USB Power supply noise reduction question

[jimcroisdale]

Hiya all,

I'm building a pedal for a friend, partly as a fun project, and partly to see if we can improve what he has.

It's a PSU for a Radial Tonebone, which requires 15v and approx 400ma. I thought building a pedal that could run off a 5v USB power bank would be a fun place to start.

The circuit is quite simple so far. From the input socket, the 5v goes to a DC-DC boost module as in the pic attached. They come with a small multi-turn resistor to adjust the voltage, but I'm going to replace this will a fullsize multiturn pot that i can mount on the box. This module will be connected to one of those panel mount Volt/Ammeter LED units that are cheaply available. This will make the pedal fully adjustable for voltage. From the module's output, the power will go through a polarity switch, and then to the output socket.

1) Does this sound to you like it would work? The module can handle over 1A output (and good USB power banks can supply 2+A)
2) The issue of noise. Yes, I am expecting some from the DC-DC booster. What is a simple filtering circuit I can add to make the unit as quiet as possible that will work with the max output voltage of 28v?

Thanks in advance,
Jim

[jimcroisdale]

[iainpunk]

Hey, welcome to the forum!!

This sounds like a cool idea, and i believe it should work fine.
For noise reduction, i recommend big, fat (not electrolitic) capacitor parallel to the output. Idk what you have access to, but i have used 4,7μF foil caps in my amp project, together with a 10.000μF electrolitic one. My project also uses a switching power supply.

Iain

[jimcroisdale]

Hi Iain!

So that would be:

https://www.ebay.co.uk/itm/10-x-4-7uF-100V-Metallised-Polyester-Film-Capacitor-10/282654568856?ssPageName=STRK%3AMEBIDX%3AIT&_trksid=p2060353.m1438.l2649

https://www.ebay.co.uk/itm/10000uF-25V-105C-Radial-Electrolytic-Capacitor-Nichicon-UVZ1E103MRD/111660233476?ssPageName=STRK%3AMEBIDX%3AIT&_trksid=p2060353.m1438.l2649

One each of those, strapped across the output socket?

Jim

[iainpunk]

Well, if your current draw is about 5.5A then, yes. But be aware of the polarity of the electrolytic. I think you (aiming for 1A, right?) Can get away with one or two of those film caps across the leads in combination with an electrolytic of about 3300μF. Im not an expert, and didn't calculate these exact numbers, so It might be over kill, but it worked out good for me.

[DIY Bass]

My biggest concern if I was making something like this for myself is that lots of options such as polarity switching and voltage control would mean that sooner or later I would cook something by plugging in the wrong voltage or polarity by mistake.  I tend towards a keep it simple philosophy so that I can't break something as easily just by a moment of inattention :-)

[jimcroisdale]

DIY Bass - I figure the caps can come before the polarity switch, so that shouldnt be an issue.

Iainpunk - Could you give me an idea of what those caps do (AC to ground I guess?) and why those values/types of cap?

[DIY Bass]

The caps are to filter as much noise and ripple out of the DC lines as possible.  Possibly the better approach may be to set your USB power module to a set output voltage using the onboard trimmer, and then use a pot to control the output voltage of an LM317.  That should give nice clean regulated DC.

[jimcroisdale]

So, use the bost modul I currently have to boost the 5v up to 30v ~(for example) and then use the 317 to trim it down again to where I want it?

[DIY Bass]

I think that's what I would do.  The LM317 is a regulator and will give you clean regulated DC without any noise.  I have a bench power supply that uses one and just switches in different resistors to give the different voltages, but you could use a pot to give more continuous variability.  check out the datasheet or example circuits on the web.

[jimcroisdale]

I would agree, if not for the fact that Ive seen a lot of articles about how to reduce noise from the output of a 317 based circuit - which seems to be another can of worms all of its own! :-D

[Prehistoricman]

The most common approach to filtering noise is a lot of capacitance. The issue with these filters is that the impedance of the incoming noise is low. This means that you need a very low impedance capacitor (read: big value, low ESR) to reduce the noise. Essentially, the noise source and the capacitor are in a potential divider like this:



You can make the filter more effective by increasing the noise source's impedance with a resistor or inductor.



