Power supply noise filter

[Halban]

Hello, hope you are doing well!

Today I want to start a new project for my hobby: eurorack synthesizers.

The problem

Eurorack is an electronic musical instrument standard, you can buy (or DIY) analog or digital instruments from a lot of brand. They are all placed into a rack, and they have share the same power supply, the PSU of the rack, delivering a rail of -12V / GND / +12.

Some instruments are well designed of not. Sometimes in the sound output of an instrument you get 10khz noise, 15khz noise... it can be caused by ground loops, how instruments are placed in the chain, the intensity of leds of one instrument that you can "hear" on an other instrument, the "CPU" noise, ...

The difficulty is that the source of noise can be multiple and depend of what instruments you have.



My idea

I want to make a power supply filter, that can be placed between the noisy instrument and the power supply rail, to lower the noise level.

After some research based on examples, I've create this design :



I've tested it today and it does not improve the noise.

I'm not sure of my design, neither the value I've used. Maybe you have some guidelines for me?

Thanks a lot, have a nice day!

[madires]

You could add a common mode choke. However, the filter doesn't help against ground loops.

[ZeTeX]

Just use a simple linear voltage regulator, e.g 7812/7912 and the recommend output capacitors as seen on the datasheet.

[Jan Audio]

Just wanto tell when creating a module :
use only 1 ground line, or 1 ground line per section, dont connect them together to avoid a ground loop.

For connections groundless cables, dont connect 2 systems together.

[Jan Audio]

Also place digital modules first in chain, and make sure all your ribbon-cables are equal length.
What PSU are you using ?

[MikeK]

Capacitance multiplier?

[David Hess]

My preferred solution would be a floating +/-12 volt supply for each instrument and then tying the grounds together at the instruments.  The supply does not even need to be regulated on the secondary side if primary side regulation is used with an inverter.

What I might try as an alternative to this if I did not mind the complexity is implementing a separate regulated output for each instrument with remote sense.

[Jan Audio]

You can have like 100 "instruments" modules in 1 setup.

[Halban]

Thanks for your tips!

- using linear regulators was something I had in mind, so I think I will try it

- I will also add common mode chokes to improve HF noise. Should I add them behind or in front of the linear regulators?

- Capacitance multiplier looks interesting. I will maybe try it if the above solution does not work.

- Indeed, as I can have like 20 or 30 instruments, it will be expensive to have separated +-12V supply for each instruments.

- The PSU I use depend on the rack I use, the components insides are often protected, I've no idea how they do the master +-12V (example of a rack: https://www.thomann.de/intl/arturia_rackbrute_3u.htm )

- @Jan Audio : what do you mean by "use only 1 ground line, or 1 ground line per section" ?

[Kibabalu]

A proposal for a solution with belt and braces ;-)

1. inverse-polarity protection --> 2. transient protection --> common mode choke --> pi filter --> capacitance multiplier --> linear voltage controller

First the common mode noise will be damped, then the higher frequencies differential mode noise, then the lower frequencies differential mode noise, and finally the voltage is stabilized.

You can feed in whatever you want, transformer with bridge rectifier or a dirty switched-mode power supply, the output will be quite smooth.

[Jan Audio]

- @Jan Audio : what do you mean by "use only 1 ground line, or 1 ground line per section" ?

In the ribbon cable you have 6 ground lines.

[Halban]

Thanks a lot Kibabalu for this awesome schematic! It's a lot for inspiration for me.

There is 2 components I have difficulties to find because there is so much references : the common mode choke and the inductor.

Can you please advice me a part number or some specifications to refine my search?

[Kibabalu]

Hi Halban,

I used the following components, because the power supply was designed for max. current of 1A:

common mode choke: Würth Elektronik WE-CMB 744825320 Drossel Ringkern, sektionell radial bedrahtet Rastermaß 25 mm 20000 µH 160 mΩ 3 A 1 S
inductors: COIL0912-0.1 Inductor: wire THT 100uH 2A 140mΩ±10% Ø10.5x13.5mm vertical FER

[Halban]

Thank you so much! I will soon begin the schematic and routing this concept.

Have a nice day!

[Halban]

I'have done the schematic, I'm routing it, I will soon test the result!

I have a newbie question: is that kind of circuit partially explain why professional audio mixing tables are way more expensive than cheap mixers? They had a lot of power filtering stages?

[prosper]

Don't linear regulators struggle with high frequency noise? They do well with mains-frequency rejection, but if you've got a lot of PWM and digital noise on the rails from PWM and switching stuff, wouldn't that still need some filtering? Something right on the digital modules (and/or the digital portions of your modules)

[David Hess]

Don't linear regulators struggle with high frequency noise? They do well with mains-frequency rejection, but if you've got a lot of PWM and digital noise on the rails from PWM and switching stuff, wouldn't that still need some filtering? Something right on the digital modules (and/or the digital portions of your modules)

Linear regulators do because open loop gain of the error amplifier necessarily falls with frequency to achieve stability, but this applies just as much if not more to switching regulators where it is less noticeable because they are noisier to start with.

Various things can be done to improve this.  A faster pass transistor will allow for a faster error amplifier.  Greater attention can be paid to input and output filtering.  Proper ground routing, really common, can be used to prevent noise from bypassing the regulator and input and output filtering.

[srb1954]

A proposal for a solution with belt and braces ;-)

1. inverse-polarity protection --> 2. transient protection --> common mode choke --> pi filter --> capacitance multiplier --> linear voltage controller

First the common mode noise will be damped, then the higher frequencies differential mode noise, then the lower frequencies differential mode noise, and finally the voltage is stabilized.

You can feed in whatever you want, transformer with bridge rectifier or a dirty switched-mode power supply, the output will be quite smooth.

There is a risk with this circuit that the regulator output voltage will go excessively high (>25V depending on regulator input voltage) if a trimpot wiper goes open circuit. This has the potential to damage circuitry powered by this PSU module.

I would replace the 5k trimpots with a series combination of 1K trimpot and 2k2 fixed resistor. Then, even if the wiper of the trimpot does go open circuit, the regulator output voltage will still be limited to a reasonably safe level for most downstream circuitry. There is also the benefit that accurately trimming the output voltage will be significantly easier as the trimpot setting won't be so touchy.

[jonpaul]

 Bonjour Monsieur, je suis a La France, vous êtes prêt a Paris?

1/ Any consumer device like EuroRack should be assumed to have the lowest cost Chinese  PSU, design , connectors etc. Many have little if any regard for noise, safety or regulation.

2/ The best solution is a systems viewpoint, check the PSU, linear? Switcher? PARD on 12V + /- under various loads?

Replace the PSU with a decent linear or high quality SMPS?

3/ Filters are effective only if the levels and bands of interference are know and if the noise is identified as CM or DM.

Otherwise one is just guessing.

Bon chance,

Jon