BTW the high side sense amplifier can also be built with npn transistors.

True. I have seen many variations. Let me stay confused about one circuit at a time.
I figure if can get one to make sense to me the others will follow.

 

This means the voltage at the diode must be a diode drop below ground

That makes sense. Thank you.

In the first example the op amp generates an output voltage such that the current through the diode equals the input current through R1.
This is to keep the minus input voltage equal to the plus input voltage (ground) as op amps do in a closed loop.
This means the voltage at the diode must be a diode drop below ground (which is .47V in this low-current case).

.


The second example doesn't work because there's no source of current for the emitter- collector through R4.
(Where would it get that current)?
To work, the emitter needs a resistor and a voltage source as the original circuit had.

You need to think about how op amps and transistors work and the flow of the current in all the nodes.
Ask the question: What does the op amp output have to do to keep both inputs at the same voltage?
Random connection of devices gives random results.

Okay, R4 returned to -12 V but I can't make sense of the result as I vary the input.
0.5 V in (0.25 mA) I get -8.6 V out (0.715 mA)
1.0 V in (0.5 mA) I get -5.7 V out (0.475 mA)
1.5 V in (0.75 mA) I get -2.47 V out (0.20 mA)
What is going on???

 

....................
Okay, R4 returned to -12 V but I can't make sense of the result as I vary the input.
0.5 V in (0.25 mA) I get -8.6 V out (0.715 mA)
1.0 V in (0.5 mA) I get -5.7 V out (0.475 mA)
1.5 V in (9.75 mA) I get -2.47 V out (0.20 mA)
What is going on???

Cockpit error.
The R4 current equals the voltage drop across the resistor (the difference between the transistor collector voltage and -12V) divided by 12k, not the collector voltage divided by 12k.

 

Here is a high side current monitor that I have used, it works very well. It is very similar to the circuit you are talking about, but is much easier to power.

 

Cockpit error.
The R4 current equals the voltage drop across the resistor (the difference between the transistor collector voltage and -12V) divided by 12k, not the collector voltage divided by 12k.

Of course. Pardon my brain fart. Okay, now I see a V to I converter.

Here is a high side current monitor that I have used, it works very well. It is very similar to the circuit you are talking about, but is much easier to power.

It certainly is. It just isn't in my inventory. A good example of using an NPN. Same basic idea. The non-inverting input drives the output until currents at the inverting input are cancelled.