So I have recently completed my first chipamp ( https://www.diyaudio.com/community/threads/a-first-lm3886-project.386817/#post-7040805 ). To the ear it is whisper quiet but I do like to verify things and so was seeking to measure the output without any signal. I have a Levell Microvoltmeter and it serves me well.
Initially I just connected it up with nothing at all connected to the inputs and got an impressively low noise of 210muV - both channels the same.
But I thought the inputs should be shorted to get a proper measurement and so knocked up a shorting RCA plug (soldered shield to core of a very short lead - photo attached. It did have some shrink tubing on but I removed that just to make sure it really was soldered under there. Also tried with another similar lead) and used that. The measured noise goes up! I measure 1.2mV with that shorted lead attached. Of course in neither case is anything audible without sticking and ear almost in the speaker driver - but I would like to know what is going on.
I am using the 20Hz to 100kHz bandwidth setting on the Levell but all other settings are consistent (obviously noise goes down for narrower bandwidth and up for wider.
I am mystified. Is that even very short lead picking up something? Why isn't the 'something' just shorted to ground?
Initially I just connected it up with nothing at all connected to the inputs and got an impressively low noise of 210muV - both channels the same.
But I thought the inputs should be shorted to get a proper measurement and so knocked up a shorting RCA plug (soldered shield to core of a very short lead - photo attached. It did have some shrink tubing on but I removed that just to make sure it really was soldered under there. Also tried with another similar lead) and used that. The measured noise goes up! I measure 1.2mV with that shorted lead attached. Of course in neither case is anything audible without sticking and ear almost in the speaker driver - but I would like to know what is going on.
I am using the 20Hz to 100kHz bandwidth setting on the Levell but all other settings are consistent (obviously noise goes down for narrower bandwidth and up for wider.
I am mystified. Is that even very short lead picking up something? Why isn't the 'something' just shorted to ground?
A quick verification would be to short it right at the input plug, just temporarily tack a wire across the terminals of the connector.
If the noise again goes down, your 'shorting plug' picks up a signal, maybe due to insufficient screening from the cheap cable.
If the noise stays, your amp may oscillate with shorted inputs, then try to short with say 47 ohms across the input connector terminal.
Jan
If the noise again goes down, your 'shorting plug' picks up a signal, maybe due to insufficient screening from the cheap cable.
If the noise stays, your amp may oscillate with shorted inputs, then try to short with say 47 ohms across the input connector terminal.
Jan
Thanks @jan.didden So I have now shorted the plug itself (chopped off the cable). No difference. So I will go for the resistor approach and see what happens. Its all a learning experience!
Hi m10chalk,
Would you post a schematic?
I'm wondering if installing the short might be raising the circuit's "noise gain" compared to the open-circuit gain. If so, this would be normal behavior.
Would you post a schematic?
I'm wondering if installing the short might be raising the circuit's "noise gain" compared to the open-circuit gain. If so, this would be normal behavior.
Good point, that is what you would get with an inverting amplifier.
Is it really noise that you get with shorted input or rather hum?
Is it really noise that you get with shorted input or rather hum?
@BSST With apologies for amateur hour scribble here is the simplified schematic. The input is intended to RF filter then DC block and then it runs into a nearly datasheet standard implementation for the LM3886. I see I have omitted the Zobel components but they are installed.
Hi @MarcelvdG - have posted schematic above now. Non-inverting. I cannot hear any hum. The added noise is still pretty much below hearing threshold unless ear pressed to speaker.
Your "amateur hour scribble" is far better than many of my posts.
I'm concerned that I don't see a bias path for the amp's + input. Is it present but not shown in the sketch? What DC voltage do you observe at the amp output terminal? Assuming bias is present, my speculation about noise gain doesn't seem to be the issue.
I'm concerned that I don't see a bias path for the amp's + input. Is it present but not shown in the sketch? What DC voltage do you observe at the amp output terminal? Assuming bias is present, my speculation about noise gain doesn't seem to be the issue.
Proper shorting plugs are made - they should be in every amp-builders toolbox! - they're simply shorted RCA plugs, no wire exposed.
If hum or noise exists while shorted, then an obvious problem could be the internal layout, or shielding, or wiring of the home-made amp itself.
This can also happen if the inputs are directly coupled to the input chip/transistors without a blocking capacitor, which would isolate things.
Residual Noise is a product of electronics, tube or transistor.
If the noise level is low enough as to not be heard, but only measured, then I would not obsess over it.
Besides, putting your ear right up to a speaker is not normal in normal listening situations involving music, etc.
Now, if you can hear hum from a few feet away, that's a different story.
If the noise level is low enough as to not be heard, but only measured, then I would not obsess over it.
Besides, putting your ear right up to a speaker is not normal in normal listening situations involving music, etc.
Now, if you can hear hum from a few feet away, that's a different story.
My chief suspicion is now minute hum voltage hum voltages across chassis between input connectors and amp board. Does observed noise drop with reduced front panel volume setting? Trouble shooting technique would detach input wiring wiring and test for reference noise level with short installed at board edge. But it sounds like residual hum/noise is already very low. Maybe time to enjoy your efforts? 😉
Cheers,
Steve
Cheers,
Steve
Indeed, a 1.2 millivolt noise level at the speaker is not an issue to worry about.
1mV on 8 ohm is ~70dB down compared to 1W. With ordinary, 80dB@1W speakers, it doesn't matter: 10dB sound level is well below any normal environment. With highly efficient speakers, like 95dB@1W, it could become an issue, assuming the noise is within the audibility range, which doesn't seem to be the case for you.
This probably means that the noise is ultrasonic, probably coming from a digital device or a SMPS, and it gets magnified by some mechanism when the input is shorted.
Try to use a more relevant bandwidth on your Levell (I also own one): 10Hz to 10kHz seems adequate. The meter also has a monitoring output, which you can use to visualize the signal on an oscilloscope, spectrum analyzer or sound card
When interference gets worse with shorted input, that usually means it couples in magnetically / inductively. Stray field from an old-fashioned mains transformer could be the cause, but indeed also ultrasonic stray field from an SMPS. It would be nice to measure the spectrum somehow.
Thanks everyone for your comments. All excellent and helpful. @Elvee - absolutely right - noise goes down on an approximate A-weighting (bandwidth 10 -10kHz). I have a very low spec picoscope so I won't be able to see any really high frequency stuff but following what @BSST says I do think the difference between shorted and not shorted input is in fact some hum component and even my rubbish oscilloscope is good for 50Hz! I will have a look when I get a chance. Connecting a real source (my preamp) and muting it reveals that in practise the noise at speakers is well less than 1mV. Looking at some of the numbers it looks like I have a better than 90bB SNR for all practical purposes. So as @BSST says its time to enjoy and stop obsessing! This has really been an excellent project. The amp is sounding really good. Again thanks all.