Tim Wescott wrote:

> Hey Piotr, why are you doing this with a current sense resistor?

Because the the ACS758 datasheet says that the resistance of
the current path is ~100uOhm. So, since there is this "resistor"
already, I simply decided to use this fact. I am still learning
Spice, so I didn't know about its more advanced primitive blocks.
Let alone about how to use them correctly.

> If the Hall sensor has some specified internal resistance (that 100u-ohm?
> ), I think I'd still simulate it as a resistor in series with the sensing
> voltage source, and use an "H" block.

Since I've already mastered "E", there is time to learn "H". ;-)

> Just a simple H block won't simulate any output impedance, frequency-
> dependence, nonlinear behaviors, etc.  But you knew that.

Yes, but that doesn't matter here. The frequency is 100Hz,
the currents are within the specified range and I simply
wanted to check whether the Hall sensor could be used to
correctly drive a synchronous rectifier. And Spice says it
can, much better than the standard comparator-based approach,
as there are no oscillations. Its primary purpose was
overcurrent protection, but since the signal is already
available, I wanted to see what would happen. I set the
threshold to 0.25A and combined its output with the outputs
of the sign detector -- voila, it works and seems to be very robust.

One stupid thing about the ACS75* and similar sensors design:
three output terminals. The output voltage is specified to be
VCC/2 + V(I), but they do not expose their VCC/2 in order to
allow accurate differential measurements. And they say that
the drift can be +-25mV, which is half an amp of an error...

	Best regards, Piotr

On Sun, 22 Feb 2015 21:27:23 +0100, Piotr Wyderski wrote:

> Hello all,
> 
> I am trying to simulate in LTSpice a circuit which contains a Hall
> effect current sensor (ACS758 for that matter). I don't have its model,
> but the simulation doesn't need to cover all the corner cases, even a
> simplified model would be enough:
> 
> There is a 100uOhm resistor R. The voltage drop across it should produce
> isolated voltage 2.5V+40[mV/A]*I(R).
> 
> I can handicraft an opamp-based network which does more or less what
> needed, but I consider it to be a workaround. So, what is the simplest
> way to express such a circuit in LTSpice? The "voltage" library node
> doesn't contain the "expression" mode or whatever its name should be.
> Is there a four-terminal abstract block to do this kind of
> transformations?

Hey Piotr, why are you doing this with a current sense resistor?  AFAIK 
the Hall sensor more or less acts like a current-controlled voltage 
source, which Spice provides, and calls "H".  IIRC, you put a voltage 
source in series, with the voltage set to zero, then you put the "H" block 
wherever you need it, with a reference to the voltage source you placed.

If the Hall sensor has some specified internal resistance (that 100u-ohm?
), I think I'd still simulate it as a resistor in series with the sensing 
voltage source, and use an "H" block.

Just a simple H block won't simulate any output impedance, frequency-
dependence, nonlinear behaviors, etc.  But you knew that.

-- 

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Piotr Wyderski wrote:

> There is a 100uOhm resistor R. The voltage drop across it should
> produce isolated voltage 2.5V+40[mV/A]*I(R).

Found it under the very descriptive name "E", LOL! :-D

	Best regard, Piotr

Hello all,

I am trying to simulate in LTSpice a circuit which contains a Hall 
effect current sensor (ACS758 for that matter). I don't have its
model, but the simulation doesn't need to cover all the corner
cases, even a simplified model would be enough:

There is a 100uOhm resistor R. The voltage drop across it should
produce isolated voltage 2.5V+40[mV/A]*I(R).

I can handicraft an opamp-based network which does more or less
what needed, but I consider it to be a workaround. So, what is the simplest
way to express such a circuit in LTSpice? The "voltage" library node
doesn't contain the "expression" mode or whatever its name should be.
Is there a four-terminal abstract block to do this kind of transformations?

	Best regards, Piotr