reducing noise in a preamp powered with dc step up module

[Marcos - Munky]

I built a preamp for a friend's bass, and sinde I have a few 18650 batteries, we had the idea on powering it using one of those and have a "usb rechargeable bass". I tested the preamp powered with a 9V battery and it's dead quiet. Then I installed the 18650 battery along with a charge module and a MT3608 based step up module to get 9v out of the 18650 battery.

Now I'm hearing a noise, which is probably caused by the step up module. Based on the MT3608 datasheet (link below), the switching frequency is 1.2MHz, which is above the audible range, but I'm still hearing some noise. Any suggestions on how I can get rid of it?

https://prom-electric.ru/media/MT3608.pdf

[Rob Strand]

Did you use a shielded inductor?

One test is to move the converter board away from the preamp, guitar pickups, guitar wiring, by connecting the power between the two with some long twisted wires.   If you still get the noise then the problem is on the power rail.    If the noise goes away then the noise is being coupled electromagnetically.

[Marcos - Munky]

Did you use a shielded inductor?

No, it's a regular smd inductor. This is the module: http://www.icstation.com/icstation-mt3608-step-converter-module-boost-converter-power-supply-module-icsa004a-p-3448.html

One test is to move the converter board away from the preamp, guitar pickups, guitar wiring, by connecting the power between the two with some long twisted wires.

I'll do the test tomorrow. How long should be the wires? "as long as possible" is a good answer but I mean, should 50cm be enough? 1m? Also, why twisted wires since the current is DC?

[Rob Strand]

No, it's a regular smd inductor.

That inductor does limit the amount of field getting out.     There's only a small gap between the inside and outside.
At least it isn't  bobbin (which wouldn't be good for efficiency at 1.2MHz anyway).

If you pack that board right next to the preamp you might get issues.   It can depend on which way the boards are facing and how far apart.

I'll do the test tomorrow. How long should be the wires? "as long as possible" is a good answer but I mean, should 50cm be enough? 1m? Also, why twisted wires since the current is DC?

If you cut to 1m then twist it will be fine.    The main aim is to see if the noise drops at all.  With the wires present you can experiment by pointing the PSU board in different directions and placing it at different distances.   Try not touching the board with your fingers. It should give you an idea where you stand.

Twisting the wires is just a precaution. You don't want current pulses in the wires generating magnetic fields.  When debugging stuff like this it's best be heavy handed on precautions.  It's easy to untwist them later to see if it makes the problem worse.     If possible twist the input wires as well.

If you can't see any change with the position and orientation then that means you might need to investigate filtering the output and/or input power rails.

These tests are more about trying to work out the cause.   Once you know the cause you can play around with solutions.

[Marcos - Munky]

So, I did the test. Changing the step up module position compared to the pickups and preamp board had an effect. For example, if the module board was in a 90 degree angle with the pickups, the noise is dropped (but not gone). The distance didn't had too much noticeable effect, so it'll be need some filtering on the power rails.

Also, turning down the treble pot of the preamp reduced the noise to almost zero.

Should I try a "eletrolytic + ceramic cap to ground" filter?

[rankot]

You may also try some kind of small inductor with large cap to filter out AC from DC.

I don't know about that particular IC, but all those power boost chips are very sensitive to PCB layout, grounding etc.

[Rob Strand]

So, I did the test. Changing the step up module position compared to the pickups and preamp board had an effect. For example, if the module board was in a 90 degree angle with the pickups, the noise is dropped (but not gone). The distance didn't had too much noticeable effect, so it'll be need some filtering on the power rails.

That a weird result.   If it depends on then angle then that means there's junk radiating off the board.  However, if there's junk radiating off you would expect  the noise to drop when you moved the board away.

Maybe there junk radiating off the power cable?   Can you see if putting the input side cable near the preamp and then try output side cable near the preamp and see if you can get a cause an effect.   The idea is to find the thing which make the noise the worst when it is located near the preamp.

You might also want to try the pickups.   Maybe noise is getting through the pickups and not the preamp.    Is the tone control located before the preamp?  If that is the case, then the fact the noise goes down with the tone control confirms the noise is getting in through the pickups.

[Rob Strand]

One thing that might be possible is to wrap the converter board in an aluminum shield.  Since the switch frequency is so high it might be enough to stop any field getting out.   Basically the stray field off the inductor and board will get shorted out by the shield due to eddy currents in the shield.   

