3
\$\begingroup\$

I've designed an audio low power (20W) guitar amplifier circuit whose pre-amp section has somewhat high input impedance, as is best for a magnetic guitar pickup. Since it operates from a battery supply, I'd hoped to employ a boost regulator so that the maximum output I have available before clipping would stay more constant as the battery voltage faded. My circuit also employs a simple compressor, which means there is more gain at lower volumes then at higher ones.

Unfortunately while everything was quiet and worked very well on the bench, I'm getting what I guess I'd call hashing noise as the components are brought closer together. The culprit is obviously the boost regulator, and the enclosure size makes it impossible to place that board more than about 6 inches away from the sensitive pre-amp circuits. Not surprisingly, the noise is worst when the amplifiers active treble control is maximized, and/or my compressor circuit is active. The boost regulator's oscillator operates at around 100Khz and of course generates additional harmonics, but I was surprised how much audible noise it added.

Which of the following do any of you think has the best chance of minimizing this problem (feel free to add more or tell me which ones likely won't help)

  • Shielding around the boost regulator (what material... grounded or floating
  • Adding more low value ceramic caps across the boost reg output
  • adding inductors to the boost reg output
  • Building a simple shield completely surrounding the amplifiers pre-amp electronics

At this point I'm strongly considering just ditching the boost regulator idea, if I can't drastically decrease this problem. FYI, the boost regulator is a fairly common one good for moderate power, Max6A 4.5-32V to 5-60V, often sold on ebay or ali express. You may recognize it from the picture... enter image description here

As requested by Andy, I've added a picture of my basic wiring diagram, but of course its not truly "physical". Ignore the BTEST board, which is just a battery test board only active when a button is pushed. Note that my power amp board is separate from the pre-amp/control board, which is where also the PCB mount controls are soldered down.

enter image description here

Addendum: user96037 had the most helpful info here. This photo of the boost converter output is very telling...

Boost Converter output

There is 20mV (!!!) of noise at 176Khz here, much more if you count those spikes. The harmonic content is huge, so all my wires are behaving as antennas. It is obvious to me that if the amplitude can be cut and the waveform softened, I just may be able to salvage my use of this boost regulator. Already I've seen that adding a "random" pi-filter helps a LOT. (By random I mean my inductor was just 100 turns around a nail). I've ordered some better (toroidal) 100uH inductors and will try to come up with the best capacitors. Of course my next challenge will be keeping my inductor far enough away from the coil on the boost converter, and figuring out the highest value capacitors that still offer decent high frequency behavior.

Thanks again everyone... I'll update this thread with my progress for the benefit of anyone else dealing with this issue.

\$\endgroup\$
7
  • 2
    \$\begingroup\$ You said it worked well on your bench - Did you on your bench also use the boost regulator already? If yes, then you should be able to solve it by proper shielding (you should shield around your amplifier circuit and around your power supply, by this you protect is from other noise as well). As shielding any conductive material is OK and it should be connected grounded. \$\endgroup\$
    – KarlKarlsom
    Jun 11, 2016 at 2:53
  • \$\begingroup\$ Your Alibaba Reg may not be so cheap after you have made it quiet .You may be better to start with a less noisey approach. \$\endgroup\$
    – Autistic
    Jun 11, 2016 at 6:10
  • \$\begingroup\$ Almost certainly some (if not most or all) is caused by bad wiring i.e. earth loops that pass power current via input wiring. Please draw how you connect the various sub circuits together. \$\endgroup\$
    – Andy aka
    Jun 11, 2016 at 10:09
  • \$\begingroup\$ Are you sure the noise is transmitted over the air and not in the power lines itself? Usually it is advisable to have a linear voltage controller to smooth the voltage in addition to the boost converter. \$\endgroup\$
    – caconyrn
    Jun 11, 2016 at 14:10
  • \$\begingroup\$ Karl... Yes... the bench test included the boost regulator. Autistic... Not sure how to make it quiet, or what approach would be better. I'm still surprised something with a 100Khze oscillator would generate noise in the audio spectrum. Andy... though the diagram is not physical, you will notice that I do have separate coax runs to carry signal/ground. Good point about the ground loops, but I have experimenting with connecting the shield at one end only on the run between the pre-amp and power amp, and shockingly that made it worse. \$\endgroup\$
    – Randy
    Jun 11, 2016 at 16:24

3 Answers 3

Reset to default
3
\$\begingroup\$

There are two main types of electromagnetic interference. Conducted emissions and radiated emissions.

