1
\$\begingroup\$

I've read about a thing called a "curve tracer" that you can use to compare known good circuits with bad circuits. It consists of an AC voltage with a resistor that sweeps from -5 to +5V (or similar) and reads the voltage and current going to the component to plot a XY graph. The resulting graph will be the "signature" of the component.

Below is the basic circuit that makes this work:

And here are some example graphs:

Image source (both)

This is all nice, neat and easy to build, but my question is:

While you can hook up a simple LED or capacitor and get a nice graph out of it, won't any modern IC be damaged if you put -5V (AC) through its VCC rail, even if low current, considering that most ICs have a very low reverse voltage maximum rating.

I saw a couple of videos of people testing and making whole databases from a single board, but isn't that playing russian roulette, in a way? In the motto of: "Let's put AC through until something breaks?"

EDIT: There are some commercial devices that rely on this principle, for example the "UCE-CT220S". Of course it's much more sophisticated, but it still relies on passing AC through a component that could be sensible.

Thanks for any help in advance!

\$\endgroup\$
4
  • 1
    \$\begingroup\$ The source link says 1V at 1mA. The resistor values back that up. Also note most cmos chips have protection diodes on the i/o pins. \$\endgroup\$
    – Kartman
    Commented May 18, 2021 at 13:07
  • 1
    \$\begingroup\$ Note that some components require special care. Such as tantalum capacitors - 10µA current in reverse can destroy them. I'd say that a curve tracer is used for testing known components, in a known way, for expected results (not blindly testing every component on a board to build a "database.") \$\endgroup\$
    – rdtsc
    Commented May 18, 2021 at 13:15
  • 2
    \$\begingroup\$ As someone who uses a "curve tracer" with +/- 100V channels, you can absolutely destroy components if you configure it incorrectly. At the same time, I can also configure mine as needed so that I don't (usually). \$\endgroup\$
    – W5VO
    Commented May 18, 2021 at 14:42
  • \$\begingroup\$ Any tool used improperly has the opportunity to destroy. Tools are not idiot-proof! \$\endgroup\$ Commented May 18, 2021 at 15:08

3 Answers 3

3
\$\begingroup\$

While you can hook up a simple LED or capacitor and get a nice graph out of it, won't any modern IC be damaged if you put -5V (AC) through its VCC rail, even if low current, considering that most ICs have a very low reverse voltage maximum rating.

You're right. You can't use this testing method, neither safely nor sensibly, for just arbitrary circuitry.

I saw a couple of videos of people testing and making whole databases from a single board, but isn't that playing russian roulette, in a way? In the motto of: "Let's put AC through until something breaks?"

Videos (esp. Youtube) are a usually a pretty bad idea.

\$\endgroup\$
2
  • \$\begingroup\$ There are several commercial devices that do this. Are they all rather unsafe, since they rely on the same principle? Example: "UCE-CT220S" \$\endgroup\$
    – Fusseldieb
    Commented May 18, 2021 at 13:14
  • 1
    \$\begingroup\$ There also are commercial power supplies able to deliver more than 5V and more than a few mA. You should't connect them directly to 3.3V logic inputs as well. Would you consider power supplies to be "rather unsafe"? It is always up to the professional user of the equipment to make sure that it is used in a proper way. Using curve tracers on circuits that can't handle the necessary input voltages/currents is just wrong handling ;) \$\endgroup\$
    – jusaca
    Commented May 18, 2021 at 13:23
2
\$\begingroup\$

In whatever activity involving electricity, you are expected to have at least basic idea of what you are doing.

The circuit in your question is hardly capable of putting out more than 3V and 3 mA. I don't remember working with a component that can be killed with that little effort.

If you are not careful, you may kill something (e.g. MOS transistor or IC) with ESD while playing with the circuit, but not with the power of the circuit itself.

Then again, what can go wrong?

  1. Something miswired, capable of outing more volts or amperes. It happens even in simple circuits.
  2. Testing a circuit that is simultaneously powered by another (e.g. its own) power source. It may as well be filter capacitors that are not fully discharged yet. Anything with push-pull topology (power amplifiers, etc) is particularily vulnerable because by opening a random transistor you can create short.
  3. A circuit designed to save the energy for a while and then do something nasty at once. Not really a common occurence and requires a big capacitor or even bigger inductor.
  4. A circuit capable of transforming this low voltage into dangerous (for the experimenter) high voltage (e.g. a transformer).

The list doesn't pretend to be exhaustive, of course.

\$\endgroup\$
2
\$\begingroup\$

A proper curve tracer, such as a Tektronix 370A or the more modern (and much more expensive) Agilent/Keysight B1505, doesn't just blindly sweep the voltage from -5 to 5. Instead, you tell it what voltage range to sweep, and it does. Of course, if you're measuring a part that can be damaged by reverse voltage, you tell it to only look at positive voltages.

Such devices will also have some form of current or power limiting; a resistor may be placed in series with the DUT to limit how much current can flow, or the measurement may be pulsed, only applying power to the DUT for microseconds at a time to limit average power dissipation. This is particularly important when measuring, for instance, reverse breakdown voltage of a diode, which may require thousands of volts; if significant current flows for any significant length of time, that diode is not going to be happy.

The circuit you've shown has none of these protection features other than the series resistor (and can only do two-terminal devices, to boot), so it's not as suitable for device characterization as a proper curve tracer would be--but there's a reason curve tracers have five- or six-digit price tags.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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