Confused about oscilloscope reading

[jimjam]

I've connected my oscilloscope (Siglent SDS2104X) to an ESP8266 (powered by a USB port and also tried powering it with a lab power supply SDS3303D)

The oscilloscope GND goes to GND and the probe input (set to 10X) goes to SCL output of the ESP8266, which is running an I2C protocol at 50kHz,.

I found the following output from the scope.

What I don't understand is, how come the scope detected a negative voltage from the overshoot of the falling edge of the clock signal?

I also hooked up a bench multimeter to the same SCL pin, set to digitizing mode (1 million samples per second). It can display statistics, and it says the minimum voltage that it detected was 21mV, so it didn't detect this negative voltage that the scope showed.

How is it possible that a point in the circuit can go more negative than GND? Is the scope incorrect?

[rstofer]

It's called 'ringing' and is caused by parasitic inductance and capacitance of the circuit, primarily the wiring.  I'm just guessing that this was rigged up on a breadboard.

Google for 'ringing signal'.  I'm having trouble posting a Wikipedia link

[jimjam]

It's called 'ringing' and is caused by parasitic inductance and capacitance of the circuit, primarily the wiring.  I'm just guessing that this was rigged up on a breadboard.

Google for 'ringing signal'.  I'm having trouble posting a Wikipedia link

Thanks! TIL about ringing! Yes, it was rigged up on a breadboard with jumper wires.

Curious that the DMM didn't pick it up.

Looking at the time, the oscillation / ringing occurred for about 80ns. The DMM can only resolve 1/1000,000 s = 1us so I guess it wouldn't be able to see it.

[MrAl]

I've connected my oscilloscope (Siglent SDS2104X) to an ESP8266 (powered by a USB port and also tried powering it with a lab power supply SDS3303D)

The oscilloscope GND goes to GND and the probe input (set to 10X) goes to SCL output of the ESP8266, which is running an I2C protocol at 50kHz,.

I found the following output from the scope.

What I don't understand is, how come the scope detected a negative voltage from the overshoot of the falling edge of the clock signal?

I also hooked up a bench multimeter to the same SCL pin, set to digitizing mode (1 million samples per second). It can display statistics, and it says the minimum voltage that it detected was 21mV, so it didn't detect this negative voltage that the scope showed.

How is it possible that a point in the circuit can go more negative than GND? Is the scope incorrect?

Hi,

In a word, energy storage elements such as inductance and capacitance.
For example, if you charge up a capacitor to 5 volts and then connect the positive lead to ground, the negative lead will measure minus 5 volts, at least for a short time until the energy in the capacitor dissipates.  If you connect an inductor in series with a cap and charge the cap up to 5 volts and connect the negative lead of the cap to ground, and the open lead of the inductor to ground, you will see the negative lead of the cap oscillate negative and positive until all the energy dissipates and the cap and inductor are discharged. This would show a similar trace as what you are seeing.

The thing is, every electrical component has some capacitance and some inductance even though that is not part of the specification.  That is because every parallel distance has an associated capacitance and every series distance has an associated inductance, and that includes just about everything including free space.  This kind of L and C are usually referred to as parasitic inductance and parasitic capacitance.  In IC chips though there could be components like this built right into the chip on purpose for use in the circuit.  The parasitic parts are there mostly due to nature itself and are not put there on purpose.

[tggzzz]

Thanks! TIL about ringing! Yes, it was rigged up on a breadboard with jumper wires.
...
Looking at the time, the oscillation / ringing occurred for about 80ns.

Look up the equation for an LCR circuit. You will find:

• the period/frequency of the oscillation is determined by the L and C (You don't mention the period, but it looks like maybe 20ns)
• the duration of the oscillation is determined by R. (You mention 80ns, lower R=>shorter duration)

Do a back-of-the-envelope calculation to find the expected oscillation frequency. The "C" will be the sum of the probe tip capacitance (15pF?) and the capacitance of the loads on the wire (5pF per load). The "L" will be ~1nH per mm length of wire (including scope probe ground lead).

You will see that the L can be a killer. Even a bog-standard probe plus ground wire will oscillate/ring at ~90MHz. For an example plus workaround see https://entertaininghacks.wordpress.com/2015/04/23/scope-probe-accessory-improves-signal-fidelity/

Solderless breadboard exacerbates such problems. For alternative prototyping techniques, see https://entertaininghacks.wordpress.com/2020/07/22/prototyping-circuits-easy-cheap-fast-reliable-techniques/

[TimFox]

A DMM may not pick this up since the "DC" voltage (technically the average voltage over time) is still positive, even though the ringing of the reactive circuit temporarily takes the voltage negative with respect to ground.
In our case, apparently the sampling rate of the DMM is too slow to see enough of the negative voltage excursions to take that short-term average negative.
The DSO shows the voltage as a function of time, within the limits of the DSO bandwidth and sampling rate.

[jimjam]

Thanks for all the tips. I switched the probe's GND alligator clip with the springy thing that attached to the tip, and stick the probe directly into the breadboard, and the ringing is much reduced.

[MrAl]

Thanks for all the tips. I switched the probe's GND alligator clip with the springy thing that attached to the tip, and stick the probe directly into the breadboard, and the ringing is much reduced.

Oh yes we should have mentioned that, that is the best way to connect to ground.

[tggzzz]

Thanks for all the tips. I switched the probe's GND alligator clip with the springy thing that attached to the tip, and stick the probe directly into the breadboard, and the ringing is much reduced.

Good.

Note that the scope probe tip is probably larger than most through hole component leads, and so will open the breadboard springs more. The next time you use that hole the springs might not grip the component as well. No, I don't like solderless breadboards, and refuse to debug anything made on them.