2
\$\begingroup\$

I recently built a logic probe adapting a circuit from an old magazine. One of the modifications I introduced is adding a polarity protection diode (1N4007) in series with the positive power rail, because the probe is intended to be powered from the circuit under test and I want to avoid damaging the probe should I mistakenly connect the power rails to the wrong place.

This has led to a small error in the definition of the voltage levels the probe recognizes as HIGH and LOW. That's because the thresholds are defined by a 3-resistors divider placed across the rails, which in the original design sensed the actual supply voltage of the circuit under test, and which now suffers from a 0.7V drop due to the protection diode.

I didn't want to modify heavily the circuit, so I thought of connecting the upper leg of the divider directly to the power rail of the circuit under test, bypassing the protection diode. I simulated the circuit with LTspice and everything is fine.

Here are the schematics for the input stage and the power supply stage:

enter image description here

enter image description here

My question is: is this design decision risky in case of a polarity inversion in the rails?

In this case I reckon that the current that may flow along "wrong paths" will be heavily limited by the resistors of the divider, which are rather large. Moreover the divider is connected to the inputs of an LM358 opamp, which should be quite insensitive to reverse voltages spikes of modest energy content (it's a bipolar design, so no MOSFET gates that can be damaged by voltage spikes even if current-limited).

Am I missing something? Is my reasoning correct?

P.S.: I assume a max reverse voltage of 20V (worse scenarios are less likely, and I won't bother to design a foolproof protection scheme for those).

\$\endgroup\$
4
  • 1
    \$\begingroup\$ Why not use MOSFET polarity protection instead? \$\endgroup\$ – Ignacio Vazquez-Abrams Jul 18 '16 at 2:46
  • \$\begingroup\$ @IgnacioVazquez-Abrams As I said, I didn't want to modify the circuit too much. I happened to notice the small error after I had already built the circuit. The mod I plan is just a "bodge wire". Moreover I have no depletion mosfet at hand. This is just a "toy" project, so I didn't want to bother too much. Moreover the planned solution intrigued my analysis/design skills: I'm curious to know if the solution I devised is viable or is a crappy idea :-) \$\endgroup\$ – Lorenzo Donati -- Codidact.com Jul 18 '16 at 2:52
  • 1
    \$\begingroup\$ @IgnacioVazquez-Abrams Please, ditch that reference to "depletion mosfet". It was just a brain-fart! Too late at night here :-) I just double checked and found that enhancement MOSFETs can actually be used as polarity protection devices. The other reasons still stand. \$\endgroup\$ – Lorenzo Donati -- Codidact.com Jul 18 '16 at 2:57
  • \$\begingroup\$ @IgnacioVazquez-Abrams in the end I followed your advice; having realized that I didn't need a depletion mode MOSFET and given that I've recently put my hands on a batch of 100 dirty cheap Fairchild 2N7000 MOSFETs, I decided to try. Of course I fried one in the attempt when testing if the protection worked :-) Darned body diode, I knew I had to put it the other way around! Lesson learned and so much fun! \$\endgroup\$ – Lorenzo Donati -- Codidact.com Jul 20 '16 at 21:37
1
\$\begingroup\$

The LM358 will allow this, it will not destroy anything. Look at the datasheet, at Absolute Maximum Ratings (I'm using the TI datahseet):

  • Input voltage: −0.3 to 32V. So that would be a problem, because with inverted supply, input would be more negative than this.
  • However: Input Current (VIN < −0.3V): 50mA max. So you're fine unless the resistors are low enough to let 50mA of current through the input. There is even a note expliciting the fact that it is not destructive.

However, I see a potential problem with your design!

The LM358 input common-mode range goes from 0 to Vcc-1.5V. With the additional diode drop, it makes 0 to Vpow-2.2V. In your application and given your inputs, I guess you can easily have a high-level voltage above Vpow-2.2V. In this case, the LM358 may output wrong results. You should use another opamp that has rail-to-rail input range instead. Check in the datasheet (with similar reasoning as above) that it can withstand negative input voltage without being destroyed.

\$\endgroup\$
3
  • \$\begingroup\$ Thank you very much! Interesting point that about common mode range. Since it is not a JFET opamp I didn't consider phase reversals a possibility, but the articles you point to mention that even some bipolar types are susceptible. I must investigate the issue (or make some more measurements on the thing!). IF phase reversals can be excluded, I guess there wouldn't be problem for a small violation of common mode range: it should only affect performance, I assume, but since I use the opamp as a comparator and its speed is not important, I guess it will still work, hopefully. \$\endgroup\$ – Lorenzo Donati -- Codidact.com Jul 18 '16 at 12:30
  • \$\begingroup\$ @LorenzoDonati To be honest, I don't really know what could be the consequences of a common-mode range violation in your specific case (since, by definition, it is not specified in the datasheet). Maybe phase reversal can be excluded, but still, I wouldn't do it. RR input opamps are cheap enough, and you can most certainly find some that has the same pinout as LM358 (if you're worried about having to redo a layout). I'd really switch to a more appropriate part if I were you. And it's not a "small" violation: for example, with 3.3V CMOS logic, you'll be more than 1V above the CMR limit. \$\endgroup\$ – dim Jul 18 '16 at 12:42
  • \$\begingroup\$ I completely agree with you: in the context of any serious project (even for one-off prototype) you definitely don't want to violate the specs. But as I said in another comment, it is just almost a "toy" built on a matrix-board with parts from my junk box. BTW, I didn't mention it is intended only for "high voltage" logic families: min nominal supply 5V. Therefore if I can stretch the limits of the devices a bit, that's OK, as long as nothing "catches fire" :-) If, in the end, I hit a showstopper flaw in the design, oh well, I had fun building it and it was an occasion for learning something! \$\endgroup\$ – Lorenzo Donati -- Codidact.com Jul 18 '16 at 13:08

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.