I see a lot of ringing and overshoot at the output of my amplifier. I’ve followed the data sheet guidelines and think that the layout is clean. What could I be doing wrong?
This is the type of problem that perplexes and frustrates us. Engineering is a science, so the result of doing A and B should be C. If you’ve been designing circuits for a while, you know that engineering is also an art. Bob Pease, God rest his soul, sent me an autographed copy of his book Troubleshooting Analog Circuits. He inscribed it with: “May all your troubles be middle sized so you can find them.” I have always liked this quote, and wish that it could always be true.
This engineer had read the data sheet. It's always a good start, but you'd be surprised by how many times it doesn't happen. So we started to delve into the problem. The first thing we looked at was the schematic, “eyeballs in motion” as a former colleague called it. We looked for the usual suspects: amplifier noise gain, bypass capacitors, load, and supply voltages. Why these?
The noise gain determines the amplifier’s stability; and if the phase margin is low, the output can ring and overshoot. Bypass capacitors keep noise out of the amplifier and store charge right at the power supply pins. This is especially important when the amplifier needs a stiff supply with ample current because its output is changing rapidly. If the power supply voltage changes while the output is slewing, the variation will surely find its way to the output. The load can cause problems if the capacitance or inductance gets too big, or if the load resistance gets too small. The performance of some amplifiers degrades when supply voltages get too big or small, so check the supply voltages against those shown in the data sheet.
If all those look good, what do you do? More troubleshooting. Next, we looked at the layout. Were there long traces with parasitic inductance? Were any bypass capacitors far from the supply pins, allowing parasitic inductance to form a tank circuit with the capacitors? Did the ground plane creep under the input and output pins, forming parasitic capacitors that could lead to ringing and overshoot? In this case, the layout looked good as well.
Ok, what next? How were they testing it? Was the input clean, and was it properly terminated? The engineer saw a little ringing on the input, but not much. As we all know, garbage in equals garbage out, so we tried to clean up the input. It was properly terminated, so we swapped generators to see if that was the problem. The new generator was a little better, but the input and output were still ringing. Then the light bulb went off. I asked the engineer if he was using a cable or scope probe to check the signal. He was using a scope probe, so I asked if it had a ground clip. It did, and it was about 3 inches long. I guessed that this was the problem, and instructed him to remove the clip lead, unscrew the plastic barrel that covered the probe tip, and to use the metal liner of the scope probe to pick up the ground close to the signal. When he did this, the ringing disappeared. Voila! So what was going on?
The ground clip had series inductance, the probe had capacitance, and the traces at the probe site had parasitic capacitance. The capacitance and inductance formed a tank circuit that oscillated when energized by fast rising edges from the circuit, causing ringing and overshoot on the input and output. Another tip: always calibrate the scope probe before making measurements, as this can help reduce the peaking too. Another case closed!
Troubleshooting is a methodical approach to finding a problem, but there’s also an art to it. Looking for trouble in all the wrong places eventually leads you to the right place!