Headscratcher - inaudible 'noise' in project

[rob_gould]

This is confusing the hell out of me. 

I've built a BBD stereo analogue chorus project.  It seems to work OK.  Calibration seems to go pretty much as per the (sparse) instructions.  When I run audio through the unit, I hear chorus.  Happy days.

But!  When I plug the output of the unit into the input of my DAW, I see signal (noise?) at -4dB on the LH channel meter on the channel which the unit is plugged into.  The RH channel has no such  mystery noise.  Nothing is plugged into the inputs of the chorus unit so nothing upstream can be to blame.

Some observations about this so called 'noise'

- the 'noise' is completely inaudible on my system.  This made me think it was very HF clock noise from the BBD chips perhaps.

- I tried to record some of the clock noise in Cubase.  A recording appeared, but when I zoomed in past a certain point, the 'waveform' just disappeared.  Weird!

- I opened the recorded file in SounddForge, and ah - now I see that the waveform isn't a waveform at all; it's just a straight line.  You can see it in the attachment

So my first question : this makes the 'noise' a DC voltage doesn't it?

And second question : I can hear that this issue is throwing the stereo imaging of the unit off, but actually it still sounds OK.  How can this 'noise' in one channel be so loud without massively affecting the performance of the unit?

Cheers,

Rob

 

[JohnRoberts]

This is confusing the hell out of me. 

I've built a BBD stereo analogue chorus project.  It seems to work OK.  Calibration seems to go pretty much as per the (sparse) instructions.  When I run audio through the unit, I hear chorus.  Happy days.

I haven't used BBDs in new designs since the late '70s... SOTA for back then. Now not so much..

But!  When I plug the output of the unit into the input of my DAW, I see signal (noise?) at -4dB on the LH channel meter on the channel which the unit is plugged into.  The RH channel has no such  mystery noise.  Nothing is plugged into the inputs of the chorus unit so nothing upstream can be to blame.

BBD are "bucket brigade devices" analog shift registers where discrete analog samples get clocked down a (string of capacitors) delay line. This stream of clocked discrete samples will have sample (clock) frequency step perturbations in the output waveform.  Even worse the BBD only shifts out a new sample during half of the clock period, so they generally repeat the last sample 2x to fill the whole clock period... This repeated sample uses a different mosfet follower, so there will be a DC step error between the original and repeated samples.  Better designs will generally use a trim pot to adjust out this DC error reducing clock frequency noise.  Also prudent design will have several poles of low pass filtering after the BBD to filter out clock frequency noise.

Some observations about this so called 'noise'

- the 'noise' is completely inaudible on my system.  This made me think it was very HF clock noise from the BBD chips perhaps.

ding ding ding.. we have a winner

- I tried to record some of the clock noise in Cubase.  A recording appeared, but when I zoomed in past a certain point, the 'waveform' just disappeared.  Weird!

no they have LPF....  the whobase recording medium doesn't have the bandwidth to record and playback clock frequency artifacts that are above the audio bandpass.

- I opened the recorded file in SounddForge, and ah - now I see that the waveform isn't a waveform at all; it's just a straight line.  You can see it in the attachment

So my first question : this makes the 'noise' a DC voltage doesn't it?

maybe, maybe not..

And second question : I can hear that this issue is throwing the stereo imaging of the unit off, but actually it still sounds OK.  How can this 'noise' in one channel be so loud without massively affecting the performance of the unit?

Cheers,

Rob

Do you have access to actual test/measurement equipment like an oscilloscope... Look at the signal coming from your BBD gadget. I suspect there is some out of band content that the scope will reveal.

JR

PS: regarding the stereo image, the BBD delay is probably a mono path, so unclear how the stereo image is managed.

 

[PRR]

> this makes the 'noise' a DC voltage doesn't it?

You could ask a DC Voltmeter.

 

[rob_gould]

Do you have access to actual test/measurement equipment like an oscilloscope... Look at the signal coming from your BBD gadget. I suspect there is some out of band content that the scope will reveal.

The best I can come up with is this.  I can't get my 50MHz scope to display anything steadier...

> this makes the 'noise' a DC voltage doesn't it?

You could ask a DC Voltmeter.

It's about -3.4V, varying continuously by a few mV

Anyway, not expecting help from the forum on this one cos I don't have much info to share - not even a schematic. 

I'm in touch with the creator of the project so I'll see what he comes up with.

Cheers,

Rob

 

[JohnRoberts]

Do you have access to actual test/measurement equipment like an oscilloscope... Look at the signal coming from your BBD gadget. I suspect there is some out of band content that the scope will reveal.

The best I can come up with is this.  I can't get my 50MHz scope to display anything steadier...

> this makes the 'noise' a DC voltage doesn't it?

You could ask a DC Voltmeter.

It's about -3.4V, varying continuously by a few mV

Anyway, not expecting help from the forum on this one cos I don't have much info to share - not even a schematic. 

I'm in touch with the creator of the project so I'll see what he comes up with.

Cheers,

Rob

Those spikes are probably the clock noise I was talking about. Play with the triggering threshold to get a more stable image.  If that display notation means 20mV per division and trigger threshold is at -4mV, there is a lot of noise there, raising or lowering the threshold voltage higher (lower) should sync on just the clock noise spikes.

That doesn't look like high enough noise spikes to rectify to -3V so that may be a different circuit error.

JR

 

[Andy Peters]

Do you have access to actual test/measurement equipment like an oscilloscope... Look at the signal coming from your BBD gadget. I suspect there is some out of band content that the scope will reveal.

The best I can come up with is this.  I can't get my 50MHz scope to display anything steadier...

> this makes the 'noise' a DC voltage doesn't it?

You could ask a DC Voltmeter.

It's about -3.4V, varying continuously by a few mV

Those spikes are likely the clock noise mentioned by JR, but I wonder how much of your display is spurious. Where is your scope probe's ground connected? With the vertical amplifier set to 20 mV/cm, you've got a lot of gain and you're likely amplifying noise picked up by the probe more than an actual signal.

-

 

[PRR]

> It's about -3.4V

The DC voltage on audio interconnect points should be essentially ZERO.

Shoving DC voltage into a next-stage is rude.

You have a build error. And surely unrelated to any HF ticks/clocks you may also have.