I say, that damn microvolts are everywhere. 🙂
I’ve made a new connection & load box for amp output interface. There is 8 Ω load inside and BNC connector for measurement gear. Here is what noise is there when everything in vicinity is powered down and amplifier is turned off.
Next, noise with just power bar, 1 m from the amplifier turned on with no load at all. That 50Hz spike is there.
Last, noise with the amp turned on.
This is rhe point where investigation begins. Endless sessions trying everytrhing. The fundamental question is the coupling mechanism: is i electrostastic, i.e. capacitive, or magnetic/inductive, or even both?
50 Hz noise is not related to the decoupling, so there will be no change. Anyway, decoupling seems to be ‘perfect’ which is proven by 100 kHz square wave response and slew rate measurement.
Now, just stop for a moment and think how we are all easily fooled with measurement scale that there is something to improve.
Should I care for 1 uV spikes here and there in a power amplifier? 😀
btw it the harmonic spectrum is interesting. The last plot shows significant 100Hz/H2 component, i.e. a heavily, unsymmetrically distorted 50Hz signal. That may originate from a rectified 50hHz circuit.
I recommend building some magnetic sniffer. This can be a simple current loop shorting the input of the oscilloscope probe. More sensitive is a coil full of magnet wire that you connect to the input of a high gain audio amp, any guitar practice amp will do the trick.
With a battery powered practice amp and a small coil I investigated the premises of a customer complaining of hum problems all over the house. And found a hum loop spread over the hole ground by double wired electrical installation in the garden which we the fixed afterwards by one simple wire disruption.
The area of that magnetic loop was really huge!
Magnetic hum hunting with such setup can be a quite interesting experience. You may detect power cables buried in the ground as well as ground currents in the vivinity of power plants.
With a battery powered practice amp and a small coil I investigated the premises of a customer complaining of hum problems all over the house. And found a hum loop spread over the hole ground by double wired electrical installation in the garden which we the fixed afterwards by one simple wire disruption.
The area of that magnetic loop was really huge!
Magnetic hum hunting with such setup can be a quite interesting experience. You may detect power cables buried in the ground as well as ground currents in the vivinity of power plants.
Last edited:
Magnetic properties of stainless steel are not that great, I would not expect any significant 50Hz magnetic fileld reduction even with ordinary steel workbench.
Absolutely not. If you connect speakers with 100dB/W sensitivity, those 6uV would produce less than 0dB hum @1m. To be more precise, ~13dB below the hearing threshold.
Speaker cables alone will pick much more hum than it comes out of this amp. So further improvment in this department is wasting time.
More than this hearing sensity at 50Hz drops significantly referring to 1kHz. Certainly you are measuring far below audibility.
Last edited:
Is there a reason you changed the vertical axis to V instead of V/rtHz? Also if you used a BW of 10Hz-100kHz the total noise would be comparable to e.g. regulator specs.
Yes, to show actual calibrated noise voltage levels. Looks like that fact how low those values are is not recognized by all participants.
Yes, it is.
Total amplifier’s output noise in the 22 Hz – 22 kHz bandwidth is ridiculously low 5 uV. That’s because of design and great PSSR that amplifier has, in the first place.
Voltage regulators are good as well with 120 dB PSSR and some 2 uV total noise.
If you look at the levels, it's clear where it comes from.
You did some good measurements, now act on them.
Jan
That's 120dB below 1V. Depends on where the other noise and distortion levels are.
For a SE tube amp with -80dB THD+N clearly 1uV hum won't be an issue.
In a very clean class D it could.
Jan
I agree that reason for that harmonic content is rectification process. As harmonics are related to the voltage waveform deviation from the ideal form, any mains waveform deformation will be seen as harmonics. Here is what mains voltage looks like at my place. Nasty.
I see symmetric distortion, kind of clipped peaks. This is normal nowadays. And gives H3,H5 etc, but not H2,H4 harmonics.
Best in class actually is about 20uVrms noise at speaker terminals, so hum of 1uVrms is expected to be buried in noise