------||------------+
. | |
. | Vtip,bias
. ---
. /// tip bias voltage applied through resistor bridge
.
. bridge adjusted for zero local potential between
.
. tip and sample - then variation of Itip vs V (Isample)
. x,y
. can be acquired.
.
.
.
.
I think that in Vx,y(Isample) mode- the instrument is servoing Vtip,bias
in order to maintain Itip constant and also tip potential at virtual
ground- but I can't be sure because of the ambiguous non-technical
descriptions at hand.
 
In article <c8qi0g$5bn$1@news.epidc.co.kr>,
Jan Panteltje <pNaonStpealmtje@yahoo.com> wrote:

OK, if we go this way, you have some cross-over distortion,
as the Q1 and Q2 need 0.7V to bias.

Since the be voltage is in the direction of the opposite
supply, perhaps you could add a diode in series with the 'cold'
sie (side connected to the supply lines) o fsec1 and sec3,
with a resistor to the opposite supply, to create a .7V bias.
(I have only drawn the negative side in the diagram).
Yes. I was going to put a resistor between the two
bases to get a small amount of Class A biasing.

But the OP's further explanation of the requirements
probably makes the circuit inappropriate.

--
Tony Williams.
 
"alan" <no-longer-valid@yahoo.com> wrote in message news:c8prk0

Yeah, that sounds about right...

Anyway, I was hoping that there would be a way to do it without
physically isolating things.
It just seems that it shouldn't be that hard to take an existing
circuit for a current source with one end grounded, but then make that
end float as well. Of course, I could build two opposite current
sources, but I don't think I could match up the source and sink currents
exactly.
How about an ordinary floaing voltage-controlled voltage source,
with a series resistor? You could add a feedback resistor, and
a couple of optoisolators for compensated feedback...

Cheers!
Rich
 
"Jan Panteltje" <pNaonStpealmtje@yahoo.com> wrote in message
news:c8oerj$smq$1@news.epidc.co.kr...
On a sunny day (Sat, 22 May 2004 20:17:51 GMT) it happened Spehro Pefhany
speffSNIP@interlogDOTyou.knowwhat> wrote in
6pdva01h4np6dcmc0k32r30fcatkktkrgv@4ax.com>:
One op-amp and a series voltage reference and a resistor, at a
minimum, once the power and input signal isolation is dealt with
(which can be either a project to sink your teeth into or a "pick out
the building blocks from the catalogs" exercise, depending on the
specifics of the requirements.

We probably need a lot more data on what is going on.
JP

Bulbs?
Light bulbs, as current source..
When the wire in the buklb gets hot, its resistance increases (a lot),
so you get some current stabilization.
JP
How about a battery and a pot?

All he said, after all, is a "floating current source".

Cheers!
Rich
 
"Jim Thompson" <thegreatone@example.com> wrote in message
news:1jl1b0df0ftm09f9eagcqkbaei5vmurbl8@4ax.com...
On Sun, 23 May 2004 16:46:44 GMT, James Meyer <jmeyer@nowhere.com
wrote:

On Sun, 23 May 2004 04:07:59 -0700, alan <no-longer-valid@yahoo.com
posted
this:



The setup is a four-point AC measurement across a sample, with some
extra goodies thrown in. There is an additional scanning probe that
touches down somewhere along the sample. Another lead that is connected
to the end of the sample pushes the sample to whatever voltage such that
the current coming out of the first lead I meantioned is zero. (In
other words, zeros out the voltage in the middle of the sample) That is
why the current souce needs to be floating. This is a circuit for
scanning potentiometry.

It might not need to "float" in the sense you are
thinking.

That's true - I'll have to think of alternatives.

If you took an op-amp and added a current sense resistor to its output
and voltage sensors before the sense resistor and after the sense
resistor, the
output of the whole thing could be programmed, or "servoed", to look like
a
floating current source and a voltage sensing probe at the same time.
Use as
many of those as you need to make your measurements. Of course there
will be
common mode restraints associated with the op-amps, but all you have to
do is
power them all from the same plus and minus supplies with the common of
the
supplies connected anywhere on the sample.

Programmed properly, the output currents of the op-amps could be made to
follow each other closely enough, some sourcing and some sinking, so that
no
current would flow in the sample-to-common connection lead and the whole
thing
would look like it was completely floating as far as the sample were
concerned.

Jim

Absolutely... pretty much what I have been thinking. I don't even
think the extra probe is needed to establish "zeros out the voltage in
the middle of the sample".

