Interesting OpAmp substraction circuit

[homebrew]

I reacently worked with a high side current sensing IC (LT6106) and thereby learned about an interesting way of subtracting two voltages.
The basic operation is shown in the datasheet (http://cds.linear.com/docs/en/datasheet/6106fb.pdf) but I couldn't resist trying this out with LTSpice. And sure enough it works.

So, while it is quite easy to understand, I've never seen it before in any book.

Does anybody know how that way of subtraction is called?

Edit: Added Link to Datasheet

[T3sl4co1l]

Not really called anything, but it has a voltage-to-current characteristic, thus it has gain in units of current/voltage, i.e., a transconductance amplifier.  The output load resistor thus sets gain (Vg = R*Gm).

Tim

[Bukurat]

Your circuit is not the same as the IC. It has a pnp transistor, yours has an npn.

Swap the transistor to the correct one and in the simulation again.

[dom0]

I don't think there is a set-in-stone name for this circuit, but it's been known and applied widely for high side shunt amplifiers and is usually just called that.

One issue with it is that it requires an PNP/PMOS output, which makes it increasingly difficult to apply to voltages beyond 100 V. There is also a static error depending on the alpha of the PNP / PMOS (PMOS alpha is ~1). When used with a PMOS therefore that error will become rather negligible. Further errors are mainly caused by gain limitation, so transconductance of the PMOS matters as well.

[bobaruni]

It's called an I/V converter (current to voltage).

[dom0]

There are dozen(s?) of I/V converter circuits.

[homebrew]

Your circuit is not the same as the IC. It has a pnp transistor, yours has an npn.

Swap the transistor to the correct one and in the simulation again.

 
WTF? Unbelievable how stupidity combined with simulation can cause even more stupidity.

Ok, I probably was not yet completely woken up to realize that the values of 4.95V was two orders of magnitude of .. should have been 49.5 mV. But BY ACCIDENT the rest of the component values EXACTLY caused a current of 495 mA to flow. What a coincidence. Altering the values lead of course to something completely different ...

Anyway, using a PNP transistor it works fine, of course.

[rfeecs]

It works with an NPN as well.  Refer to Figure 5 of this app note:
http://cds.linear.com/docs/en/application-note/an105fa.pdf