reducing noise in a preamp powered with dc step up module

[Marcos - Munky]

I just tested putting a 100n ceramic cap right at the input terminals of the converter. Now I got zero noise! Well, I'm still getting a very low volume noise sometimes, but I found out it has something to do with the jack. If I pull the cable just a little bit, the noise stops. Guess it's time to put in a new jack.

Thanks again for your help!

[rankot]

This must be bad grounding connection with a cable.

[Rob Strand]

I just tested putting a 100n ceramic cap right at the input terminals of the converter. Now I got zero noise! Well, I'm still getting a very low volume noise sometimes, but I found out it has something to do with the jack. If I pull the cable just a little bit, the noise stops. Guess it's time to put in a new jack.

Thanks again for your help!

That's a good result.  Maybe try 1uF or 10uF ceramic or at least 2x100nF in parallel.

I'm beginning to understand the details of your problem.   If you are connecting the Li-ion battery via the jack, then *any* current pulses on the input of the converter will appear as noise across the jack "switch" connection and this get added onto any audio at the jack.   The jack will always have some contact resistance so it is unavoidable some noise will creep in.   

One solution is the input cap on the converter board like you have done as it keeps the current pulse on the board not down the converter ground wire.     An LC input filter could help as well.

A much better solution is to use the input caps but use the jack connection to turn on a MOSFET switch.  The MOSFET switch would then switches the power from the battery ground to the converter 0V input.  The wiring for Battery -ve  -> MOSFET -> converter ground can be kept short.  That way the high current pulses don't go through the ground at the jack.

[amptramp]

Nice to see a problem get solved!

Were there any application notes (or links to app notes on the net) for this device?  If the cure was that simple, maybe there is some guidance somewhere that tells you how to use it, much like linear regulators have app notes telling you what capacitance to put to ground at the input and output.  Somebody somewhere knows this for this unit!

[Rob Strand]

Were there any application notes (or links to app notes on the net) for this device?

It's a module, see link in Reply #3.  IMHO, the module probably has a skimped input cap (or long tracks to the cap).

[Marcos - Munky]

Yeah, I'm using the jack to connect the battery to the step up module. It's probably related to the contact resistance of the jack, because I can get rid of the noise by pulling the cable just a little bit, and that probably changes the plug position and the resistance enough to affect the noise.

Altough I don't really know how to use a mosfet as a switch, I agree that's a better solution. I found a simple 2N7000 switch that probably would do the job. Simple enough to quickly test it. Tomorrow I'll try to add another ceramic cap to the input of the module because that's easier than the mosfet switch , and if I still get noise, I'll try the switch.

[Rob Strand]

Altough I don't really know how to use a mosfet as a switch, I agree that's a better solution. I found a simple 2N7000 switch that probably would do the job. Simple enough to quickly test it. Tomorrow I'll try to add another ceramic cap to the input of the module because that's easier than the mosfet switch , and if I still get noise, I'll try the switch.

RG has already done it,
http://www.geofex.com/Article_Folders/mosswitch/mosswitch.htm

I'm not sure about magnitude of the current or what current waveform looks like going into your module.   If it's high it could take-out the MOSFET.    Adding a cap to the MOSFET to make it "soft start" may help but it won't help if the module currents are high.

[Marcos - Munky]

So, I finally tested the mosfet switch. The noise is still there, but it's more stable now. I mean, with the negative supply switched by the jack, a little cable movement was responsable for the noise to appear or disappear. Now I can move the cable a little bit more, but sometimes the noise is still there. Guess it's time to change the jack for a new one.

[Rob Strand]

So, I finally tested the mosfet switch. The noise is still there, but it's more stable now. I mean, with the negative supply switched by the jack, a little cable movement was responsable for the noise to appear or disappear. Now I can move the cable a little bit more, but sometimes the noise is still there. Guess it's time to change the jack for a new one.

Something must still be quite noisy for that to happen.  Maybe try increasing the size of the input cap on the converter, the place where you added the cap.

Unless you can see the problems with measurements, the next best alternative is to simply keep bashing away at the problem.  Eventually you will have a containment field around all the leaks.    Unfortunately these problems end-up being caused by specific things.  At the start the list of possible culprits is long and you just have to work through them until you get to the cause(s).

[Marcos - Munky]

I added more 2 100n caps in parallel on the input of the module, so now I have 300n. The preamp is dead quiet almost all the time, but I'm still having that issue with the jack/cable, when I move the cable a little bit the noise is back. I'll try to put in another jack.

[Rob Strand]

I added more 2 100n caps in parallel on the input of the module, so now I have 300n. The preamp is dead quiet almost all the time, but I'm still having that issue with the jack/cable, when I move the cable a little bit the noise is back. I'll try to put in another jack.

See how you go.   IMHO, the input jack shouldn't have an effect unless there's something still not 100%.   Clearly the input caps improve things.  You might need a large value like 10uF there.   The input current pulses could be very high and you need to keep them local to the converter board.  Another thing would be to increase the output filter caps and add LC filters.  I suspect you might be sick of this by now. 

[Marcos - Munky]

Yeah, I'm thinking it'll be way easier to just use a 9V battery . I can't find ceramics bigger than 100n here, so I can't try a bigger cap value. Well, I can use a lot of them in parallel, but I will need a hundred to make a 10u cap

[Rob Strand]

Yeah, I'm thinking it'll be way easier to just use a 9V battery . I can't find ceramics bigger than 100n here, so I can't try a bigger cap value. Well, I can use a lot of them in parallel, but I will need a hundred to make a 10u cap

If you leave the 3x100nF at the board like you have you might be able to use a normal cap 10uF to 100uF.

While the converter is 1.2MHz, when you run with virtually no load the switch frequency can drop.   Modern converters change from operating at constant frequency to constant on time and variable frequency (or perhaps in bursts).   If the load is light, like in your case, the switch frequency might be so low it is actually in the audio band.

Check-out figure 2:
http://file.scirp.org/pdf/ENG_2013011608512789.pdf

With that thinking, you can see why a 10uF to 100uF at the input might work.  Also, it probably explains why the RC filters you added at the converter output helped.   

Yet another thing to try is to add a small load to the output.  That might cause the switch frequency to shift up higher.    I wouldn't worry too much about battery drain since the converter efficiency is going to be poor anyway.   You could just decrease the load resistor until the noise moves out of the audio band.   After that measure the battery current with and without the load.   If the battery drain hasn't significantly increase then just leave the load in place.

Yeah, I'm thinking it'll be way easier to just use a 9V battery

Yes, it's far easier.    I think maybe that converter board isn't suited for the light load you have.  Ideally you want something where you can guarantee the switch frequency is outside of the audio band.  That why I was thinking a charge pump might be better for the light load case.

[Marcos - Munky]

Tried a 10uF eletrolytic cap today, the noise is still there. I'll just go the easiest way: remove everything and use a common 9V battery

[Rob Strand]

Tried a 10uF eletrolytic cap today, the noise is still there. I'll just go the easiest way: remove everything and use a common 9V battery

Yes, it seems like too much of a hard slog getting that switcher to run quietly.
IMHO a charge pump might be less trouble.  It won't run efficient from 3.6V but the switcher's not that efficient at low currents so you aren't losing anything anyway.   The good thing about the charge pump is you can set the switch frequency above audio.

[Marcos - Munky]

Yeah, a charge pump probably will do a better job. ICL7660 works with voltages down to 1.5 V, so I'll probably give it a try when I get some free time. I have a voltage doubler board laying around somewhere here, that'll make 7.4 V with a 18600 cell. Probably enough for a quick test.