CONDUCTED EMISSIONS:

The first thing I would try is to look at the output of the boost regulator on an oscilloscope and see how much ripple voltage there is.

If you see lots of ripple then I would add a PI filter at the output of the regulator using a small inductor and two ceramic caps, and a small damping resistor in series with the inductor, where R>SQRT(4L/C). The resistor prevents resonance that could boost the noise at the resonance frequency of the filter 1/(2*pi*SQRT(LC)) Hz.

The cutoff frequency 1/(2*piSQRT(LC)) Hz should be several times lower than the 100kHz frequency of your regulator. Remember that the PI filter is second order, so if you make the cutoff 10x lower (at say 10KHz) you would squash any noise by a factor of 100X.

Also, keep the inductor in your PI filter away from that large toroid or they may couple together which defeats the purpose of the filter.

RADIATED EMISSIONS:

If you think that electromagnetic fields are radiating then steel sheet metal will block both electric and magnetic fields. It needs to be a magnetic type of steel. You basically just need to form it in a box around your noise source (in this case the power supply).

\$\endgroup\$
7
  • 1
    \$\begingroup\$ it onlt needs to be magnetic metal if you want to block DC magnetic fields. copper or aluminium will block AC fields. \$\endgroup\$
    – Jasen Слава Україні
    Jun 11, 2016 at 3:46
  • \$\begingroup\$ I will say that conducted emissions, particularly ground loops, can be very confusing, even to a veteran musician like myself who has dealt with them a lot. But honestly, at no load (when the noise is worst), the ripple is very hard to see (just a few millivolts), and is mainly far above the audio band. I suppose I will have to try some shielding, maybe be soldering together a small box out of spare PCB stock. (good point Jason, steel would be a lot harder to fabricate ). If it works, I'll "up" your answer to "solved". \$\endgroup\$
    – Randy
    Jun 11, 2016 at 16:30
  • \$\begingroup\$ I've not solved this 100%. Reconfiguring the input stage of my pre-amp as a differential amp helped a little. But so far creating a pi-filter has offered the most progress here. My Pi filter was just random though, and I'm still concerned that making a 10kHz pi filter might require a pretty huge inductor. But at least i have a starting point. \$\endgroup\$
    – Randy
    Jun 13, 2016 at 22:02
  • \$\begingroup\$ SWhoul;dn't the cutoff frequency be F = 1/(2*PISQRT(LC)) ? \$\endgroup\$
    – Randy
    Jun 13, 2016 at 22:50
  • 1
    \$\begingroup\$ @Randy the pi filter need not be as low as 10kHz. Even a 31kHz filter will offer a 10x noise reduction. A 50kHz cutoff would give roughly 4x noise reduction, and if you cascaded two of them you could get nearly 16x, etc. \$\endgroup\$
    – user4574
    Jun 14, 2016 at 14:07
1
\$\begingroup\$

Typical practice is two steps:

  1. Fully ENCLOSE the offending circuit (your switch-mode booster) in a Faraday Cage. i.e. a metal enclosure/box.
  2. Use FEEDTHROUGH FILTER CAPACITORS where any voltage goes INTO the Faraday Cage, or comes OUT of the Faraday Cage. The exact value of the capacitors depends on the operating frequency of your switch-mode booster.

Feed-thru capacitor

\$\endgroup\$
1
  • \$\begingroup\$ thanks Richard. Note my updates. That waveform is awful, and i see that my first line of attack has to be a better filter than what the boost reg has on board. That my solve my problem without the added hassle of completely enclosing the regulator. \$\endgroup\$
    – Randy
    Jun 14, 2016 at 13:50
0
\$\begingroup\$

I got nervous just by reading "boost converter" and "preamp" in adjacent sentences.

Use a linear regulator for the preamp and leave the power amp directly powered. You will get both less noise and more battery life. They make rather good batteries nowadays so you can get some 10% voltage drop from a fresh battery to 80% depleted one, esp. if you use rechargeables (and you should).

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.