As a test I'd put the board in a plastic bag.  As is it's probably not final solution as I'd be *very* concerned getting shorts with the Li-Ion battery.  A better solution could be to sandwich the converter PCB between two un-etched PCB's with some spacers.

[amptramp]

It looks like the converter could get a separate enclosure like a 1590A.  Maybe the output cable could be shielded twisted pair or coax.  Then you have the problem of what to ground it to.

[R.G.]

As a systems kind of approach, noise can be conducted or radiated.

Radiated noise is affected by shielding and orientation (that is, rearrangement of the antennas, or parts that are acting like antennas). Note that wires can be picking up radiated noise and then conducting it into the circuit.

Conducted noise can be noise conducted down any wire, but with this setup, it's probably power and/or ground. conducted noise is attacked by
(1) reducing the noise at the source; in this case, maybe better filtering on or right next to the up-converter
(2) filtering between the noise source and circuit
(3) making the circuit relatively immune to the frequency spectrum of the noise.

As noted, the fact that mechanical re-orientation of the pieces changes the noise points to it being at least partially radiated. Shielding the circuit or power module, or both might reduce this.

If that doesn't work, or doesn't reduce it enough, you will probably need to look into filtering. Filtering gets complicated when high frequencies are involved, as the noise can actually be made worse by using the wrong filter, missing the frequencies of the noise, or actually tuning in the noise with L and C parts that are accidentally resonant at part of the noise frequencies.

One really frustrating thing is when you have high frequency noise - 1.2MHz qualifies - and it's coupled into the circuit, "detected" in the radio sense by the semiconductors, converting it down into audio, then amplified. It might help to try to make the circuit more immune to the noise by trying to cut its high frequency response. This can often be done by inserting a resistance in series with the input leads, and a small capacitor from the input to ground. Using an inductor instead of a resistor can actually tune RF. The resistor eats the radio energy.

Your prospective input filter could start rolling off at the top of audio, say, 25kHz. Just picking 100pF for the cap, that makes the series R be 6.4k or so if I didn't miss a decimal point. Try 1K and 100pF.

[amptramp]

R.G. is right to bring up one thing - a circuit with only L and C and no R is likely to trade response at one frequency with response at another frequency and you end up playing whack-a-mole with the filter response.  You cut out a resonance at one frequency only to have it reappear at another.  Resistance is one method of limiting this effect because the frequencies are going to be damped.

You may need a filter at the input to the converter and one thing to understand here is that a regulated feedback converter is a negative resistance.  If you have an inverter that puts out 12 volts at 2 amps, the regulation will look like a negative resistance at the input.  If you have an input of 9 volts, the current (minus any losses) is 2.666  amps to get the 24 watt output.  If the battery drops to 6 volts , the current rises to 4 amps.  As the voltage drops, the current increases, so the converter looks like a negative resistance.  The impedance of an input L-C filter looks like sqrt (L/C).  The magnitude of the filter positive resistance has to be lower than the negative resistance of the converter.  If the input is 12 volts at 2 amps, the resistance is 6 ohms.  If the input is 6 volts at 4 amps, the resistance is 1.5 ohms.  If the negative resistance of the converter is less than the positive resistance of the filter, you have a negative resistance oscillator.  If the negative resistance of the oscillator is in the stable region but close to it, the transient response will suffer.

[Rob Strand]

IMHO, we haven't identified the cause enough to propose a solution.   By the same token you often have to try things which could be solutions in order to identify the cause!

We don't know if the problems are cause by 1.2MHz or the harmonics.   Like RG mentioned the semiconductors can act like detectors.   You have the option of removing the problem at the source (a filter on the PSU or shield) or on the circuit (add filters to the ckt or a shield).

R.G. is right to bring up one thing - a circuit with only L and C and no R is likely to trade response at one frequency with response at another frequency and you end up playing whack-a-mole with the filter response.  You cut out a resonance at one frequency only to have it reappear at another.  Resistance is one method of limiting this effect because the frequencies are going to be damped.

All is not lost.

This is FYI to others (not nit-picking):

For frequencies say above 30MHz you often choose the ferrite material so the inductor acts a resistor instead of an inductor - that might twist some people's minds but it's true.   

At 1.2MHz it's more likely we end-up in the wack-a-mole region.   Some damping come from the cap ESR and some from the inductor loss.    For converters in the 1.2MHz the caps are likely to be ceramics which have low ESR.  One trick is to add ESR.   Another trick is to add a parallel cap with a high ESR or an added ESR.   Even a parallel load resistor can help.

