Electronic Design

# What's All This Three-Level Grader Stuff, Anyhow?

About 40 years ago, when I was at Philbrick, I worked a lot on operational-amplifier (we NEVER said op-amp) applications. Every once in a while, a customer would ask about a "zero-crossing detector," or more simply, just a "crossing detector." We never thought of calling this a "comparator." We usually made the customer happy with an op amp, a couple resistors, and maybe a few diodes in the feedback, so he could get a response in a few dozen milliseconds, perhaps. (When Widlar brought out the µA710, we were all astonished by the sub-microsecond response!)

But I did figure out how one op amp can perform two comparisons. Back when an op amp cost \$20 and UP, that was NOT insignificant savings (see Fig. 1). When IIN is around zero, the output is around zero. If IIN exceeds I2, the output will switch to a large negative level. If IIN reverses and exceeds I1, the output will switch to a large positive level. So here is a sort of discriminator with three output levels, which provides nice advantages. This went into the Philbrick 1966 Amplifier Applications Handbook. I never heard of anybody else who invented a three-level grader using just one amplifier.

Recently (38 years later), a guy asked me if we could help him with a circuit to detect a fairly small current (20 mA) even if the current may at other times exceed a couple amperes. He also had to detect when the current exceeded +20 mA, or −20 mA. And the common-mode range could exceed 10, 20, or 40 V. Therefore, most linear, direct-coupled circuits were not going to do the job for him. What's a mother to do?

Figure 2 is a basic three-level grader, with two optically coupled outputs that are brought down to ground by 4N28's. The LMV2011's low VOFFSET of 3 µV (20 µV max) is quite small compared to the 400-µV signal. Neither output is ON for input currents −16 mA < IIN < +16 mA, but each output responds when the input signal exceeds plus or minus 22 mA. Response is a few milliseconds for small input overdrive, falling below 20 µs for large currents. I had to build this, to prove it would work—and it does. So in some cases, getting one amplifier to do the work of two is NOT a silly deal. Getting signal information down to ground from a high voltage level is not that hard, if you know the tricks.

Mail Stop D2597A, National Semiconductor
P.O. Box 58090, Santa Clara, CA 95052-8090