Lunetta power supply noise - (edit: possibly solved)

[ianbax]

I've been playing with the active mixer design on the logic noise series

https://hackaday.com/2015/07/02/logic-noise-ping-pong-stereo-mixers-and-more/

which uses a 4069 (I'm actually using a 4049 because that's what I have but I don't think it makes much difference - happy to be corrected)

But I've found that - unlike passive mixing - as soon as I introduce this chip I start to get a low-level "chirping" digital noise which is my clock generator (4093 NAND oscillators with large capacitors) being amplified. It's very low level and not audible at all when the oscillators are on but in gaps between it's definitely there.

If I passively mix it disappears. So it's something to do with using the 4049 as a summing amplifier.

I know lunettas aren't hi-fi but it's really annoying to fall at this final hurdle. I know I can passively mix but it's not ideal if you have more than one oscillator going and they start to interact. Perhaps that's just the nature of it. Would a 'proper' op amp in place of the 4049 make a huge difference?

Is there a basic concept I'm missing - a bypass capacitor or something? Intuitively it feels like something to do with shared grounds like the switching noise is in the power supply? When I first built my power in I put a big capacitor across the rails to stop ripple but that doesn't seem to be having an effect.

Thanks in advance! Any help gratefully received.

[ianbax]

So I did some tinkering with the circuit last night and may have solved my problem. Just in case anyone turns up this thread looking with a similar problem here's what I did.

I had a good read up on this thread:

https://electro-music.com/forum/topic-55327.html

and this page of schematics

http://www.all-electric.com/schematic/simp_mix.htm

(number 3 - simple active mixer with attenuation)

I intended to get an op amp at some point (noting long and good explanations of using or not using virtual ground on the thread) but tried it anyway with the 4049 I had to hand.

The clock noise I had bleeding through was cancelled once I re-inverted the output - trial and error but what's Nicolas Collins rule - if it works and doesn't smoke then don't worry if you don't understand it?

Guesses:

Some sort of phase cancellation going on?

Something to do with impedance? I did find that the cancellation was improved if the first inverter used 100k for feedback and input resistors and the second a pair of 1k resistors.

[ianbax]

Wrong on all counts.

It turned the gain right up on my output mixer yesterday and there was the noise again.

So I started unplugging modules one by one - including the 'clock' module that I thought was the source of the noise.

The noise was still there but now settled down into a steady oscillation.

It's the (admittedly very cheap) switching AC-DC 5v power supply I'm using.

I still don't know why it's the mixer circuit that's set it off particularly - the gain involved I guess.

So now I have a load of questions about power supplies - what will be quietest?

I'm thinking a higher end AC-DC adapter - a power brick style one - at a higher voltage (9v, say) put through a LM7805 regulator. Do folks reckon that would cure the switching noise coming though? There's a couple of designs floating about with a protection diode and large cap to cut ripple. It seems straightforward.

Related to this - I read the LM7805 can supply about 1.5A max. How do I go about measuring - or at least estimating - how much each lunetta 'module' is consuming. I've got a case with room for 10 so I guess I need to be around 100ma per module to stay within bounds (annoying actually - my cheap DC adapter is rated for 3A which I thought would be plenty).

And before someone says it - I know 5v is the lower limit for CMOS BUT I've said before, I wanted to interface easily with Arduino further down the line, so it makes sense that my logic = high is 5v.

[zaphod betamax]

Do you have power supply decoupling, both .01u and 1nf caps. (per chip)

[ianbax]

[blue hell]

[zaphod betamax]

I picked up some blowout 5V regulated PSUs for $1.50CAN,
took the chance and got 6 of them.

Switching and regulated, they are clean as a whistle (for switching PSUs)
and have barely any HF hash on my B&K scope.

When I put my CD4093BE oscillators on the breadboard and in oscillation,
they generate way more hash than the PSU.

As the hash is quite high frequency, I used a .1uF and .001uF and this really mitigates the hash I see on the scope.

The circuits I run are all 5V single sided, but with the decoupling caps it gets rid of the noise.

I agree to get a high quality switching power supply.

[ianbax]

Thanks Zaphod - I'm going to be liberal with the capacitors on each module.

I bought the Mean Well supply linked above and it was marginally less noisy but definitely still there (at a lower frequency so even more noticeable to the ear!)

Final roll of the dice I dug out a 12v power brick (also marked switching but for whatever reason quieter) and breadboarded up a LM7805 to provide the steady 5v I need.

The problem there is I don't know what solved the problem! But I now have a quiet power line when nothing is sounding.

When I move off breadboard I'll probably follow these instructions https://www.youtube.com/watch?v=7vA24EC8d6s to add diode protection and filtering caps.

[Grumble]

I have powered active audio monitors connected to my pc, and I use a glassfiber to decouple the inputs from my pc because they tend to hum/hiss caused by creeping current from the pc.
That worked remarkably well, until I rearranged the cables of my setup what reintroduced the noise.
It appeared that the usb supply for the glassfiber to DAC module caused the noise, so changing the usb out connector cured the noise problem.
So the lesson is: if you experience noise, changeing the usb connector can make a difference 😀👍🏻