[Marcos - Munky]

Sorry for the delay on the reply, I was out of home for the weekend.

That a weird result.   If it depends on then angle then that means there's junk radiating off the board.  However, if there's junk radiating off you would expect  the noise to drop when you moved the board away.

You might also want to try the pickups.   Maybe noise is getting through the pickups and not the preamp.    Is the tone control located before the preamp?  If that is the case, then the fact the noise goes down with the tone control confirms the noise is getting in through the pickups.

The noise is always there, but it gets worse if the step up module is closer to the pickups, based on the angle. So it's probably being both radiated and conducted. The tone control is part of the preamp. I'm using Baja's bass preamp from the other forum, which is basically a 2N5457+BC557 stage, tone control and another 2N5457+BC557 stage.

One thing that might be possible is to wrap the converter board in an aluminum shield.

I'll try this tonight. To avoid shortings with the li-ion, I'll wrap everything on duct tape.

Your prospective input filter could start rolling off at the top of audio, say, 25kHz. Just picking 100pF for the cap, that makes the series R be 6.4k or so if I didn't miss a decimal point. Try 1K and 100pF.

I'll give a try tonight.

[Rob Strand]

The noise is always there, but it gets worse if the step up module is closer to the pickups, based on the angle. So it's probably being both radiated and conducted. The tone control is part of the preamp. I'm using Baja's bass preamp from the other forum, which is basically a 2N5457+BC557 stage, tone control and another 2N5457+BC557 stage.

Ah, OK.  So basically turning the tone knob down is just attenuating the noise wherever it comes from.

From what you are saying here it seems like a lot is getting straight into the pickups.  We don't know if it's audio band or RF band (getting demodulated by the preamp ckt).

[Marcos - Munky]

Ok, so I tried both approaches (ground shielding around the board and 1K + 100pF filter at the input), and the noise is still there. Well, I tried them one at a time, not both at the same time.

Maybe it'll help if I describe better my setup. It's a P/J bass, with the pickups in parallel. First thing on the signal is a "balance" pot (one pickup to each end, output on wiper), then a single volume pot. Wiper of the volume pot is going to this preamp:

Then it goes to the output jack. No noise at all with a 9V battery, noise with the step up module, and no noise at all if I roll down the treble pot.

I tried the filter right after the volume pot. If I got it right, it's a 1K resistor from wiper of the volume pot to the input of the preamp, and the 100pF cap goes from the input of the preamp to ground. Should I try other values, or try to put it somewhere else?

Or maybe I should just forget about it and go with a 9V alkaline battery?

[Rob Strand]

Well, I tried them one at a time, not both at the same time.

If you couldn't tell the difference then neither of those might be the cause.

Maybe it'll help if I describe better my setup.

OK very helpful.

The main thing I see is the high-impedances throughout the circuit.   This can promote capacitive coupling.  the other thing is no supply filter.   So maybe the next thing to try is a supply filter, say 1k in series with the power and 10uF or even 100uF across the preamp side power.  It's something to try even to see if it has an effect.

Another thing you could try is to simply wire the volume control to the output jack *but* leave the preamp in place and running.   That may tell you if the junk is getting into the pickup/balance/volume part of the circuit before the preamp.

tried the filter right after the volume pot. If I got it right, it's a 1K resistor from wiper of the volume pot to the input of the preamp, and the 100pF cap goes from the input of the preamp to ground. Should I try other values, or try to put it somewhere else?

That would be a good place.  Notice the next stage kind of has that built-in as well.

Or maybe I should just forget about it and go with a 9V alkaline battery?

What about,

https://duckduckgo.com/?q=rechargeable+lithium+ion+9V&ia=products

No guarantee these will be noise free and I suspect you will get vast variations across different brands.

[Marcos - Munky]

So maybe the next thing to try is a supply filter, say 1k in series with the power and 10uF or even 100uF across the preamp side power.  It's something to try even to see if it has an effect.

So, if I got this right, both the resistor and cap should be put between the power module and the preamp, correct? Should I put the filter right at the module output or as close as possible to the preamp?

Another thing you could try is to simply wire the volume control to the output jack *but* leave the preamp in place and running.   That may tell you if the junk is getting into the pickup/balance/volume part of the circuit before the preamp.

Great idea. I'll try it tomorrow.

