Simulating my amplifier with ngspice shows intermittent self oscillation if I choose a time step of 10nS. LTSpice, on the other hand, does not show any self oscillation. This disagreement is motivating me to test the amplifier with a digital oscilloscope. I can use as a signal a digital audio file I created containing a sine wave starting from a few tens of hertz to around 20kHz. Since, I never made such a measurement, I would like to ask how I should do such a test. I am somewhat worried as high frequency self oscillation even if it is intermittent damages tweeters. I would like to clarify that the self oscillation is smaller than the signal and is superposed on it. This is what ngspice is simulating.
Do a single sweep storage. Adjust the amplitude and time base for best display.
Keep manually triggering the sweep until you capture the oscillation.
You probably won't be able to easily trigger on the oscillation if it's small compared to the signal.
But always use a dummy 8R resistor load for such testing.
Keep manually triggering the sweep until you capture the oscillation.
You probably won't be able to easily trigger on the oscillation if it's small compared to the signal.
But always use a dummy 8R resistor load for such testing.
Set your scope's trigger mode to "norm", not "auto". You also will need to be able to trigger on the burst so a high-pass filter will be needed -- or perhaps your scope has a trigger mode that rejects low frequencies. Do not put the scope into an "averaging" mode.
The "norm" setting retains the captured waveform in-between trigger events so you can more easily see it. For an alternative approach, put the scope into its single-shot mode and arm it when your amp is running.
For any of these approaches you will need to set the trigger level correctly so it can capture that oscillation burst.
The "norm" setting retains the captured waveform in-between trigger events so you can more easily see it. For an alternative approach, put the scope into its single-shot mode and arm it when your amp is running.
For any of these approaches you will need to set the trigger level correctly so it can capture that oscillation burst.
The oscilloscope is a Siglent SDS1104X-E. Since this is a new oscilloscope I am somewhat hesitant to use it. I already used it to test isolated circuits. I am worried because the amplifier has an earth connection.
I have the same oscilloscope and i use it with a Sigilent SDG1032 to make bode plots.
The generator has a lot of junk in the output specially at low levels so i have made passive damping kits 1/10, 1/100 and 1/1000
for the output so i can be near the 4 volt max out.
I would absolutly break the ground for the amplifier. I use an extension cord without ground that i only use in tests. It is only a foot long so there is no risk of using it anywhere else.
The oscilloscope has a extra ground plug beside the calibration output. A crododile clip and a cord to amplifier ground is also possible but on mv signals the ground loop can affect your measurements.
With bode plots up to 5MHz you should find your problem.
The generator has a lot of junk in the output specially at low levels so i have made passive damping kits 1/10, 1/100 and 1/1000
for the output so i can be near the 4 volt max out.
I would absolutly break the ground for the amplifier. I use an extension cord without ground that i only use in tests. It is only a foot long so there is no risk of using it anywhere else.
The oscilloscope has a extra ground plug beside the calibration output. A crododile clip and a cord to amplifier ground is also possible but on mv signals the ground loop can affect your measurements.
With bode plots up to 5MHz you should find your problem.
Use a DVM to check for high voltage between the two grounds. I would expect the voltage to be pretty low, particularly if the scope and amp are plugged into the same circuit. All the earth grounds should go back to a common point in the breaker box so if you DO observe a significant level of AC you have something that needs to be looked at by an electrician.
Square waves tend to uncover oscillations. Excessive ringing after rising an falling edges reveals design issues. Use 1 or 10 kHz and gradually increase power output.
I always measure the ground difference between a device under test and any grounded test equipment. Be safe , be smart and preserve your equipment, device and the hookup cables. Don't learn the hard way like I did!
You could try simulating it with the trapezium rule integration method. Trapezium rule has no numerical damping, like many other numerical integration methods have. Unfortunately it messes up the current waveforms. You get this weird triangle wave superimposed on the currents: one time step too low, next time step too high, next too low and so on. The voltages are usually not affected.