And stack them for multiplying the filter effect.



Voltage regulators, opamps, and pass transistors all suffer from the same thing with regard to filtering: high frequencies. Most of these devices have decreased gain at high frequencies.

[jimcroisdale]

Thanks for the info! :-)

Would anyone care to suggest some values for a stacked filter such as the one in the last pic?

Basically, it needs to be able to cope with a DC-DC module output of 2A max, between 7v and 26v

I have also read that ferrite beads are useful in reducing the spikes from a switching power module such as the one im using?

[jimcroisdale]

le bump...

[Prehistoricman]

Ferrite beads are very small inductors.
I'd say 100-2000uF are good values for capacitors. It's also a good idea to use parallel caps, one big value electrolytic, and one small value ceramic. This reduces the ESR.
Inductor value as high as you can get while handing 2A. You probably want to choose something with a 50% higher current rating to be sure. 47uH inductors should be an acceptable size.

[jimcroisdale]

So say a 1000uf electrolytic and a 100nf ceramic, in parallel, across the + and - output of the PSU module?

[Rob Strand]

So say a 1000uf electrolytic and a 100nf ceramic, in parallel, across the + and - output of the PSU module?

There's no recipe it may or may not work.  Some switchmode modules specify a maximum external capacitance load.   The capacitance can affect the operation of the switchmode.  Some switchmodes have an inductor in series with the output to separate the outside world from the switchmode.  These will often handle caps added to the output.  The whole situation is quite complicated and variable.

The source of ripple voltage in a switchmode comes from current pulses through the output cap.   Often the magnitude of the ripple is determined by the capacitor's ESR and not the capacitance.  The capacitance can help with load transients.  Often small 10uF ceramic caps will work better than 1000uF electrolytics.    These days the switching frequencies are quite high 200kHz up and this is where the 1000uF+ caps start to fall appart.  You need different thinking at 200kHz than at 50Hz.   The big caps might look bigger and more impressive but are they doing anything?

If you look at the diagram Prehistoricman posted, in order for a capacitor to reduce noise in a filter situation there must be a source impedance.    Putting a 1000uF cap at the end of the long wires, like in an effects pedal, might work as the wires provide the source impedance and the source impedance is high.   On the other hand an LC filter can provide much greater filtering because the L in effect increases the source impedance at high frequencies.   An LC filter can get better filtering and is less dependent on the the cap ESR.

Noise can get in through other mechanisms.  The switchmode has current pulse in the input.  If those are not filtered, and the switchmode connects to input long wires, the input wires can radiate junk.    The magnetic field from inductors on the switchmode can couple into audio circuits.

All the stuff about ferrite beads is more to do with EMC issues.  Emissions which may affect radio frequencies.  Sometimes the ferrite can help audio problems but they are usually more complicated scenarios.

The thing is you have to attack specifics of *your* problem.
Maybe you just power-up what you have and it works fine without adding anything.

EDIT:  Think about this.   When you look at smps waveforms on a CRO you see all the spikes and angry stuff.   If the ripple is 200kHz and it's gets into your audio what do you hear?   200kHz?   What you see and what you hear are different things.   If you have one of those old charge-pumps at 5kHz then you do see and hear but at 200kHz it's a different story.

[jimcroisdale]

Noise aside...

Last night i wired it all up and couldn't get much above 12v and everything was getting rather hot - my large ceramic resistor dummy load was boiling! Nothing blew, and it ran okay for an hour, but the voltage shortfall is a problem. I'm assuming 15v is too much of a boost from a 5v input?

[Rob Strand]

I'm assuming 15v is too much of a boost from a 5v input?

With a fixed load resistor the load current increases as you wind-up the output voltage.  If the load current is high that translates to an even higher input current and your 5V input might be going into current limit.     Try a lighter load.    The input current for a 15V output voltage will be more than (15V/5V) times the output load.  So if you have a 1A output load the input current will be more than 3A; probably somewhere in the 3.5A to 4A zone.

[jimcroisdale]

When i initially connected the module to the power source with no load and measured the output, I could get 15v easily. So like you say, I am evidently trying to draw too much power from the power source. With only 2A available from most USB style power banks, this means my idea is fundamentally flawed I think.