On the rechargable batteries, the idea even crossed my mind. Actually, for this circuit, I'm not worried on using an alkaline battery - they're usually rated from 400mAh to 600mAh and this preamp is described to draw around 80uA (yep, that low), so a battery would last too long to worry about. I'm mostly checking how hard is to fix the noise so I can try to use the 18650 + step up module setup in other audio sittuations like powering pedals or a little amp.

[Rob Strand]

So, if I got this right, both the resistor and cap should be put between the power module and the preamp, correct? Should I put the filter right at the module output or as close as possible to the preamp?

Yes.  I'd try it at the power module first.    The best end  depends on what junk is on the wires.  If it doesn't work try the other end.

I'm mostly checking how hard is to fix the noise so I can try to use the 18650 + step up module setup in other audio sittuations like powering pedals or a little amp.

Nothing wrong with trying to get it to go.   If you use a different converter later you might find you get a different set of problems.

[Marcos - Munky]

Ok, so I tried the filter (1K + 100uF) right after the step up module, and it did reduced the noise. I still had that long twisted pair of wires connecting the module to the preamp. Then I tried the filter (this time with a 220uF cap) right on the preamp board, and it reduced even more the noise, to a level it's still there but maybe it's acceptable. Then I connected the short wires back, put everything inside the bass and closed the lid...

Now, the noise is back . So I opened it again to play with the module position compared to the pickups, and found a place where there was low noise. But then I found out there's another source of noise that is position-based: the 18650 battery itself! When I changed it's position, the noise changed. And even at the same exact spot but rotated 180 degrees (just swapping which way the polarities were facing), I got different amounts of noise.

Maybe now the best idea is to just use a 9V battery.

[Rob Strand]

Ok, so I tried the filter (1K + 100uF) right after the step up module, and it did reduced the noise. I still had that long twisted pair of wires connecting the module to the preamp. Then I tried the filter (this time with a 220uF cap) right on the preamp board, and it reduced even more the noise, to a level it's still there but maybe it's acceptable.

OK, to me that means there is ripple on the output of the converter.  Probably a lot.   I fact I'm even thinking the output cap on the converter board isn't doing its job.   Do you have an oscilloscope to look at the ripple at the output of the converter?
You might get a small improvement putting another 1k between the 100uF and the 220uF.

Then I connected the short wires back, put everything inside the bass and closed the lid...

Now, the noise is back . So I opened it again to play with the module position compared to the pickups, and found a place where there was low noise. But then I found out there's another source of noise that is position-based: the 18650 battery itself! When I changed it's position, the noise changed. And even at the same exact spot but rotated 180 degrees (just swapping which way the polarities were facing), I got different amounts of noise.

OK that's annoying but what is says to me is there are strong current pulses at the input of the converter.  That is normal for a converter but it is also normal to try to stop them getting out by using a cap or filter.   Problems occur when the cap across the input terminals is too small, or the layout is bad ; the layout for a 1.2MHz switcher is *very* critical.     The current pulses also go through the battery and that can make the battery appear like a noise source.  Twisted battery wires can help but you can't twist the battery .

A full blown input filter looks like this.   If you *need* a large L you can get into trouble with all the stuff amptramp mentioned.
https://www.digikey.com/en/articles/techzone/2012/jun/~/media/Images/Article%20Library/TechZone%20Articles/2012/June/Managing%20Converter%20In-Rush%20Current/article-2012june-managing-converter-in-rush-fig1.jpg

So I'd probably try putting a large smd ceramic cap across the input terminals of the converter, located right at the input terminals.    If that doesn't work you might need a small inductor then a second cap like in that picture.

It crossed my mind that perhaps the biggest source of problems is the fact that converter board is rated at very high currents.    The current pulses on the input (and through the output caps) usually scale-up roughly proportional to the converter rating.  You are running at tiny currents so the current pulse are astonomical compared to your operating current.   It's probably the wrong converter for the job.

At such low load currents, and such a large rating, and with the high switch frequency, the efficiency of the converter is going to be very low.   So you shouldn't think the switcher is giving you high efficiency.

Some options are:
- find a lower power converter
- use a charge pump.   Even if it is lossy the efficiency will probably be better than you switcher.
- Redesign the switcher for lower currents.   You might be able to reduce the noise but the efficiency
  could still be bad because the switcher has a large MOSFET and the losses will be large compared your load current.

I'd look at the charge pump!