...Jim Thompson
I did this in a product I designed about 10 years ago, and it worked well -
a current limited voltage source - it was a switchable 0-10V/+-10V/0-20mA
output stage for user I/O in an ac drive. Interestingly enough, the
differential current measurement set upper bounds on the accuracy - a pair
of voltage dividers off each end of Rsense, connected to +ve and -ve opamp
inputs. this is of course a resistor mismatch detector, and with 1% 50ppm/C
resistors throughout the circuit, gave us an overall accuracy of 5%. We
later changed to the dual of this topology, ie a voltage limited current
source (our old friend, mr modified howland). Sensitivity analysis proved
the dual was less critical than the original - about 3% vs 5% - and actually
used a few less parts (had to do some silly things when outputting -ve
voltages, so had diodes on the current sense amp dividers, which had to be
low volt-drop (ie schottky) and low leakage (ie not schottky). We could have
just used 0.5% or 0.2% resistors, but $$$ is important. The dual cost about
$0.20 less than the original, mostly because the diodes used in the voltage
clamp were LL4148s and we didnt care about Vf or Ileak.

Terry
 
alan wrote:
Spero Pefany wrote:
On Sat, 22 May 2004 01:41:06 -0700, the renowned alan
no-longer-valid@yahoo.com> wrote:


Hello,

I am looking to design a voltage controlled current source. I would
like the current source to be bipolar and fully floating, i.e. both the
source and sink are totally floating. Or in other words, the current
that goes out the source HAS to come back through the sink. The input
is an AC voltage source (not really floating). This circuit is used in
a four-point measurement, but some other part of the circuit "pushes the
voltages around", thus the need for the current source to be floating.

The only circuits for (voltage controlled) current sources that I have
seen have one end of the load grounded. Also, I am not sure how to feed
in the controlling voltage and still have the circuit floating.


What kind of bandwidth?

1kHz

Galvanic isolation is one obvious way. If you
tranformer-isolate the power supply (for example, use an isolated
DC-DC converter) and use an isolation amplfier to get the input
voltage across then it's pretty straightforward, at least for
relatively low frequencies.

Yeah, that sounds about right...

Anyway, I was hoping that there would be a way to do it without
physically isolating things.
It just seems that it shouldn't be that hard to take an existing
circuit for a current source with one end grounded, but then make that
end float as well. Of course, I could build two opposite current
sources, but I don't think I could match up the source and sink currents
exactly.
Actually, precise matching is not needed.
Another alternate method is to use a 1X constant source, and a
switchable 2X sink.
 
On a sunny day (Mon, 24 May 2004 01:19:22 GMT) it happened "Rich Grise"
<null@example.net> wrote in <u4csc.18709$ZQ.14150@nwrddc03.gnilink.net>:

"Jan Panteltje" <pNaonStpealmtje@yahoo.com> wrote in message
news:c8oerj$smq$1@news.epidc.co.kr...
On a sunny day (Sat, 22 May 2004 20:17:51 GMT) it happened Spehro Pefhany
speffSNIP@interlogDOTyou.knowwhat> wrote in
6pdva01h4np6dcmc0k32r30fcatkktkrgv@4ax.com>:
One op-amp and a series voltage reference and a resistor, at a
minimum, once the power and input signal isolation is dealt with
(which can be either a project to sink your teeth into or a "pick out
the building blocks from the catalogs" exercise, depending on the
specifics of the requirements.

We probably need a lot more data on what is going on.
JP

Bulbs?
Light bulbs, as current source..
When the wire in the buklb gets hot, its resistance increases (a lot),
so you get some current stabilization.
JP

How about a battery and a pot?

All he said, after all, is a "floating current source".

Cheers!
Rich
That wil work, but the current will not be very stable when there
are load variations.
When you say 'current source' the idea is that the current is
stabilized somewhat.
But you are right, take a high enough voltage for the battery and a
high value resistor, and it will be close.
JP
 
On a sunny day (Mon, 24 May 2004 01:17:53 GMT) it happened "Rich Grise"
<null@example.net> wrote in <53csc.18695$ZQ.16149@nwrddc03.gnilink.net>:

"alan" <no-longer-valid@yahoo.com> wrote in message news:c8prk0

Yeah, that sounds about right...

Anyway, I was hoping that there would be a way to do it without
physically isolating things.
It just seems that it shouldn't be that hard to take an existing
circuit for a current source with one end grounded, but then make that
end float as well. Of course, I could build two opposite current
sources, but I don't think I could match up the source and sink currents
exactly.

How about an ordinary floaing voltage-controlled voltage source,
with a series resistor? You could add a feedback resistor, and
a couple of optoisolators for compensated feedback...

Cheers!
Rich
It has to be AC
100kV transformer with 100 MOhm gives stable 1mA
 

Welcome to EDABoard.com

Sponsor

Back
Top