Practical current mirror

You're cracked.

A classical Excel problem.

Yeah. It's about time that software recognizes the SI multipliers when reading numbers. Not just Excel, everything!

Jeroen Belleman

When confronted with that situation I usually go with an inverting op-amp stage to create a positive signal. But you may want a high-impedance measurement solution.

There are a number of instrumentation amplifiers (INA) that handle voltages below 0.0 volts. Aoe III page 363, has Table 5.8. Thirteen parts qualify as single-supply and seven parts rail-to-rail. Many of these can grab negative signals and translate to positive voltages.

Pages 368 & 369 show seven INA circuit configurations, named A to F2, see 2nd column in the table, providing insight into negative below-the-rail operation. E.g., configuration C2 has PNP transistors to lift the signal by 600mV before amplification. The AD627 INA should do a fine job translating and amplifying negative signals to -300 to -400mV. Configuration D has input switches with flying capacitors making measurements below ground, two choices. BTW, specs are often overly conservative, to allow for operation to +125C, etc., where the "diode drop" goes down to a low voltage, Just ignore that.

Need AoE III, surely buy a nice bound copy, or check out

Thanks Win ! I already bought the nice bound copy but this is handy for when I only have my smart-ass-phone !

So very nice ! boB

I was going to suggest that but he wants to measure around ground (range +/-200mV). You'd need to shift the voltage, too, which is more difficult. One might be able to do something with the ADC (using a reference to both shift and as the ADC reference). Or possibly an absolute value configuration (which polarity?).

What about the positive voltages, then?

The INA reference pin lets you shift the output voltage.

But you need an accurate reference, at least as "accurate" as you want "ground (0V)" to be. Given that, the problem becomes trivial with just resistors (buffered with an opamp or summing point, if you must).

The original intent, I thought, was to use current sources (high impedance) and matched pullup resistors; that's a no-reference-required level translation for a differential signal (as well as the currents match from the mirror). Any noise in the reference current is common-mode on the output.

It is not strictly necessary, as the shunt itself is low impedance, so its voltage couldn't be distorted much by the load.

Thanks, Win, it is very generous to put this book in public domain. I've been reading it for 3 days and I've learnt a lot. But I dare to say that two important chapters are missing: one devoted to high-reliability circuits design with the analysis of the masters' circuits (NASA has a lot of wonderful materials on the subject, one can also check General Electric Semiconductor Data Handbook). How to built, say, a rectifier that can automatically isolate one failed semiconductor component and continue operation till the scheduled maintenance? Or just still work, if the maintenance visit is pretty hard to schedule (try to fix Voyager 2...). The folks did it, the gear has been in continuous operation for 40 years: IMHO this is the *true* art of electronics. The next chapter could be more theoretical and devoted to failure mode analysis, exactly as you did with the chapter on noise.

Best regards, Piotr

OK, I didn't understand it to be a differential signal. I thought it was single-ended in the range of -200mV to +200mV.

ritto:

Thank you!!

(I already have a hardcopy, but it's nice to have the pdf).

Bye Jack